Research On Precoding For MIMO Relay Systems

Relays can be deployed in cellular wireless networks to extend coverage and combat shadowing, and thus significantly increase system throughput. Meanwhile, multiple-input multiple-output (MIMO) technology, which provides multiplexing and diversity gains by equipping multiple antennas at the transmitter and receiver, could improve spectral efficiency and transmission reliability without increasing bandwidth or transmit power. MIMO relay systems incorporate the benefits of MIMO into relay systems by equipping multiple antennas at the relay node. If channel state information (CSI) is available at the transmitter, advanced MIMO transmit signal processing technologies such as precoding can be applied to further improve system performance. However, in practice only imperfect channel state information can be obtained. In Frequency-Division Duplex (FDD) systems, the receiver first quantizes CSI and then feeds back the quantized CSI to the transmitter, but since quantization or feedback bits are limited, quantization errors always exit. In Time-Division Duplex (TDD) systems, although CSI can be easily obtained at the transmitter by channel reciprocity, it needs to be estimated using pilot signals. Due to noise and the time varying nature of wireless channels, channel estimation errors are inevitable. Therefore, we investigate MIMO relay precoder designs with imperfect channel state information to provide robustness against CSI errors and improve system performance.In a multi-user MIMO relay system with limited feedback, we consider that only quantized channel direction information (CDI) is available at the base station (BS) and relay station (RS), and jointly design the BS and RS precoders to minimize the expected mean squared errors conditioned on quantized CDI. Since the original problem is a complex non-convex problem, we decompose it into two more tractable sub-problems and derive their closed-form solutions respectively. An iterative algorithm is proposed to jointly optimize the BS and RS precoders. Simulation results show that the proposed robust scheme achieves better bit-error rate (BER) performance than the non-robust scheme that does not take CDI quantization errors into account.In MIMO relay systems with imperfect channel estimation, we model the CSI errors as Gaussian random variables and calculate the mean squared errors (MSE) averaged over the CSI errors. We derive a closed-form solution for the optimal relay precoder to minimize the average MSE. Simulation results verify the superior performance of the proposed robust relay precoder compared to the non-robust precoder in terms of both MSE and BER. This advantage becomes more evident when signal-to-noise ratio (SNR) increases. The performances of both the robust and non-robust precoders improve with smaller CSI errors.In a two-way MIMO relay system with imperfect channel estimation, we address the transceiver design problem. Since only estimated channel state information (CSI) is available, the self-interference (SI) cannot be completely canceled at destination nodes. With both channel estimation errors and residual self-interference considered, we propose two robust schemes to minimize the average sum mean squared error (MSE). In the iterative scheme, the relay precoder and destination receivers are alternately optimized until convergence. In the constrained structure scheme, a specific structure is imposed on the relay precoder to reduce complexity. Then the relay precoder design is reduced to the simple power allocation problem. Simulation results show that the proposed iterative scheme provides better robustness against channel uncertainties than the non-robust iterative scheme, and the constrained structure scheme achieves close performance with the proposed iterative scheme.There are still many problems to be investigated for precoding designs in MIMO relay systems. In practice, relay could work in different modes, and there could be different requirements, so different precoding designs need to be proposed. Besides, it is worth studying robust relay precoding schemes under different channel quantization and channel estimation methods.