Weighted consensus algorithms for distributed localization in cooperative wireless networks

In this paper we consider distributed localization in a wireless peer-to-peer network. Each node is required to estimate the geographical configuration of the whole network (i.e., the positions of all nodes regardless of link availability) based on local processing and iterated information exchange with neighbors. We propose a new weighted-average consensus method based on received-signal-strength peer-to-peer measurements. To handle the non-linearity of the measurement model, an iterative Gauss-Newton location estimator is integrated into the consensus algorithm. Consensus processing is designed so as to account for the different degree of reliability of the location information provided by different neighbors and boost the convergence process. Performances are analysed and compared to those of conventional consensus algorithms, in terms of accuracy and convergence rate, using fundamental performance bounds as benchmarks.