Complex Environment Of Multi-antenna Non-coherent Communication

Applying multiple antennas to transmitter and receiver can improve channel capacity, system communication rate and capability to mitigate fading. However, in practice, channel characteristic is changeable. Terminal movement causes channel fast fading, which makes channel estimation obsolete and influences receiver signal detection. Moreover, channel spatial correlation, which is raised by the limited space to install multiple antennas at mobile terminal, also impacts system performance. In this paper, we discuss some research results and corresponding algorithm on the topic of multiple antennas communication in fast fading and spatial correlated channel.To get perfect channel state information (CSI) in fast fading channel, a large number of training blocks are required to send to aid channel estimation. For multiple antenna system, this training cost is futher increased considering the fact that multiple channel taps are required to estimate. If channel estimation component is discarded at receiver, the resources can be saved. We denote the system with no CSI at receiver as noncoherent system. Because it is relatively easy to get channel statistical information in fast fading channel, we consider first in this paper of a noncoherent system with channel spatial correlation information at both link ends. When channel is extremely fast fading (the coherent interval is approximately one or two channel use), eigenvector modulation and orthogonal signal design are proposed to utilize channel spatial correlation information. Eigenvector modulation modulates information bit one to the maximuim eigenvector of transmit spatial correlation matrix and information bit zero to zero eigenvector. Orthogonal signal design projects two information bits to two mutual orthogonal vectors, and transmits signal in the direction of maximuim eigenvector of spatial correlation matrix. It is shown in the study that these two communication schemes receive performance gain on spatial correlation, which is regarded as detrimental to system performance. The higher spatial correlation is, the better performance results. Next, a novel spatial correlation estimation algorithm is proposed in this paper to get channel statistical information.Differential transmission using space time code is another effective approach for noncoherent communication. Differetial space time code (DSTC) has the advantage that the receiver does not need channel estimation component and the detection complexity is low. Also, DSTC achieves high diversity gain. However, the bit error rate performance of DSTC is 3 dB lower than that of coherent detector. Thus, how to improve the performance of DSCT is one of the important points in noncoherent communication research. First, by regarding differential detector as the combination of combined channel estimation and coherent detector, we propose a frame structure of generalized differential transmission (GDT). GDT frame composes reference signal part and normal signal part. The space time block code in both reference signal part and normal signal part is differentially encoded based on previous reference signal block. By power loading between reference signal part and normal signal part, the optimal GDT scheme surpass conventional differential transmission scheme by 3 dB and approaches coherent system. In addition, we study the approach to improve DSTC performance using channel spatial correlation information. An optimal transmit beamformer for DSTC is designed on the basis of correspondence between multiple input multiple output (MIMO) space time processing unit and multiple input single output (MISO) space time processing unit. Simulation shows the optimal transmit beamformer improves DSTC performance to 1 dB when channel spatial correlation is high.