Wireless Ad Hoc and Sensor Networks (Energy Management) Outline Energy Management Issue in ad hoc networks WS 2009/2010 Main Reasons for Energy Management in ad hoc networks Classification of Energy Management Summary Prof. Dr. Dieter Hogrefe / Prof. Dr. Xiaoming Fu Dr. Omar Alfandi 2 Energy Management Issue in Ad Hoc Networks In ad hoc networks the devices are battery powered So the computation and communication capacity of each device is constrained Devices that expend their whole energy can be recharged when they leave the network Energy resources and computation workloads have different distributions ib ti within the network Therefore it is beneficial to redistribute spare energy resources to satisfy varying workloads in the network Outline Energy Management Issue in ad hoc networks Main Reasons for Energy Management in ad hoc networks Classification of Energy Management Summary 3 4
Main Reasons for Energy Management in Ad Hoc Networks (1/2) Main Reasons for Energy Management in Ad Hoc Networks (2/2) Limited Energy Reserve: The improvement in battery technologies is very slow Difficulties in replacing the batteries: E.g. in battlefields or emergency applications Lack of central coordination: In ad hoc networks as distributed networks some nodes may work as relay nodes; when relay traffic is heavy the power consumption is high Constraints on the battery source: The batteries should be small and not heavy; So low power is available at each node Selection of optimal transmission power: Higher transmission power results in higher energy consumption and higher interference between nodes 5 6 In General Energy management deals with the process of managing energy resources by means of : Controlling the battery discharge Adjusting the transmission i power Scheduling of power sources Outline Energy Management Issue in Ad Hoc Networks Main Reasons for Energy Management in ad hoc networks Classification of Energy Management To increase the lifetime of the nodes of an ad hoc wireless network Summary 7 8
Classification of Energy Management (1/2) Energy Management Classification of Energy Management (2/2) Battery management : Concerned with problems that lie in the selection of battery technologies, finding the optimal capacity of the battery Battery management schemes Device dependent schemes: e.g. battery scheduling Transmission power Data link layer: e.g. Dynamic power adjustment System power Processor power management schemes: e.g. Power saving modes Transmission power management: Attempt to find an optimum power level for the nodes in the ad hoc wireless network Data link layer: e.g. Lazy packet scheduling Network layer: e.g. Routing based on battery status Network layer: Power aware routings Higher layers: Congestion control, Transmission policies at TCP/IP Device management schems: e.g. Low power design of hardware System power management: Deals mainly with minimizing the power required by hardware peripherals of a node (such as CPU, DRAM and LCD display) 9 10 Classification of Energy Management Battery management schemes Device dependent schemes: e.g. battery scheduling Data link layer: e.g. Lazy packet scheduling Network layer: e.g. Routing based on battery status Energy Management Transmission power Data link layer: e.g. Dynamic power adjustment Network layer: Power aware routings Higher layers: Congestion control, Transmission policies at TCP/IP System power Processor power management schemes: e.g. Power saving modes Device management schems: e.g. Low power design of hardware 1. Battery Management The lifetime of a node is determined by the capacity of its energy source and the energy required by the node. There are some device dependent d approaches that t increase the battery lifetime by exploiting its internal characteristics. Key Fact: Batteries recover their charge when idle Use some batteries and leave others to idle/recover Device depending schemes: I. Battery scheduling techniques: In a battery package of L cells, a subset of batteries can be scheduled for transmitting a given packet leaving other cells to recover their charge. There are some approaches to select the subset of cells, e.g.: 11 12
Battery Management 1. Joint technique: The same amount of current is drawn equally from all the cells which are connected in parallel. 2. Round robin technique: The current is drawn from the batteries in turn by switching from one to the next one. 3. Random technique: any one of the cells is chosen at random with a uniform probability. Data link Layer Battery Management I. Lazy Packet Scheduling: Reduce the power Increase the transmission time (lower bit rate) But this may not suit practical wireless environment packets a transmission schedule is designed taking into account the delay constraints of the packets Battery Management II. Battery-Aware MAC Protocol: to provide uniform discharge of the batteries of the nodes that contend for the common channel Lower back off interval for nodes with higher charge Network Layer Battery Management Goal: Increase the lifetime of the network I. Shaping algorithm: introducing delay slots in the battery discharge process If battery charge becomes below threshold, stop next transmission allowing battery to recover through idling The remaining requests arriving at the system are queued up at a buffer As soon as the battery recovers its charge and enters state higher than the threshold, it starts servicing the queued-up requests 13 14 Classification of Energy Management Battery management schemes Energy Management Transmission power System power 2. TRANSMISSION POWER MANAGEMENT SCHEMES Link layer solutions: I. Power Save in IEEE 802.11 Ad Hoc Mode Time is divided into beacon intervals Each beacon interval begins with an ATIM (ad hoc traffic indication message) window Device dependent schemes: e.g. battery scheduling Data link layer: e.g. Dynamic power adjustment Processor power management schemes: e.g. Power saving modes Data link layer: e.g. Lazy packet scheduling Network layer: Power aware routings Device management schems: e.g. Low power design of hardware Network layer: e.g. Routing based on battery status Higher layers: Congestion control, Transmission policies at TCP/IP 15 16
If host A has a packet to transmit to B, A must send an ATIM Request to B during an ATIM Window On receipt of ATIM Request from A, B will send an ATIM Ack, and stay up during the rest of the beacon interval If a host does not receive an ATIM Request during an ATIM window, and has no pending packets to transmit, it may sleep during rest of the beacon interval Size of ATIM window and beacon interval affects performance: If ATIM window is too large, energy saving is reduced d and may not have enough time to transmit buffered data If ATIM window is too small, not enough time to send ATIM request 17 E.g. A has some data packets to send to B; C is idle 18 II. Power Control in IEEE 802.11 Ad Hoc Mode A power control MAC protocol allows nodes to vary transmit power level on a per-packet packet basis When C transmits to D at a high power level, B cannot receive A s transmission due to interference from C If C reduces transmit power, it can still communicate to D Reduces energy consumption at node C Allows B to receive A s transmission (spatial reuse) But difference in transmit power can lead to increased collisions In following example suppose nodes A and B use lower power level than nodes C and D When A is transmitting to B, C and D may not sense the transmission When C and D transmit to each other using higher h power, their transmission may collide with the on-going transmission from A to B 19 20
As a solution to this problem, RTS-CTS are transmitted at the highest possible power level but DATA and ACK at tthe minimum i power level lnecessary to communicate In figure nodes A and B send RTS and CTS respectively with highest power level such that node C receives the CTS and defers its transmission By using a lower power level for DATA and ACK packets, nodes can save energy 21 In the previous scheme, RTS-CTS handshake is used to decide the transmission power for subsequent DATA and ACK which h can be achieved in two different ways 1. Suppose node A wants to send a packet to node B. Node A transmit RTS at power level p max (maximum possible). When B receives the RTS from A with signal level p r, B calculates the minimum necessary transmission power level, p desired. For the DATA packet based on received power level, p r, transmitted power level, p max, and noise level at the receiver B. Node B specifies p desired in its CTS to node A. After receiving CTS, node A sends DATA using power level p desired. 22 2. When a destination node receives an RTS, it responds by sending a CTS (at power level p max ). When source node receives CTS, it calculates p desired based on received power level, p r, and transmitted power level (p max ) as P desired = (p max /p r ) x Rx thresh x c where Rx thresh is minimum necessary received signal strength and c is a constant III. Centralized Topology Control: The power of each node is reduced until it has single connectivity, i.e. there is one path between each pair of nodes or bi-connectivity it 23 Network layer solutions: I. Common Power Control: Given reachability of each node as a function of power, find the min power level that provides network connectivity If the common power level is selected too high, this may lead to interference (fig. a); if too low, the reachability of nodes may become weak (fig. b) choosing an optimum value is a difficult task. 24
II. Min Variance in Node Power Levels: The motivation is to ensure that all the nodes are given equal importance and no node is drained at a faster rate compared to other nodes in the network For transmitting a packet, a node selects the next-hop node so that it has the least amount of traffic among all neighbours of the node III. Min Battery Cost Routing: Minimize sum of battery cost along a path Does not ensure that lower charge nodes are not used» The lower path is used IV. Conditional Min-Max Battery Cost Routing: Using only nodes that have battery charge over a threshold, Find the min total power path. V. Minimum Energy Disjoint Path Routing: The important need for having disjoint paths in ad hoc networks is because of the need for reliable packet transmission i and energy efficiency. Ad hoc networks are highly unreliable due to the mobility of nodes and hence the probability bili of link failure is quite high h in such networks. This problem can be overcome by means of link disjoint routing. Also, since the ad hoc nodes have stringent battery constraints, node disjoint routing considerably increases the lifetime of the network by choosing different routes for transmitting packets at different points of time. 25 26 Classification of Energy Management 3. SYSTEM Power Management Battery management schemes Device dependent schemes: e.g. battery scheduling Data link layer: e.g. Lazy packet scheduling Energy Management Transmission power Data link layer: e.g. Dynamic power adjustment Network layer: Power aware routings System power Processor power management schemes: e.g. Power saving modes Device management schems: e.g. Low power design of hardware Efficient design of the hardware brings about significant reduction in the power consumed. This can be effected by operating some of the peripheral devices in power-saving mode by turning them off under idle conditions. System power consists of the power used by all hardware units of the node. This power can be conserved significantly by applying the following schemes: Processor power Device power Network layer: e.g. Routing based on battery status Higher layers: Congestion control, Transmission policies at TCP/IP 27 28
SYSTEM Power Management Processor Power Management : Deals with techniques that try to reduce the power consumed by the processor, e.g. reducing the number of calculations performed. I. Power-Saving Modes: The nodes consume a substantial amount of power even when they are in an idle state since they keep listening to the channel, awaiting request packets from the neighbours. As a solution, the nodes are switched off during idle conditions and switched on only when there is an arrival of a request packet. Since the arrival of request packets is not known a priori, it becomes difficult to calculate the time duration for which the node has to be switched off. One solution to this problem calculates the node's switch-off time based on the quality of service (QoS) requirements. Hard QoS requirements make the node stay active most of the time. SYSTEM Power Management As soon as the node enters the idle state, it is switched off by the remote activated switch RAS. The receiver of the RAS switch still listens to the channel. The remote neighbours send the wake-up signal and a sequence. The receiver, on receiving the wake-up signal, detects the sequence (the waking-up signal). The logic circuit compares it with the standard sequence for the node. It switches on the node only if both the sequences match. 29 30 SYSTEM Power Management II. Power-Aware Multi-Access Signaling (PAMAS): Power-Aware Multi-Access Signaling is another approach for determining the time duration for which the node should be turned off. Conditions under which the node enters the power-off mode: Condition 1: The node has no packets for transmission. Condition 2: A neighbour node is transmitting or receiving packets, that is, the channel is busy. SYSTEM Power Management Device Power Management : Some of the major consumers of power in ad hoc wireless networks are the hardware devices present in the nodes. Device power minimize the power consumption. 31 32
SYSTEM Power Management I. Low-Power Design of Hardware: varying clock speed CPUs, disk spin down, and flash memory II. CPU Power Consumption: by changing the clock frequency, etc. III. Power-Aware CPU Scheduling: a small reduction in the value of the voltage produces a quadratic in the power consumed, so clock rate has to be also reduced. IV. Hard Disk Drive (HDD) Power Consumption: by bring down the speed of spinning on the disc drives Outline Energy Management Issue in Ad Hoc Networks Main Reasons for Energy Management in ad hoc networks Classification of Energy Management Summary 33 34 Summary Three major divisions in energy management Battery Management: When idling increases the capacity of the battery Transmission Power Management: Distance vs. Power tradeoff System Power Management: Put system/components to sleep whenever possible 35