於通道偏差下單載波及多載波區塊式傳輸系統之強健式接收機設計

Abstract

多輸入多輸出正交分頻多工（MIMO-OFDM）系統及多輸入多輸出單載波迴旋前綴（MIMO SC-CP）系統能非常有效地補償頻率選擇性衰減及支援高資料傳輸率，因此已獲得許多系統設計者的注意，在本篇論文中，吾人將分別針對上述兩種系統在特定的通訊環境中設計接收機架構。在MIMO-OFDM系統部分，吾人考慮CP長度比通道階數（Channel Order）短的通訊環境，首先針對單輸入多出輸出（SIMO）模式進行設計，吾人聯合利用接收端的空間及頻率資源提出一個能有效地消除CP不足所造成之內符碼干擾（ISI）及內載波干擾（ICI）的強制最佳化（Constrained Optimization）線性等化器，其最佳化問題在等效的無強制（Unconstrained）廣義旁波帶消除（GSC）機制下進行求解，之後吾人再將所提出之GSC等化器架構推廣至MIMO模式。除此之外，吾人進一步假設接收端無法得知精確之通道參數而需採用最小平方（LS）技術進行估測，並提出利用擾動分析（Perturbation Analysis）技術將通道估計錯誤之效應明確地併入GSC系統模型，這使得LS通道誤差之特性能被應用，藉以推導出一個能對抗通道估計錯誤的封閉（Closed-Form）強健解。吾人亦推導出此強健式等化器的近似輸出訊號干擾雜訊比（SINR），由此結果可看出相較於非強解健解的一些優點。 在MIMO SC-CP系統部分，吾人考慮一個時變通道環境，同時假設通道參數亦採用LS技術估測而得。因為通道在時間上的變化會破壞頻域訊號間的正交性，使得低複雜度之各頻（Per-Tone）等化機制無法實現，所以在頻域處理訊號將不再具有優勢。因此，吾人提出直接於時域中處理訊號，於時域中，吾人發現訊號特徵矩陣能夠被分為數個擁有正交元素的群組，能自然地被用來設計群組式（Group-Wise）訊號偵測技術，為了實現此特性吾人提出一個GSC接收機，其能同時抑制通道時間變化與通道估計錯誤所產生之通道偏差效應。由驗證顯示吾人於MIMO-OFDM系統中針對GSC所設計的擾動分析數學架構亦能將本部分所考慮的通道偏差併入系統方程式中，基於通道時間變化與通道估計錯誤的統計假設吾人亦可推導出一個封閉（Closed-Form）強健解。透過一些數值範例可證實吾人所提出之接收機架構在所考慮的通訊環境中效能明顯優於現存的方法。

MIMO orthogonal frequency division multiplexing (MIMO-OFDM) and MIMO single-carrier with cyclic prefix (MIMO SC-CP) have drawn a lot of attention since they can effectively compensate frequency selective channels and can support high data rates. In this dissertation, we will design receiver architectures for both of them, each under a specific communication environment. For MIMO-OFDM systems, we consider a scenario that the adopted CP length is shorter than the channel order. By jointly exploiting the receiver spatial and frequency resources, we first propose a constrained optimization based linear equalizer, which can mitigate the resultant inter-symbol interference and inter-carrier interference (ICI) incurred by the insufficient CP insertion, for the SIMO case. The optimization problem is solved under an equivalent unconstrained generalized sidelobe canceller (GSC) setup. Then the proposed GSC-based equalization framework is generalized to the MIMO case. Moreover, in this case we further assume that the channel parameters are not exactly known but are estimated using the least-squares (LS) training technique. We propose to apply the perturbation analysis technique to explicitly incorporate the channel parameter error into the GSC system model; this allows us to exploit the presumed LS channel error properties for deriving a closed-form robust solution against the net detrimental effects caused by the channel estimation errors. A closed-form approximate output SINR expression of the proposed robust equalizer is also derived, based on which some appealing advantages over the non-robust counterpart can be inferred. For MIMO SC-CP systems, we consider a communication environment that the channel is time-varying, under the assumption that the channel parameters are also estimated via the LS training technique. Since the channel temporal variation destroys the orthogonality between signal components in frequency domain, low-complexity per-tone based equalizations can no longer be realized. As a result, there are no specific advantages of processing signals in frequency domain, and we propose to directly process signals in time domain. By this way, it is observed that in time domain the signal signatures can be arranged into groups of orthogonal components, leading to a very natural group-wise symbol recovery scheme. To realize this figure of merit, we propose a GSC-based receiver, which also takes into account the mitigation of channel mismatch effects caused by the channel temporal variation and the imperfect estimation. It is shown that the proposed perturbation analysis framework as well enables us to model the channel mismatch effects into the system equation and, in turn, to further exploit the statistical assumptions on the channel temporal variation and the estimation error for deriving a closed-form robust solution. By some numerical examples, it is confirmed that the proposed receiver architectures outperform the existing methods under the respectively considered communication environments.