Optimization of Localization Error in Multi-Agent Systems through Cooperative Positioning: Autonomous Navigation in Partially Denied GNSS Environments

This paper proposes a novel navigation technique in multi-agent autonomous vehicles to improve network localization in a partially denied GNSS (Global Navigation Satellite System) environment, where at least one agent (referred to as the leader) has access to its actual position, and the GNSS denied part of the network tends to achieve consensus over the position of individual agents, based on the position of the leader. In a large network of autonomous vehicles (as in the Internet of Drones (IoD), or UGV traffic control) a fully-denied GNSS environment is rarely the case, as on one hand GNSS cyberattacks (including malicious jamming, spoofing, and meaconing) cannot affect extensive spaces (due to limited energy resources of adversaries), and on the other hand naturally obscured GNSS signals are often locally restricted features, and rarely span over vast geographic areas. This paper argues that if any anonymous agent in a connected network has access to GNSS, then all other agents could estimate their own locations by sensing their relative distance with respect to adjacent agents, and also exchanging their own estimated locations with respective neighbours, hence, accomplishing "cooperative positioning". It is shown that network localization is improved through optimization of its topology.