Channel Prediction On Grassmannian Manifold In Wireless Communication System

Multiple-input multiple-output(MIMO) and orthogonal frequency division multiplexing(OFDM) as the key technology of the fourth generation mobile communication system, have been widely used in all kinds of wireless communication system.However, the good performance of the system depends on known channel state information at the transmitter. Limited feedback is a practical technique to obtain partial channel state information at the transmitter used for preprocessing. The feedback delay in practical systems is inevitable due to signal processing, propagation delay and network delay. For time-varying channel, the feedback channel state information may become outdated before its use at the transmitter, degrading the system performance seriously. The traditional limited feedback technology focuses on quantifying the channel state information, without considering feedback delay. The prediction of the channel state information is the effective method to solve the feedback delay. Through a certain prediction algorithm to predict the future channel state information, compensating feedback delay, can greatly improve the system performance. The prediction algorithm has become a research hotspot in wireless communications.This thesis, exploiting the differential geometric properties of the Grassman manifold, proposes a new Grassmannian predictive coding algorithm for delayed limited feedback systems. We make the evolution of channel state information in temporally correlated channel be modeled as a time series on the Grassmann manifold. In order to obtain the high quantization resolution, this paper raises a new adaptive quantization technique based on the tangent predictive error approximate to Gaussian. The simulation results show that, the algorithm can indeed enhance the sum rate of the system..In order to resolve the quantization error, feedback delays, clustering feedback in multiple-input multiple-output–orthogonal frequency division multiplexing(MIMO-OFDM) communications, this thesis put forward the geodesic-based prediction and interpolation algorithm. The classical Grassmannian prediction algorithm is performed at the receiver in time domain to mitigate the feedback delay effect. At the transmitter, geodesic-based interpolation is performed in frequency domain to obtain other subcarriers’ channel state information. We quantize the direction and the magnitude of the tangent predictive error separately, using two different dynamic codebooks to do adaptive quantization. The effectiveness of the proposed approach is demonstrated using extensive numerical simulations.