An Incremental Sensor Deployment Strategy For Wireless Sensor Networks Based On Bayesian Estimation

Wireless sensor networks (WSNs), which comprise a large number of tiny sensor nodes, are formed through wireless communication technologies. This kind of networks is mainly used to sense, collect and process interested information objects to accomplish hign-level application tasks. As more and more research is being done and more applications are being applied to our life, WSN will gradually be used in all fields of our lives.Wireless sensor networks, which refer to many subjects, have become a hot research area. But there are still lots of key technologies which neet to be found and discussed. Sensor node deployment is one of the fundamental issues in wireless sensor networks, which affects the performance and effectiveness of the networks. As a kind of sensor deployment, sensor redeployment is to deploy additional sensor nodes in a low energy level of a wireless sensor network, which is an efficient way to prolong the lifetime of the networks.Recently, sensor deployment is increasingly drawing more and more attention and has proposed some usefull algorithms. Howerer, a majority of algorithms fail to take event occurrence probability into account for sensor redeployment, a key factor which affects the performance of sensor deployment, or just make some assumptions. In consideration of the randomicity of events and experiential factors, this article formulated the event occurrence probability distribution of the redeployment problem in WSNs as Binomial Distribution rather than Uniform Distribution and use Bayesian Estimation to estimate the probability of each node's sensing area.In consideration of bayies estimation and incremental sensor deployment strategy, we proposed a novel algorithm called SD-B (Sensor Deployment using Bayies Estimation) to solve the redeployment problem under static routing schemes. All the events which are generated before the total network energy reaches to a low level are regarded as sample information and then estimate each sensing area's event probability. As a result, these estimated probability values are used in SD-B algorithm.Finally, the proposed algorithm was tested with a commonly-used tool GCC compiler on Redhat operating system. We compare SD-B with SD-U and SD-N in terms of network lifetime and successful hit numbers. Experimental results demonstrated that our proposed algorithm greatly outperformed the one using uniform distribution algorithm (SD-U) and SD-N algorithm which doesn't take any sensor deployment measures, in the percentage of successfully predicting the earliest energy drain out node and the networks lifetime had been extended about thirty percentages.