序列串接時空碼在衰褪環境下之盲蔽式解碼

Abstract

在無線通訊環境中，由於人車的移動、建築物的遮蔽及天氣的影響，往往會造成接收端在估計接收訊號上有很大的困難性。因此在接收端通常需要藉著分集(diversity)的技術，來增加接收訊號的準確性。 然而在無線通訊環境中，在接收端要做分集會有一定的困難度。原因是一般的接收端可能都是所謂的手機、PDA 或是其他一些行動裝置。運算功能及功率可能都不是很充足。而一般的傳送端如基地台 (mobile station)，運算功能及功率都相對較強。因此在傳送端採用分集的技術其可行性也相對較高。如時空碼 (Space-Time codes) 技術的一種編碼調變方式。 本篇論主要探討時空碼的兩種基本架構—方塊碼 (block code)及籬柵碼(trellis code)的編碼調變方式。藉著結合串接碼連接交錯排列器(interleaver)的形式並利用反覆解碼 (iterative decoding)的方式，來增加系統的編碼增益 (coding gain)。 除此之外，我們會把無線環境中所會碰到的頻率選擇性複衰減通道 (frequency selective complex fading channel)的因素考慮進來。因此必須先估計出通道參數，因為在做反覆解碼時，通道參數的已知是必要的。 首先我們利用一些的引導位元 (pilot symbol)，結合虛擬最大可能性 (pseudo maximum likelihood)演算法及維特比 (viterbi)演算碼的概念來估計通道參數。 由於利用引導位元的方式會降低頻寬效率及增加功率消耗，所以我們另外採取了盲蔽式 (blind)的方法。藉著分支與限定法 (branch and bound)及論文中所推出的結果，大大地降低原本需要全面性搜尋 (exhaustive search)的次數，增加了此盲蔽式方法的可行性。最後我們將利用數值模擬加以驗證分析及討論我們所提出的各種架構。

Space-time (ST) codes arrange the transmitted coded symbols in both time and spatial domains in such ways so as to achieve high coding gain while taking full advantage of the transmit and receive diversity. This thesis proposes decoder structures for ST block codes and ST trellis codes. The decoders use the turbo principle to decode the "equivalent" serially-concatenated ST codes, assuming a frequency selective fading environment. We first examine a serially concatenated ST block coded system. It consists of a nonrecursive convolutional code as the outer code and a ST block codes as the inner codes with an interleaver inserted in between them. The second system is a ST trellis coded scheme. The receiver regards the received samples as if they are serially-concatenated coded with a ST trellis codes as the outer code and the equivalent frequency selective FIR channel model as the inner codes with an interleaver placed at the output of the ST encoder. The performance of both systems under perfect channel state information is examined first. We then consider the more practical case issue when channel state has to be estimated. To begin with, we discuss the scenario when pilot symbols are available. Pseudo-maximum-likelihood (PML) and block least mean square (BLMS) methods are used to derive proper solutions. In order to increase bandwidth e?ciency, i.e., reduce the length of transmitted pilot symbols, Viterbi algorithm (VA) will be associated with them for channel estimation. The second scenario is the blind case where there is no pilot symbols. We make use of the concept of tree search together with a branch and bound algorithm to reduce the complexity and make the overall algorithm practical. Numerical system performance for various receiver structures is obtained and compared via computer simulation. We show that a multiple-antenna system does deliver diversity gain it promises and iterative decoding procedure of (equivalent) serially-concatenated codes results in performance improvement. Numerical results also indicate that all our proposed detector structures are efficient and practical.