位元分配預測應用於時間及空間相關多輸入多輸出通道

Abstract

本論文中，考慮時間與空間相關多輸入多輸出系統之有限回授適 應性位元分配。藉由時間相關性及預測編碼， 設計出一本適應性碼 書， 可以動態調整每一個子通道所分配之位元數。實驗結果顯示當 通道變化較慢的情況下， 位元分配預測編碼有很突出的表現， 即使 是在回授的位元數很少的情況下。在具空間相關性的通道，根據統計 特性，使用能最小化平均錯誤率下限之傳送器，使其在具相關性通道 能有效使用較少回授的位元數而有不錯的錯誤率表現。

In this thesis, we consider the adaptive rate transmission for a MIMO channel that is correlated in time and space. According to the channel condition, we adapt the number of bits loading to each subchannel of the MIMO system dynamically. The bit loading is predictively coded based on previously channel condition for low feedback rate. Assuming the channel is a first-order Gauss-Markov random process, and the statistics of channel is known to the transmitter. A statistical precoder is used to take advantage of the spatial correlation of the channel. The optimal predictor is derived for the next bit loading and the corresponding prediction error variance is analyzed for a slowly varying channel. Using the prediction error variance, the quantizers of the prediction errors can be adapted to get a smaller quantization error. Furthermore, we show that the prediction error variance is dominated by time correlation coefficient. This leads to the conclusion that a codebook designed for a different time coefficient can be easily modified to a codebook for different time correlation coefficient without redesign. Simulations are presented to demonstrate that the proposed predictive coding can achieve transmission rate very close to the unquantized case with a low feedback rate.