Optimization of a Modular Ad Hoc Land Wireless System via Distributed Joint Source-Network Coding for Correlated Sensors

This paper outlines a proposition of a framework optimizing the communication scheme from the physical part of the transmitters to the data gathered at the receiver in a wireless sensor network. We propose a coding scheme able to take into account the correlation between measurements obtained by the sensors. This scheme consists of joint distributed source encoding and linear network coding. Several coding strategies are compared. (1) linear source coding, based on the low density generator matrix (LDGM) codes, where the compression process is performed by every sensor independently (2) distributed source coding, where the correlation between sources is taken into account, without cooperation between sensors. The compression ratios costs are evaluated analytically for each strategy, with respect to the communication costs between the nodes of the network. The results show a significant improvement in terms of compression rate and distortion compared to linear source encoding.