Compressive network coding for error control in wireless sensor networks

Siguang Chen, Meng Wu, Kun Wang, Zhixin Sun

Abstract: Since the observed signals of nearby sensors are known to be correlated, this paper firstly investigates the connection between network coding and compression concept of compressed sensing and then makes an in-depth combination between these two powerful concepts for error control in wireless sensor networks. Thus, a joint scheme is developed to achieve the maximum gain by exploiting the temporal and spatial correlations simultaneously. This scheme overcomes drawbacks of network coding theory by injecting the corresponding distributed compressed sensing concept into network coding, i.e., the scheme possesses good compression gain and graceful degradation of precision in the reconstruction process. Meanwhile, it can tolerate finite erasures and errors as well as reconstruct the original information as precise as possible when the rank of error matrix (induced by erasures and errors) doesn’t exceed the upper boundary. Finally, the reliability analysis and numeric results show that the compressive network coding scheme (i.e., the joint scheme) outperforms the conventional network coding scheme in robustness and performance.