Robust Cross-layer Design of Wireless Multimedia Sensor Networks with Correlation and Uncertainty

Due to content-enriched sensing information and flexibility, Wireless Multimedia Sensor Network (WMSN) has a lot of potential applications. Low channel capacity, limited resource, correlation between sensing sources and uncertain factors make the design and optimization of WMSN challenging. In a densely deployed WMSN, there generally exist correlation and redundancy in the multimedia information collected by sensors with overlapped sensing area. In this paper, we adopt a crosslayer method to deal with the robust lifetime optimization of WMSN with correlated sources, which also has energy consumption uncertainty in the transmission of wireless links. To reduce the number of constraints of the source rate region, a pairwise Distributed Source Coding (DSC) scheme is proposed by matching sensor nodes based their correlation. A Distributed Pairwise Matching (DPM) algorithm is thus proposed. With a polyhedral set modeling the uncertain energy consumption, a robust cross-layer problem, which maximizes the lifetime of the WMSN through allocating source rate and flow rate on the link (i.e., routing) simultaneously under all the possible uncertainty, is formulated. The counterpart of the problem is showed to be a convex problem with linear constraints, which can be divided into several separate subproblems in different layers by dual decomposition. Based on the subgradient method and DPM, a partially distributed optimization algorithm is proposed. The algorithm can be implemented in small scale WMSNs with sink node only responsible for the calculation and distribution of the value of the lifetime. Simulation results verify the performance of the proposed algorithm and show its robustness under uncertainty.