Precoding Techniques for Concurrently Improving PAPR and Spectrum Performance of OFDM Signals

[[abstract]]多重輸入/多重輸出正交多重載波(MIMO-OFDM)系統已經被視為提供多媒體、視訊、資料、語音、與高速網際網路接取等高品質服務 4G寬頻無線通訊與室內網路系統的主角。然而 OFDM 的主要缺點就是具有較高的峰均功率比值(PAPR)；當大量的同相位子載波加在一起時，OFDM 訊號就會造成很大的波封浮動，而具有很大波封浮動的 MIMO-正交多載波訊號會產生高峰均功率比 (PAPR)值。本研究專題將專注在對 OFDM 系統發展各種不同的前置編碼矩陣技術以及相關調整演算法來同時達成降減 OFDM訊號的 PAPR值以及改善 OFDM頻帶外圍功率頻譜密度的效能為目標。由於前置編碼技術具有控制區塊資料之間的相關性又不會破壞子載波之間正交性的優點，因此極適合依據不同的需求發展演算法來調整矩陣特性。對於計畫的執行，我們首先考量選擇相關編碼序列與固定振幅零自相關 (Constant Amplitude Zero Auto Correlation，CAZAC)序列兩種技術做為起點，相關編碼序列技術能夠改善 OFDM 訊號頻帶外圍的頻譜效能，但是造成訊號 PAPR值比一般 OFDM訊號的 PAPR值還要高。依據序列我們建構相對應的 OFDM 系統前置編碼矩陣，然後根據各種降減 OFDM 訊號PAPR 值的方法與理論分別發展最小-最大-調整演算法(Min-Max-Adaptation Algorithm)與組合演算法(Combined Algorithm)來修改前置編碼矩陣特性，使得前置編碼系統可以同時達到改善 OFDM訊號頻帶外圍的頻譜效能以及降減 PAPR值。 另一個方法是從既有的 OFDM訊號 PAPR值降減技術出發，降減 OFDM訊號 PAPR值的技術很多，主要可分為頻率領域技術與時間訊號技術兩類，頻率領域技術降減 PAPR 值是藉由增加反離散傅立葉轉換(Inverse Discrete Fourier Transform, IDFT)的輸入資料交相關性，降低 IDFT 輸出訊號的峰值，或是增加 IDFT 輸出訊號的平均值等。時間訊號技術降減 PAPR值是直接在訊號放大之前將它失真，降低發射機輸出訊號的峰值，或是藉由加入額外的訊號來增加訊號的平均值等等。基於我們的研究主題，我們選擇脈波整形訊號法、單式矩陣轉換法、與最佳前端濾波器與後端濾波器等三種方法，這些方法能有效率地區降低 OFDM 訊號的 PAPR 值但是會升高頻繁帶外圍的功率譜密度值。針對每種方法我們建立相對應的 OFDM 系統前置編碼矩陣，然後根據各種改善正交分頻多工訊號頻譜效能的理論來發展分解調整演算法(Decomposition-adaptation Algorithms)以修改 PAPR降減前置編碼矩陣特性，讓前置編碼系統可以同時達到 PAPR值降減以及頻帶外圍的功率譜密度改善的效能。。我們將撰寫 MATLAB 與 C/C++程式對本計畫案中的每種機制進行 Monte-Carlo 效能模擬，將分析模擬結果和其他方法所獲得的結果做比較。最後我們將對本研究的成果撰寫數篇學術論文發表，提供學界與業界參考，做為本計畫的總結。[[abstract]]The multi-input multi-output orthogonal frequency-division multiplexing (MIMO-OFDM) system is a potential candidate for 4G broadband wireless communication and indoor networks for high quality services in multimedia, video, data, voice, and high-speed internet access. However, one major drawback of OFDM is its high peak-to-average power ratio (PAPR). As a large number of subcarriers are added with the same phase, OFDM signals will have a large envelope fluctuation such that the MIMO-OFDM signals incur a large envelope fluctuation would have a high PAPR value. In this research project, we focus on the precoding techniques and related adaptation algorithms for concurrently improving out-of-band power spectral density and PAPR reduction performance of the OFDM signals. Since the precoding technique has the convenience advantage in controlling and adapting the correlation between the OFDM block samples without destroying the orthogonal property between the subcarriers, it was suitable for developing various algorithms to adjust the matrix for different requirements. To reach the goal, we first take the techniques like correlatively coded sequences and constant amplitude zero auto correlation sequences (CAZAC) for spectrum improvement of OFDM signals as the starting point. In generally, a correlative code sequence significantly improves the spectral performance of OFDM signals but results in much larger PAPR values than general OFDM signals. A procedure to construct the corresponding precoding matrix is developed. Then the Min-max-adaptation Algorithm and Combined Algorithm will be developed respectively for adjusting the precoding matrix for PAPR reduction and hence we obtain the precoding matrix able to simultaneously improve out-of-band spectrum and PAPR reduction performance of OFDM signals. On the other hand, we take the PAPR reduction techniques as the starting point. Many methods have been proposed to reduce the PAPR values. These techniques can be divided into two categories: frequency domain techniques and signal time domain techniques. Frequency domain techniques reduce PAPR by increasing the cross-correlation of the input data of the inverse discrete Fourier transform (IDFT), decreasing the peak values of output signal of IDFT, or increasing the average value of the output signals. Time domain techniques reduce PAPR directly by distorting the signal prior to amplification, decreasing the peak values of output signal of transmitter, or increasing the average power of the signals by adding extra signals. The methods of pulse shaping, unitary matrix transform, and optimum front-and-post-end filters for reducing the PAPR values of OFDM signals were selected for our purposes. The corresponding precoding matrix for each method is established. These methods can significantly reduce PAPR values of OFDM signals, however, accompanied by much higher out-of-band power spectral density. We then develop the Decomposition-adaptation Algorithms for adjusting the precoding matrix for PAPR reduction such that power spectral density and PAPR reduction performance of OFDM signals can be improved concurrently. Monte-Carlo simulations programmed in Matlab and C/C++ will be executed for various schemes proposed in this project. Simulations results will be analyzed and compared with those from other methods. Finally, the achievements obtained in this research will be presented in several articles as the conclusion of this research.[[note]]MOST103-2221-E327-01