Cooperative Localization in Massive Networks

Abstract: Network localization is capable of providing accurate and ubiquitous position information for numerous wireless applications. This paper studies the accuracy of cooperative network localization in large-scale wireless networks. Based on a decomposition of the equivalent Fisher information matrix (EFIM), we develop a random-walk-inspired approach for the analysis of EFIM, and propose a position information routing interpretation of cooperative network localization. Using this approach, we show that in large lattice and stochastic geometric networks, when anchors are uniformly distributed, the average localization error of agents grows logarithmically with the reciprocal of anchor density in an asymptotic regime. The results are further illustrated using numerical examples.