Research On Data Compression Algorithm For Wireless Sensor Networks

With the development of computer and correlative technologies, it is becoming realizable to integrate functions of sensing, communicating and computing into a tiny modular device. So the development and extensive applications of wireless sensor networks have been accelerated greatly. People can obtain information about objective world through wireless sensor networks rather than site investigation, which expands the function of existing networks and improves people's ability of recognizing the world. However, some problems are still needed to research and solve because of the specificities of wireless sensor networks.For the limits of computing ability, storing capacity, communication band, battery capacity and the existence of data redundancy, it is indispensable to design suitable data compression algorithm to adapt above mentioned characters of wireless sensor networks. The data redundancy not only exists in data obtained by the same node in sequent instants but also exists in data obtained by adjacent nodes in the same instant. If nodes directly send data which includes redundancy, the limited communication band and energy won't be used efficiently and the delay of system will be prolonged and the ability of the whole system will be affected seriously. For eliminating data redundancy of wireless sensor networks, three new kinds of algorithms have been proposed in this paper. These algorithms are summarized as following.1. For reducing the complexity of computing in nodes and the average energy cost of nodes, the algorithm of data compression for wireless sensor networks base on optimal order estimation and distributed coding has been proposed. The algorithm can compute correlation coefficients in sink according to temporal-spatial correlation of data and optimal order estimation. Nodes only need to do modular arithmetic operation for coding obtained data. And the sink can restore data according to correlation coefficients. So the redundancy of data can be reduced and the average energy cost of nodes can be decreased correspondingly.2. Respect to the temporal-spatial redundancy which exist in data obtained by single node in sequent instants and adjacent nodes in the same instant respectively, the algorithm of data compression for sensor networks base on suboptimal clustering and virtual landmark routing within cluster has been proposed. Time redundancy can be eliminated firstly in single node according to the time correlation. And then virtual landmark of nodes in cluster can be built base on cluster head and routing in cluster can be realized correspondingly. A global structure tree will be established base on cluster heads. In the courses of data transmitted from nodes to cluster head and from cluster head to sink, space redundancy exists in data can be eliminated.3. Regard to the character of dynamic change of obtained data and actual working nodes of networks, data compression algorithm for wireless sensor networks based on Huffman coding and stochastic optimizing policy has been proposed. For eliminating temporal-spatial data redundancy, coding threshold value can be regulated dynamically through stochastic optimizing policy. Nodes only need to transmit part of original data and Huffman code. And sink can restore original data according to Huffman code. So the algorithm can decrease average energy cost of node and extend the longevity of networks through reducing the amount of data which need to transmit.Some researches have been done from different views for reducing temporal-spatial redundancy of data and system delay and improving the efficiency of energy consuming of nodes. The aim of getting information of object lastingly, timely and accurately by limited energy and computing capacity of nodes can be achieved effectively.