Research Of Distributed Compressive Sensing Based On Unbalanced Expander

Compressive sensing (CS) is a novel sampling theory which can recover a sparse signal x of large n dimension via the measurement whose dimension is much small than n. As we know, most of the nature signals can be sparse projected to some basis (generally wavelet basis). The compressive sensing can project the signal into the measure matrix and reconstruct the wavelet coefficients from the measurements and measure matrix.For large-scale distributed wireless sensor networks, WSN nodes usually have a very limited source of power and bandwidth. If the sensor nodes transmit the data to the fusion center directly without any in-network communication, it will not only cause network congestion but also waste energy. Therefore, the design of energy-efficient and reducing data transmission wireless sensor network architecture becomes urgent need. Relative to other data fusion technology, compressive sensing is suitable for appling in wireless sensor networks, its advantages are as follows: (1) Encoding is simple, using the non-adaptive random measurement matrix for data amplitude linear transformation can achieve the purpose of sampling and compression at the same time. (2) Encoding and decoding independence, the decoder only needs the measurement matrix and measurement to reconstruction the original signal adopting different reconstruction algorithms. (3) Excellent robust performance, even though lost part of the data, it still can reconstruct the original signal exactly.Considering the advantage of the compressive sensing and the wireless sensor network energy is limited. This paper we transplant the unbalance expander to distributed compressive sensing, propose the distributed compressive sensing based on unbalanced expander and then lead to distributed algorithm. The key idea is that the highly sparsity of binary random projections greatly reduces the communication cost of preprocessing the data. For distributed data values which is sparsity in one translate domain, fusion center can perfect reconstruct distributed data values by receiving m(m(?)n) data values from sensor nodesThis paper simulates in open infinite environment, the point source produce diffusion of gas to spread around at certain moments. We construct the distributed compressive sensing architecture which based on the unbalanced expander. The simulation results prove the architecture reduces the amount of data transmission without compromising on reconstruction performance.