In recent years, the interest for millimeter-wave (mm-Wave) frequency bands for cellular systems and for Vehicle-to-Everything (V2X) applications is rapidly increased thanks to huge spectrum availability and recent advances in the hardware. The main issue at mm-Wave is the much higher path-loss compared to current wireless systems. In this regard, Massive Multiple Input Multiple Output (mMIMO) systems are a viable and widespread solution, beamforming gains ensure transmission links with sufficient Signal-to-Noise Ratio (SNR). The high mobility, characterizing the vehicular communications scenarios, and the constraints of Ultra-Reliable Low-Latency Communications V2X (URLLC V2X) make the channel estimation a challenging task. Among the existing approaches to improve the mMIMO channel estimation, algebraic approaches as Low-Rank (LR)-based provide a suitable trade-off between complexity of the implementation, performance and robustness to hardware imperfections, operating a modal filtering on the LS-estimated channel. Unfortunately, the conventional LR channel estimation, based on the invariance of channel space-time (ST) features over multiple consecutive transmission slots (MS-LR), is characterized by a long convergence time, especially for mMIMO settings. Moreover, the training procedure to retrieve the channel modes must be repeated for each vehicle in the Base Station (BS) coverage area. In this thesis, we first review the LR channel estimation algorithms proposed in the literature for MIMO-OFDM and then we propose a new approach called Multi-Vehicular Low Rank (MV-LR) that overcome MS-LR limits exploiting multiple passages of different vehicles over the same location in space. Finally, to face high cost and complexity of full digital mMIMO systems, we introduce a new Hybrid MIMO systems designing precoders and combiners properly to perform MV-LR channel estimation. In order to validate the proposed techniques, we evaluate the performance over the geometrical stochastic model and over a deterministic channel model obtained from a ray-tracer. Numerical results show the feasibility of the proposed channel estimation method for a mMIMO mobility scenario.
Negli ultimi anni l'interesse per la bande di frequenze ad onde millimetriche (mm-Wave) applicate a sistemi cellulari e ad applicazioni Vehicle-to-Everything (V2X) è cresciuto grazie alla disponibilità di spettro e ai recenti progressi dell'hardware. Il problema principale delle mm-Wave sono le intense perdite durante la propagazione, maggiori rispetto agli attuali sistemi wireless. I sistemi Massive Multiple Input Multiple Output (mMIMO) sono una possibile e diffusa soluzione al problema, infatti il guadagno dato dal beamforming garantisce un Signal-to-Noise Ratio (SNR) accettabile. La mobilità degli utenti che caratterizza le comunicazioni veicolari e i vincoli delle Ultra-Reliable Low-Latency Communications V2X (URLLC V2X) rendono difficile stimare il canale. Tra gli approcci già esistenti per sistemi mMIMO, metodi algebrici come il Low Rank (LR) forniscono un trade-off tra complessità nell'implementazione, performance e solidità alle imperfezioni dell'hardware, operando un filtraggio sulle stime di canale ottenute tramite LS. Sfortunatamente, i metodi LR standard, basati sull'invarianza delle componenti spazio-tempo (ST) del canale per consecutivi slots di trasmissione (MS-LR), sono caratterizzati da lenta convergenza specialmente per sistemi mMIMO. Inoltre, la procedura per ottenere i modi del canale deve essere ripetuta per ogni veicolo all'interno della cella della Base Station (BS). In questa tesi verrà riportato l'algoritmo LR per sistemi MIMO-OFDM presente in letteratura, poi verrà presentato un nuovo approccio, il Multi-Vehicular Low Rank (MV-LR), che risolve i limiti del MS-LR sfruttando molteplici passaggi sopra la stessa posizione di diversi veicoli. Infine, per sopperire agli alti costi e alla complessità dei sistemi mMIMO completamente digitali, sarà introdotto un sistema MIMO ibrido in cui precoders/combiners sono progettati per eseguire la stima di canale usando l'approccio MV-LR. Per validarne le prestazioni, questi verranno testati sia con un modello di canale stocastico che deterministico, quest'ultimo ottenuto attraverso un ray-tracer. I risultati numerici proposti mostrano la fattibilità del MV-LR applicato a uno scenario mMIMO dinamico.
Multi-vehicular low-rank channel estimation for mm-wave hybrid MIMO s ystem
RINALDI, LUCA
2019/2020
Abstract
In recent years, the interest for millimeter-wave (mm-Wave) frequency bands for cellular systems and for Vehicle-to-Everything (V2X) applications is rapidly increased thanks to huge spectrum availability and recent advances in the hardware. The main issue at mm-Wave is the much higher path-loss compared to current wireless systems. In this regard, Massive Multiple Input Multiple Output (mMIMO) systems are a viable and widespread solution, beamforming gains ensure transmission links with sufficient Signal-to-Noise Ratio (SNR). The high mobility, characterizing the vehicular communications scenarios, and the constraints of Ultra-Reliable Low-Latency Communications V2X (URLLC V2X) make the channel estimation a challenging task. Among the existing approaches to improve the mMIMO channel estimation, algebraic approaches as Low-Rank (LR)-based provide a suitable trade-off between complexity of the implementation, performance and robustness to hardware imperfections, operating a modal filtering on the LS-estimated channel. Unfortunately, the conventional LR channel estimation, based on the invariance of channel space-time (ST) features over multiple consecutive transmission slots (MS-LR), is characterized by a long convergence time, especially for mMIMO settings. Moreover, the training procedure to retrieve the channel modes must be repeated for each vehicle in the Base Station (BS) coverage area. In this thesis, we first review the LR channel estimation algorithms proposed in the literature for MIMO-OFDM and then we propose a new approach called Multi-Vehicular Low Rank (MV-LR) that overcome MS-LR limits exploiting multiple passages of different vehicles over the same location in space. Finally, to face high cost and complexity of full digital mMIMO systems, we introduce a new Hybrid MIMO systems designing precoders and combiners properly to perform MV-LR channel estimation. In order to validate the proposed techniques, we evaluate the performance over the geometrical stochastic model and over a deterministic channel model obtained from a ray-tracer. Numerical results show the feasibility of the proposed channel estimation method for a mMIMO mobility scenario.| File | Dimensione | Formato | |
|---|---|---|---|
|
2020_10_Rinaldi.pdf
non accessibile
Descrizione: Testo tesi
Dimensione
3.49 MB
Formato
Adobe PDF
|
3.49 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/167111