Emitter Localisation from Reception Timestamps in Asynchronous Networks (1407.8056v3)

Abstract: We address the problem of localising a mobile terminal ("blind" node) in unknown position from a set of "anchor" nodes in known positions. The proposed method does not require any form of node synchronisation nor measurement (or control) of the transmission times, which is difficult or anyway costly to achieve in practice. It relies exclusively on reception timestamps collected by the anchor nodes, according to their local clocks, that overhear packets transmitted by the blind node and by (at least one) other transmitting node(s) in known position, e.g., other anchors. The clock differences between the nodes are not eliminated ex ante through clock synchronisation, as in traditional ToA and TDoA methods. Instead, they are counteracted ex post, during the data processing stage, leveraging the data redundancy that is intrinsic to the multiple reception of the same packet by different (anchor) nodes. We validate the proposed method in different experimental settings, indoor and outdoor, using exclusively low-cost Commercial-Off-The-Shelf WiFi devices, achieving sub-meter accuracy in full Line-of-Sight conditions and meter-level accuracy in mild Non-line-of-sight environment. The proposed method does not require that the blind node participate actively to the localisation procedure and can use "opportunistically" any legacy signal or packet available over-the-air for communication purposes. Considering the very minimal requirement on the system - basically, only that anchors in known positions are able to collect and share reception timestamps - the proposed approach can enable practical adoption of opportunistic and/or cooperative localisation on top of existing radio communication systems.