Distributed Compressive Sensing For Wireless Sensor Networks
Sun,Xinyao,Wang,Xue,Wang,Sheng,Bi,Daowei
2010-01-01
Abstract:In industrial applications of wireless sensor networks (WSNs), due to the constraints of limited bandwidth and power, it is not easy to transmit vibration signals in a real-time manner. To address this problem, compressive sensing is exploited to perform vibration signal compression with low computational complexity, which reduces volume of transmitted data and increase transmission efficiency and real-time performance. The fundamentals of compressive sensing enabled signal communication include exploring the inherent sparsity of vibration signals in the frequency domain, compressing the signal with a random Gaussian sampling matrix, and reconstructing the original signal by L-1 norm optimization. A distributed computing paradigm is also proposed to improve the performance of reconstruction by fusing the information of multiple sensor nodes. Experiments show that the proposed distributed compressive sensing enabled signal communication can decrease computational load, improve real-time performance and satisfy the constraints, making it a promising technology for communication in WSNs.