Simple Approximate Message Passing for Network Synchronization and Localization

This paper revisits the problem of time synchronization and localization in wireless networks with approximate message passing. A simple yet efficient, approximate Gaussian Belief Propagation (GBP) algorithm for cooperative localization is proposed. Compared to prior art, the algorithm exploits only one-way measurements, as well as simplified equations, not offered before in the literature. The joint time offset and location estimation problem is then studied, and a non-cooperative approximate GBP algorithm is proposed, where agents communicate only with anchors, again with one-way (and not round-trip) measurements. Both proposed algorithms perform close to state-of-the-art methods while offering minimal total communication overhead, which for localization, is smaller by a factor of 100 compared to the prior art.