Near Field Channel Modelling through Large Aperture Massive Array Systems and Low-Complexity Pre-processing Robust to Local Mobility

Nielsen, Rasmus Birkelund ; Cadenas Buelga, Pelayu

4. term

Wireless Communication Systems, Master

2014

2014-06-20

222

This thesis addresses two related challenges in large-aperture massive MIMO systems: (1) developing a point-to-point (P2P) MIMO channel model that explicitly captures near-field effects, and (2) designing low-complexity multi-user pre-processing that is robust to users’ local mobility. The channel model is based on a simplified COST2100 approach, reduced to 2D and narrowband operation with a near-field focus, and is validated using relevant performance metrics, showing consistency with similar models. For pre-processing, linear precoders based on instantaneous CSIT (including matched filter, zero forcing, and MMSE) are examined to confirm the expected optimality of matched filtering with asymptotically many antennas; however, discrepancies in inter-user correlation caused by the massive array configuration are observed. This motivates statistical, subspace-based precoding methods that exploit channel statistics and users’ local movement. While most statistical schemes are suboptimal compared with full CSIT, results indicate they can achieve performance comparable to matched filtering, and one proposed statistical approach can outperform it under specific conditions. Overall, the work delivers a consistent near-field channel model for large apertures and practical, low-complexity beamforming strategies that enhance robustness to local mobility.

Denne afhandling undersøger to sammenhængende problemstillinger inden for massive MIMO-systemer med store aperturer: (1) at udvikle en passende punkt-til-punkt (P2P) MIMO-kanalmodel, der eksplicit tager højde for nærfeltet, og (2) at designe lav-kompleksitets forbehandling til multi-bruger adgang, som er robust over for brugernes lokale mobilitet. Kanalmodellen bygger på en forenklet COST2100-tilgang, nedgraderet til 2D og smalbåndet drift med nærfeltsfokus, og den valideres med relevante ydelsesmålinger, der viser overensstemmelse med beslægtede modeller. På forbehandlingssiden undersøges lineære forkodere baseret på øjeblikkelig CSIT (herunder matched filter, zero forcing og MMSE) for at afdække forventet optimalitet af matched filter ved asymptotisk mange antenner; dog afsløres afvigelser i bruger-til-bruger korrelation, som kan opstå på grund af systemets massive array-geometri. Dette motiverer statistiske, subrumsbaserede forkodningsmetoder, der udnytter kanalstatistik og brugernes lokale bevægelse. Selv om de fleste statistiske metoder er suboptimale sammenlignet med fuld CSIT, viser resultaterne, at ydeevne på niveau med matched filter kan opnås, og at en foreslået statistisk tilgang under visse forhold kan overgå matched filter. Arbejdet giver dermed en konsistent nærfeltskanalmodel for store aperturer og praktiske, lav-kompleksitets beamforming-strategier, der øger robustheden over for lokal mobilitet.

[This apstract has been generated with the help of AI directly from the project full text]

Massive ; MIMO ; Very Large Aperture ; channel ; cost2100 ; MMSE ; Minimum mean square error ; ZF ; zero forcing ; MF ; matched filter ; precoding ; beamforming ; subspace

Download PDF
View record in AAU Student Projects

Nielsen, Rasmus Birkelund:

• Physical Layer measurements in 3GPP LTE (2012)