Towards a classification of energy aware MAC protocols for wireless sensor networks

Transcription

1 WIRELESS COMMUNICATIONS AND MOBILE COMPUTING Wirel. Commun. Mob. Comput. 2009; 9: Published online 4 February 2009 in Wiley InterScience ( Towards a classification of energy aware MAC protocols for wireless sensor networks Bashir Yahya and Jalel Ben-Othman, Department of Computer Science, PRiSM Laboratory, University of Versailles Saint Quentin, 45 Avenue des Etats-Unis, Versailles, France Summary Power management is an important issue in wireless sensor networks (WSNs) because wireless sensor nodes are usually battery powered, and an efficient use of the available battery power becomes an important concern specially for those applications where the system is expected to operate for long durations. This necessity for energy efficient operation of a WSN has prompted the development of new protocols in all layers of the communication stack. Provided that, the radio transceiver is the most power consuming component of a typical sensor node, large gains can be achieved at the link layer where the medium access control (MAC) protocol controls the usage of the radio transceiver unit. MAC protocols for sensor networks differ greatly from typical wireless networks access protocols in many issues. MAC protocols for sensor networks must have built-in power conservation, mobility management, and failure recovery strategies. Furthermore, sensor MAC protocols should make performance trade-off between latency and throughput for a reduction in energy consumption to maximize the lifetime of the network. This is in general achieved through duty cycling the radio transceiver. Many MAC protocols with different objectives were proposed for wireless sensor networks in the literature. Most of these protocols take into account the energy efficiency as a main objective. There is much more innovative work should be done at the MAC layer to address the hard unsolved problems. In this paper, we first outline and discuss the specific requirements and design trade-offs of a typical wireless sensor MAC protocol by describing the properties of WSN that affect the design of MAC layer protocols. Then, a typical collection of wireless sensor MAC protocols presented in the literature are surveyed, classified, and described emphasizing their advantages and disadvantages whenever possible. Finally, we present research directions and identify open issues for future medium access research. Copyright 2009 John Wiley & Sons, Ltd. KEY WORDS: wireless sensor networks; medium access protocols; energy efficiency; survey 1. Introduction Recent advances in micro-electro-mechanical systems, low power highly integrated digital electronics, tiny microprocessors, and low power radio technologies have created low-cost, low-power, and multi-functional sensor devices, which can observe and react to changes in physical phenomena of their surrounding Correspondence to: Jalel Ben-Othman, Department of Computer Science, PRiSM Laboratory, University of Versailles Saint Quentin, 45 Avenue des Etats-Unis, Versailles, France. jalel.ben-othman@prism.uvsq.fr Copyright 2009 John Wiley & Sons, Ltd.

2 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1573 environments. These sensor devices are equipped with a small battery, a radio transceiver, and a set of transducers that used to acquire information about the surrounding environment. This emergence of such sensors has led engineers to envision networking of a large set of sensors scattered over a wide area of interest [1 6]. A typical wireless sensor network consists of a number of sensor devices that collaborate to accomplish a common task such as environment monitoring and report the collected data, using the radio, to a center node (sink node). Wireless sensor networks (WSNs) can serve many civil and military applications that include target tracking in battlefields [7], habitat monitoring [8,9], civil structure monitoring [10], and factory maintenance [11],... etc. In many applications sensor nodes should be deployed in an ad hoc fashion without careful planning [12]. They must organize themselves to form a multi-hop, wireless communication network to communicate with each other and with one or more sink nodes [13]. Multihop communication in sensor network is expected to consume less power than traditional single-hop communication. In addition, the transmission power levels can be kept low, which is highly desired in covert operations. To control the operation of the sensor network, a remote user can issue commands to the sensor network through a control center (sink) to assign data collection, processing, and transfer tasks to the sensors, and it can later receive the sensed data through the sink. Provided that sensor nodes carry limited, generally irreplaceable, power source, then wireless sensor networks must have built-in trade-off mechanisms that enable the sensor network to conserve power and give the end user the ability of prolonging network lifetime at the cost of lower throughput and/or higher latencies [1]. The energy constraints of sensor nodes and the need for energy efficient operation of a wireless sensor network have motivated a lot of research on sensor networks which led to the development of a novel communication protocols in all layers of the communication stack [14 28,29]. Given that the radio is the most power consuming component of a typical sensor node, then large gain can be achieved at the link layer where the medium access control (MAC) protocol is controlling the usage of the radio unit. MAC protocols have been extensively studied in traditional wireless networks. Time division multiple access (TDMA), frequency division multiple access (FDMA), and code division multiple access (CDMA) are MAC protocols that are widely used in modern cellular communication systems. Their principle idea is to avoid interference by scheduling nodes onto different sub-channels that are divided either by time, frequency or orthogonal codes. Since these sub-channels do not interfere with each other, MAC protocols in this group are largely collision free. Another class of MAC protocols is based on contention. Rather than pre-allocated transmission, nodes share the same channel. Collusion happens during the contention procedure in such systems. Classical examples of contention based MAC protocols include ALOHA [30] and carrier sense multiple access (CSMA) [31]. In ALOHA, a node simply transmits a packet when it is generated (pure ALOHA) or at the next available slot (slotted ALOHA) [32]. Collide packets are discarded and will re-transmitted latter. In CSMA, a node listens to the channel before transmitting. If it detects a busy channel, it delays access and retries latter. As sensor networks differ from traditional wireless networks in many points. First of all, most nodes in sensor networks are likely to be battery powered, and it is often very difficult to change batteries for all nodes. Second, nodes are often deployed in an ad hoc manner rather than with careful pre-planning; they must then organize themselves into a communication network. Third, many applications require large number of sensor nodes, and node density will vary in different places and times. Finally, most traffic in the network is triggered by sensing events, and it can be extremely bursty. All of these characteristics make that traditional MAC protocols are not suitable for wireless sensor networks. Various aspects of MAC protocols for wireless sensor networks are discussed in several surveys [1,3,14 28,33]. However, having the special characteristics of wireless sensor networks in mind, this paper gives a comprehensive review on most recent developments and challenging issues that sensor network s MAC protocol should overcome and discusses some solutions proposed in the literature. The rest of the paper is organized as follows. Section 2 presents and discusses the attributes of MAC protocols and design trade-offs for sensor networks. The proposed MAC protocols for sensor networks are classified in Section 3, and a typical set of specific sensor networks specific MAC protocols are described and discussed listing their advantages and disadvantages whenever possible. Moreover, we investigate also in this section, MAC protocols that designed based on cross-layer architecture

3 1574 B. YAHYA AND J. BEN-OTHMAN (i.e., merging layers of OSI communication stack). As well some quality of service oriented MAC protocols are presented in discussed in this section. We present future research directions for open issues that have not been studied thoroughly in Section 4. In Section 5, we conclude the paper and provide a set of general recommendations should be taken into account in future research on energy efficient MAC protocols. 2. MAC Attributes and Trade-offs in MAC Layer Design for Sensor Networks This section discusses important attributes of MAC protocols and how the design trade-offs could be made to meet the needs of the wireless sensor networks and their applications. Because sensor nodes are often battery constrained, we emphasize energy efficiency on MAC design Requirements and Design Constraints of a Wireless Sensor MAC Protocol Since sensor nodes are battery powered and the use of large batteries is impossible because of the space and cost constraints. Additionally, it is often not feasible to change batteries on a regular basis. It is therefore essential to make sensor nodes save as much energy as possible and, hence prolonging the network lifetime. Given that the radio unit is the most power consumer within the sensor node, then a significant amount of energy could be saved through controlling the radio operation. An energy efficient MAC protocol possesses the greatest capability to decrease the energy consumption of the radio unit since it directly controls the radio unit operation. MAC protocols are influenced by a number of constraints. A well-designed MAC protocol should consider a set of performance attributes and make tradeoffs among them. The most important performance attributes that are required for wireless sensor MAC protocols are described in References [14,34]: Collision avoidance is the principal task of all MAC protocols. It determines when and how a node can access the medium and sends its data. Collisions are not always completely avoided in regular operation; contention based MAC protocols accept some level of collisions. But all MAC protocols avoid frequent collisions. Scalability and adaptability are closely related attributes of MAC protocol that accommodate changes in network size, node density, and topology. Some of the reasons behind these issues are limited node lifetime, addition of new nodes to the network, and varying interference which may alter the connectivity and hence the network topology. A good MAC protocol should deal with and accommodate such network changes. Mobility in wireless sensor networks poses a challenge to the MAC protocol design. MAC protocol should adapt itself to changes in mobility patterns, making it suitable for sensor environments with both high and low mobility. Latency is the time required to send a packet by the sender until the packet is successfully received by the receiver. In sensor networks, the importance of latency is application dependent. The speed of the sensed object places a bound on how rapidly the network must react. Channel utilization refers to how the entire bandwidth of the channel is utilized in communications. Channel utilization is normally a secondary goal in sensor networks. Throughput refers to the amount of data successfully transferred from a sender to a receiver in a given time. Many factors affect the throughput in sensor network including efficiency of collision avoidance, latency, channel utilization, and control over-head. As with latency, the importance of throughput is an application dependent. Reliability, reliable delivery of data, is a classical design goal for all network infrastructures. The problem of reliability is of a grand importance in wireless sensor networks, where guaranteed data packet delivery should be ensured. In wireless networks, packet drops are mainly caused by buffer overflow and signal interference. Buffer overflow could be avoided through employing a buffer management strategy at MAC protocol to stop the number of backlogged packets from exceeding the maximum buffer size. Such buffer control could be achieved using traffic prioritization and/or filtering and aggregation. Knowing that data readings of neighboring sensors are highly correlated, then employing a data filtering and aggregation mechanism may yield reduction in data traffic and hence reducing communication, avoiding buffer overflow, and saving energy [35]. Packet drops due to signal interference can be minimized through the use of sufficiently high transmission power and the prevention of contention for medium access among nodes. Energy-efficiency, as explained earlier, energy is a scarce resource for sensor networks, and as the radio is the major consumer of sensor node s battery, especially for long range transmission and when the radio is

4 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1575 kept on all the time. Therefore energy aware MAC protocol can save transmission and reception energy by limiting the potential for collisions, minimizing the use of control messages, utilizing most of the available frequency band to shorten the transmission time, turning the radio into low power sleep state when it is idle, and finally avoiding the excessive transitions among active and sleep states. Fairness reflects the ability of different users, nodes, or applications to share the channel equally. It is an important attribute in traditional voice and data networks. However, in sensor networks, all nodes cooperate for a single common task. At a particular time, one node may have more data to send than some other nodes. Thus, rather than treating each node equally, success is measured by the performance of the application as a whole, and per-node or per-user fairness becomes less important. In short, the above attributes reflect the characteristics of a MAC protocol. For wireless sensor networks, the most important factors are effective collision avoidance, energy efficiency, mobility, scalability, and adaptability to densities and numbers of nodes. Other attributes are normally of secondary importance Sources of Energy Wastage Prolonging the sensor node life time and keeping network operation viable as long as possible are the most important issues in sensor networks. Energy efficient MAC protocol should consider a set of reasons that cause energy wastage and make sensor s battery drains quickly [36 39]. In this section we list and discuss a number of sources that should be tackled when designing MAC protocols. Packets collisions is the most dominant source of energy waste. When two packets are transmitted at the same time and collide, they become corrupted and must be discarded, and the retransmission of these packets are required which increase the energy consumption. Another important source that causes energy wastes energy in wireless domain is the Overhearing. Overhearing means that a node receives packets that are destined to other nodes. Overhearing unnecessary traffic can be a dominant factor of energy waste especially in heavy traffic load environments and dense networks. Dense sensor network deployments are common because the sensing range of many physical parameters is much smaller than the communication range. A typically radio unit can operate in four distinct modes of operation; idle, receive, transmit, and sleep. As it is expected that the radio consumes more energy in transmit and receive modes, running in the idle mode is also costly especially when there is no data to send during the period when nothing is sensed, this is commonly named as idle-listening. It is thus desirable to completely shut down the radio rather than switching into the idle mode. However, frequent switching between modes, especially switching from sleep mode to an active mode, leads to more energy consumption, than leaving the radio transceiver unit in idle mode because of the start-up power [40]. Control packet overhead is also a major source of energy that we consider here. Sending, receiving, and listening for control packets consume energy. Since, control packets do not directly convey useful application data; they also reduce the effective throughput. Minimal number of control packets should be used to make a data transmission. Avoiding over-emitting in the sensor network improves the energy efficiency. Over-emitting is caused by the transmission of a message when the destination node is sleep or not ready to receive. This again results in a waste of system s energy resources and ought to be avoided. Traffic fluctuations, wireless sensor networks usually generate traffic that fluctuates in place and time, which results in peak loads that may drive the sensor network into congestion which consequently raises the collisions probability, hence, much time and energy are wasted on waiting in the random back-off procedure [20]. Choosing the appropriate packet size is also an important issue from the energy point of view. As the packet size gets smaller, the transition energy becomes dominant to the energy consumed during receiving and transmitting of packets [41]. Most of these overheads are incurred by MAC protocol that based on contention technique. When turning to MAC protocols that based on scheduled techniques such as TDMA, may seem attractive at the first glance because idle-listening, overhearing, and collision simply do not occur, as sensor nodes are prescheduled, and each node knows clearly in which slots should transmit and receive, before any data transmissions. But, these advantages come at the cost of protocol complexity which leads to reduced flexibility to handle traffic fluctuations and network topology changes, as well a significant increase in protocol overhead. One solution for these problems is to apply some kind of over-provisioning and use a frame size that is large enough to handle peak loads. Another approach is dynamically adapting the frame size, but

5 1576 B. YAHYA AND J. BEN-OTHMAN this largely increases the complexity of the protocol and, hence, is considered to be an unsuitable option for resource limited devices such as sensor nodes. Wireless sensor network hardware as well as communication protocols should be designed to achieve their goals with a minimum of energy consumption through avoiding or reducing the energy waste sources listed above. A complete energy management scheme must consider all hardware units of the sensor node that consume energy not only the radio transceiver unit. However, at MAC layer level, energy efficiency can be improved through avoiding or minimization of idle listening, retransmissions, unwanted overhearing, and over-emitting. Turning off the radio when it is not needed is an important strategy for energy conservation. Based-on the under laying access technique, MAC protocols could be classified into: Unscheduled Protocols Scheduled Protocols Hybrid Protocols Multi-Channel based Application-Oriented Multi-path Data Propagation Rendezvous based Preamble based Slotted Contention based Time Division based Reservation based Priority based Preamble based Reservation Based Traffic sensitive based Clustering Based 3. Classification of MAC Protocols In this survey we cannot cover all MAC protocols proposed in the literature because of the space. Instead of that many typical protocols are included and discussed in this section. According to the underplaying mechanism of collision avoidance and the organization of sensor nodes, MAC protocols for wireless sensor networks can be classified into three general groups: scheduled, unscheduled (or random), and hybrid protocols. Scheduled MAC protocols organize the communication between sensor nodes in an ordered way. The most common scheduling method organizes sensor nodes using the TDMA technique, where a single sensor node utilizes a time slot. Organizing sensor nodes provides the capability to reduce collisions and message retransmission at the cost of synchronization and state distribution. Unscheduled protocols attempt to conserve energy by allowing sensor nodes to operate independently with minimum of complexity. While collisions and idle listening may occur and cause energy loss, the unscheduled MAC protocols typically do not share information or maintain state. Hybrid MAC protocols combine the strengths of scheduled and unscheduled MAC protocols while offsetting their weakness to design an efficient MAC protocol. Hybrid protocols use different techniques to conserve energy within nodes; some protocols differentiate between small and long data messages. Long data messages are assigned scheduled slots with not contention, whilst small periodic control messages are assigned random access slots. Other Cross-Layer Protocols Multi-Frequency Based PHY + MAC MAC + NETWORK PHY + NETWORK PHY + TRANSPORT Three-Layer Solutions Fig. 1. Classification of wireless sensor networks MAC protocols. hybrid protocols adjust the behavior of MAC protocol between CSMA and TDMA depending on the level of the contention in the network. The greatest advantage of the hybrid MAC protocols comes from its easy and rapid adaptability to traffic conditions which can save a large amount of energy, but this advantage comes at the cost of the protocol complexity which limits its range of applications. Some proposed MAC protocols do not easily fit into this classification scheme and may other classifications exist. Figure 1 presents a classification tree for wireless sensor networks MAC protocols. Most sensor network MAC protocols have some similarities in their effort to reduce energy consumption. The most common and effective way to conserve energy to turn the sensor node s radio transceiver and processor units into a low power sleep state when these resources have nothing to perform. Using this method, the sensor node can conserve a significant amount of energy than leaving the radio transceiver entering an idle state and/or the processor entering a busy loop. Sensor network MAC protocols may make nodes sleep periodically for a known fixed durations or may make

6 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1577 them sleep for random lengths of time depending on how a sensor node interacts with other sensor nodes. The duty cycle of a sensor node corresponds to the fraction of time the sensor node remains in an active state. MAC protocols that give nodes high duty cycle can respond to traffic and network changes more quickly, but consume energy at a higher rate. From the other hand a lower duty cycle MAC protocol can save energy, but low activity levels put a limit on the protocol s complexity, the possible network capacity, and the message latency Unscheduled Based MAC Protocols A common MAC paradigm in wireless networks is CSMA. It is popular because of its simplicity, flexibility, and robustness. It does not require much infrastructure support: no clock synchronization and global topology information are required, and dynamic node joining and leaving are well handled without extra operation. These advantages, however, come at the cost of trial and error a trail may cost access collision where more than two conflicting nodes transmit at the same time causing signal fidelity degradation at destinations. Collision can happen in any two hop neighborhood of a node. While collision among one hop neighbors can be greatly reduced by carrier sensing before transmission, carrier sensing does not work beyond one hop. This gives rise to the well-known hidden/exposed terminal problems [42], which cause serious throughput degradation especially in high data rate sensor network applications. With reference to Figure 2a, hidden-terminal problem could be explained as follow: assume (a), (b), (c), and (d) are sensor nodes with transmission range indicated by the circles. Node (a) starts its transmission to node (b). Node (c) does not catch the transmission of node (a) and start its transmission to node (d). The two transmissions collide at node (b). For the exposed-terminal problem, consider Figure 2b, here, node (b) defers its transmission to node (a) because it hears transmission of node (c) to node (d), even if there would be no collision at node (a). To overcome these problems, a collision avoidance technique like RTS/CTS handshaking is used. According to Figure 2c, the RTS/CTS handshaking works as follow: node (a) sends RTS, which it blocks any possible transmission of all nodes within its radio range. Node (b) catches RTS of node (a) and it responds with CTS. With CTS, node (b) blocks its neighbors and it let node (a) transmit. If the data packet is correctly received, node (b) sends an acknowledge packet to node (a). Fig. 2. (a) Hidden node problem (circles represent the transmission range of each node), (b) exposed node problem (circles represent the transmission range of each node), and (c) RTS/CTS handshaking for collision avoidance. Unscheduled MAC protocols have several advantages. Unscheduled protocols allocate resources on-demand; they can scale more easily and flexibly across changes in node density or network topology because they do not have to obtain the current schedule or join another sensor node group. Furthermore, unscheduled MAC protocols also allow sensor nodes to adapt more easily to changing in traffic conditions because channel reservation can occur with finer granularity and sensor nodes can adaptively contend for the channel. Additionally, unscheduled protocols do not require fine-grained time synchronization as in TDMA protocols. However, unscheduled MAC protocols experience in general many drawbacks as it has all sources of energy waste, a higher rate of collision, idle listening, and overhearing because

7 1578 B. YAHYA AND J. BEN-OTHMAN the sensor nodes do not coordinate transmission. In addition, Fairness becomes an issue in unscheduled MAC protocols because no mechanism implicitly exists that equalizes the channel usage, unlike in a scheduled MAC protocol. Based on the idea behind the design of the MAC protocol, more specific classification of unscheduled protocols is provided in this section Multi-channel based MAC protocols Using multiple radio transceivers in a single sensor node may seem a bad choice to conserve the energy of a sensor node, but several design approaches based on this technique could yield a significant energy reduction for the sensor node. However using multiple radio channels enables the sensor node to communicate simultaneously on separate channels if needed to increase bandwidth or response time. These benefits come at the cost of additional hardware requirements. First, radio transceivers constantly consume energy, even while asleep, so adding radio transceivers increases the energy consumption which lowers the overall energy consumption of the node. Second, a multiple radio transceiver system must possess the computational capability to receive and process data from multiple channels. Then, multiple radio transceivers system requires higher performance communication mechanisms and processor capabilities then single radio transceiver system. A typical protocol of this type is PAMAS protocol, which is detailed below. Power aware multi-access with signaling protocol The power aware multi-access with signaling (PAMAS) protocol [43] attempts to conserve battery power by switching off nodes that are not transmitting or receiving. PAMAS is using two transceivers: one for data messages and the other for control messages. Using this separation can prevent collisions of the larger data messages and save power. Control channel exchanges use RTS, CTS messages, and a busy tone. The busy tone is used to indicate that the data channel is in use by the receiving device. Figure 3 shows the message transfer in PAMAS. Message transfer in PAMAS starts by the source sending an RTS message to the destination on the control channel. The destination then decides if it should transmit a CTS by examining the data and control channels. If the destination does not detect any activity on the data channel and has not heard an RTS or CTS message recently it responds with a CTS message. Fig. 3. PAMAS data transfer. A source that does not receive a CTS in time will back off using a binary exponential algorithm. Once the source receives a CTS message it transmits the data message over the data channel. The destination starts transmitting a busy tone over the control channel once it starts receiving the data message so that nearby nodes realize that they may not use the data channel. PAMAS implements a busy tone as a message twice the length of an RTS or CTS message. Furthermore, during the data reception the destination will transmit a busy tone any time it receives an RTS message or detects noise on the control channel to corrupt possible CTS message replies and prevent further data transmissions. The main drawback of PAMAS is the inclusion of multiple radios which will greatly increases the energy consumption and the device cost of the sensor network. Additionally, controlling access to two wireless mediums increases the MAC protocol complexity. Most sensor networks have the nature that data messages are too small, which leads to decrease the benefits behind the separation of the data and control channels. However, ideas such as those proposed through PAMAS may work for sensor networks with large data messages like multimedia sensor networks Event-oriented MAC protocols The application characteristics may be used to enable the MAC protocol to conserve energy. For example, a monitoring based sensor network will have very little traffic most of the time, but may produce relatively large volumes of data when an event of interest occurs. MAC protocols that operate based on the assumption of constant traffic generation would waste energy when the sensor network has no data to manipulate. Typical protocols that utilize the application characteristics to save a considerable amount of energy are CC-MAC, and STIF. Collaborative MAC (CC-MAC) CC-MAC protocol attempts to conserve energy [44], while fulfilling application requirements, by utilizing the fact that sensor nodes located near each other generate correlated measurements. To achieve energy savings,

8 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1579 CC-MAC filters measurements from highly correlated sensor nodes in an effort to reduce the number of messages the sensor network must handle. Minimizing the message volume leads to a reduction in the contention of wireless medium, as consequence fewer collisions occur, and hence the number of messages sensor network must transmit or receive is reduced which allows sensor nodes to use lower duty cycles. CC-MAC consists of two components: the event MAC (E-MAC), which filters sensor node measurements to reduce traffic and network MAC (N-MAC), which forwards the filtered measurement to the sensor network sink. E-MAC reduces the traffic generated in an area by having only sensor nodes separated by at least the correlation distance generate measurements. Other nodes periodically sleep to save energy and awake to forward messages. Correlated sensor nodes rotate the role of generating measurements to balance energy consumption throughout the network. E-MAC slightly modifies the standard RTS/CTS/DATA/ACK scheme in the IEEE standard by introducing a first hop (FH) bit into the control packet header. The sensor node actively reporting measurements sets the FH bit when it transmits messages so that other nodes can decide to generate measurements or not. If a sensor node lies further than the correlation radius from all other sensor nodes generating measurements, then it will begin to generate measurements as well. Once the originating sensor node has transmitted the measurement, the FH bit gets cleared and the message becomes a forwarding message for the N- MAC protocol. N-MAC forwards messages from senor nodes generating measurements to the sensor network sink, but since the E-MAC protocol has removed most of the redundancy present in multiple measurements the forwarded traffic becomes more important. The main disadvantage of CC-MAC that it requires sensor nodes posses or obtain ranging information about their neighbors in order for N-MAC to filter data from correlated sensor nodes. Furthermore the complexity of the CC-MAC protocol may limit the application of the protocol. In addition, as the number of sensing events increase, especially if the sensing conditions change with time, the overhead associated with computing the correlation radius and distributing throughout the network increases. For large networks this overhead may become significant. SIFT SIFT [45] is a MAC protocol proposed for event-driven sensor network environments. The motivation behind SIFT is that when an event is sensed, the first R of N potential reports is the most crucial part of messaging and has to be relayed with low latency. Authors of the protocols use a non-uniform probability distribution function of picking a slot within the slotted contention window. If no node starts to transmit in the first slot of the window, then each node increases its transmission probability exponentially for the next slot assuming that the number of competing nodes is small. Comparing SIFT with MAC protocol showed that SIFT decreases latency considerably when there are many nodes trying to send a report. Since SIFT is a method for contention slot assignment algorithm, it is proposed to co-exist with other MAC protocols like SMAC [36]. SIFT has an important advantage, where low latency is achieved with many traffic sources. However, when the latency is an important parameter of the system, slightly increase in energy consumption must be accepted. Contrary, one of the main drawbacks of SIFT is the increase in the idle listening time caused by listening to all slots before sending. The second drawback is the increase in the overhearing. When there is an ongoing transmission, nodes must listen till the end in order to contend for the next transmission which causes overhearing. Beside that, system-wide time synchronization is needed for slotted contention windows, which increases the implementation complexity of the SIFT protocol. Alert MAC protocol Alert [46] is a MAC protocol for collecting event-triggered urgent messages from a group of sensor nodes with minimum latency and without requiring any cooperation or pre-scheduling among the senders or between senders and receivers during protocol execution. Alert minimizes contention among nodes by using a combination of time and frequency multiplexing. Multiple frequency channels are used within time slots and contention is minimized by controlling the selection probability of each channel by the nodes. The Alert protocol divides the time into slots which are called Alert slots. Each alert slot can be used to exchange one data packet and its acknowledgment between a sender receiver pair. In each alert slot, multiple frequency channels (denoted by M) can be used by the senders and receivers. These channels have different priorities. The receiver samples them one by one based on their priority level and tries to receive a packet from one of the senders. Each sensor selects a frequency channel randomly and independently of all other senders. The channels are selected with different probabilities. Less chance is given to select a higher priority channel. Once a sender has selected a

9 1580 B. YAHYA AND J. BEN-OTHMAN frequency channel, it switches to the selected frequency and sends a long preamble before sending its data packet. After the data packet, the node expects an acknowledgment packet (ACK) from the receiver. If the ACK packet is received correctly, the sender stops, otherwise, it tries to send its message again in the next time slot. At the beginning of each time slot, the receiver samples the signal level on each of the M frequency channels starting with the highest priority channel. If high signal level is sensed by the receiver, it stays on the same frequency and stops sampling and more channels. Then receiver waits to receive a packet. If a packet is received correctly, it sends an ACK packet back in response, otherwise, after some fixed timeout period, the receiver stops and continues to the next alert slot. The important features of alert protocol are the following: (a) minimizes delay of collecting first message as well as all messages; (b) non-carrier sense protocol; it thus eliminates hidden terminal collision problems; (c) dynamic shifting of frequency channels to provide robustness against interference; (d) adaptive characteristic enables operation without knowledge of number of contending nodes. In spite of these features, the protocol is greatly affected by the traffic patterns, as it uses preamble sampling concept. STEM protocol The sparse topology and energy management (STEM) [47] protocol conserves energy by leaving all sensor nodes in a sleep state while monitoring the environment, and allowing only the sensor nodes with a message to transmit to wake up neighboring sensor nodes that may have entered the sleep state. When data packets are generated, the sensor generating the traffic uses a paging channel (separate from the data channel) to awaken its downstream neighbors. Two versions of STEM have been proposed STEM-T, which uses a tone on a separate channel to wake neighboring nodes, and STEM-B, in which the traffic generating node sends beacons on a paging channel and sleeping nodes turn on their radios with a low duty cycle to receive the messages (the paging channel simply consists of synchronized time slots within the main communication channel). While STEM-T guarantees that minimal delay will be met (since receivers are turned on nearly instantaneously after data is generated), it requires more overhead than STEM-B since the receivers on the channel where the tones are sent must be idle, listening all of the time. Moreover, STEM-T may require extra hardware as a separate radio is needed for this channel [48] Preamble based MAC protocols In unscheduled MAC protocols, sensor nodes might not know the sleeping schedule of their neighbors, so they must somehow probe with messages until the neighbor awakes. As the communicating sensor nodes capture the messages of each other on time, they can begin the message transfer. This type of protocols is called preamble sampling or channel polling MAC protocols. The energy savings provided by preamble technique come from only synchronizing nearby sensor nodes when needed and only for the duration of the transmission. However, Energy conservation of preamble based protocols is greatly effected by the traffic patterns. Additionally, long preambles used in most preamble based protocols may cause performance degradation through the increase in latency which limits the deployment of this type of protocols on real time or latency sensitive applications. Some preamble based protocols are discussed below. Berkeley MAC (B-MAC) In B-MAC [49] protocol, sensor nodes independently follow a sleeping schedule based on the target duty cycle for the sensor network. As the sensor nodes operate on independent schedules, B-MAC uses very long beacons or preambles for message transmission. The source sensor node transmits a beacon long enough that the destination, which periodically senses the channel, has enough time to wake up and sense activity. Sensor nodes that sense activity on the channel remain awake to receive the message following the beacon or return to sleep if they do not detect activity on the channel. Figure 4 shows the message transfer in B-MAC. B-MAC is flexible, that means through the protocol interface, the network designer can tweak many operating variables in the protocol, such as delay and back-off values. From other side, B-MAC does not provide any protection mechanism against traditional wireless problems, such as the hidden terminal problem. Furthermore long pre-ambles in B-MAC protocol may Fig. 4. B-MAC message transfer.

10 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1581 introduce an additional latency; this problem could be controlled through the B-MAC interface. SyncWUF SyncWUF [50] combines both the simple signaling concept and a new wake-up frame (WUF) technique together, where the simple signaling is a meaningless signal and WUF contains meaningful information. The idea of SyncWUF is that the sender records the receivers schedules. To transmit a data packet, a sender node checks the receiver s schedule first. If the schedule is up-to-date, a short WUP is used as in the Wise-MAC protocol. If the schedule is out-of-date, a long WUF are used. Since a WUF is comprised of multiple short wake-up frames (SWUFs), each of which contains information like destination MAC address and the current SWUF position in the whole WUF, a receiver can decide when to turn on radio to receive data for reducing unnecessary waiting time. In SyncWUF, if a sender misses the receiver s active period, it must wait until next period to send data. The transmission delay of SyncWUF may thus be long. TrawMAC protocol TrawMAC is a traffic aware medium access control protocol for WSNs [51]. TrawMAC is a preamble sampling, which divides the monolithic preamble into small micro-frames each containing the information of the destination node and the time of the data transmission. TrawMAC nodes also maintain a sleep schedule of the neighbors, which is announced in the micro-frames of preamble. After exchanging enough packets in a network, each node is able to know the sleep schedule of all of its neighbors. Optimization on the preamble length based on the gathered sleep schedule information of the neighboring nodes is applied in the case of unicast traffic. Furthermore, TrawMAC uses strobed preamble technique for unicast transmission, where after transmitting a micro-frame preamble, the transmitting node waits for its acknowledgment from the potential receiver. In the best case, only one micro-frame of the preamble needs to be transmitted. If the transmitter receives an acknowledgment of a micro-frame preamble, it immediately sends the data. Otherwise, it keeps on sending the subsequent micro-frames and waiting for their acknowledgments. In broadcast traffic, the transmitter usually consumes a lot more energy since its long preamble cannot be shortened. Therefore, frequent broadcasts can deplete the energy on the transmitters. In certain applications where broadcasts are carried out very frequently, it is sometimes more energy efficient to do multiple unicasts instead of a single broadcast. TrawMAC protocol has the intelligence to decide in different circumstances whether or not to replace a broadcast with multiple unicasts based on the node density, knowledge of the wake-up schedule, node mobility, etc. In order to support high traffic loads and bursty traffic, TrawMAC transmits multiple back-to-back data frames with a single reservation. Support for backto-back data frames transmission leads to less energy consumption for all nodes due to the elimination of multiple preamble transmission. On the other hand, for small data sizes, the data is piggybacked in the micro-frames, which are called data-frame preambles. However, using micro-frames is more energy efficient than repeating large data packets back-to-back Rendezvous based MAC protocols Communication between any two nodes is possible only if both of them are powered simultaneously. Hence a method to put nodes wishing to communicate on time is necessary. This method typically called a rendezvous scheme. There are many ways to accomplish rendezvous between wireless nodes. The most popular strategy is called cycled receiver. In this scheme, nodes are powered on and off periodically, and a beaconing approach is used to express the desire to communicate. A typical example is TICER and RICER protocols. TICER and RICER The transmitted initiated cycled receiver (TICER) and Receiver initiated cycled receiver (RICER) are two similar protocols presents in [52]. The TICER protocol make sensor nodes with data to periodically transmit RTS control packet followed by a sensing period. Receivers periodically listen to the wireless channel and if they detect an RTS message, reply with a CTS message. The sensor nodes can then transfer the data message. RICER reverses the operation, so receivers periodically transmit beacons when they awake from their normally scheduled sleep time. A Sensor node with data to transmit stays awake and monitors the channel until it hears a wake up beacon from the intended destination. Upon reception, it starts transmitting the data packet. The session ends with an acknowledgment (ACK) signal transmitted from the destination node to the source node, after correctly receiving the data packet. Authors of the protocol mentioned that some protocol parameters such as time between control messages and the channel characteristics play an

11 1582 B. YAHYA AND J. BEN-OTHMAN Fig. 5. (a) TICER scheme and (b) RICER scheme. important role in the protocol overall performance. Figures 5(a) and 5(b) show the TICER and RICER schemes Scheduled MAC Protocols Scheduled MAC protocols attempt to reduce energy consumption by coordinating sensor nodes with a common schedule. Most proposed protocols use some form of TDMA since other forms of multiple access, such as FDMA or CDMA, would increase the cost and power requirement of the sensor nodes. By producing a schedule, the MAC protocol clarifies which sensor nodes should utilize the channel at any time and thus limits or eliminates collisions, idle listening, and overhearing. Nodes not participating in message communication may enter in a sleep mode until they have work to perform or need to receive a message. Additionally, the MAC protocol can share traffic or status information so that individual sensor nodes can optimize energy consumption over a collection of sensor nodes instead of at just a single sensor node. However, these advantages come at the cost of increased messages to create and maintain a schedule. Node mobility, node redeployment, and node death all complicate schedule maintenance. Sensor nodes that enter the network must wait until they learn, and possibly join, the schedule in order to use the channel. Additionally, some delay exists between the time a sensor node dies and the time neighboring sensor node reassigns its resources, so some resources may go unused and lead to unnecessary delays or packet loss. Scheduled MAC protocols must also operate properly under situations where sensor nodes posses incorrect state. A segmentation of the MAC state may lead to conditions where collisions cancel the benefits provided by the scheduled protocols. Synchronization becomes an important problem for a scheduled protocol and may occur through a periodic beacon, which increases the transceiver utilization, or by using higher precision oscillators, which increases the sensor node cost. Scheduled MAC protocols must also minimize the effect of added latency and limited throughput. Typically, added sensor node can only access the wireless channel for a fraction of the possible time. With a TDMA based MAC protocol the time a sensor node may access the channel depends heavily on the time of slot length. Typically, only one sensor node may transmit during the interval, so any unused time goes to waste. Reducing the time slot length may decrease the waste, but also decreases the maximum message length without fragmentation. Several schedule based MAC protocols attempt to overcome the limitations on throughput and latency at the cost of sharing additional information in messages or higher duty cycle. Scheduled MAC protocols could be further classified into two main sub-classes; scheduled contention based and scheduled contention-less based. Following sections present a set of typical scheduled protocols, where they are classified according to their method of operation Slotted contention based MAC protocols Slotted contention MAC protocols attempt to conserve energy by having nodes agree on a common sleep/listen pattern allowing them to use the radio transceiver at arbitrarily low duty cycles. Slotted protocols divide time into frames, and each frame is subdivided into a certain number of slots. Sensor nodes that have data to send wake up at the beginning of each frame and contend for the channel. This channel contention leads to a high probability of packet collisions because all communications being grouped into the listen part of the slot. To overcome this problem and enhance the performance of slotted MAC protocols, collision avoidance technique like RTS/CTS handshaking is used.

12 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1583 Fig. 6. S-MAC frame format. Sensor-MAC protocol (S-MAC) The basic idea behind the Sensor-MAC (S-MAC) protocol is based on periodic sleep-listen schedules and locally managed synchronizations [34]. Neighboring nodes form virtual clusters to set up a common sleep schedule. If two neighboring nodes reside in two different virtual clusters, they wake up at listen periods of both clusters. Schedule exchanges are accomplished by periodical SYNC packet broadcasts to immediate neighbors. The period for each node to send a SYNC packet is called the synchronization period. Figure 6 shows the S- MAC frame format. Collision avoidance is achieved by a carrier sense. Furthermore, RTS/CTS packet exchanges are used for unicast type data packets. An important feature of S-MAC is the concept of messagepassing where long messages are divided into frames and sent in a burst. With this technique, one may achieve energy savings by minimizing communication overhead at the expense of unfairness in medium access. Periodic sleep may result in high latency especially for multi-hop routing algorithms, since all immediate nodes have their own sleep schedules. Adaptive listening technique is proposed to improve the sleep delay, and thus the overall latency [36]. In this technique, the node which overhears its neighbor s transmissions wakes up for a short time at the end of the transmission. Hence, if the node is the next-hop node, its neighbor could pass data immediately. The end of the transmissions is known by the duration field of RTS/CTS packets. The important advantage that offered by sleeping schedules is the reduction on idle listening time, which consequently leads to more energy conservation in sensor nodes. But this comes at the cost of latency. In addition, the adaptive listening which is proposed to improve the sleeping delay may cause overhearing or idle listening if the packet is not destined to the listening node. Furthermore, sleep and listen periods are predefined and constant, which decreases the efficiency of the algorithm under variable traffic load. DMAC The principal objective of DMAC [53] is to achieve very low latency, but still to be energy Fig. 7. DMAC gathering tree and its implementation. efficient. DMAC could be viewed as an improved slotted ALOHA scheme where slots are assigned to the sets of nodes based on a data gathering tree rooted at a single data sink, the direction of packets arriving at a node are fairly stable and predictable. DMAC takes advantage of this by staggering the wake up times for nodes based on their distance from the data sink. By staggering the wake up times in such a way, DMAC reduces the large delays that can be observed in packets that are forwarded for more than a few hops when synchronizing schedules as in S-MAC. The wake up scheme consists of a receiving period and send period, each of length µ (set to accommodate a single transmission), followed by a long sleep period. Nodes on the data gathering tree begin their receiving period after an offset of d*µ, where d represents the node s depth on the tree. In this way, a node s receiving period lines up with its upstream neighbor s send period and a node simply sends during downstream neighbors receive periods, as shown in Figure 7. Contention within a sending period is accomplished through a simple random back-off scheme, after which a node sends its packet without a preceding RTS-CTS exchange. DMAC achieves very good latency and a significant reduction in energy consumption compared to other sleep/listen period assignment methods like S-MAC. Collision avoidance methods are not utilized, hence when a number of nodes that has the same schedule (same level in the tree) try to send to the same node, collisions will occur. This is a possible scenario in event-triggered sensor networks. Besides, the data

13 1584 B. YAHYA AND J. BEN-OTHMAN transmission paths may not be known in advance, which precludes the formation of the data gathering tree. Time out MAC (T-MAC) Several protocols have been developed based on S-MAC that offers solutions for various deficiencies and limitations of the original S-MAC protocol. T-MAC seeks to eliminate idle energy further by adaptively setting the length of the active portion of the frames [54]. Rather than allowing messages to be sent throughout a predetermined active period, as in S-MAC, messages are transmitted in bursts at the beginning of the frame. If no activation events have occurred after a certain length of time, sensor nodes set their radios into sleep mode until the next scheduled active frame. Activation events include the firing of the frame timer or any radio activity, including received or transmitted data, the sensing of radio communication, or the knowledge of neighboring sensors data exchanges, implied through overheard RTS and CTS packets. Figure 8 shows the frame format of T-MAC, where the first active period has the sensor node involved in a message transmission and the second active period has only SYNC transmission. By adaptively ending the active period, T-MAC protocol nodes may save energy by reducing the amount of time they spend idle listening and also adapt to changes in traffic conditions. The time-out period is set to span a small contention period and an RTS/CTS exchange. If a node does not detect any activity within the time-out interval, it can safely assume that no neighbor wants to communicate with it and goes to sleep. On the other hand, if the node engages or overhears a communication, it simply starts a new time-out after that communication finishes. To save energy, a node turns off its radio while waiting for other communications to finish. To improve message latency in T-MAC, the future request to send control message (FRTS) is proposed. Sensor nodes can use FRTS to inform the next hop that it has a future message transfer. The FRTS is an attempt to solve the problem of early sleeping inherited in T-MAC. T-MAC also considers the buffer size of the sensor nodes when calculating the contention period. Sensor nodes that have a full buffer may take higher priority and control the channel. This way limits buffer overflow of sensor nodes. Dynamic sensor MAC (DSMAC) Dynamic sensor-mac (DSMAC) extends S-MAC by adding dynamic duty cycle feature [55]. The objective of DSMAC is to decrease the latency for delay-sensitive applications. Figure 9 depicts the frame format of DSMAC. Within the SYNC period, all nodes share their one-hop latency values (time between the reception of a packet into the queue and its transmission). All nodes start with the same duty cycle. When a receiver node notices that average one-hop latency value is high, it decides to shorten its sleep time and announces it within SYNC period. Accordingly, after a sender node receives this sleep period decrement signal, it checks its queue for packets destined to that receiver node. If there is one, it decides to double its duty cycle when its battery level is above a specified threshold. The duty cycle is doubled so that the schedules of the neighbors will not be affected. The latency observed with DSMAC is better than the one observed with S- MAC. Moreover, it is also shown to have better average power consumption per packet. Mobility aware MAC protocol (MS-MAC) The mobility aware MAC protocol for sensor networks (MS-MAC) works similar to S-MAC to conserve energy when nodes are stationary. At the other extreme, this medium access scheme may also switches to work similarly to IEEE for a mobile ad hoc scenario [56]. Fig. 8. T-MAC message transfer. Fig. 9. Doubling of duty cycle in DSMAC.

14 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1585 MS-MAC introduces a new mechanism to handle mobility based on actual mobility status of nodes. Each node discovers the presence of mobility within its neighborhood based on the received signal levels of periodical SYNC messages from its neighbors. If there is a change in the signal received from a neighbor, it presumes that the neighbor or itself are moving. The level of change in the received signals also predicts the level of the mobile node s speed. Instead of storing only information on the schedule of the sender node as for SMAC, the SYNC message in MS-MAC also includes information on the estimated speed of its mobile neighbor or mobility information. If there is more than one mobile neighbor, then the SYNC message only includes the maximum estimated speed among all neighbors. This mobility information is used by neighbors to create an active zone around a mobile node when it moves from one cluster to another cluster, so that the mobile node can expedite connection setup with new neighbors before it loses all its neighbors. In the active zone, nodes run the synchronization periods more often resulting in higher energy consumption, but the time it takes to create new connections is lower. The reason behind the formation of active zone based on mobile speed is that the faster the mobile sensor moves, the less time it takes for it to cross the border between virtual clusters. Getting into a new virtual cluster without knowing the new schedule is a disastrous situation for the mobile sensor as it has to wait for long time for the next synchronization period (by default it is 2 minutes) during which the mobile is disconnected from the entire network. Whereas connection set up among nodes in the same virtual cluster is only 10 seconds which is the standard interval between periodical SYNC messages. In other words, active zones are only required when there are mobile nodes crossing from one cluster to the other. To create active zones this way, the mobility information in the SYNC message is set to be empty when a node does not discover any change in received signal levels from neighbors or when the node is not a border node. If a border node detects a change of received signal levels, it will add the mobility information in the SYNC message it is about to broadcast. After receiving such SYNC messages with mobility information, its neighbors will expedite the frequency of their synchronization periods according to the speed of the mobile The physically mobile node itself also receives SYNC messages with mobility information from the border node and it also quickens the synchronization periods, ready for new connections with new neighbors. This mobility aware mechanism of MS-MAC protocol allows nodes to work efficiently in both stationary and mobile scenarios. Under a stationary scenario or when mobile nodes only move within a single virtual cluster, all nodes work in a very energy efficient mode. No active zone is formed. When there is a mobile node crossing cluster borders, the mobile node and surrounding nodes form an active zone two hops away around the mobile node. In the active zones, nodes stay awake longer. The nodes in the active zones can be awake at all time if the mobile s speed exceeds a certain threshold, which is an application specific parameter. This high duty cycle mode allows nodes to set up connections with new neighbors in a timely basis. Optimized MAC for wireless sensor network Recently Yadav, Varma, and Malaviya [57] have proposed the optimised medium access control for wireless sensor networks. The proposed MAC protocol originally is based on the S-MAC protocol [34]. The proposed MAC protocol adapts the sensor duty cycle according to the network traffic conditions. If the traffic is high the duty cycle will increase, and for low traffic the duty cycle will be less. The network traffic status (i.e., high or low) is identified based on the number of messages in the queue of a particular sensor. The control packet overhead is minimized by reducing the number and size of the control packets where the control packets such as SYNC and RTS have been combined into one control packet SYNCrts. This contributes in the reduction of the energy consumption and latency. The synchronization of the sensor nodes in the proposed MAC is done using the SYNC packet as is done in S-MAC protocol. The SYNC packet contains the time of its next sleep. After deployment a sensor node starts by waiting and listening. If it hears nothing for a certain amount of time, it chooses a frame schedule and transmits a SYNC packet. If the node, during startup, hears a SYNC packet from another node, it follows the schedule in that SYNC packet and transmits its own SYNC accordingly. The synchronization table is maintained by all sensors for its neighboring nodes. Upon reception of the SYNC packet the synchronization table is updated and the timer is adjusted accordingly. The duty cycle is adaptive and updated according to the traffic status. When a message is received, the counter is increased and when it is transmitted the counter is decreased. If the message counter is greater than a threshold (COUNTthres), then the duty cycle

15 1586 B. YAHYA AND J. BEN-OTHMAN is increased and the changed duty cycle is reported to the neighboring sensor in the SYNCrts packet. The neighboring sensor on the reception of SYNCrts packet checks if its queue also contains message more than the COUNTthres, then it also increases it duty cycle. If not, it simply updates the synchronization table and continues with the original duty cycle. When the traffic is less and so when message counter is less than COUNTthres, the duty cycle is decreased. The sensor node intimates the changed duty cycle to its neighboring nodes. Through simulations, the proposed MAC is evaluated and compared against S-MAC and T-MAC [54]. Results have shown that the proposed MAC slightly improves the energy consumption as compared to S-MAC, but slightly higher than T-MAC. For latency, the proposed MAC protocol is less than both S-MAC and T-MAC protocols because of the adaptive adjustment of the duty cycle based on the network traffic. Pattern MAC (PMAC) protocol Pattern MAC, or PMAC is an adaptive energy efficient MAC protocol for wireless sensor networks [58]. PMAC protocol divides time into frames, each of which consists of two parts: the pattern repeat part and the pattern exchange part. Both parts are divided into slots. During the pattern exchange part, nodes advertise their intended sleep/wake patterns, which represent one slot by one bit (0 for sleep mode and 1 for active mode) and can be dynamically adjusted based on traffic conditions. Each slot in the pattern exchange part is long enough to send a node s pattern information and nodes have to contend for these slots. To enable all nodes to send their pattern information, the pattern exchange part has as many slots as the maximum number of neighbors a node is expected to have. During the pattern repeat part, nodes follow the sleep/wake pattern they have advertised. A node also wakes up at a time slot t, if one of its neighbors has advertised to be awake at the time slot t and it has data for sending to the node. The sleep/wake patterns are adapted to the traffic going through a node, to achieve maximal energy savings. Since a node decides its tentative sleep/wake schedule based only on its own traffic, P-MAC has the drawback that a receiver node may have a low duty cycle even though it has a lot of data to receive, which lengthens the transmission delay and decreases the throughput. ADCA MAC protocol ADCA [59] protocol is an asynchronous sleep/wake up schedule protocol which does not need synchronize nodes timers and allows each node to set its own sleep/wake up schedule independently. When a node starts up, it first decides its own sleep/wake up schedule, broadcasts the schedule and collects all neighbors schedules within an initial period of arbitrary length. It then starts executing its sleep/wake up schedule individually. The sleep/wake up schedule of a node is composed of repeated and fixlength cycle periods, each of which in turn consists of a contention period,asyn period,anextended period and a sleep period. In ADCA, a node listens to the channel for possible incoming packets at the contention period and broadcasts a SYN packet to the intended senders at the SYN period. An extended period immediately follows the SYN period to prolong the active time. A node turns its radio into sleeping mode to save energy in the sleeping period. When a node has a packet to send, it checks its neighbor-schedule table and contends to send the data packet in the receiver s contention period. The node then goes into sleeping mode after the transmission. If a sender fails to send the data packet in the receiver s contention period, it switches the radio into the receiving mode to wait for the receiver s SYN packet and tries to retransmit the data packet in the receiver s extended period. If the transmission still fails during the receiver s extended period, the sender waits for the contention period in the receiver s next cycle period. Since nodes maintain their schedules asynchronously, the schedules are staggered and the successful transmission rate and channel utilization are thus increased. Furthermore, ADCA allows nodes to dynamically adjust the contention period and the extended period based on current transmission statuses and/or traffic loads. In this way, the throughput is increased and the transmission delay is decreased without scanting energy. However, under light traffic conditions ADCA protocol experiences more delay compared to T-MAC protocol, which limits its application in real time environments. Furthermore, the complexity of the protocol is high Time division based MAC protocols TDMA provides an attractive solution for sensor network MAC protocols, because reducing collisions and idle listening can save considerable amounts of energy. TDMA divides the channel access time into a repeated frames; each frame is subdivided into N time slots, as shown in Figure 10. In each slot, only one node is allowed to transmit.

16 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1587 Fig. 10. TDMA divides the channel into N time slots. When designing a TDMA based protocol, many complications arise. Time slot assignment becomes difficult because sensor nodes cannot coordinate on large scales without introducing large overhead. Synchronization functionality must exist to correct timing errors caused by clock drift within each sensor node. Strict TDMA protocols also suffer from utilization problems during periods of light traffic generation. However, TDMA scheme has some disadvantages that limit its use in wireless sensor networks. TDMA normally requires nodes to form clusters, analogous to the cells in the cellular communication systems. One of nodes within the cluster is selected as the cluster head, and acts as the base station. This hierarchical organization has several implications. Nodes are normally restricted to communicate with the cluster head within a cluster; peer to peer communication is not directly supported. More importantly, TDMA based protocols have limited scalability and adaptability to the changes on number of nodes. When new nodes join or old nodes leave a cluster, the base station must adjust the frame length or slot allocation. In addition, frame length and static slot allocation can limit the available throughput for any given node, and the maximum number of active nodes in any cluster may be limited. Finally, TDMA based protocols depends on distributed, fine-grained time synchronization to align slot boundaries. Many variations on this basic TDMA protocols are possible. Rather than scheduling slots for node transmissions, slots may be assigned for reception with some mechanism for contention within each slot. The base station may dynamically allocate slot assignments on a frame by frame basis. In ad hoc settings, regular nodes may assume the role of base station, and this role may rotate to balance energy consumption [60]. In general, TDMA based protocols can provide good energy efficiency, but they are not flexible to changes in node density or mobility, and lack of peer to peer communication. Lightweight MAC (LMAC) Light weight MAC protocol (LMAC) is based on the time division multiple access (TDMA) paradigm [61]. Time is divided into Fig. 11. LMAC frame format. time slots, which nodes can use to transfer data without having to content for the medium or having to deal with energy wasting collisions of transmissions. Only one time slot is assigned to each node and give this node control over this time slot. Unlike traditional TDMA-based systems, the time slots in LMAC protocol are not divided among the networking nodes by a central manager. Instead a distributed algorithm is used. During its time slot, a node will always transmit a message which consists of two parts: control message and a data unit. Figure 11 illustrates the frame format of LMAC protocol. Because a time slot can only be controlled by a single node, then this node can communicate collision-free. The control message has a fixed size and is used for several purposes. It carries the ID of the time slot controller, it indicates the distance of the node to the gateway in hops for simple routing to a gateway in the network, and it addresses the intended receiver and reports the length of the data unit. The control data is used also to maintain synchronization between the nodes and therefore the nodes also transmit the sequence number of their time slot in the frame. All neighboring nodes put effort in receiving the control messages of their neighboring nodes. When a node is not addressed in that message or the message is not addressed as an omni-cast message, the nodes will switch off their power consuming transceivers only to wake up at the next time slot. If a node is addressed, it will listen to the data unit which might not fill the entire remainder of the time slot. Both transmitter and receiver(s) turn off their transceivers after the message transfer has completed. A short time out interval ensures that nodes do not waste energy for idle listening in time slots that are not controlled. In this protocol, it is only possible for a node to transmit a single message per frame.

17 1588 B. YAHYA AND J. BEN-OTHMAN Fig. 12. EMACS frame format. EMACS EMACS is a TDMA based protocol [62]. EMACS protocol divides the time slot into three sections as shown in Figure 12: communication request, traffic control, and data. Sensor nodes use the traffic control section to transmit their periodic control information. Every sensor node must transmit this information during their time slot and neighboring sensor nodes listen for the control packet of the neighbors. A sensor node may request to use the data section of a time slot it does not own by transmitting a request during the communication request section. The time slot owner can give ownership to the requesting sensor node within its control message. All data transmissions occur within the data section. Sensor nodes within the network using EMACS protocol operate in one of three possible modes. Active nodes co-operate fully in the communications, own a slot, and transmit a control message within each slot they own. Passive sensor nodes do not own a slot and only transmit messages after requesting a slot from an active sensor node. Finally, dormant sensor nodes do not participate in the sensor network and sleep until they wish to participate in an active or passive role. Providing varying levels of functionality allows the senor nodes to conserve energy when the application does not need them and activate only the minimum number of sensor nodes to perform the application functionality Clustering based MAC protocols Gathering sensor nodes into clusters offers many advantages. First, clustering enables to differentiate between local traffic from global traffic to conserve energy. Second, locally sharing information provides a trade-off between global state distributions, which would consume too much energy for the dynamic nature of sensor networks, and greedy algorithms that optimize sensor node behavior independent of other sensor nodes. Third, clustering also allow protocols to scale more easily since the protocol might view a cluster as a single entity. Finally, clustering may allow sensor nodes to perform some functionality, such as synchronization, on a local scale that would consume too much energy on a global scale. However, these advantages come at the cost of coordination messages overhead. The Cluster head which managing the cluster must coordinate the sensor nodes to ensure that the cluster reduces energy on average. Protocols often rotate the cluster head functionality among sensor nodes to evenly distribute the additional energy consumption caused by managing operation. Node dynamics further complicates clustering protocols since cluster formation and cluster head assignment algorithms must adapt to redeployment or sensor node death. LEACH Low energy adaptive clustering hierarchy (LEACH) is a clustering based protocol that minimizes energy dissipation in sensor networks [63]. The purpose of LEACH is to randomly select sensor nodes as cluster heads, so the high-energy dissipation in communicating with the base station is spread to all sensor nodes in the sensor network. The operation of LEACH is separated into two phases, the setup phase, and the steady phase. The duration of the steady phase is longer than the duration of the setup phase in order to minimize the overhead. During the setup phase, a sensor node chooses a random number between 0 and 1. If this random number is less than a certain threshold, the sensor node is selected as a cluster head. After the cluster heads are selected, the cluster heads advertise to all sensor nodes in the network that they are the new cluster-heads. Once the senor nodes receive the advertisement, they determine the cluster that they want to belong based on the signal strength of the advertisement from the cluster-heads to the sensor nodes. The sensor nodes inform the appropriate cluster-heads that they will be a member of the cluster. Afterwards, the clusterheads assign the time on which the sensor nodes can send data to the cluster-heads based on a TDMA approach. During the steady phase, the sensor nodes can begin sensing and transmitting data to the cluster-heads. The Cluster-heads also aggregate data from the nodes in their cluster before sending these data to the base

18 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1589 station. After a certain period of time spent on the steady phase, the network goes into the setup phase again and entering into another round of selecting the cluster heads. LEACH possesses several disadvantages in its design. First, it requires a complex radio capable of a direct sequence spread spectrum (DSSS) and power scaling, which increases the energy consumption and the sensor node cost. Second, cluster formation and restructuring can take a long time during which the sensor nodes consume energy and cannot perform any useful work. Third, LEACH assumes that each sensor node can communicate directly with the base station. Requiring this would either cause sensor nodes to consume large amounts of energy transmitting messages or limit the geographical area a sensor network can cover. To overcome some of drawbacks of LEACH protocol, the authors described a variant version, called LEACH-C, which uses the base station to select the optimal cluster heads. During the setup phase of operation each sensor node transmits its location and energy levels to the base station. After computing the optimal selection of clusters for energy savings, the base station transmits a list of sensor nodes that will act as cluster heads. Cluster formation then continues similar to LEACH with sensor nodes transmitting join messages and cluster heads setting and distributing schedules. However, using LEACH-C requires nodes that can determine their location. Localization support would increase the node cost and power consumption for either hardware support (e.g., GPS) or protocol support (range estimation algorithms) Priority based MAC protocols By using of a random function, the access to the wireless channel is controlled by assigning priorities to sensor nodes or links to destinations. Senor node with highest priority is given the chance to access the channel. Sensor node IDs and time slots numbers provide an input to a random function that establishes the priority within a two neighborhood. One example of this type of protocols is the series of protocols proposed by Garcia-Luna-Aceves [64]. Node activation multiple access (NAMA) protocol NAMA protocol [64] activates individual nodes to transmit a single message in each slot. NAMA protocol uses TDMA with time divided into blocks of Sb sections. Ps parts constitute each section and Fig. 13. Time division structure of NAMA protocol. the parts contain Tp time slots (see Figure 13). Each node selects a single part, chosen to balance channel utilization across the parts, and contends with the other sensor nodes that select the same part. NAMA reserves the last section of each block for signaling messages that allow sensor nodes to join the network. Each sensor node computes its priority along with the priority of its neighbors and uses these to determine who has access to the current time slot within the senor node s chosen part. A senor node gets assigned a particular slot within a section based on its priority. If a senor node has the highest priority among its two hop neighbors for the given time slot, then the sensor node may transmit. If no sensor node s priority maps to a time slot, then the sensor node with the highest priority may use the time slot. Link activation multiple access (LAMA) protocol LAMA [64] protocol activates links to destinations sensor nodes based on the direct sequence spread spectrum (DSSS) code assigned to the receiver and the priority of the transmitter. Each senor node gets a code assigned from a finite set of pseudo-noise codes. During each time slot the sensor node with the highest priority in a two hop neighborhood calculated based on sensor node ID as in NAMA may activate a link by using the code assigned to the receiver. Using orthogonal codes allows sensor nodes to communicate when they would normally interfere and using the topology information prevents collisions at the receiver. PAMA protocol The pairwise-link activation multiple access (PAMA) protocol [64] activates links between sensor nodes by assigning priorities to the

19 1590 B. YAHYA AND J. BEN-OTHMAN links and by varying the codes and priorities of links based on the current time slot. A communication link between two sensor nodes, u the source and v the destination, gets activated if the link (u, v) has the highest priority among all links of nodes u and v and node u has the highest priority of its two hop neighbors using the code assigned to link (u, v). Similar to LAMA, the use of DSSS allows nodes to communicate on different codes without interruption and the protocol algorithm prevents collisions on the same code. The largest drawback to the NAMA, LAMA, and PAMA protocols arise from the resources required. All the protocols require a sensor node to compute the priorities of each neighboring sensor node for each time slot. Constantly calculating sensor node priorities may consume energy resources quickly and degrade the network lifetime to unacceptable levels. Additionally, LAMA and PAMA require the sensor nodes have radios with spread spectrum capabilities, which increases sensor node cost. Dynamic slot assignment also prevents sensor nodes from developing a regular sleep schedule since the priorities vary based on the current slot number Reservation based MAC protocols Time based medium access has the potential of capturing most of the opportunities for energy optimization in sensor networks. Energy wastage due to overhearing, collision, idle mode, and transitions between different states can be minimized if the medium access is shared on a time basis. In addition, time based medium arbitration can enhance delay predictability and limit packet drops due to interference and buffer overflow. However the problem of scheduling access to the medium is NP-hard making the scalability of time based MAC scheme a major concern. Moreover, distributed time based medium arbitration typically introduces excessive overhead. In addition, maintaining clock synchrony among nodes is essential to enforce the schedule which is a non-trivial problem for the resource-constrained sensor nodes. Most of the time based MAC protocols proposed in the literature have focused on addressing these issues either using reservation requests over preset data routes or pursuing simplified heuristics to tackle the complexity of medium access scheduling. Energy efficient TDMA MAC protocol scheduling Energy efficient TDMA MAC protocol scheduling [65], the use of reservation requests has been explored for tackling the scalability of time based medium arbitration. Nodes that have data to transmit make a reservation request to a base station, which responds with a traffic control message indicating medium access schedule. Nodes that are not included in the traffic control message can turn off their radios transceivers. The nodes that have been assigned slots transmit in the order scheduling by the base station. The base station trades off latency with energy efficiency. While it is better to bundle all transmissions from a node in consecutive time slots, the transmission of other nodes will be delayed Hybrid Protocols Hybrid MAC protocols combines the strengths of the contention based protocols and scheduled protocols while offsetting their weakness to design a good and efficient MAC protocol. Hybrid MAC protocols use different techniques to conserve more energy within nodes. Some protocols adjust the behavior of MAC protocol between CSMA and TDMA depending on the level of the contention in the network. Other types of protocols try to differentiate between small and long data messages. Long data messages are assigned scheduled slots with no contention, whilst small periodic control messages are assigned random access slots. The greatest advantage of the hybrid MAC protocols comes from its easy and rapid adaptability to traffic conditions which can save a large amount of energy. Following some protocols use hybrid techniques Preamble based hybrid protocols By combing the strengths of scheduled and unscheduled MAC protocols, preamble based protocols may be made more energy efficient and more sensitive to changes in traffic type. The reduction in energy consumption made by preamble hybrid techniques comes from two sources: first, synchronizing nearby sensor nodes when needed and only for the duration of the transmission; second, by utilizing the fact the preamble based protocol are very sensitive to traffic patterns, then the protocol can alter its operation based on changes in traffic to make a significant save in energy resources. However, the most important drawback of most preamble based protocols and their use of long preambles which causes increase in latency. High latencies inherited in these types of protocols limits the deployment of this type of protocols on real time or latency sensitive applications. Additionally,

20 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1591 as these protocols use mixed technique of scheduled and unscheduled protocols, the complexity of protocol becomes very high. Wireless MAC protocol (WiseMAC) A hybrid TDMA/CSMA with Preamble Sampling protocol is proposed in Reference [66] where all sensor nodes are defined to have two communication channels. Data channel is accessed with TDMA method, whereas the control channel is accessed with CSMA method. WiseMAC protocol uses non-persistent CSMA (NP- CSMA) with preamble sampling to decrease idle listening. In the preamble sampling technique, a preamble precedes each data packet for alerting the receiving node. All nodes in a network sample the medium with a common period, but their relative schedule offsets are independent. If a node finds the medium busy after it wakes up and samples the medium, it continues to listen until it receives a data packet or the medium becomes idle again. The size of the preamble is initially set to be equal to the sampling period. However, the receiver may not be ready at the end of the preamble, due to reasons like interference, which causes the possibility of over-emitting type energy waste. Moreover, over-emitting is increased with the length of the preamble and the data packet, since no handshake is done with the intended receiver. To reduce the power consumption incurred by the re-determined fixed-length preamble, WiseMAC offers a method to dynamically determine the length of the preamble. That method uses the knowledge of the sleep schedules of the transmitter node s direct neighbors. The nodes learn and refresh their neighbor s sleep schedule during every data exchange as part of the acknowledgment message. In that way, every node keeps a table of sleep schedules of its neighbors. Based on neighbors sleep schedule table, WiseMAC schedules transmissions so that the destination node s sampling time corresponds to the middle of the sender s preamble. To decrease the possibility of collisions caused by that specific start time of wake-up preamble, a random wake-up preamble is advised. Another parameter affecting the choice of the wakeup preamble length is the potential clock drift between the source and the destination. A lower bound for the preamble length is calculated as the minimum of destination s sampling period, Tw, and the potential clock drift with the destination which is a multiple of the time since the last ACK packet arrival. Considering this lower bound, a preamble length, Tp, is chosen randomly. Figure 14 depicts the WiseMAC data message transfer. Fig. 14. Data Message transfer in WiseMAC protocol. The simulation results show that WiseMAC performs better than one of the S-MAC variants. Besides, its dynamic preamble length adjustment results in better performance under variable traffic conditions. In addition, clock drifts are handled in the protocol definition which mitigates the external time synchronization requirement. Main drawback of WiseMAC is that decentralized sleep listen scheduling results in different sleep and wake-up times for each neighbor of a node. This is especially an important problem for broadcast type of communication, since broadcasted packet will be buffered for neighbors in sleep mode and delivered many times as each neighbor wakes up. However, this redundant transmission will result in higher latency and power consumption. In addition, the hidden terminal problem comes along with WiseMAC. This is because WiseMAC is based on non-persistent CSMA. This problem will result in collisions when one node starts to transmit the preamble to a node that is already receiving another node s transmission where the preamble sender is not within the range Traffic sensitive based protocols Traffic types and conditions in the sensor network have a direct effect on the energy consumption at the sensor node. A MAC protocol can utilize this fact to make a significant save in energy resource by adapting itself to network conditions. Sensor networks that generate large volume of traffic provide a good case for MAC protocols that adapt their operation based on traffic conditions. Additionally, the differentiation in traffic characteristics between control traffic and data traffic could be utilized to make the MAC protocol to alter its operation according to the traffic type in order to provide a considerable reduction in energy resource consumption. However, to realize and implement these benefits come from the nature of the traffic, MAC protocols should keep track of traffic characteristics within the sensor network and share traffic information among sensor nodes within the sensor network.

21 1592 B. YAHYA AND J. BEN-OTHMAN Hybrid MAC (ZMAC) Z-MAC (or Zebra MAC) [67] is a hybrid MAC protocol for sensor networks. Z-MAC combines the strengths of TDMA and CSMA while offsetting their weaknesses. The main feature of Z-MAC is its adaptability to the level of contention in the network so that under low contention, it behaves like CSMA, and under high contention, like TDMA. Z-MAC uses CSMA as the baseline MAC scheme, but uses a TDMA schedule as a hint to enhance contention resolution. In Z-MAC, a time slot assignment is performed at the time of deployment higher overhead is incurred at the beginning. Its design philosophy is that the high initial overhead is amortized over a long period of network operation, eventually compensated by improved throughput and energy efficiency. Z-MAC uses DRAND [68], an efficient scalable channel-scheduling algorithm. After the slot assignment, each node reuses its assigned slot periodically in every predetermined period, called frame. Any node assigned to a time slot is called an owner of that slot and the others the non-owners of that slot. There can be more than one owner per slot because DRAND allows any two nodes beyond their two-hop neighborhoods to own the same time slot. Unlike TDMA, a node may transmit during any time slot in Z-MAC. Before a node transmits during a slot (not necessarily at the beginning of the slot), it always performs carriersensing and transmits a packet when the channel is clear. However, an owner of that slot always has higher priority over its non-owners in accessing the channel. The priority is implemented by adjusting the initial contention window size in such a way that the owners are always given earlier chances to transmit than nonowners. The goal is that during the slots where owners have data to transmit. Z-MAC reduces the chance of collision since owners are given earlier chances to transmit and their slots are scheduled a priori to avoid collision, but when a slot is not in use by its owners, non-owners can steal the slot. This priority scheme has an effect of implicitly switching between CSMA and TDMA depending on the level of contention. An important feature of this priority scheme is that the probability of owners accessing the channel can be adjusted independently from that of non-owners. This feature contributes to increasing the robustness of the protocol to synchronization and slot assignment failures while enhancing its scalability to contention. By mixing CSMA and TDMA, Z-MAC becomes more robust to timing failures, time-varying channel conditions, slot assignment failures, and topology changes than a standalone TDMA; in the worst case, it always falls back to CSMA. However, the slot assignment and synchronization may lead to high costs especially when significant network changes occur frequently. TRAMA protocol The traffic adaptive medium access (TRAMA) [69] protocol attempts to balance the benefits of scheduled and unscheduled protocols by providing scheduled slots with no contention for longer data messages and random access slots for small, periodic control messages. Additionally, sensor nodes adapt to traffic and network conditions by sharing traffic needs with neighbors and learning the two hope topology of their neighbors. TRAMA consists of three sub-protocols: the neighbor protocol (NP), which shares the topology information; the scheduled exchange protocol (SEP), which allows nodes to share what traffic they have queued; and adaptive election algorithm (AEA), which selects the slots to use for data transfer based on topology and traffic conditions. Frames within TRAMA protocol consists of several slots, where the random access control slots occur together at the beginning of the frame and the scheduled data slots occur at the end as illustrated in Figure 15. Random-access slots are used to establish twohop topology information where channel access is contention-based. A basic assumption is that, by the information passed by the application layer, MAC layer can calculate the transmission duration needed. Then at time t, the node calculates the number of slots for which it will have the highest priority among twohop neighbors within the period [t, t + scheduled slot interval]. The node announces the slots it will use as well as the intended receivers for these slots with a schedule packet. Additionally, the node announces the slots for which it has the highest priority but will not be used. The schedule packet indicates the intended receivers using a bitmap whose length is equal to the number of its neighbors. Bits correspond to one-hop neighbors ordered by their identities. Since the receivers of those messages have the exact list and identities of the one hop neighbors, they find out the intended receiver. When the vacant slots are announced, potential senders are evaluated for re-use Fig. 15. TRAMA time slot organization.

22 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1593 of those slots. Priority of a node on a slot is calculated with a hash function of node s and slot s identities. The main advantages of TRAMA protocol is that higher percentage of sleep time and less collision probability is achieved compared to contention based protocols. Since intended receivers are indicated with a bitmap, less communication is performed for multicast and broadcast type of communication patterns compared to other protocols. The drawback of TRAMA protocol is that the transmission slots are set to be seven times longer than the random access period. However, all nodes are defined to be either in receive or transmit states during the random access period for schedule exchanges. This means that without considering the transmissions and receptions, the active duty cycle is at least 12.5% of the frame length, which is a considerably high value. For a time slot, every node calculates each of its twohop neighbors priorities on that slot. In addition, this calculation is repeated for each time slot, since the parameters of the calculation change with time Reservation based hybrid protocols By combining the contention and time division schemes, the performance of reservation based protocols could be improved. PARMAC protocol is based on this idea to gain a significant reduction in energy. PARMAC protocol The power aware reservation based MAC (PARMAC) protocol [70], PARMAC is an energy aware protocol primarily designed for ad hoc networks and is applicable to sensor networks as well. The approach is actually a combination of contention and reservation based medium arbitration schemes. The network is divided into grids and each node is assumed to reach all the other nodes within its grid. Time is divided into fixed frames. Grids are assigned distinct frames. Each frame is composed of reservation period (RP) and contention free period (CFP). In each RP, nodes within a grid cell exchange three messages to reserve the slots for data transmission and reception and the exchange of acknowledgments. Data is then sent in the CFP. The clocks of all nodes are assumed to be synchronized. The protocol saves energy by minimizing the idle time of the nodes and allowing the nodes to sleep during a CFP. Moreover, intra-grid control packets overhead and packet retransmissions are minimal, achieving significant energy savings. However, inter-grid contention is still possible and the efficiency of this approach can significantly diminish if the application requires data exchange among nodes in different grids. HMAC protocol HMAC protocol is an energyefficient low-latency MAC protocol for wireless sensor networks [71]. HMAC uses a slotted frame structure to achieve high energy efficiency. Each frame contains multiple short wake up slots and multiple data slots. Each node needs to choose a wake up slot and notifies all its neighbors of the chosen slot number with a technique proposed in HAMA [72] during the deployment phase, so that the wake up slot number can be received properly with high probability (>0.99). It is noted that nodes can also use specific data slots to announce its chosen slot number after the deployment phase for some special occasions. The data slots are assigned on an on-demand basis. A sender s first sends a message during the chosen wake up slot of receiver r to notify r of the data slots during which s would like to send data to r. The receiver r will then wake up during the specified data slots to receive data from s. Because a data slot may have multiple contenders, RTS/CTS/DATA/ACK mechanism is used to avoid collision. HMAC has good performance in terms of channel utilization and transmission delay. However, HMAC needs very accurate time synchronization which causes a large overhead Clustering based MAC protocols Sensor networks consist of wireless enabled sensor nodes with limited energy. As sensors could be deployed in a large area, data transmitting and receiving are energy consuming operations. One of the methods to save energy is to reduce the transmission distance of each node by grouping nodes into clusters. Each cluster has a cluster-head (CH), which communicates with all the other nodes of that cluster and transmits the data to the remote base station or to other cluster heads. Hybrid clustering based MAC protocols combine both contention and contention-less schemes to handle the channel access, where for example the contention techniques is used for intra-cluster communication and the contention-less scheme is used for inter-cluster communication such as in GANGS protocol, which is detailed below. GANGS protocol The GANGS protocol gathers sensor nodes into clusters [73]. GANGS protocol utilizes an unspecified contention protocol for intra-cluster communication and TDMA based communication protocol for transmissions between clusters. GANGS does not assume that the sensor

23 1594 B. YAHYA AND J. BEN-OTHMAN nodes can communicate with the base station, therefore the cluster heads must form a routing backbone in the sensor network using a separate routing protocol. Clusters formation in GANGS is done through two phases: cluster head election and clusters connection. During the cluster head election phase, each sensor node shares its energy resource level with its neighbors. Any node that has more energy left relative to its energy neighbors levels declares itself a cluster head and transmits a message announcing it. During the clusters connection phase, a non-cluster head sensor node may exist in one of three states: it could receive a single cluster head announcements, or it could receive multiple cluster announcements, or it does not receive any announcement. If a sensor node receives only one announcement, it joins that cluster. For those sensor nodes that receive multiple cluster announcements, the senor node with the highest energy level becomes the new cluster head. Finally, if the node does not receive any announcement, it sends a message to its neighbor that has the highest energy level requesting cluster head service and that sensor node becomes a new cluster head. Repeating this process results in a clustered sensor network with connected cluster heads, if such a network exists. As the cluster heads perform their operation they will eventually have lower energy resources than other nearby sensor nodes because of their increased functionality. When this occurs, the sensor nodes perform the cluster formation procedure again so that sensor nodes equalize energy consumption throughout the network. To assign slots, the cluster heads perform a distributed algorithm that results in each cluster head having a slot to transmit in and knowing the slots used by each neighbor. Each cluster head picks a random number between one and the number of neighbors it has plus one and transmits this number to its neighbors. If two neighboring cluster heads pick the same number they try again by picking an unused number. If no collision occurs, then the cluster head uses the chosen time slot to transmit data. After the cluster heads determine the TDMA schedule, they distribute the information within the cluster so that the other sensor nodes may use the unassigned slots at the end of the frame for sending their data. GANGS protocol has the disadvantage that cluster formation and reconstruction consumes energy resources and takes time. Furthermore, the slot organization in GANGS also introduces wasted resources since not all slots may get used. Despite these disadvantages, the GANGS protocol provides contention free traffic flow for forwarded traffic while retaining the flexibility and simplicity of a random access protocol within the cluster Hybrid multi-frequency based protocols Recently many applications of sensor networks in both mission-critical operations and wide-area surveillance like real-time streaming for voice and low-rate video delivery [4] require relatively high bandwidth utilization and throughput as well as bounded end-toend delay of a few milliseconds. Therefore, design of effective WSN medium access control (MAC) protocols has become a more challenging task given the unique set of resource constraints in these networks which result in very different design trade-offs than those in wireless ad hoc networks. An important fact to be observed in WSN MAC layer protocol design is that current WSN hardware such as MICAZ [74] and Telos [75] use CC2420 radio [76] which provides multiple frequencies. Given the limited radio bandwidth available for sensor nodes (19.2 Kbps in MICA2 [77] and 250 Kbps in MICAZ and Telos), designing MAC protocols which can exploit the available frequencies to improve parallel transmission and increase the network throughput seems to be an imperative task. The significance of such a design becomes very clear when we notice that almost all of the purposed solutions for WSN MAC layer assume the availability of only one single physical frequency. Although multi-frequency MAC protocols have been a topic of intense research in general wireless networks, the purposed protocols are a poor fit for wireless sensor networks due to the restricted sensor-net hardware, its limited bandwidth, and the small WSN MAC layer packet size. Recently some MAC layer multi-channel protocols have been proposed to improve network performance in WSNs. These protocols typically assign different channels to two hop neighbors to avoid potential interferences, and also design sophisticated MAC schemes to coordinate channel switching and transmissions among nodes. Simulation results show that they can significantly improve network throughput over MAC protocols using a single channel. Following we provide some typical examples of such protocols. MMSN protocol MMSN is a multi-frequency media access control for wireless sensor networks [78]. MMSN protocol exploits the availability of multiple

24 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1595 frequencies provided in radio sensor s hardware to take full advantage of parallel transmission to improve network throughput. The MMSN protocol consists of two aspects: frequency assignment and media access. The frequency assignment is used to assign different frequencies if enough frequencies exist, or evenly allocate available frequencies if there are more neighbors than available frequencies, to nodes that have potential communication conflicts. MMSN allows users to choose one of four available frequency assignment strategies. The first scheme called exclusive frequency assignment guarantees that nodes within two hops are assigned different frequencies but only functions when the number of available frequencies is at least as large as the number of such nodes. Furthermore, the communication overhead in this scheme is relatively high due to several broadcasts. The second scheme, implicit-consensus provides the mentioned guarantee with smaller overhead but it assumes that physical frequencies are abundant which is not the case in current real-world WSN platforms. The two other schemes, even-selection and eavesdropping do not guarantee the assignment of different frequencies to two-hop neighbors and therefore, do not avoid potential conflicts. It is important to note that although MMSN requires time synchronization during media access in order to provide efficient broadcast support, it does not take advantage of the synchronization service to resolve the conflicts and/or improve its scheme. After frequency assignment, each node gets a physical frequency for data reception. With the assigned frequencies, nodes cooperate to maximize parallel transmission among neighboring space in media access. To provide efficient broadcast support, nodes are time synchronized during media access. A time slot consists of a broadcast contention period (T bc ) and a transmission period (T tran ). During the T bc period, nodes compete for the same broadcast frequency and during the T tran period, nodes compete for shared unicast frequencies. The T tran period also provides enough time to actually transmit or receive a broadcast or unicast data packet. The time slot size depends on the number of nodes that compete for the same frequency and the data packet size. When a node intends to transmit a packet it has to listen for the incoming packets both on its own frequency and the destination s frequency. A snooping mechanism is used to detect the packets on different frequencies which makes the nodes to switch between channels frequently. The big disadvantage of MMSN protocol is that, at the start of each timeslot, all nodes are required to listen on a common channel in order to exchange control information which simply increases the protocol overhead. HyMAC protocol HyMAC [79] is a hybrid TDMA/FDMA MAC protocol suitable for WSN applications in which data gathered by sensor nodes has to be delivered to at least one base station in a timely manner. HyMAC is designed to provide high throughput and small bounded end-to-end delay for the packets exchanged between each node and the base station. The communication period in HyMAC is a fixedlength TDMA cycle composed of a number of frames. Each frame is equivalently divided into several fixed time slots where a slot duration is the time required to transmit a maximum sized packet. In addition, a fixed number of consecutive slots in each cycle starting from its beginning form the scheduled slots while the remaining slots of that cycle are its contention slots. The base station is responsible to assign an appropriate frequency as well as specific time slot(s) to each node through running a certain breadth first search (BFS) algorithm on a tree topology. Such scheduled node will be able to communicate in an energy-efficient collision free manner turning off its radio when it is not necessary. All scheduled nodes employ the low power listening (LPL) technique on contention slots during which they randomly select one slot to send a HELLO message to the base station (using flooding if specific routes are not present). On the other hand, all of the unscheduled nodes like the ones which have just joined the network only operate in contention slots sending the HELLO message in a similar way. If a node hears a HELLO message from any other node in its one-hop distance, it adds the sender to its neighbor list. The updated neighbor list will be included in the next HELLO messages sent by that node. Having received the HELLO messages sent by the nodes, the base station is able to construct the schedule and send it to each node in a SCHEDULE message. Consequently, every node will be able to send DATA messages to its parent using its assigned slot and frequency in a way that maximizes the network throughput and minimizes the overall uplink delay. Figure 16 presents the supported packet types and their format in HyMAC. However, many open questions such as how to maintain time synchronized communication, how to resolve collisions, or how a new node joins the network are not answered in HyMAC protocol.

25 1596 B. YAHYA AND J. BEN-OTHMAN Fig. 16. HyMAC packet format. TMCP protocol TMCP is a realistic and efficient multi-channel medium access protocol for wireless sensor networks [80]. TMCP is a tree-based multichannel designed specifically for data collection applications in WSNs. The idea of TMCP protocol is to firstly partition the whole network into multiple vertex-disjoint subtrees all rooted at the base station and allocate different channels to each sub-tree, and then forward each flow only along its corresponding sub-tree, shown in Figure 17. TMCP has three components, channel detection (CD), channel assignment (CA), and data communication (DC). The CD module finds available orthogonal channels which can be used in the current environment. To do this, two nodes are used to sample the link quality of each channel by transmitting packets to each other, and then among all channels with good link qualities, non-adjacent channels are selected. Assume that k represents the number of non-adjacent channels Fig. 17. The conceptual design of TMCP. selected. Given k orthogonal channels, the CA module partitions the whole network into k sub-trees and assigns one unique channel to each sub tree. This is the key part of TMCP. The goal of partitioning is to decrease potential interference as much as possible. After assigning channels, the DC component manages the data collection through each sub tree. When a node wants to send information to the base station, it just uploads packets along the sub tree it belongs to. Here, authors assume that the base station is equipped with multiple radio transceivers, each of which works on one different channel, because of the tree-based channel assignment strategy. Also, the base station can use this network structure to perform data dissemination. When the base station wants to send commands or update the code, it can send out packets through all transceivers, and then packets will go through every sub-tree and reach all nodes in networks. However, TMCP has some disadvantages. TMCP is designed to support converge-cast traffic and it is difficult to have successful broadcasts due to the partitions. Furthermore, contention inside the branches is not resolved since the nodes communicate on the same channel. Communication between nodes in different sub-trees is blocked. Adjacent channels are not used, and therefore the limit bandwidth is not fully utilized. Similar to TMCP, the protocol presented in Reference [81] uses a control theory approach to assign channels to the clusters of nodes. Initially all the nodes communicate on the same channel and when a channel becomes overloaded nodes migrate to new channels based on the feedback information from the neighbors around. Although, this protocol is lightweight enough to run on MICAZ [74] mote with small memory and code footprint, the protocol has some open issues; there is no discussion about interference among k available channels; The threshold γ is determined based on worst-case delay d, which will change during various network status. Y-MAC protocol Y-MAC is an energy-efficient multi-channel MAC protocol for dense wireless sensor networks [82]. It is based on scheduled access. However, timeslots are not assigned to the senders but to the receivers. At the beginning of each timeslot potential senders for the same receiver contend for the medium. Each timeslot is long enough to transmit one data message. If multiple packets need to be transmitted, then the sender and the receiver hop to a new channel according to a predetermined sequence. Other potential senders also follow the hopping sequence of the receiver.

26 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1597 The main drawback with Y-MAC is that the increased contention especially around the sink node with high data rate scenarios is hard to solve Cross-layer MAC Protocols All MAC protocols discussed so far improve the energy efficiency to a certain extent by exploiting the collaborative nature of wireless sensor networks and its correlation characteristics. However, the main commonality of these protocols is that follow the traditional layered protocols architecture. As these protocols may achieve very high performance in terms of the metrics related to each of these individual layers, they are not jointly optimized to maximize the overall network performance while minimizing the energy expenditure. Considering the scarce energy and processing resources of wireless sensor networks, joint optimization and design of networking layers, (i.e., cross-layer design), becomes the most promising alternative to inefficient traditional layered protocol architecture. Recently, significant work on the cross-layer development of wireless sensor network protocols has emerged. In fact, recent research on wireless sensor networks reveals that cross-layer integration and design techniques result in significant improvement in terms of energy conservation. Generally, many reasons lay behind this improvement. First, the stringent energy, storage, and processing capabilities of wireless sensor nodes necessitate such an approach. The significant overhead of layered protocols results in high inefficiency. Furthermore, recent experimental studies necessitate that the properties of low power radio transceivers and the wireless channel conditions be considered in protocols design. Finally, the eventcentric approach of wireless sensor networks requires application-aware communication protocols. In this section, we review some recent communication protocols devised for wireless sensor networks that focus on cross-layer design approach. The protocols are classified according to the interactions between different layers of OSI network stack. The classification presented in Reference [28] is followed, and other protocols are added MAC + PHY In Reference [83], the energy consumption analysis for physical and MAC layers is performed for three different MAC protocols. The authors provided analysis of energy consumption and concluded that single-hop communication can be more efficient if real radio models are used. Despite of this interesting result, the analysis is based on a linear network, which may not be practical in realistic scenarios. A cross-layer solution among MAC and PHY layers is proposed in Reference [84], a new cross layer based carrier-sensing mechanism for alleviating exposed/hidden node problem, refereed as MP scheme. This scheme uses MAC-address-based physical carrier sensing to determine if the medium is busy. In MP, the addresses of transmitter and receiver of a packet are incorporated into the PHY header. Making use of this address information for its carrier-sensing operation, a node can drastically reduce the detrimental effects of exposed/hidden node. Results show that the proposed scheme is more efficient and more effective than the previous schemes MAC + Network The MAC and routing cross-layer interaction for receiver based routing has been investigated in many research papers [85,86,87]. In these papers authors discuss the energy efficiency, latency, and multi-hop performance of the algorithm. In Reference [88], the geographic random forwarding (GeRaF) is proposed as a routing protocol, but the underlying MAC algorithm is also defined in the work which is based on CSMA/CA. The difficulty of the system proposed is its need for additional radio, which is used for busy tone announcement. Rugin et al. [89], improved GeRaF reducing it to a one-channel system. In Reference [90], the MACRO protocol is proposed, where the routing decision is performed as a result of successive competitions at the medium access level. More specifically, the next hop is selected based on a weighted progress factor and the transmit power is increased successively until the most efficient node is found. Moreover, on off schedules are used. The MAC-CROSS is proposed in Reference [91], MAC-CROSS protocol minimizes the number of nodes that should be awake to complete the communication process. The protocol utilizes the routing information to awake only the nodes that are involved in the routing path. All other nodes that are not in the routing path can stay in their sleep mode until the beginning of the next duty cycle. However the improvement achieved in energy conservation comes at the cost of protocol s latency. A joint scheduling and routing scheme is proposed in Reference [92] for periodic traffic in wireless sensor

27 1598 B. YAHYA AND J. BEN-OTHMAN networks. In this scheme, the nodes form distributed on-off schedules for each flow in the network while the routes are established such that the nodes are only awake when necessary. Since the traffic is periodic, the schedules are then maintained to favor maximum efficiency. The authors also investigate the trade-off between on off schedules and the connectivity of the network. The usage of on off schedules in a cross-layer routing and MAC framework is also investigated in Reference [93]. In this work, a TDMA-based MAC scheme is devised, where nodes distributively select their appropriate time slots based on local topology information. The routing protocol also exploits this information for route establishment. The performance evaluations of all these solutions present the advantages of cross-layer approach at the network and MAC layers. Multi-hop infrastructure network architecture (MINA) is another work for integrating MAC and routing protocols [94]. Ding et al. propose a layered multi-hop network architecture where the network nodes with the same hop count to the base station are grouped into the same layer. Channel access is a TDMA-based MAC protocol combined with CDMA or FDMA. The super-frame is composed of a control packet, a beacon frame, and a data transmission frame. Beacon and data frames are time slotted. In the clustered network architecture, all members of a cluster submit their transmission requests in beacon slots. Accordingly, the cluster-head announces the schedule of the data frame. The routing protocol is a simple multi-hop protocol where each node has a forwarder node at one nearer layer to the base station. The forwarding node is chosen from candidates based on the residual energies. Ding et al. then formulate the channel allocation problem as an NP-complete problem and propose a sub-optimal solution. Moreover, the transmission range of sensor nodes is a decision variable, since it affects the layering of the network (hop-counts change). Simulations are run to find a good range of values for a specific scenario Network + PHY A cross-layer optimization of network throughput for multi-hop wireless networks is presented in Reference [95]. Authors split the throughput optimization problem into two sub-problems; multi-hop flow routing at the network layer and power allocation at the physical layer. The throughput is tied to the per-link data flow rates, which in turn depend on the link capacities and hence, the per-node radio power level. On the other hand, the power allocation problem is tied to interference as well as the link rate. Based on this solution, a CDMA/OFDM based solution is provided such that the power control and the routing are performed in a distributed manner. In Reference [96], new forwarding strategies for geographic routing are proposed. Authors provide expressions for the optimal forwarding distance for networks with automatic repeat request (ARQ) and without ARQ. Moreover, two forwarding strategies for these cases are provided. The forwarding algorithms require the packet reception rate of each neighbor for determination of the next hop and construct routes accordingly. Although the new forwarding metrics illustrate the advantages of cross-layer forwarding techniques in WSNs, the analysis for the distribution of optimal hop distance is based on a linear network structure Transport + PHY In Reference [97], a cross-layer optimization solution for power control and congestion control is considered. Authors provided analytical analysis of power control and congestion control, and the trade-off between layered and cross-layer approach is presented. Based on this framework, a cross-layer communication protocol based on CDMA is proposed, where the transmission power and the transmission rate is controlled. However, the proposed solutions only apply to CDMA-based wireless multi-hop networks, which may not apply to WSNs that CDMA technology may not be feasible Three-layer solutions In addition to the proposed protocols that focus on pair-wise cross-layer interaction, more general crosslayer approaches among three protocol layers exist. In Reference [98], the optimization of transmission power, transmission rate, and link schedule for TDMA-based WSNs is proposed. The optimization is performed to maximize the network lifetime, instead of minimizing the total average power consumption. In Reference [99], adaptation strategies that maximize the network lifetime through joint routing, scheduling, and link layer optimization is proposed. Authors propose a variable-length TDMA scheme where the slot length is optimally assigned according to

28 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1599 the routing requirement while minimizing the energy consumption across the network. The optimization problem considers energy consumption that includes both transmission energy and circuit processing energy. Based on this analysis, it is shown that single-hop communication may be optimal in some cases where the circuit energy dominates the energy consumption instead of transmission energy. Although the optimization problems presented in the paper are insightful, no communication protocol for practical implementation is proposed. Moreover, the transport layer issues such as congestion and flow control are not considered Quality of Service Specific MAC Protocols The concepts of latency, throughput, and delay jitter were not primary concerns in most of the presented work on sensor networks. However, the increasing interest in real time applications of sensor networks has posed additional challenges on protocol design. For example, handling real time traffic of emergent event triggering in monitoring based sensor network requires that end-to-end delay is within acceptable range and the variation of such delay is acceptable [100]. Such performance metrics are usually referred to as quality of service (QoS) of the communication network. Therefore, collecting sensed real time data requires both energy and QoS aware MAC protocol in order to ensure efficient use of the energy resources of the sensor node and effective delivery of the gathered measurements [101]. However, achieving QoS guarantees in sensor network is a challenging task, because of the strict resource constraints (limited battery power and data memory) of the sensor node, and the hostile environments in which they must operate [102]. QoS provisioning in wireless sensor networks is acquiring the attention of many researchers more and more. Recently many MAC protocols that support some type of quality of service in wireless sensor networks are proposed [ ]. Some examples are discussed below QoS control for sensor networks [103] Authors explicitly exploit node redundancy. They developed an adaptive scheme for each sensor to determine independently whether to transmit or not so that a fixed total number of transmissions occur in each slot. The protocol accomplishes its task by allowing the base station to communicate QoS information to each sensor node within the network through a broadcasting channel, and by use the Gur Game mathematical paradigm to dynamically adjust the optimum number of active sensors. The protocol makes tradeoffs between the required number of sensors that should be powered-up so that enough data is being collected in order to meet the required QoS and number of sensors that should be turnedoff to save a considerable amount of battery power of sensor nodes, and hence maximizing the network s lifetime. Here, the concept of QoS is defined as the total number of transmissions that should occur in each slot in order to gather enough data (i.e., information quality) Q-MAC protocol Q-MAC is a QoS-aware medium access control protocol [104]. Q-MAC assumes an environment of multi-hop WSNs where nodes may generate packets with different priorities. The design objective of Q-MAC is to minimize energy consumption and provide QoS guarantees. Q-MAC is composed of intranode and inter-node QoS scheduling mechanisms. The intra-node QoS scheduling scheme classifies outgoing packets according to their priorities, while the inter-node QoS scheduling solution handles channel access with the objective of minimizing energy consumption via reducing collision and idle listening. The intra-node scheduling mechanism employs multiple first-in first-out (FIFO) queues with different priorities, among which an instant queue has the highest priority and its en-queued packets are always instantly served. Self-generated and relayed packets are classified to different queues with several QoS metrics, such as content importance and number of traveled hops. After a packet is scheduled for transmission, the inter-node scheduling mechanism, power conservation MACAW [110], is executed to achieve loosely prioritized random access (LPRA) between sensor nodes. In PC-MACAW, a successful transmission consists of two periods: the contention period and the packet transmission period. In the contention period, a node sends out RTS after waiting for a certain duration (contention time) and expects a CTS packet before accessing the channel. The contention time is randomly generated with a contention window size CW, where CW is determined by each node s transmission urgency including packet criticality, number of transmitted

29 1600 B. YAHYA AND J. BEN-OTHMAN hops, residual energy, and queue s proportional load. After accessing the channel, the node enters the transmission period to send data packets and waits for an ACK packet. In the case of collision, CW is doubled and the packet is retransmitted. When the difference between the current time and when the packet is generated exceeds a threshold, the packet is dropped. Q-MAC presents a combined effort of intra-node and inter-node QoS scheduling in WSNs. It is shown through simulations that Q-MAC provides the equivalent QoS while consuming less energy in comparison with an existing mechanism, S-MAC. However, complex scheduling mechanisms and relatively loosely defined QoS metrics stand out as shortcomings of this proposal EQ-MAC An energy efficient hybrid medium access control scheme for wireless sensor networks with quality of service guarantees (named as EQ-MAC) is proposed in Reference [109]. EQ-MAC is a MAC layer protocol for wireless sensor networks that reduces energy consumption and provides quality of service guarantees through the use of service differentiation concept. The EQ-MAC consists of two sub-protocols: classifier MAC (C-MAC), and channel access MAC (CA-MAC). C-MAC is responsible of classifying gathered data at sensor nodes based on its importance and then stores it in the appropriate queue of the node s queuing system. The CA-MAC is an energy efficient medium access mechanism that uses a hybrid approach of both scheduled and unscheduled schemes to gain a save in energy consumed by sensor nodes, and hence prolonging the network s lift time. The save in energy achieved by CA-MAC is coming from the differentiation between small and long data messages. Long data messages are assigned scheduled slots with no contention (CA-MAC assigns time slots only to those nodes which have data to send, this allows an efficient energy use of the TDMA slots), while small periodic control messages are assigned random access slots CoCo protocol CoCo is a coloring-based real-time communication scheduling [111]. CoCo is designed for multi-hop WSN environments that use IEEE MAC protocol, where all communication is unicast. It is assumed that node locations are available at all times, and a central scheduler running CoCo is in charge of communication scheduling. CoCo aims to schedule real-time communication avoiding collisions and minimizing the overall packet transmission time. In CoCo, a set of messages waiting for transmission at various sensors are modeled with a weighted, directed graph G = (V, E), where a vertex denotes a sensor node, a directed edge from vertex vi to vj denotes a message to be sent from sensor vi to vj, and the weight of an edge denotes the transmission time. The communication problem is equivalent to assigning a color to each edge. Here, each color represents a set of simultaneous communication during disjoint time periods, and the weight of a color equals the maximum weight of the edges assigned with this color. CoCo aims to find edge color assignment such that (i) no adjacent edges share the same color, (ii) no two edges with the same color interfere with each other, and (iii) the overall weight of used colors is minimized. Since the optimal coloring problem is NP-complete, a coloring heuristic is presented in Reference [111]. First, the edges of the vertex with the maximum degree are assigned different colors. Once a color is assigned to an edge, it is removed from the palettes of all adjacent edges, and its weight is updated. Then, the following steps are repeated until all edges are colored: the edge with the smallest palette is chosen. A color from the available palette is assigned to the edge such that no other edge with that color interferes with the chosen edge. Then, the chosen color is removed from the palettes of all uncolored adjacent edges. Three heuristics are presented for selecting a color from an edge s palette: the random color selection heuristic randomly picks a color from the palette that does not cause interference. The least used color (LUC) heuristic chooses the color with the smallest number of colors. The minimal weight color (MWC) heuristic first checks whether there are colors in the palette whose weights are higher than the edge. If so, among these colors, the color with the smallest weight is selected. Otherwise, the color with the maximum weight is assigned from the palette. CoCo aims to schedule a set of communication events with the minimum communication time in real-time WSNs. According to the simulations, MWC-based CoCo provides performance superior to that of the other two-color selection heuristics, and its performance is close to the optimal solution. The central computation requirement limits the applicability of CoCo in large-scale sensor networks.

30 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS Future Research Directions and Open Issues Although many medium access schemes have been proposed for sensor networks, MAC layer protocol design is still largely open to research. In this section, we identify these open issues, and present future research directions. Standardization, there is a lack in the MAC protocol standardization for sensor networks. Up to now there is no protocol accepted as standard. The main reason behind this is that the choice of MAC protocol in general is application dependant, which means that there will many standard MAC protocols for sensor networks. Another reason is the lake in standardization in lower layers (physical layer) and the physical sensor hardware. TDMA has a natural advantage of collision-free medium access. However, it includes clock drift problems and decrease throughput at low traffic loads due to idle slots. The difficulty with TDMA systems are the synchronization of the nodes and adaptation to topology changes where these changes are caused by insertion of new nodes, exhaustion of battery capacities, broken links because of interference, sleep schedules of relay nodes, and scheduling caused by clustering algorithms. The slot assignments, therefore, should be done regarding such possibilities. However, it is not easy to change the slot assignment within a decentralized environment for traditional TDMA, since all nodes must agree on the slot assignment. FDMA is another scheme that offers a collision free medium. Though, it brings an additional circuitry requirement to dynamically communicate with different radio channels. This increases the cost of the sensor nodes which is contrary to the objective of the sensor network systems. CDMA also offers collision free medium, but its high computational requirement is a major obstacle for less energy consumption objective of the sensor node. If it is shown that the high computational complexity of CDMA could be traded with its collision avoidance feature, CDMA protocols could also be considered as good solutions for wireless sensor networks. CSMA based protocols have a lower delay and promising throughput potential at low traffic loads, which generally happens to the case in wireless sensor networks. However, additional collision avoidance or collision detection methods should be employed to handle the collision possibilities. Lake of comparison of TDMA, CSAM, or other medium access protocols in a common framework is a crucial deficiency of the literature. To date, the primary design goal for sensor networks in general and MAC in particular has been energy efficiency. However, as new applications of sensor networks emerge other optimization criteria (or quality of service parameters) such as latency and compliance with real time constraints, or reliable data delivery may gain importance. One particular issue is that many applications need to be optimized for multiple, conflicting criteria. Hence, applications need a way to implement particular trade-offs between these conflicting goals. Mobility Management, considering mobility in MAC protocol design has been identified as an open research challenge in sensor networks for quite some time and yet even the most recent MAC protocols appearing in the literature do not explicitly consider mobility at the MAC layer except few ones such as those described in References [56,112,113,114]. Recently, there has been an increased interest in medical care and disaster response applications of sensor networks and these environments make use of mobile sensor nodes. So, there is much room for research in this area. Scalability issues are most notable among possible future research directions given the growing ambition for very large deployments of tiny sensors and the widening scope of the use in many applications. Adaptability to traffic and topology changes, to conserve more energy MAC protocol should adapt to changes in both network topology and traffic characteristics. However, adaptability to changes usually increases the protocol complexity, which brings other disadvantages and leads to consuming sensor node resources. More research is still required to address the adaptability issue. Coverage is another fundamental problem in the emerging area of wireless sensor networks. Due to the large variety of sensors and applications, coverage is subject to a wide range of interpretations [115,116]. In general, coverage can be considered as the measure of quality of service of a sensor network. Furthermore, coverage formulations can try to find weak points in a sensor field and suggest future deployment or reconfiguration schemes for improving the overall quality of service. Despite of the existing work that tackles the coverage problem in wireless sensor networks [116,117], still there is a large horizon of possibilities for future researches on this subject. The solution for obtaining a ubiquitous context is to

31 1602 B. YAHYA AND J. BEN-OTHMAN Table I. Summary of general classes of MAC protocols for wireless sensor networks. Classification of WSNs MAC protocols Advantages Disadvantages General classification Specific classification Unscheduled protocols Scheduled protocols Multi-channel based Use separate channels to increase Additional hardware requirements bandwidth and hence more energy consumption Suitable for delay sensitive applications Event-oriented Sensitive to traffic changes and High overhead due to parameters adaptable to application requirements computations and higher complexity Preamble based Simple complexity Affected by the traffic patterns Asynchronous approach Long preambles may cause Good energy efficiency performance degradation in terms of Rendezvous based Low energy consumption Simple protocols Scalable Slotted contention based latency Greatly affected by traffic patterns, channel characteristics Range of applications range is limited Energy conservative protocols Collisions occur frequently Scalable Control overhead is high Time division based Collision free Tight synchronization is required Reduce idle listening Limited scalability and adaptability Energy Conservative protocols to node changes Clustering based Coordinate sensor nodes for energy conservation Enables differentiation between local traffic from global traffic to conserve energy Priority based Suitable for real time applications, and delay sensitive applications Reservation based Minimizes overhearing, collision, idle listening. Transitions between different states of the sensor s node can be minimized as well Hybrid Protocols Preamble based Nearby sensor nodes are synchronized and only for the duration of the transmission which conserves energy Sensitive to traffic patterns Traffic sensitive The protocol adapts itself to traffic types based and conditions for a conserve in energy Reservation based Idle listening is greatly minimized Enable the application to balance traffic and energy requirements Low power consumption Clustering based Good coordination of sensor nodes for energy conservation Clustering reduces the transmission distance and hence use low transmission power Multi-frequency Good throughput and good channel based utilization Suitable for real time streaming Cross-layer protocols Jointly optimized to maximize the overall network performance while minimizing the energy expenditure Exploit dependency between layers of the WSN protocol stack to achieve some performance gains Protocol complexity is high Clusters formation and maintenance consumes sensor resources Control overhead is high Latency is high Priorities calculation consumes sensor node resources Need tight synchronization Limited scalability and adaptability to node changes Protocol complexity is high Excessive overhead Long preambles causes increase in latency Protocol complexity is high High overhead due to exchange of traffic characteristics between Tight synchronization is needed which increases the protocol complexity Inter-grid contention leads to diminish in efficiency Cluster formation and reconstruction consumes energy resources and time Slot organization introduces wasted resources Protocol control overhead is high Implementation complexity is high Cross layer design decreases interchangeability, flexibility, modularity Protocols based on cross-layer designs with tight coupling between the layers become hard to review and redesign

32 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1603 assemble information from a combination of related services. Such information fusion is similar in intent to the related and well-researched area of sensor fusion. For the context of coverage, negotiation and resolution strategies are needed to integrate information from this stage to be used in related contexts such as tracking mobile objects in the network and handling obstacles. Quality of service handling, despite the approaches discussed above, the area of QoS in WSNs is still rather unexplored field. The problem here is how to balance between application s QoS requirement and energy constraints [26]. Hybrid approach in protocol design is a promising methodology. While there have been significant works in this domain, extensive research is still required to address the wireless sensor MAC protocol requirement discussed above. Cross Layer Interaction, despite of the existing research on developing new communication protocol based on cross layer interactions, there is still much to be gained by rethinking the protocol functions of network layers in a unified way so as to provide a single communication module for efficient communication in WSNs. There are several open research problems towards the development of systematic techniques for cross layer design of wireless sensor network protocols. For more details refer to Reference [28]. Realistic Simulation Models, assumptions made in most simulation environments do not necessarily reflect the real-world conditions (e.g., a radio s transmission area is circular, all radios have equal range... etc). In order to fully understand the complexity of designing a MAC protocol and to develop solutions which work in real life, it is necessary to develop more realistic radio and energy simulation models. It is important to revisit Kotz s mistaken axioms of wireless network research (see Reference [118]), to understand why MAC protocols that yield extremely accurate results in simulation fail in real life deployments. In order to obtain more realistic insight into MAC layer performance, sensor networks researchers should move from simulation to prototype or real-world experiments. In summary, existing wireless MAC protocols focus on optimizing system energy with considering the MAC performances, yet still do not adequately consider all of the requirements of sensor networks. The key challenge remains to provide predictable delay and/or prioritization guarantees while minimizing overhead packets and energy consumption. 5. Conclusion Medium access control protocols in wireless sensor networks have attracted a lot of attention in the recent years and introduced unique challenges compared with traditional MAC in other wireless networks. In this paper, we have discussed the special requirements of wireless sensor network MAC protocols, classified the research on MAC in wireless senor networks and covered many proposed protocols, and discussed the open research issues as well. In Table I, we have summarized the general classes of MAC protocols discussed in the paper. Although many MAC schemes have been proposed for wireless sensor networks, the area is still largely open to research and there is no clear single direction in which future efforts should be directed as discussed in the previous section. It remains an open question of great interest, does a general, and flexible MAC protocol exist that supports various applications and operating environments while consuming minimal power and offering acceptable traffic characteristics? However, Along with the current research of MAC protocols in wireless sensor networks, we encourage more insight into the problems and more development in solutions to the open research issues presented in the previous section. References 1. Akyildiz F, Su W, Sankarasubramaniam Y, Cayirci E. Wireless sensor networks: a survey. Computer Networks 2002; 38: Min R, Bhardwaj M, Cho SH, et al. Low power wireless sensor networks. In The Proceedings of International Conference on VLSI Design, Bangalore, India, January Tilak S, Abu-Ghazaleh NB, Heinzelman W. A taxonomy of wireless microsensor network models. ACM SIGMOBILE Mobile Computing and Communication Review (MC2R) 2002; 6(2): Rabaey JM, Ammer MJ, da Silva JL, Patel D, Roundy S. PicoRadio supports ad hoc ultra low power wireless networking. IEEE Computer 2000; 33(7): Pottie GJ, Kaiser WJ. Wireless integrated network sensors. Communication of the ACM 2000; 43(5): Katz RH, Kahn JM, Pister, Mobile KSJ. Networking for smart dust. In Proceedings of the 5th Annual ACM/IEEE International Conference on Mobile Computing and Networking (MobiCom 99), Seattle, WA, August Bokareva T, Hu W, Kanhere S, Ristic B, Gordon N, Bessell T, Rutten M, Jha S. Wireless Sensor Networks for Battlefield Surveillance. In Proceedings of the LWC Mainwaring A, Polastre J, Szewczyk R, Culler D, Anderson J. Wireless sensor networks for habitat monitoring. In Proceedings of the ACM WSNA Conference, September, 2002.

33 1604 B. YAHYA AND J. BEN-OTHMAN 9. Tolle G, Polastre J, Szewczyk R, Turner N, Tu K, Burgess S, Gay D, Buonadonna P, Hong W, Dawason T, Culler D. A macroscope in the redwoods, In Proceedings of the ACM SenSys, Xu N, Rangwala S, Chintalapudi K, Ganesan D, Broad A, Govindan R, Estrin D. A wireless sensor network for structural monitoring, In Proceedings of the ACM SenSys Conference, November Srinivasan K, Ndoh M, Nie H, Xia H, Kaluri K, Ingraham D. Wireless technologies for condition-based maintenance (CBM) in petroleum plants. In Proceedings of the 1st International Conference on Distributed Computing in Sensor Systems (Poster Session), Sohrabi K, Gao J, Ailawadhi V, Pottie GJ. Protocols for selforganization of a wireless sensor network. IEEE Personal Communications 2000; 7(5): Buczak A, Jamalabad V. Self-organization of a heterogeneous sensor network by genetic algorithms. In Intelligent Engineering Systems Through Artificial Neural Networks, Vol. 8, Dagli CCH, Akay M, Buczak AL, Ersoy O, Fernandez BR (eds). ASME Press: New York, 1998; Demirkol I, Ersoy C, Alagoz F. MAC protocols for wireless sensor networks: a survey. IEEE Communications Magazine 2006; 44(4): Ye W, Heidemann J. Medium access control in wireless sensor networks. USC/ISI Technical Report ISI-TR-580, Kredo K II, Mohapatra P. Medium access control in wireless sensor networks. Journal of Computer Networks 2007; 51: Bohm A. State of the art on energy-efficient and latency constrained networking protocols for wireless sensor networks, HH. Technical Report, IDE0749, June Langendoen K. Meduim access control in wireless sensor networks. In Practice and Standards (Volume II), Wu H, Pan Y (eds). Nova Science Publishers: Hauppauge, NY, Baronti P, Pillai P, Chook VWC, Chessa S, Gotta A, Hu YF. Wireless sensor networks: a survey on the state of the art and the and ZigBee standards. Journal of Computer Communication 2007; 30(7): Czapski PP. A survey: MAC protocols for applications of wireless sensor networks. TENCON 2006, Langendoen K, Halkes G. Energy-efficient medium access control. In The Embedded Systems Handbook, Zurawski R (ed.). CRC press: USA, ISBN: Jurdak R, Lopes CV, Baldi P. A Survey, classification and comparative analysis of medium access control protocols for ad hoc networks. IEEE Communications Surveys and Tutorials 2004; 6(1): Naik P, Sivalingam KM. A survey of MAC protocols for sensor networks. Wireless Sensor Networks. Kluwer Academic Publishers Norwell: MA, USA, 2004; Ganesan D, Cerpa A, Ye W, Yu Y, Zhao J, Estrin D. Networking issues in wireless sensor networks. Journal on Parallel and Distributed Computing 2004; 64(7): Al-Karaki JN, Kamal AE. Routing mechniques in wireless sensor networks: a survey. IEEE Journal on Wireless Communications 2004; 11(6): Chen D, Varshney PK. QoS support in wireless sensor networks: a survey. Proceedings of the International Conference on Wireless Networks, Las Vegas, Nevada, USA, June Ali M, Saif U, Dunkels A, Voigt T, Römer K, Langendoen K, Polastre J, Uzmi ZA. Medium access control issues in sensor networks. ACM Computer Communication Review Melodia T, Vuran MC, Pompili D, The state of the art in crosslayer design for wireless sensor networks. Network Architect 2005, LNCS 3883, 78 92, Springer-Verlag, Berlin, Heidelberg, Boukerche A, Richard W., Nelem Pazzi, Regina Borges de A. Fault-tolerant wireless sensor network routing protocols for the supervision of context-aware physical environments. Journal of Parallel and Distributed Computing (JPDC) 2006; 66(4): Abramson N. The Aloha system: another alternative for computer communications. In AFIPS Conference Proceedings, 36, 1970; Kleinrock L, Tobagi F. Packet switching in radio channels: part I carrier sense multiple access modes and their throughputdelay characteristics. IEEE Transactions on Communications 1975; 23(12): Roberts L. Aloha packet system with and without slots and capture. Stanford Research Institute, Network Information Center, Tech. Rep. ASS Note 8, Boukerche A (Ed.). Algorithms and protocols for wireless sensor networks, Weley Series on Parallel and Distributed Computing. John Wiley & Sons, Inc.: Hoboken, New Jersey, US, ISBN: Ye W, Heidemann J, Estrin D. An energy-efficient MAC protocol for wireless sensor networks. In Proceedings of the IEEE INFOCOM, New York City, NY (June 2002). 35. Boukerche A, Samarah S. A Novel algorithm for mining association rules in wireless ad hoc sensor networks. IEEE Transactions on Parallel and Distributed Systems 2008; 19(7): Ye W, Heidemann J, Estrin D. Medium access control with coordinated adaptive sleeping for wireless sensor networks. IEEE/ACM Transactions on Networking 2004; 12(3): Jones C, Sivalingam K, Agrawal P, Chen JC. A survey of energy efficient network protocols. Wireless Networks 2001; Lettieri P, Srivastava B. Advances in wireless terminals (I). IEEE Personal Communications 1999; 6(1): Havinga P, Smit G, Bos M. Energy efficient wireless ATM design. In Proceedings of wmatm 99 (June 1999). 40. Miller M, Vaidya N. Minimizing energy consumption in sensor networks using a wake up radio. In Proceedings of the IEEE International Conference on Wireless Communications and Networks (WCNC 04), Atlanta, GA, (March 2004). 41. Shih E, Cho S, Ickes N, Min R, Sinha A, Wang AA, Chandrakasan A. Physical layer driven protocol and algorithm design for energy efficient wireless sensor networks. In Proceedings of the 7th ACM/IEEE Conference on Mobile Computing and Networks, Rome, Italy (July 2001). 42. Kleinrock L, Tobagi F. Packet switching in radio channels: part II the hidden terminal problem in CSMA and busy tone solutions. IEEE Transactions on Communications 1975; 23: Singh S, Raghavendra C. PAMAS: power aware multi-access protocol with signaling for ad hoc networks. SIGCOMM Computer Communications Review 1998; 28(3): Vuran MC, Akyildiz IF. Spatial correlation-based collaborative medium access control in wireless sensor networks. IEEE/ACM Transactions on Networking 2006; 14(2): Jamieson K, Balakrishnan H, Tay YC. Sift: a MAC protocol for event-driven wireless sensor networks. MIT Laboratory for Computer Science, Tech. Rep., 2003; Namboodiri V, Keshavarzian A. Alert: an adaptive low-latency event-driven MAC protocol for wireless sensor networks. In Proceedings of ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN 08), April Schurgers C, Tsiatsis V, Ganeriwal S, Srivastava M. Optimizing sensor networks in the energy-latency-density design space. IEEE Transactions on Mobile Computing 2002; 1(1):

34 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS Boukerche A (Ed.), Handbook of Algorithms for Wireless Networking and Mobile Computing. Chapman and Hall/CRC, Taylor & Francis Group: Boca Raton, New York, US, 2005; ISBN: Polastre J, Hill J, Culler D. Versatile low power media access for wireless sensor networks. In Proceedings of the International Conference on Embedded Networked Sensor Systems (SenSys), 2004; Shi X, Stromberg G. SyncWUF: an ultra low-power MAC protocol for wireless sensor networks. IEEE Transactions on Mobile Computing 2007; 6: Ansari J, Zhang X, Mähönen P. Poster abstract: traffic aware medium access control protocol for wireless sensor networks. In Proceedings of EuroSys2008, Glasgow, Scotland, 31st March 4th April Lin E-YA, Rabaey JM, Wolisz A. Power-efficient rendezvous schemes for dense wireless sensor networks, Proceedings of the IEEE International Conference on Communications (ICC), 2004; 7: Lu G, Krishnamachari B, Raghavendra CS. An adaptive energy efficient and low-latency MAC for data gathering in wireless sensor networks. Proceedings of the 18th International Parallel and Distributed Processing Symposium, April Dam TV, Langendoen K. An adaptive energy-efficient MAC protocol for wireless sensor networks. The First ACM Conference on Embedded Networked Sensor Systems (SenSys 03), Los Angeles, CA, USA, November, Lin P, Qiao C, Wang X. Medium access control with a dynamic duty cycle for sensor networks, IEEE Wireless Communications and Networking Conference, Volume 3, , March Pham H, Jha S. An adaptive mobility aware MAC protocol for sensor networks. The 1st IEEE international Conference on Mobile ad hoc and Sensor Networks, October 24 27, 2004, Florida, USA. 57. Yadav R, Varma S, Malaviya N. Optimized medium access control for wireless sensor network. International Journal of Computer Science and Network Security (IJCSNS), 2008; 8(2): Zheng T, Radhakrishnan S, Sarangan V. PMAC: an adaptive energy-efficient MAC protocol for wireless sensor networks. In Proceedings of the 19th IEEE International Parallel and Distributed Processing Symposium (IPDPS 05). 2005; Chang Y-C, Jiang J-R, Sheu J-P, Shih H-Y. ADCA: an asynchronous duty cycle adjustment MAC protocol for wireless sensor networks. In The proceedings of GLOBECOM 08, New Orleans, LA, USA, 30 November 4 December Sagduyu YE, Ephremides A. The problem of medium access control in wireless sensor networks. IEEE Wireless Communications 2004; 11(6): van Hoesel L, Havinga P. A lightweight medium access protocol (LMAC) for wireless sensor networks. In 1st International Workshop on Networked Sensing Systems (INSS 2004), Tokyo, Japan, June Dulman S, Hoesel L, Nieberg T, Havinga P. Collaborative communication protocols for wireless sensor networks. European Research on Middleware and Architectures for Complex and Embedded Systems Workshop, Italy, April Handy MJ, Haase M, Timmermann D. Low energy adaptive clustering hierarchy with deterministic cluster-head selection. In Proceedings of Fourth IEEE Conference on Mobile and Wireless Communications Networks, Stockholm, Erschienen, Bao L, Garcia-Luna-Aceves J. A new approach to channel access scheduling for ad hoc networks. In Proceedings of the International Conference on Mobile Computing and Networking (MobiCom), 2001; Havinga P, Smit G. Energy-efficient TDMA medium access control protocol scheduling. In Proceedings of the Asian International Mobile Computing Conference (AMOC 2000). 66. El-Hoiydi A. Spatial TDMA and CSMA with preamble sampling for low power ad hoc wireless sensor networks. In Proceedings of ISCC 2002, 7th International Symposium on Computers and Communications, , 1 4 July Rhee I, Warrier A, Aia M, Min J. Z-MAC: a hybrid MAC for wireless sensor networks. In Technical Report, Department of Computer Science, North Carolina State University, April Rhee I, Warrier A, Xu L. Randomized dining philosophers to TDMA scheduling in wireless sensor networks. In Technical Report, Computer Science Department, North Carolina State University, Raleigh, NC, Rajendran V, Obraczka K, Garcia-Luna-Aceves J. Energyefficient, collision free medium access control for wireless sensor networks. In Proceedings of the International Conference on Embedded Networked Sensor Systems (SenSys), 2003, Adamou M, Lee I, Shin I. An energy efficient real-time medium access control protocol for wireless ad-hoc networks. In WIP session of IEEE Real-time systems Symposium (RTSS 01), London, UK (December 2001). 71. Wang H, Zhang X, Khokhar A. An energy-efficient lowlatency MAC protocol for wireless sensor networks. In Proceedings of the IEEE GLOBECOM 06, 1 5, November Bao L. Hybrid channel access scheduling in ad hoc networks. In Proceedings of the 10th IEEE International Conference on Network Protocols (ICNP), Biaz S, Barowski YD. GANGS: an energy efficient MAC protocol for sensor networks. In Proceedings of the Annual Southeast Regional Conference, 2004; XBOW MICAz Mote Specifications Polastre J, Szewczyk R, Culler D. Telos: enabling ultra-low power wireless research. In ACM/IEEE IPSN/SPOTS 2005, April CC GHz IEEE /ZigBee-ready RF Transceiver Hill J, Szewczyk R, Woo A, Hollar S, Culler D, Pister K. System architecture directions for networked sensors. In The 9th International Conference on Architectural Support for Programming Languages and Operating Systems, November 2000, Zhou G, Huang C, Yan T, He T, Stankovic JA, Abdelzaher TF. MMSN: multi-frequency media access control for wireless sensor networks. Proceedings of IEEE INFOCOM, Mastooreh S, Hamed S, Antonis K. HyMAC: hybrid TDMA/FDMA medium access control protocol for wireless sensor networks. In Proceedings of PIMRC 2007, September 2007, Wu Y, Stankovic J, He T, Lin S. Realistic and efficient multichannel communications in wireless sensor networks. In Proceedings of IEEE INFOCOM 2008, 2008; Le HK, Henriksson D, Abdelzaher T. A practical multichannel media access control protocol for wireless sensor networks. In Proceedings of IPSN 08, April 2008; Kim Y, Shin H, Cha H. Y-MAC: an energy-efficient multichannel MAC protocol for dense wireless sensor networks. In Proceedings of IPSN 08, April 2008; Haapola J, Shelby Z, Pomalaza-Raez C, Mahonen P. Crosslayer energy analysis of multi-hop wireless sensor networks. In EWSN 05, 2005; Chan A, Chang S. LiewMerit of PHY-MAC cross-layer carrier sensing: A MAC-address-based physical carrier sensing scheme for solving hidden-node and exposed-node problems in large-scale Wi-Fi networks. In Proceedings of the 31st IEEE

35 1606 B. YAHYA AND J. BEN-OTHMAN Conference on Local Computer Networks, November 2006; Skraba P, Aghajan H, Bahai A. Cross-layer optimization for high density sensor networks: distributed passive routing decisions. In Proceedings of 3rd International Conference on AD-HOC Networks & Wireless, Vancouver, Canada, July Zorzi M, Rao R. Geographic random forwarding (GeRaF) for ad hoc and sensor networks: multihop performance. IEEE Transactions on Mobile Computing 2003; 2(4): Zorzi M, Rao R. Geographic random forwarding (GeRaF) for ad hoc and sensor networks: energy and latency performance. IEEE Transactions on Mobile Computing 2003; 2(4): Zorzi M. A new contention-based MAC protocol for geographic forwarding in ad hoc and sensor networks. Proceedings of the IEEE International Conference on Communications 2004; 6: Rugin R, Mazzini G. A simple and efficient MAC-routing integrated algorithm for sensor network. IEEE International Conference on Communications 2004; 6: Ferrara D, Galluccio L, Leonardi A, Morabito G, Palazzo S. MACRO: an integrated MAC/routing protocol for geographical forwarding in wireless sensor networks. In Proceedings of the IEEE Infocom 05, Vol. 3, March 2005; Suh C, Ko YB, Son DM. An energy efficient cross-layer MAC protocol for wireless sensor networks. In Proceedings of the International Workshop in Sensor Networks (IWSN 2006), in conjunction with the 8th Asia-Pacific Web Conference (APWeb 2006), Harbin, China, Shen, HT, Li J, Li M, Ni J, Wang W (eds). Springer-Verlag: Berlin, Heidelberg LNCS 3842, Sichitiu ML. Cross-layer scheduling for power efficiency in wireless sensor networks. In Proceedings of the IEEE INFOCOM conference, Hong Kong, 2004; 3: van Hoesel L, Nieberg T, Wu J, Havinga PJM. Prolonging the lifetime of wireless sensor networks by cross-layer interaction. IEEE Wireless Communications 2004; 11(6): Ding J, Sivalingam K, Kashyapa R, Chuan LJ. A multilayered architecture and protocols for large-scale wireless sensor networks. IEEE 58th Vehicular Technology Conference, VTC Fall 2003, Volume 3, 6 9 October 2003; Yuan J, Li Z, Yu W, Li B. A Cross-Layer optimization framework for multicast in multi-hop wireless networks wireless internet. In Proceedings of WICON 05, July 2005; Seada K, Zuniga M, Helmy A, Krishnamachari B. Energyefficient forwarding strategies for geographic routing in lossy wireless sensor networks. In Proceedings of ACM Sensys 04, November Chiang M. Balancing transport and physical Layers in wireless multihop networks: jointly optimal congestion control and power control. IEEE JSAC 2005; 23(1): Madan R, Cui S, Lall S, Goldsmith A. Cross-layer design for lifetime maximization in interference-limited wireless sensor networks. In Proceedings of the IEEE INFOCOM 05, Vol. 3, March 2005; Cui S, Madan R, Goldsmith A, Lall S. Joint routing, MAC, and link layer optimization in sensor networks with energy constraints. In Proceedings of the ICC 05, Vol. 2, May 2005; Younis M, Akkaya K, Eltoweissy M, Wadaa A. On handling QoS traffic in wireless sensor networks. In Proceedings of the 37th Annual Hawaii International Conference on System Sciences, HICSS04, Boukerche A, Chatzigiannakis I, Nikoletseas S. Powerefficient data propagation protocols for wireless sensor networks. Simulation Special Issue: Transactions of the Society for Modeling and Simulation International 2005; 81(6): Chen D, Varshney PK. QoS support in wireless sensor networks: a survey. In International Conference on Wireless Networks, 2004; Iyer R, Kleinrock L. QoS Control for Sensor Networks. In Proceedings of the international Conference on Communications, ICC 03, May Yang L, Elhanany I, Hairong Q. An energy-efficient QoSaware media access control protocol for wireless sensor networks. In The Proceedings of Mobile ad hoc and Sensor Systems Conference, Lee WL, Datta A, Cardell-Oliver R. QMAC: a quality of service-oriented medium access control protocol for data gathering in wireless sensor networks. In Technical Report UWA-CSSE , The University of Western Australia, Zhao Q, Tong L. QoS specific medium access control for wireless sensor networks with fading. In ACSP Technical Report TR , School of Electrical and Computer Engineering, Cornell University, June Wu Y, Fahmy S, Shroff NB. Optimal sleep/wake scheduling for time-synchronized sensor networks with QoS guarantees. In Proceeding off IEEE IWQoS, June Frolik J. QoS control for random access wireless sensor networks. In Fifth IEEE Wireless Communications and Networking Conference, Atlanta GA, USA, 2004; Yahya B, Ben-Othman J. An energy efficient hybrid medium access control scheme for wireless sensor networks with quality of service guarantees. In The Proceedings of GLOBECOM 08, New Orleans, LA, USA, November 30 December Bharghavan V, Demers A, Shenker S, Zhang L. MACAW: a media access protocol for wireless LANS. In Proceedings of the ACM SIGCOMM, Vol. 24, No. 4, Li H, Shenoy P, Ramamritham K. Scheduling communication in real-time sensor applications. In Proceedings of the 10th IEEE Real-Time and Embedded Technology and Applications Symposium (RTAS 04), May 2004; Ali M, Suleman T, Uzmi ZA. MMAC: a mobility-adaptive, collision-free MAC protocol for wireless sensor networks. In Proceedings of 24th IEEE IPCCC 05, Phoenix, Arizona, USA, April Jhumka A. Kulkarni S. On the design of mobility-tolerant TDMA-based media access control (MAC) protocol for mobile sensor networks. LNCS Book Series: Distributed Computing and Internet Technology. Springer Berlin: Heidelberg, 2007, ISBN: Boukerche A, Oliveira H, Nakamura E, Loureiro A. Novel location-free greedy forward algorithm for wireless sensor networks. In ICC 2008, Wireless Networking Symposium, Meguerdichian S, Koushanfar F, Potkonjak M, Srivastava MB. Coverage problems in wireless ad-hoc sensor networks. In Proceedings of IEEE INFOCOM 2001, Vol. 3, April 2001; Boukerche A, Fei X. A coverage-preserving scheme for wireless sensor network with irregular sensing range. Ad Hoc Networks 2007; 5(8): Chai E, Chan MC, Ananda AL. Coverage aware buffer management and scheduling for wireless sensor networks. In IEEE SECON, Kotz D, Newport C, Gray RS, Liu J, Yuan Y, Elliott C. Experimental evaluation of wireless simulation assumptions. In Proceedings of the 7th ACM/IEEE International Symposium on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWiM 04), October 4 6, Venice, Italy, 2004;

36 ENERGY AWARE MAC PROTOCOLS FOR WIRELESS SENSOR NETWORKS 1607 Authors Biographies Bashir Yahya received a B.Sc. degree in Computer Engineering from Al-Fateh University (Tripoli, Libya) in 1994, and an M.Sc. degree in Computer Networks from INT (Institut National des Télécommunications, Evry, France) in Currently he is a member of the Network Architectures and Services (ASR) research group at PRiSM Laboratory and pursing his Ph.D. at the University of Versailles, Versailles, France. Mr. Yahya research interests include energy efficient medium access protocols for wireless sensor networks, quality of service in WSNs, cross layer optimization and design, mobility management. He is a member of IEEE and IEEE Communication Society. Marie Curie (Paris 6), in 1998 and 1999 respectively. He is now an Associate Professor at the University of Versailles since Ben-Othman research interests are in the area of wireless ad hoc and sensor networks, Broadband Wireless Networks, multi-services bandwidth management in WLAN (IEEE ), WMAN (IEEE ), WWAN (LTE), security in wireless networks in general and wireless sensor and ad hoc networks in particular. He has supervised and co-supervised several graduate students in these areas. His widely known for his wok on wireless ad hoc and sensor network, in particular security. Since 2002, he has served as Technical committee of more than 20 international IEEE/ACM conferences and workshops including ICC, Globecom, MSWIM, LCN, etc... He is a member of IEEE and ACM. He is active member of IEEE TC AHSN. He is IEEE/ACM member. Jalel Ben-Othman received his B.Sc. and M.Sc. degrees both in Computer Science from the University of Pierre et Marie Curie, (Paris VI) France in 1992, and 1994 respectively. He received his Ph.D. from the University of Versailles, France, in He was an Assistant Professor at the University of Orsay (Paris 11) and University of Pierre et