Distributed Fast Coherence Filtering Of Wireless Sensor Networks

Wireless sensor network is a multi-hop and self-organized network composed of a large amount of wireless sensor nodes with sensing, communication and computational capability. For owning the advantages of low cost, remote control, high precision, fault tolerance, easy diagnosis and maintenance, it has been used broadly in the fields of military, environmental science, health, space exploration, smart home and intelligent transportation systems. As a novel distributed information fusion algorithm in wireless sensor networks (WSNs), Kalman-consensus filtering do not need fusion center, the sensor nodes only communicate with their neighbors, and all the nodes can obtain approximate and high accuracy estimation value. However, in many practical problems with real-time requirements, convergence speed is a vital performance index to judge the quality of a protocol. Consequently, considering the computational ability and energy are limited, an important goal for wireless sensor network is how to design an effective distributed consensus filtering algorithm to realize the fast fusion.Several types of distributed consensus filtering algorithms are proposed from convergence speed in this paper to deal with real-time requirements. The main research content structure is as follows:1) A new type of fast distributed Kalman-consensus filtering algorithm with double gains regulation is proposed to deal with filtering problems in wireless sensor networks (WSNs). By using the Lyapunov’s second method, this fast filtering problem can be changed into the stability issue of the dynamic estimation errors. Moreover, a sufficient condition for the convergence rate of this algorithm is presented. Finally, a simulation example is given to show the effectiveness of the algorithm.2) A fast distributed Kalman-consensus filtering algorithm based on incremental PID is discussed. Considering the feedback control is similar with conventional proportional controller, an incremental PID algorithm is introduced to replace the feedback information for making full use of current and past limited information to improve the convergence speed. Moreover, a genetic algorithm is added to obtain the fastest consensus convergence rate through optimizing PID parameters.3) A fast distributed cluster-based Kalman-consensus filtering algorithm with local information feedback is investigated. In order to reduce the amount of data transmission, the network is divided into several different clusters. When a cluster node fuse its neighbors’information in a consensus manner, a feedback mechanism between a cluster head and its neighbors is introduced to obtain more useful information so as to achieve faster convergence rate. Finally, a simulation example is given to show the effectiveness of the algorithm.