Compressive Sensing-Based Adaptive Top-k Query over Compression Domain in Wireless Sensor Networks

Data query and energy saving in wireless sensor networks (WSNs) are crucial and have been widely studied. The compressive sensing (CS) technique is a candidate for reserving energy in communications and prolonging the lifetime of WSNs. However, the decompression of CS makes sensor nodes computationally inefficient. An intuitive approach to resolving this problem is to preserve sensing data in the compression domain, but this would limit the utility of stored data, such as data queries in WSNs. In this paper, we introduce the CS theory on sensing nodes to compress sensing data and store compressed data on storage nodes so that the communication and space overhead can be significantly reduced. More importantly, we allow querists to launch Top-k queries to the storage nodes, which can apply adaptive compressed data reduction (ACDR) to derive required data from the compression domain in accordance with the k ranks and return query results to the querists. ACDR enables storage nodes to dynamically change the size of compressed data according to the k ranks launched by querists so that the communication overhead is lower than that of other methods in literature.