| 摘要: | 本論文主要探討三個主題: 調適性解調系統、區塊式相差編碼及用於非
同調區塊碼之籬柵碼搜尋。
在第二章，本文提出了可加性白高斯雜訊(additive white Gaussian
noise, AWGN)通道及準穩態衰減通道(quasi-static fading channel)上使用無率碼(rateless code)之調適性解調系統的一個新檢測器。本文提出的檢測器能依照最大聯合機率找最可靠位元，而且等效於使用16DPSK(di erential phase-shift keying)調變的檢測器[2]。本文將使用決策邊界比較本文提出的用於調適性解調系統的16DPSK檢測器與傳統DPSK的錯誤效能。對於使用16DAPSK ( di erential amplitude phase-shift keying)調變，本文提出的檢測器比檢測器[2]有較佳的錯誤效能，而且除了解出1個可靠位元的情況下效能勝過16DPSK檢測器[2]。對於使用16PSK (phase-shift keying)調變，本文提出的檢測器會和16PSK檢測器[1]有相同的錯誤效能。對於使用16QAM (quadrature amplitude modulation)調變，本文提出的檢測器在解出2個最可靠位元時有較佳錯誤效能。本文另外研究調適性解調使用多重訊號差分檢測，而得到的結論是多重訊號差分檢測不適用於調適性解調系統。
在第三章中，本文提出用於區塊編碼16QAM及16TAPSK的區塊式相差編碼方法。由此方法可以使傳送區塊碼字的第一個符號和前一個傳送區塊碼字的最後一個符號相同，因此可以省去不傳。最後比較用於區塊式相差編碼的區塊編碼16QAM及16TAPSK和非同調區塊編碼16QAM及16TAPSK的錯誤效能比較。
在第四章中，本文提出使用非同調籬柵碼視作非同調區塊碼，找出不同狀態數下之最佳的非同調籬柵碼。此外將所搜尋的非同調籬柵碼和非同調區塊編碼調變在多個碼率上比較錯誤效能。所搜尋到的非同調籬柵碼可以提供系統多個選擇在不同碼率上。;In this dissertation, we aim to study on three topics: adaptive demodulation (ADM) system, block-wise differential encoding and trellis searches for noncoherent block codes.
In Chapter 2, we propose a new detector of adaptive demodulation (ADM) for rateless code over the additive white Gaussian noise (AWGN) channel or the quasi-static fading channel. For 16-DPSK, we show that the proposed detector is able to find bits with the largest joint probability, and equivalent to the detector in [2]. Besides, the error performance of ADM is compared with conventional DPSK by decision boundaries. For 16-DAPSK, the proposed detector has better error performance than the detector in [2], and also outperform 16-DPSK except that the only one bit is recovered. For 16-PSK, the proposed detector has the same error performance as the detector in [1]. For 16-QAM, the proposed detector has better error performance than the detector in [1] except that three bits are recovered. In addition, we research ADM using multiple-symbol differential detection (MSDD) and conclude that MSDD is not suitable for ADM.
In Chapter 3, we propose a block-wise differential encoding method for noncoherent block-coded 16TAPSK and 16QAM. By this method, the first symbol of the transmitted codeword is the same as the last symbol of the previously transmitted codeword, so it can be omitted. We analyze the minimum noncoherent distance of noncoherent block-coded QAM/TAPSK, and also modify the bit labeling of 16QAM and 16TAPSK to achieve better error performance for noncoherent block-coded QAM/TAPSK.
In Chapter 4, we propose to use noncoherent trellis codes as noncoherent block codes. Trellis codes for different number of trellis states are searched. Besides, we compare the search trellis codes with codes of noncoherent block-coded modulation over various rate. The search codes can offer various trellis states for various rates. |
