Global Optimisation for Time of Arrival-Based Localisation

Abstract

Synchronous and asynchronous time of arrival-based localisation problems are considered. The likelihood functions in these problems are non-convex and can have issues of local extrema. Typical approaches therefore approximate maximum likelihood estimation by something easier to compute. We aim for global optimisation with guarantees, which we achieve by partitioning the search space into regions, each containing at most one critical point.