Routing Algorithm Research For Sensor Networks Based On Self-adaptation

Sensor network, which consists of the convergence of sensor, micro-electro-mechanism system and network technologies, has a wide application prospect in national defense and environmental monitoring and so on. Yet the sheer numbers of these sensors and the expected dynamics in these environments present unique challenges in the design of unattended autonomous sensor networks. Therefore, self-adptive routing algorithm is studied in this paper.Considering the limited energy and transmission range of sensor nodes in sensor networks, an innovative routing algorithm named self-adaptive Flooding based on delay is proposed. In this algorithm, Sources deliver Route Request Packets (RREQ) and Sink answers Route Reply Packets (RREP) to build routes. In order not to miss optimum routes with the best metrics, the algorithm is to wait a specific amount of time to ensure that almost the RREQ near the best metrics is received. And the waiting time is decided self-adaptively. Then nodes will tranmit the RREQ with the best metrics. After the packet arrives at Sink, Sink sends RREP to let other nodes know the routes. Finally data packets are transmitted through them. Theoretic analysis and simulation results prove it's more efficient and much better than Flooding and it better sloves the problem of redundancy of packets.For the sake of prolonging the lifespan of nodes and improving the performance of stable and routing, a self-adaptive routing algorithm based on Analytic Hierarchy Process(AHP) is proposed.The new algorithm induces AHP to cluster self-adaptively, and then it makes virtual backbone network by clusters and sinks, after that it creates routes in the virtual backbone network. Also theoretic analysis and simulation results prove it is more effective than LEACH.On the basis of the above research, a self-adaptive clustering routing algorithm sensitive to position is proposed. In this algorithm it selects cluster-heads by considering the working order synthetically instead of periodically. And cluster is stable for a while until it obtains certain conditions whose cluster-head changes in-cluster. Furthermore data is forwarded via routes build among virtual backbone networks instead of being sent directly to sinks which may be faraway. Therefore it supports lower-class sensors and it is distributed, low complexity and can balance the energy expenditure which will render the total energy expenditure lower and increase the throughput. Also theoretic analysis and simulation results prove it is effective and scalable.