Impact of Channel Asymmetry on Performance of Channel Estimation and Precoding for Downlink Base Station Cooperative Transmission

TL;DR

This paper investigates how channel asymmetry affects channel estimation and precoding in downlink base station cooperative systems, showing that certain estimation methods are robust and that rate loss remains minimal despite asymmetry.

Contribution

It introduces three linear estimators for joint channel estimation in asymmetric cooperative systems and analyzes their robustness and impact on system performance.

Findings

Using large scale channel information reduces performance loss with non-orthogonal training.

Estimation errors in weak links cause minor rate loss due to weighting by SNR.

Rate loss per user remains nearly constant regardless of user location.

Abstract

Base station (BS) cooperative transmission can improve the spectrum efficiency of cellular systems, whereas using which the channels will become asymmetry. In this paper, we study the impact of the asymmetry on the performance of channel estimation and precoding in downlink BS cooperative multiple-antenna multiple-carrier systems. We first present three linear estimators which jointly estimate the channel coefficients from users in different cells with minimum mean square error, robust design and least square criterion, and then study the impact of uplink channel asymmetry on their performance. It is shown that when the large scale channel information is exploited for channel estimation, using non-orthogonal training sequences among users in different cells leads to minor performance loss. Next, we analyze the impact of downlink channel asymmetry on the performance of precoding with channel estimation errors. Our analysis shows that although the estimation errors of weak cross links are large, the resulting rate loss is minor because their contributions are weighted by the receive signal to noise ratio. The simulation results verify our analysis and show that the rate loss per user is almost constant no matter where the user is located, when the channel estimators exploiting the large scale fading gains.