Research On Limited Feedback Channel Under High Mobility Scenarios

Closed-loop multi-antenna technology can significantly improve the performance of the system by using channel state information (Channel State Information, CSI) at transmitter. In actual communication systems, TDD systems can use the reciprocity between uplink and downlink channels to obtain CSI. However in FDD systems, the reciprocity between uplink and downlink channels is unavailable because they use different transmission frequencies. The CSI is obtained by the feedback of the transmitter through the rate-limited uplink. This thesis studies the performance of multi-antenna systems with perfect CSI and deduces the effects of feedback delay on the BER. At last, several channel prediction techniques which can elevate the impact of the feedback delay on the system performance are introduced.Firstly, this thesis introduces the basic theory of matrices which will be used many times in the text, and then derives the channel capacity expressions of MIMO system when the CSI is unknown as well as when the transmitter has perfect CSI. The power allocation scheme is also discussed. Finally, simulation results show that the usage of CSI can significantly improve the capacity of system and the correlation between channels will lead to reduction of the system capacity.Secondly, this thesis investigates the multi-antenna spatial-multiplexing technique, including its implementation schemes and detection algorithms of mixed signals. Based on the analysis of the performance of zero forcing detection and the minimum mean square error detection schemes, we get the conclusion that the usage of the CSI can improve of the performance of the system. It is because the use of pre-equalization avoids causing the impact of enhanced noise that the improvement of SNR/SINR of the received signal can be achieved.Thirdly, we investigate spatial diversity technique and its usage of the CSI. In this section, spatial-time coding technique and its pair wise error probability are mainly discussed. The pair wise error probability is the basic principle of the code word designing. After the analysis, we get the conclusion that the performance of technique with pre-coding is much better than the technique without the usage of the CSI. At last, we introduce the antenna selection scheme which can reduce the cost of hardware while still maintaining the diversity gain. The analysis shows that the spatial-time coding technique joint with the antenna selection can provide the diversity gain which is extremely close to the pre-coding method.Finally, in high mobility scenarios, we investigate the impact of the feedback delay on the multi-antenna system with finite-feedback and multi-carrier channel prediction techniques which exploit the spatial-temporal correlation. Under the situation of feedback delay, we derive the closed-form expression of the BER by using the MGF of the SNR of the received signal. The simulation results proof that when the system is in high mobility scenarios the feedback delay will lead to a significant deterioration of the performance. This is because the channel varies rapidly and the coherent time of the channel is relatively short. Channel prediction technique is an effective measure against feedback delay, since it possesses the ability to track the time varying channel and to reduce the severity of performance deterioration of system. Most schemes predict the channel state by exploiting the temporal correlation of the channels, but when the channel varies rapidly, the temporal correlation between channels will be very low, thus if we still predict the future state using the SISO method and without considering the spatial correlation, we probably cannot get the desired result, even when the order of the predictor is very high. Therefore, for fast time-varying channel, we should consider the temporal correlation as well as the spatial correlation of the channels. This thesis analyzes a multicarrier channel prediction scheme which exploits both the spatial correlation and the temporal correlation. On this basis, we analyze a two-step prediction algorithm which ensures a good prediction performance while maintains a relatively low computational complexity. Finally, we simulate the mean square error (MSE) performance and improvements on the BER of the system of several prediction methods.