Compressed sensing for efficient random routing in multi-hop wireless sensor networks

Compressed sensing (CS) is a novel theory based on the fact that certain signals can be recovered from a relatively small number of non-adaptive linear projections, when the original signals and the compression matrix own certain properties. In virtue of these advantages, compressed sensing, as a promising technique to deal with large amount of data, is attracting ever-increasing interests in the areas of wireless sensor networks where most of the sensing data are the same besides a few deviant ones. However, the applications of traditional CS in such settings are limited by the huge transport cost caused by dense measurement. To solve this problem, we propose several ameliorated random routing methods executed with sparse measurement based CS for efficient data gathering corresponding to different networking topologies in typical wireless sensor networking environment, and analyze the relevant performances comparing with those of the existing data gathering schemes, obtaining the conclusion that the proposed schemes are effective in signal reconstruction and efficient in reducing energy consumption cost by routing. Our proposed schemes are also available in heterogeneous networks, for the data to be dealt with in CS are not necessarily homogeneous.