Research On Data Collection For Meteorological Sensor Networks Based On Distributed Compressed Sensing

Wireless meteorological sensor network consist of a large number of micro sensor nodes with advantages of distributed widely, easy deployment,small impact for the environmental, high precision, low cost. It is suitable for regional meteorological environment monitoring. Due to the sensor nodes are deployed densely and the meteorological elements such as temperature and humidity have high spatial-temporal correlation, the data collected by meteorological sensor network is very bulky, highly correlated and redundant.If these nodes transmit these high redundancy data without processing, the network will consume most of energy. It can efficiently reduce the network traffic by distributed random compressed sampling. The main work in this paper is data collection technology based on distributed compressed sensing in wireless meteorological sensor network by exploit the spatial-temporal correlation of meteorological data.The basic structure and characteristics of wireless meteorological sensor network are introduced firstly. Then, this paper outlines the data collection technology in sensor network, compressed sending and distributed compressed sensing technology. Lastly, the DCS based data collection technology of sensor network is analyzed. All of these studies provide preparation for further research on the DCS based collection in meteorological sensor network.In order to solve excessive traffic expense caused by high spatial-temporal correlation and redundant of sensed data in meteorological sensor network, a Data Preprocessing Model based on Joint Sparsity(DPMJS) was proposed. The DPMJS uses a regional common portion calculated by the detection area meteorological forecast value and the cluster head elements to preprocess sensed data. The DCS based data collection schema was applied in clustered meteorological sensor network, sensed data was measured in multiple rounds by Gauss random observation sub-matrix constructed in cluster head node,the measure values were send to Sink through multiple hopes to recovery the original data. This data collection schema can reduce the network traffic and achieve load balancing. By taking the impact of abnormal data on data sparse and reconstruct, a common portion based schema was designed. Simulation results shows that compared with using CS,the collection schema proposed above can enhance data sparsity by exploiting spatial-temporal correlation of meteorological information and compression performance efficiently, improve data recovery rate obviously. And this schema can also reconstruct abnormal data with high rate successfully.Aiming at the limited resource of sensor node and too many nodes participating in distributed compressed sampling in clustered meteorological sensor network,a Participating Nodes Optimization model based on Energy Analysis (PNOEA) was proposed. Each node in same cluster achieves their own energy consumption analysis according to the residual energy and estimated communication cost. Finally cluster system determines the amount of nodes participating in data collection according to the energy analysis level. In simulation, compared with the DCS and DPMJS data collection schema, the PNOEA can not only ensure the data recovery accuracy, but also control and reduce the sampling nodes number efficiently, save a large amount of resource consumption, prolong the whole network operation cycle.