Research Of Coverage Problem Based On Probability Sensing Model In WMSNs

Wireless sensor network is an important means to perceive the physical world.The traditional sensor network can no longer satisfy people's needs.So the multimedia sensor network arises at this moment.Compared with the traditional sensor network,multimedia sensor network can obtain image,video,audio and other multimedia information.Effective coverage is the important guarantee of the sensor performance,and the traditional sensor network model is no longer suitable for multimedia sensor network.Therefore,it's necessary to combine the characteristics of multimedia sensor networks with key technologies of coverage to do new research.First of all,this thesis introduces the concept of multimedia sensor network,as well as its key technology and application.And this thesis carries on an intensive analysis about research status of multimedia sensor network coverage problem.Then,aiming at the existing multimedia sensor network sensing model,the multimedia sensor network perception model by introducing the focal length factor,resolution factor and combining with Raleigh criterion is established.This model has taken the actual resolving ability of camera into consideration.It helps to further promote the practical application of multimedia sensor networks.Combining multimedia sensor network deployment and coverage theory,this thesis carries out research in the coverage hole of network.The coverage hole detection methods based on Levy Flight and intelligent hole repair strategy based on TSP methods are proposed.In addition,this thesis also adopts the simulation experiment to prove the effectiveness of this method.Experimental results show that the method can quickly and effectively repair coverage holes.In order to achieve the optimized coverage effect,this thesis introduces the mechanism of elite into cuckoo algorithm to enhance its network coverage,optimizing the camera rotation angle by introduce a multidimensional optimization strategy,and further introducing the learning feedback strategy to modify step length factor to improve the network performance.Simulation experiments prove the algorithm's performance is better than the simulated annealing algorithm.Taking into account the stability of the algorithm,this thesis also studies factors which affect network performance as well.Finally,this thesis carries out the research of the multimedia sensor network nodes based on embedded systems.Firstly,the frame differential method and the particle filter tracking algorithm are introduced;secondly,the embedded system platform is constructed,including the construction of hardware platform and software platform;finally,the target detection and tracking experiment which based on ARM embedded system is carried on.Together this thesis carries on the comparative experiment with Kalman filter algorithm.The results indicate that the frame differential method and the particle filter algorithm can effectively detect and track the target,and the performance is better than Kalman filtering algorithm.