適用於前瞻協同式多天線通訊系統之高效最佳化演算法

Abstract

為了滿足第五代（Fifth Generation, 5G）無線通訊系統中的高性能要求，並突破單一裝置在硬體與能量消耗方面的限制，協同式通訊（Cooperative Communication）技術中的低成本智能反射面板（Intelligent Reflecting Surface, IRS）以及放大傳輸中繼站（Amplify-and-Forward Relay, AF Relay）近年來逐漸受到關注。由於各類協同裝置的效能與成本相異，本論文分成兩部分深入探討智能反射面板與放大傳輸中繼站在各應用場景中的適用性，並提出對應之低複雜度演算法以最佳化系統配置，提升整體傳輸效能。 首先，智能反射面透過控制反射元件的相位來達成波束成形（Beamforming）的指向性優勢。為了聯合優化反射係數與預編碼以達到頻譜效率最大化，我們提出了一種基於最小均方誤差（Minimum Mean Square Error, MMSE）的低複雜度與低延遲的交替優化（Alternating Optimization, AO）演算法。此外，我們的MMSE-AO演算法可以根據估計的子通道進行更新，以便在實際應用中使用。為了增強由於雙重衰減效應而導致的智能反射面有限覆蓋範圍，我們將MMSE-AO擴展到室內外情境中的多智能反射面系統。 另一方面，為了解決便攜裝置在天線數量與功率上的限制，同時符合延展實境（Extended reality, XR）應用的高傳輸需求，本論文提出中繼輔助載波聚合（Relay-assisted Carrier Aggregation, RACA）系統，將訊號傳輸於高頻段並利用附近的裝置作為中繼站，從而將訊號轉換至低頻段後傳輸，提高系統的覆蓋範圍與效能。為最大化延展實境裝置的上行傳輸速率，我們提出基於加權最小均方誤差（Weighted MMSE, WMMSE）與基於奇異值分解（Singular Value Decomposition, SVD）之演算法於集中式與分散式傳輸協議，相較於傳統載波聚合系統能提升32%的傳輸速率。最後，針對傳輸速率增長隨著能耗增加逐漸趨緩的問題，我們提出基於Dinkelbach轉換之加權最小均方誤差（Dinkelbach’s-transformed WMMSE, DWMMSE）框架，專注於最大化能源效率的分式規劃（Fractional Programming, FP）問題而非單純提升傳輸速率，以利輕量裝置於延展實境應用中的長時間使用。

To meet the high-performance requirements of Fifth Generation (5G) wireless communication systems and overcome the limitations of hardware and energy consumption in individual devices, cooperative communication technologies such as low-cost intelligent reflecting surfaces (IRS) and amplify-and-forward relays (AF relay) have gained increasing attention in recent years. Due to the varying performance and costs of different cooperative devices, this dissertation is divided into two parts to explore the applicability of IRS and AF relay in various application scenarios and propose corresponding low-complexity algorithms to optimize system configurations and enhance overall performance. First, intelligent reflecting surfaces (IRS) achieve beamforming advantages by controlling the phase of reflecting elements to reconfigure wireless transmission environments. To jointly optimize the reflecting coefficients and precoder for maximum spectral efficiency, we propose a low-complexity and low-latency alternating optimization (AO) algorithm based on the minimum mean square error (MMSE) objective. In addition, our MMSE-AO can be updated based on estimated subchannels for practical usage. Furthermore, to enhance the limited coverage of IRS caused by the double-fading effect, we extend our MMSE-AO to multi-IRS systems in indoor and outdoor scenarios. On the other hand, to address the limitations of antenna number and power in portable devices while meeting the high transmission requirements of Extended Reality (XR) applications, we introduce a relay-assisted carrier aggregation (RACA) system, where signals are transmitted in the high-frequency band and relayed by nearby devices to convert the signals to the low-frequency band, thereby improving system coverage and performance. To maximize the uplink transmission rate of XR devices, we propose algorithms based on weighted minimum mean square error (WMMSE) and singular value decomposition (SVD) for centralized and distributed transmission protocols, which improve transmission rates by 32% compared to traditional carrier aggregation schemes. Finally, to address the issue where transmission rate growth gradually slows with increased energy consumption, we propose a Dinkelbach’s-transformed WMMSE (DWMMSE) framework, focusing on maximizing energy efficiency through fractional programming (FP) rather than simply increasing transmission rate, to support prolonged use of lightweight devices in XR applications.