Distributed Signature Learning and calibration for large-scale sensor networks

In this paper, we consider the problem of joint sensor calibration and target signature estimation using distributed measurements from a large-scale wireless sensor network with random link variations. Specifically, we propose a new Distributed Signature Learning and Node Calibration, D-SLANC, which can estimate the (constrained) parameters of interest, using measurements from the sensor nodes, in a distributed manner. Unlike a centralized algorithm that relies on pooling measurement vectors from the network, D-SLANC operates at the parameter space reducing the communication bandwidth. We model the sensor network as a connected graph and show that the gossip-based distributed consensus can be used to update the estimates at each iteration of the D-SLANC algorithm. As a result the proposed algorithm is robust to link and node failures, unlike previously suggested distributed subgradient methods that rely on formation and maintenance of a stable network infrastructure to perform iterations in parameter space. We prove the guaranteed convergence of the algorithm to the centralized data pooling solution and compare its performance with the derived Cramér-Rao bound, using simulations.