功率多樣性分配於垂直分層空時碼與多進制調變系統之效能分析與應用

Abstract

隨著人們需要更高速的無線傳輸，垂直分層空時碼提供了一種兼顧高頻譜效能與高傳輸速度的多天線架構。在本篇論文中，我們分析了瑞利衰減通道下使用多進制調變與排序迫零干擾刪除演算法的垂直分層空時碼系統之符號錯誤率，並利用此分析結果重新分配傳輸端的天線功率，以使得符號錯誤率下降至最低。在模擬結果中，我們發現所推導的最佳化功率分配在快速瑞利衰減通道下，可使得系統的符號錯誤率表現提升3.5至4分貝；而在慢速瑞利衰減通道下，則可使得系統的符號錯誤率表現提升8.5至10分貝。我們在本論文中進一步將分析結果推廣至串聯迴旋碼與垂直分層空時碼的無線傳輸系統。對迴旋碼與垂直分層空時碼分析時，我們首先推導出單一垂直分層空時碼傳輸所造成的位元錯誤數量之機率分佈，再利用此機率分部計算出迴旋碼的密碼字之成對錯誤率以及其位元錯誤率的聯合邊界。透過模擬我們發現位元錯誤率的聯合邊界與模擬結果相當吻合。此外，我們亦將最佳化功率分配應用於迴旋碼與垂直分層空時碼之無線傳輸系統，而其結果顯示平均分配功率於各傳輸端天線即接近最佳之分配分式。

As the demand of high data rate transmission in wireless communication rises persistently, the Vertical Bell Laboratories Layered Space-Time (VBLAST) systems provide a solution which attains very high spectral e±ciency. In this work, the Symbol Error Rate (SER) of the VBLAST system with M-PSK and M-QAM using zero-forcing successive interference cancellation (ZF-SIC) under Rayleigh fading are analyzed. With the optimal power allocation pattern derived from the analysis, the power of the transmit antennas can be adjusted judiciously to acquire the lowest SER. The numerical results corroborate the accuracy of the analysis and reveal that the power diversity scheme can improve the system SER performance by 3.5 to 4 dB in the fast fading environment and 8.5 to 10 dB in slow fading environment. With the methodology developed in the analysis of the SER of the VBLAST system, we further analyze the bit error rate (BER) of the system of convolutional code concatenated with VBLAST architecture using ZF-SIC detection. The union bounds of BER of the convolutional coded ZF-SIC VBLAST system using BPSK and M-QAM modulation schemes are derived. To analyze the BER of the system, we first derive the probability distribution of the bit error number in a VBLAST symbol for BPSK and M-QAM, where the VBLAST symbol stands for the modulation symbols transmitted by the transmit antennas in one VBLAST transmission. The distribution of bit error number in a convolutional codeword is calculated with the bit error number in a VBLAST symbol, and then the pairwise probability of the convolutional code is derived. In the BPSK case, the union bound of BER approximates the simulation results as the SNR is higher than 6 dB, and in the 16QAM cases, the union bounds of BER approximates the simulation results as the SNR is higher than 15 dB. Since the union bounds are tight when the simulation BER is lower than 10^(-3), the analysis in this work provides a precise evaluation of the system performance in the appropriate operating region. The optimal power allocation is also applied to minimize the BER of the convolutional coded VBLAST system, and the numerical results reveal that allocating equal power to transmit antennas is closely optimal.