Location-Unaware Node Scheduling Scheme In Wireless Sensor Networks

In Wireless sensor networks (WSNs), network coverage rate reflects the ability tomonitor the physical world, and often as an important technology indicator of thequality of service (QoS). Because of the limited energy and not easy to maintain,sensors are usually deployed densely to obtain high coverage and prolog the lifetimeof WSNs. On the other hand, the higher density of sensors will cause severe problemssuch as redundancy, radio channel contention, network congestion and so on, whichconsume more energy and shorten survival time of WSNs. So node schedulingscheme with turning off some redundant sensor alternatively are considered as aneffective measure to reduce energy consumption and prolong the network lifetime.Most of the exiting node-scheduling schemes depend on the accurate locationinformation of the nodes. In fact, since sensors are numerous and usually work incomplex environment, it is not economically feasible to accurately compute thelocation of each sensor. So in this dissertation location-unaware node schedulingschemes in wireless sensor networks are focused. We mainly studied the redundantnodes judgment without location information, the energy consume-balanced duringnode-scheduling, the node-scheduling and the fault-tolerant when event occurrenceand the event region identification algorithm without accurate location information.Our works and contributions are summarized as follows:(1) A location-unaware boundary nodes identification method and a judgmentalgorithm of redundant boundary nodes are proposed. Inequality sleep problem ofboundary nodes will lead to boundary contraction. The neighbor number of a node isless than the mathematical expectation value of a boundary node’s neighbor numberor not is used to determine whether the node is in the network boundary. And theeffective redundant area to the effective sensing area ratio can be used to judgewhether a boundary node is redundant. This judgment can be combined with variousnode-scheduling schemes. The simulation results showed that this algorithm can notonly improve the network coverage, but also prolong the network lifetime by11%,and still save network energy by7%. The boundary contraction problem can also beeffectively alleviated.(2) An optimal sleep scheduling scheme based on balanced energy consumption(ECBS) for uniform distribution is proposed. Because of the inequality sleepproblems in the process of scheduling that means energy consumption unbalanced, it leads to some number of nodes premature death that formed “coverage hole”, andthen affect the quality of coverage and communication. So the network diesprematurely. ECBS algorithm needs sensor-to sensor distance but no locationinformation. The neighbor number of a redundant node can be calculated accordingthe distance information between nodes. And the half-hop neighbor node that has lessresidual energy will be selected to sleep to balance energy consumption in network.The simulation results showed that using ECBS algorithm to schedule makes workingsensors distribute more uniform and less redundant coverage. The energy variance isnot more than0.0001and there is approximately17%residual energy when thenetwork died. It can extend the network lifetime effectively.(3) A non-uniform nodes distribution strategy based on density and an energybalanced non-uniform distribution node scheduling scheme (EBNDNS) for wirelesssensor networks are proposed. In large-scale wireless sensor networks, multi-hopcommunication is often adopted. But “energy hole” around Sink is inevitable inuniform nodes distribution, and the network dies prematurely while a large number ofnodes are still alive. To solve this problem, we put forward a feasible non-uniformdistribution strategy. The monitoring region can be divided into several sub-regions.According to a ballpark estimate of energy consumption in different sub-regions, thenode density of different sub-regions can be computed. The simulation results showthat this distribution strategy can greatly relieve the “energy hole” problem inmulti-hop communication, prolong the network lifetime more than1.68times, andimprove the coverage quality at the same time. Because the existing node-schedulingschemes are mostly based on uniform distribution, EBNDNS algorithm usessensor-to-sensor distance but no the location information to compute the redundantcoverage probability produced by multiple neighbors. If the probability meets thecoverage QoS demand of network, this node can be turned off. In order to balance theenergy consumption, the pre-sleep time is related to the residual energy. Oursimulation results showed that this algorithm is not only suitable for non-uniformdistribution, but also for uniform distribution, the energy banlaced effect is even betterthan ECBS.(4) A fault-tolerant node scheduling scheme for event monitoring in wirelesssensor networks and an event region identification method without accurate locationinformation are proposed. For the data sensed by a single node is highly uncertainty,we used space correlated characteristic of sensing data when the events occurs. The sensor that first sensed abnormally need to wake some neighbor nodes up and collectdata from neighbors, and then the joint probability of the event occurs can becalculated according to Bayesian theory. We designed the node-scheduling algorithmand routing algorithm with different value to achieve the purpose of fault tolerance.Our simulation results showed that this algorithm has higher event identification rateeven when sensor fault probability is as high as20%. And this algorithm canself-adapt the event spreading or narrowing. In event monitoring application, peoplehope to estimate the event region correctly, but the existing studies mostly rely on theaccurate location information. We only using some anchor nodes location but not eachnode location, the event polygon region can be estimated roughly according to thevoronoi polygon of anchor nodes and the joint probability that is calculated by theactive nodes located in event region. Simulation results show that more than90%areaof the event is included in the scope of our estimation.