Grid-Based Belief Propagation for Cooperative Localization

We present a novel parametric message representation for belief propagation (BP) that provides a novel grid-based way to address the cooperative localization problem in wireless networks. The proposed Grid-BP approach allows faster calculations than non-parametric representations and works well with existing grid-based coordinate systems, e.g., NATO’s military grid reference system (MGRS). This overcomes the hidden challenge inherent in all distributed localization algorithms that require a universally known global reference system (GCS), even though every node localizes using arbitrary local coordinate systems (LCSs) for a reference. Simulation results demonstrate that GridBP achieves similar accuracy at much reduced complexity when compared to common techniques that assume ideal reference.