Details

Autor(en) / Beteiligte

• Fascista, Alessio
• Coluccia, Angelo
• Ricci, Giuseppe

Titel

A Pseudo Maximum likelihood approach to position estimation in dynamic multipath environments

Ist Teil von

• Signal processing, 2021-04, Vol.181, p.107907, Article 107907

Ort / Verlag

Elsevier B.V

Erscheinungsjahr

2021

Link zum Volltext

Quelle

Elsevier ScienceDirect Journals Complete

Beschreibungen/Notizen

• •The problem of maximum likelihood (ML) direct position estimation (DPE) of a multi-antenna receiver for the case of dynamic multipath environments is addressed.•A reduced-complexity algorithm based on the pseudo ML approach is proposed, which is able to estimate via spatially-smoothed adaptive beamforming the unknown multipath parameters that are needed to maximize the likelihood function.•The proposed approach is coupled with a novel AOA-based mechanism that conditionally associates the line-of-sight (LOS) over time for a given trial position, so obtaining a performance gain.•The simulation analysis reveals that the proposed algorithm is able to provide satisfactory performance even in presence of severe multipath, outperforming natural competitors also when the number of antennas and samples is kept at the theoretical minimum. The problem of maximum likelihood (ML) direct position estimation (DPE) of a multi-antenna receiver for the case of dynamic multipath environments is addressed, exploiting narrowband broadcast radio signals, without assuming special conditions such as mmWave massive MIMO, OFDM, or large bandwidth. To overcome the dramatic complexity of the plain ML formulation, which involves a large number of unknown parameters (proportional to the number of paths times the number of observations), a reduced-complexity algorithm based on a pseudo ML approach is proposed. Unlike classical two-step approaches, where angles of arrival (AOAs) are first estimated and then used in a second step to (geometrically) estimate the unknown position, the proposed algorithm also exploits the information brought by non line-of-sight (NLOS) paths: specifically, the whole multipath parameters are estimated via spatially-smoothed MUSIC and adaptive beamforming, to reconstruct the projection matrices appearing in the ML cost function, which is ultimately maximized with respect to the unknown position (sticking to the DPE approach). In addition, a novel AOA-based mechanism that conditionally associates the LOS over time for a given trial position is designed; in doing so, a performance gain is obtained by the coherent integration of multiple observations from different channel realizations. The performance assessment shows that the proposed algorithm is very effective in (even severe) multipath conditions, outperforming natural competitors also when the number of antennas and samples is kept at the theoretical minimum, and exhibiting robustness to several types of mismatch.