基於最佳化之前瞻無線通訊系統實體層設計

Abstract

本論文旨在針對前瞻無線通訊系統之重要技術問題，在最佳化方法的輔助下，提出可硬體實現化之演算法。本論文分為三個部分，涵括的主題有：循環字首回收、考量電路耗功之設計、應用於前瞻無線通訊系統之(預)等化器。

首先，第一部分探討的主題是循環字首回收機制。吾人分別對在於「加成高斯雜訊」、「時變通道」、與「相位雜訊」存在的情形下提出最佳的循環字首重覆使用策略。相較於其他傳統啟發式的作法，本論文所提出的方法有顯著的效能改善，同時維持在一個相似的複雜度。

為因應行動裝置電池容量不足的困境，將電路耗能一併考量進行演算法設計已是通訊工程的一個趨勢。在第二部分(包含第四章與第五章)，吾人提出兩個考量電路耗能之通訊演算法。在波束形成通訊應用中，分集增益與電路耗功存在一個權衡的關係，意即，當啟動的天線數增加時，天線分集增益隨之增加，然而在另一方面，電路耗能(包括混波器、濾波器、數位類比轉換器等)也同時隨之上升。為了在天線分集增益與電路耗功兩者之間取得一個平衡點，第四章提出了一個系統化的方法來決定啟動天線的集合並同時調整波束形成之權重。接著，在第五章，我們針對能量效益(衡量的標準為「每焦耳耗能可傳送的位元數」)設計一功率分配演算法。此方法可以被運用在均一速率正交分頻多工系統以改善能量效益。

第三部分(從第六章到第八章)提出了三種低複雜度的等化器與預編碼器。在第六章，吾人提出了一適用於多路徑多用戶下行通道之線性預編碼器，並說明了其在毫米波多輸入單輸出蜂巢式通訊系統之應用。 此種預編碼器可望在未來第五代行動通訊系統(將使用毫米波頻段與巨量天線)的發展中成為有潛力的候選技術。 在第七章，吾人提出了一個適用於單載波區塊傳輸系統之低複雜度時域等化器。數值模擬結果說明了此方法可以在一可容忍的效能損失下，大幅度降低複雜度。 第八章探討的主題是多輸入多輸出系統傳送端與接收端分集技術，並嘗試設計適用於「『只有』傳送端擁有通道資訊時之多輸入多輸出通訊系統」此一情境之訊號處理技術。在此章節中，吾人設計出了匠心獨具的方法，並說明即便接收端沒有通道資訊，使用此方法仍然可以讓系統正常運作。

This dissertation aims to propose hardware-implementable solutions, with the aid of optimization techniques, for some open issues in emerging wireless communication systems. This dissertation is divided into three parts, and the covered issues include cyclic prefix (CP) recycling, circuit-power-aware communication algorithms, and (pre)-equalizers for emerging wireless communication systems.

In Part I, the CP recycling scheme is considered. In this part, I proposed the optimum CP reuse strategies in the presence of additive noise, Doppler spread, and phase noise, separately. Compared to conventional (heuristic) strategies, the proposed methods show evident improvement, while maintaining similar complexity.

As mobile equipments suffer from the insufficiency of battery capacity, system design taking circuit power consumption into consideration has been a trend in communication engineering. In Part II (including Chapter 4 and Chapter 5), circuit-power-aware communication algorithms were proposed. In beamforming applications, there is a tradeoff between diversity gain and circuit power consumption, i.e., activating more antennas results in more diversity gain but more circuit power consumption (contributed by mixers, filters, digital/analog-converters) in the meantime. In order to strike a balance between diversity gain and circuit power consumption and minimize the overall power consumption, Chapter 4 proposed a systematic algorithm that dynamically determines the set of active antennas and adjusts the beamforming weights. Afterwards, in Chapter 5, a power-loading algorithm with respect to energy-efficiency, which is evaluated by bit-per-Joule, is proposed. This method can be applied to uniform-rate orthogonal frequency division multiplexing (OFDM) systems to improve energy efficiency.

In Part III (from Chapter 6 to Chapter 8), low-complexity equalizers and precoders were proposed. In Chapter 6, a linear spatial-temporal precoder for time-dispersive broadcast channels is proposed, and the potential applications to millimeter-wave and MISO-based cellular communications were described. The proposed precoding method could be a promising candidate in the future development of fifth-generation cellular communications, where millimeter wave band is used and massive antennas are exploited. In Chapter 7, a low-complexity time-domain linear equalizer for single-carrier block transmission (SCBT) is proposed. Numerical results show that the proposed method can significantly reduce complexity under a tolerable performance degradation. Chapter 8 discusses the topic of MIMO transmitter and receiver diversity techniques and aims to design algorithms that can be applied to the following situation : `\emph{a MIMO system whose channel information is ONLY available at transmitter}'. In this chapter, I proposed sophisticated methods and showed that the system can perform well by using this method even if receiver does not have channel knowledge.