Research Paper FAULT MANAGEMENT USING MODIFIED RAT FOR WIRELESS SENSOR NETWORKS a b Sathyapriya.L Jawahar A Address for Correspondence b Professor, a Research Scholar, Department of ECE, SSN College of Engg., Kalavakkam India. ABSTRACT Wireless Sensor Networks are self-configured, infrastructureless wireless networks made of small devices with specialized sensors and wireless transceivers. This provides larger scope of research and study in this field. Faults in one or two nodes in the network should not affect the network s connectivity. So it is necessary to achieve a fault tolerant topology control protocol. Extending the lifetime of the wireless sensor networks remains the most challenging and demanding requirements that impede large-scale deployments of nodes. Putting the node s radio to sleep mode can save considerable energy. These two important issues are addressed by exploiting latency and communication redundancy. This is achieved by integrating sparse topology energy management (STEM) and Redundancy Aware Topology Management (RAT) in WSN. In STEM, energy efficiency is achieved at the cost of latency. In order to maintain connectivity, only few nodes are kept active by exploiting redundancy in RAT thereby managing the faults. We exploit the energy efficiency by introducing a parameter called sleeping ratio. Our objective is to improve metrics such as Energy efficiency, Latency, Sleeping ratio. We implement this proposed scheme using network simulator (NS 2). INDEX TERMS Sleeping ratio, communication redundancy, Life time. I. INTRODUCTION Recent advances in wireless communication technologies along with the developments in electronics have made available a new type of communication network made of battery-powered, low cost and multifunctional integrated wireless sensor devices [, 9]. The sensor network consists of several number of sensor nodes which is deployed near the phenomenon or the area surrounding it. These nodes have sensing, data processing units and communication components that make the network independent for sensing and sending information for further processing. Instead of sending raw data to the sink node, these sensor nodes locally perform some operations and send partially processed results. The field of wireless sensor networks continue to evolve and grow both in practical and research domains. Most of the researches in wireless sensor networks involve the design of energy efficient, computationally effective algorithm and Protocols [0]. Many wireless sensor networks are being used to gather information in real life applications. Sensor nodes are being applied in precision agriculture, environmental Monitoring, intelligent buildings, bridges, security mechanisms, military operations, transportation-related applications, etc. The future technology seems even more promising. More powerful wireless sensor devices will be available, and wireless sensor networks will have applicability in an endless number of application scenarios. They will be able to handle bursty traffic loads, make more computations, store more data, and live longer because of better management of energy sources. II. MOTIVATION WSNs are being used in real time applications to gather information. The sensor nodes can be used for continuous sensing, event detection, event ID, location sensing, and local control of actuators. The concept of micro-sensing and wireless connection of these nodes promises many new application areas [3,9].We categorize the applications into military, environment, health, home and other commercial areas. It is possible to expand this classification with more categories such as space exploration, chemical processing and disaster relief. This provides larger scope of research and study in this field. Faults in one or two nodes should not affect the networks connectivity. One of the constraint which affects all the challenges is energy [, 7]. The lifetime of the network depends on the energy available in individual sensor nodes. Therefore, in WSNs, the design of energy-efficient algorithms and protocols is of utmost importance. This motivated us to take up initiative to device energy efficient fault tolerant protocol. The rest of the paper is organized as follows. Related works are discussed in section 3, Problem Definition is discussed in section 4, and Section 5 deals with the proposed mrat scheme, in section 6, the performance analysis are discussed. Section 7 concludes the paper. III. RELATED WORKS The main aim of topology management schemes [6, 7, 8, and ] in wireless sensor network is to reduce the energy consumption and maintain connectivity to efficiently forward data to sink. In Sparse Topology and Energy Management scheme (STEM) [0, 5, 4], each and every node will have two radios namely wake up radio and data radio. The wake up messages is transmitted through wake up plane and data is transmitted through data plane. It is enough to keep the wake up plane in on and data plane in off state. If any event occurs, the data plane will come to on state. Here the problem is radio of next hop in the path towards sink will be in off state. To overcome this problem, each node will periodically turn on their radio for a short time to check if any other nodes want to communicate with it. In order to avoid the interference between wake up message and data message, each radio is operating at different frequency bands. Thus, energy is conserved due to efficient usage of dual radio. But the limitation of STEM is that it does not preserve network capacity and the latency is high. In Sustainable Physical Activity in Neighborhood scheme (SPAN) [, 4], only few nodes are be elected as coordinators and form a definite backbone path in the network through which data is forwarded from source to sink. The other nodes are called as noncoordinators. Here the coordinators are selected based on coordinator election algorithm. Coordinator election and withdrawal will vary for each and every particular instant of time. Here equal chance is given for all the nodes to become coordinator. Since there is a definite backbone path, the latency is less. SPAN preserves network capacity.
In Geographic Adaptive Fidelity scheme (GAF) [, 2, 5], the network is divided into several grids and in each and every grid; only one node will remain in on state. Thus if there are n grids in the network, then there will be only n nodes in on state in the network. Here energy is conserved. The main drawback of GAF is location information. Here GPS is used for location information that is very costly. Another drawback is that the nodes in the corner of the grid are not reachable. In Adaptive Self- Configuring Sensor network Topology scheme (ASCENT) [2], only few nodes will be in active state and other nodes in passive, test and sleep states. Active nodes participate in the data transmission. The nodes, which are in test state and passive state checks continuously, to become active, for successful transmission of data packets. The node that detects the loss of packets sends help message to neighboring nodes to join the network. The nodes in passive state receive these messages and check to become active node if needed. Hence data loss is reduced and successful transmission takes place. In Enhanced SPAN (E-SPAN) [3], directional antenna is used. In this method, data is sent in directional mode and hello message is sent in Omni mode. The directional antenna has greater gain than Omni directional antenna and hence the range of directional antenna beam is greater than Omni directional beam. The receiver will be in Omnidirectional mode and transmitter will be either in Omni or directional. Here only one antenna can be enabled at a time. By using directional antennas, the E-SPAN is more efficient than SPAN. STEM is orthogonal to the density domain topology management scheme GAF and additional gain is achieved by combining STEM and GAF. Thus integrating the STEM with GAF, [6,8] more energy is conserved. Also the combined scheme preserves capacity, reduces latency compared to STEM. The Lifetime improvement factor of combined scheme is more compared to GAF. IV. PROBLEM DEFINITION In Wireless Sensor Networks, the increased energy efficiency is the major design requirement. Maximum energy can be saved by putting the sensor node s radio to sleep mode. Lifetime improvement can be brought about by exploiting the fact that most of the time the node is in monitoring state and the radio can be turned off. We employ a two radio model [5] one radio of data transfer and another low power radio for exchanging control signal in the proposed mrat scheme. Fig. Radio Setup of Sensor Node [5] The wake-up messages are sent in the frequency plane f. This is referred to as wake-up plane. Once the initiator has notified the target node, the data packet is transmitted in the frequency band f2. This frequency plane is called data plane. At time t the node wants to wake up the neighboring node, thus it becomes an initiator. It sends beacon packet in f plane till it gets a response. At time t2 it gets a response from the neighbor node. After this the data plane f2 switches on and data transfer begins. After the data is sent the data plane f2 radio is switched off. At t4 some other initiator node tries to communicate to this node, so the data plane f2 radio is switched on again for data transfer Furthermore for increased energy saving maximum number of nodes should be put on to sleep state. The choice of switching the node on or putting it to sleep mode should be made keeping in mind the maximum connectivity of the network. Minimum numbers of active nodes are selected that are good enough to maintain connectivity [9]. To discuss the algorithm we have defined few parameters given below. These parameters give knowledge about the characteristics of the network. A. Set-up Latency Set-up latency (Ts) is defined as the interval from the time the initiator starts sending out beacons, to the time the target node has responded to the beacon. Typically the target and originator node are not synchronized, which means that the beacon sending process starts at a random point in the cycle of the target node. Fig 2. Latency Analysis B =Transmit time of the beacon packet in the x wakeup plane B = Inter Beacon spacing y B = Time taken for initiator node to send the beacon and Receive response B = B x + B y T b = ON Time duration of target node ( T b = ( B + B x )) T= Total time period Fig 2. Latency analysis shows the values of Ts, normalized versus B = B x + B y for different start times of the beacon sending process. This is the time it takes to send a beacon and receive the response to it. For the region that is labeled i in Figure 2, the setup latency is equal to (i. B ), since beacon i is the first one to fall entirely within the interval of length T, when the target node's radio is on. The probability of being in region i is equal to the length of that region divided by T [5] B. Sleeping ratio (τ) Sleeping ratio (τ) is the ratio of sleeping nodes to the total number of nodes [2]. No of Sleeping nodes Sleeping Ratio.. () Total no of nodes There is a tradeoff between the sleeping ratio and the connectivity of the network. When the sleeping ratio increases, the lifetime of the network increases as
more number of nodes are put on to sleep mode. But the connectivity might be affected. So the sleeping ratio should be selected such a way that the sleeping nodes do not affect the connectivity of the network. C. Redundancy measure (ρ) ρ(i,j) describes quantitatively how much communication redundancy node j can provide to node i. It describes how much responsibility can node j takes away from node i. No of Redundant nodes between i & j i, j ( No of nodes in neighbor set in i) (2) A node s responsibility can be viewed as the number of neighbor nodes it has. More the number of neighbor nodes more is the responsibility of the node. More number of nodes are required to provide the needed communication redundancy if the node wants to switch back to sleep mode. D. Replacement Nodes For a particular node, the replacement nodes are a set of nodes in neighbor set which can together cover the node s responsibility in the network. When the number of replacement nodes for a particular node is large, it can be said that the node s responsibility in the network is high. E. Example scenario The Figure 3 shows an example scenario with 0 nodes and connected by wireless links. The Table shows the calculation of redundancy measure and replacement nodes. The first column gives the node ID. The second column gives their corresponding neighbor nodes. The third column gives the redundancy measure ρ (i, j) describes quantitatively how much communication redundancy node j can provide to node i. It describes how much responsibility can node j takes away from node i. The fourth column gives the replacement nodes for maximum sleeping ratio of 80%. Fig 3. Communication Graph Table I. Calculation Redundancy Measure and Replacement Nodes Node Neighbor set Redundancy Measure ρ(i,j) Replacement Nodes 5, 2, 8, 4, ρ(5,8)=3/6 7, 3,6 8 5, 7, 4, 0, ρ(8,5)=3/5 5,9 6, 9 ρ(8,9)=2/5 9 8, 6, 0 ρ(9,8)=2/2 8 V. THE PROPOSED MRAT SCHEME Proposed mrat Scheme consists of two phases, the initialization phase and the node scheduling phase. In the initialization phase, each node becomes aware of its role (responsibility) in the multi-hop network, while in the scheduling phase, information from the initialization phase is used to put as many nodes as possible into sleep, while maintaining connectivity [2]. A. Initialization Phase The heart of the initialization phase is to obtain the neighbor sets. In proposed scheme, we exploit neighbor discovery procedure and the broadcast underlying wireless communication to aid each node obtain the shared neighbor sets with each one of its neighbors efficiently. Once the neighbor set and the shared neighbor sets are available, each node i can locally calculate the redundancy measure with each neighbor j and the replacement nodes for the required sleeping ratio. B. Scheduling Phase The proposed Scheme uses the number of replacement nodes to trigger scheduling threads. In a trial to minimize the total number of active nodes, nodes with higher number of replacement nodes switch to the active mode first and trigger a scheduling thread giving the opportunity for more nodes to collect enough neighbor nodes and switch to the sleep mode. The basic concept involved is that nodes with higher number of replacement nodes require more nodes to stay active and cover its neighbor set than a node with smaller number of replacement nodes. All nodes initiate start up timers inversely proportional to their number of replacement nodes. If the timer fires before any scheduling thread reaches that node, the node switch to active and triggers a new scheduling. On the other hand, if the node hears an active announcement before the timer fires, the node initiate timers proportional to their redundancy measure with the sender, so that nodes with least communication redundancy with the sender switch to the active mode first and so minimize the total number of active nodes in the network. All other nodes hold back their timers once they receive an active announcement. During the active announcements, any node that is able to collect enough replacement nodes becomes eligible for the sleep mode and switch to sleep mode. VI. PERFORMANCE METRICS The performance of the proposed scheme method are evaluated in terms of energy efficiency, sleeping ratio, setup latency and lifetime of the network. A. Energy efficiency The relative gain in the STEM scheme in terms of energy [5] ENode Enode Trx E Node, original T.. (3) T rx -time duration for which the f radio is switched ON T-the time interval between 2 successive ON period in Integrating STEM and RAT: For a combination of STEM and RAT, No of sleeping nodes Sleeping ratio Let, τ Total No of Nodes be the sleeping ratio S-no of sleeping nodes N-total no of nodes S. N (4) No. of ON nodes = = N ( N )... (5) For N nodes (-τ) nodes are ON. So for every / (-τ) nodes, node is switched ON an average. f
So an equivalent parameter, m /... (6) So combining STEM and RAT, E T / T (7) node This Equation gives the energy saved for each node on an average combining STEM and RAT. The design of energy-efficient protocols is of utmost importance. The lifetime of the network depends on the energy available in individual sensor nodes. Therefore, in WSNs, the design of energy-efficient algorithms and protocols is of utmost importance. We define the normalized energy as E/E0, where E is the energy consumed by the network employing different topology management schemes such as STEM, RAT and mrat and E0 is the energy 0.5 0.45 0.4 rx Energy Efficiency STEM mrat RAT consumed as the number of nodes is attributed to the neighbor discovery process that occurs in the network. C. Lifetime By comparing Equation (3) and (7), it can be inferred that the lifetime of the sensor node increases by the factor of m. m. (8) ( ) D. Sleeping Ratio (τ) Sleeping ratio (τ) is the ratio of sleeping nodes to the total number of nodes. There is a tradeoff between the sleeping ratio and the connectivity of the network. When the sleeping ratio increases, the lifetime of the network increases. 0.8 0.75 0.7 0.35 E/Eo 0.3 0.25 0.2 0.5 Sleeping ratio 0.65 0.6 0.55 0.5 0. 0.05 40 50 60 70 80 90 00 Number of nodes Fig. 4 E/E0 Comparison of STEM, RAT and Proposed mrat Scheme Figure 4 compares the performance of STEM, RAT and proposed scheme mrat, based on simulations. First of all, we observe that the energy savings of RAT are moderate, except for high network densities. The reason is that the energy savings in RAT increases as the redundant nodes increases. STEM, on the other hand, is independent of the network density. The proposed mrat scheme thus performs well even in a network with less redundant nodes as compared to STEM and RAT alone. B. Comparing energy consumed The energy consumption of the three schemes STEM, RAT, mrat are compared in this graph. Same three packets are sent in all the three protocols. The plot is between the energy consumed and the total number of nodes in the network. 0.45 40 50 60 70 80 90 00 Number of Nodes Fig. 6 Sleeping Ratio of Proposed mrat scheme From the Figure 6 it is shown that, as the node degree increases the number of redundant nodes increases. Therefore the number of nodes that can be put to sleep also increases. So the sleeping ratio of the network increases. Increased sleeping ratio can bring about increased lifetime of the network. E. Setup Latency Setup latency is defined as the interval from the time the initiator starts sending out beacons, to the time the target node has responded to the beacon. Latency in ms 4 2 0 8 6 Latency STEM mrat 4 Fig. 5 Energy consumed Comparison Figure 5 shows energy consumed in the m-rat as compared with the individual schemes STEM, RAT is reduced. The linear increase in the energy 2 40 50 60 70 80 90 00 Number of nodes Fig. 7 Latency Comparison of STEM and Proposed mrat scheme Figure 7 shows the latency introduced while sending the data. As the same packets are sent, the setup latency remains almost constant for both STEM and mrat. As it shows the setup latency reduces for mrat while comparing to STEM due to the existence of active nodes that are enough to provide sufficient connectivity. So definite path between two sensor nodes exists. So the setup latency also decreases. VII. CONCLUSION The proposed mrat scheme was implemented
exploiting latency and communication redundancy resulting in a more energy efficient and fault management topology. The two radio model in STEM and the redundancy aware concept in RAT was exploited. The simulation results were discussed. At the end of the discussion it can be concluded that the performance of the network in terms of energy efficiency, lifetime of the network, was improved. The future work involved in employing the proposed scheme mrat is three fold. First would be extending the scheme in mobile environment. Second would be employing the scheme in heterogeneous network. Thirdly the security issues in the scheme can be analyzed. REFERENCES [] Akyildiz I.F, W. Su, Y. Sankarasubramaniam and E. Cayirci, "A Survey on Sensor Networks," IEEE Communications Magazine, vol. 40, pp. 02-4, 2002. [2] Alberto Cerpa, Deborah Estrin, ASCENT: Adaptive Self-Configuring Sensor Networks Topologies. IEEE transactions on mobile computing,vol. 3, pp. 272-285,2004. [3] Th. Arampatzis, J. Lygeros and S. Manesis, A Survey of Applications of Wireless Sensors Proceedings of the 3 th Mediterranean Conference on Control and Automation,Limassol, Cyprus, June 27-29, 2005. [4] Chen Benjie, Jamieson Kyle, Balakrishnan H. and Morris Robert, SPAN: An Energy-Efficient Coordination Algorithm for Topology Maintenance in Ad Hoc Wireless Networks, ACM Wireless Networks Journa,Volume 8, Number 5, September, 2002. [5] Curt Schurgers, Vlasios Tsiatsis, Mani B. Srivastava, STEM: Topology Management for Energy Efficient Sensor Networks IEEEAC paper #260, Updated Sept 24, 200. [6] Heinzelman, W., Chandrakasan, A., Balakrishnan, H, Energy-efficient communication protocol for wireless microsensor networks In: Proceedings of the 33rd International Conference on System Sciences (HICSS), pp. 0,(2000) [7] Jawahar, A., Radha, S., and Sharth Kumar, S., Capacity- Preserved, Energy-Enhanced Hybrid Topology Management Scheme in Wireless Sensor Networks for Hazardous Applications, in International Journal of Distributed Sensor Networks, Volume 202. [8] Intanagonwiwat J C, Govindan R, Estrin D, Directed diffusion: A Scalable and Robust Communication Paradigm for Sensor Networks Proc. ACM Mobicom 00. Boston. MA. Pp. 56-67,2000. [9] Miguel A. Labrador Pedro M. Wightman, Topology Control In Wireless Sensor Networks -ISBN 978--4020-9585-6 Springer Science + Business Media B.V. 2009. [0] Mo Li, Baijian Yang, A Survey on Topology issues in Wireless Sensor Network, SPIE The International Society for Optical Engineering, Orlando, FL, pp. 229-237, April 999. [] Salfan Al-Omari and Weisong Shi, Wayne State University. Redundancy Aware Topology Management in Wireless Sensor Network, IEEE transactions on mobile computing, (2005). [2] Savvides.A, C.-C. Han, M. Srivastava, Dynamic finegrained localization in ad-hoc networks of sensors, MobiCom 200, Rome, Italy, pp. 66 79, July 200. [3] Mehdi Saeidmanesh,Mojtaha hajimoham madi and Ali Movaghar, Energy and Distance Based Clustering : An Energy Efficient Clustering Method for WSNs, World Acadamy of science, Engineering and Technology 55 2009. [4] Li N and Hou J, Topology control in heterogeneous wireless networks: Problems and solutions. In proc of IEEE Conference on Computer Communications (INFOCOM 04), March 2004. [5] Li Li Joseph Y. Helpern, Paramvir Bahl, Yi-Min Wang and Roger Wattenhofer,A cone-based distributed topology-control algorithm for wireless multi-hop networks. IEEE/ACM Trans. Netw., 3():47-59, 2005.