Research Of Sequencial Monte-Carlo Based Localization Algorithm For Mobile Sensor Networks

In mobile wireless sensor networks(MSN), sensors can move randomly or keep static temporarily. Location of sensor nodes will change at any time. Mobility makes the sensor networks better acquire information, thus it will expand the scope of MSN application.But it also makes accurate localization more difficult since the network environment changes continually and the sensor nodes only have limited hardware capability. We focus on the research of Sequential Monte Carlo Localization scheme for Mobile Sensor Networks, the main work and conclusions are as follows:(1) Study three localization algorithms based on Sequential Monte-Carlo: Range-free SMCL,MCB and Range-based SMCL, summing up all the strengths and shortcomings. Reproduce them in NS-2.(2) The energy consumption of localization and the memory size restrictions of sensor nodes limit the use of these methods. To overcome these defects, we propose an energy-efficient dynamic sampling and filtering based localization algorithm (DSL). DSL uses range measurement information to design more restricted sample region, improving the effectiveness of sample collection.(3) Design a speed prediction mechanism. Combined with the connectivity constraints, establishes a dynamic filtering mechanism to improve the localization performance and efficiency.(4) Analytical and simulation results are provided to study the localization cost and location accuracy in different mobility models and various environmental settings. The results indicate that our algorithm outperforms the best known simulation based localization schemes under a wide range of conditions, with localization accuracy improved by an average of 24% and computation cost reduced significantly for similar high localization accuracy. Moreover, DSL is able to adapt to multiple motion models including linear and nonlinear models.