Exjobbspresentation: Analysis of Phase Noise Effect on Panel-Based LIS systems

Large intelligent surfaces (LIS) extend conventional massive MIMO systems by employing a physically large and densely antenna surface, offering a promising architecture for next-generation wireless communication systems. In practical receivers, oscillator phase noise introduces time-varying phase rotations that cause the effective channel to evolve during a transmission block. This thesis investigates channel estimation, phase tracking, and coherent reception for a near-field LIS uplink in the presence of oscillator phase noise. Two LIS architectures are considered: a common-local-oscillator (CLO) architecture, where all antenna elements share one oscillator, and a panel-based architecture, where different panels are driven by independent oscillators. A discrete-time Wiener process is used to model the phase noise. The proposed receiver first applies pilot-aligned channel estimation to reduce the effect of phase variation during pilot transmission, and subsequently uses phase-tracking reference signal (PTRS) to track the residual phase drift during data transmission. Maximum-ratio combining is then used for coherent data detection.

Numerical results show that direct pilot averaging develops an error floor as phase noise becomes significant, whereas the pilot-aligned estimator maintains substantially lower channel-estimation error at moderate and high SNR. During data transmission, PTRS-aided tracking significantly reduces the channel mismatch and improves the mismatched achievable rate when the phase estimates are sufficiently reliable. The results also reveal a trade-off between phase-tracking accuracy and PTRS overhead, with the best rate obtained at an intermediate PTRS spacing. Furthermore, increasing the LIS size improves the achievable rate, but the gain is increasingly limited by residual phase noise.

The comparison between CLO and panel-based receivers further shows that the oscillator architecture strongly influences both phase-tracking behavior and the achievable coherent-combining gain.

Handledare: Ashkan Sheikhi & Juan Vidal Alegría

Examinator: Ove Edfors