正交分頻多工系統中Clipping Noise之消除

Abstract

由於在正交分頻多工(OFDM)系統中,過高的時域訊號峰值會造成傳送訊號直流偏壓過高,使其增加傳送功率之消耗.我們採以限制訊號振幅的方式來達到降低訊號峰值之目的.但這會引進額外的雜訊(Clipping Noise),進而嚴重影響系統的表現.我們考慮以下三種消除此雜訊的方法.分別為在傳送端利用超頻取樣加上低通濾波器的方式來降低雜訊;以及在接收端利用decision-aided reconstruction(DAR)或clipping noise estimation and cancellation(CNEC)兩種演算法來還原訊號.一般而言, DAR與CNEC在演算過程中我們均假設Clipping Ratio(CR)為已知.但此假設未必適用於所有的應用.因此我們亦提出一個可以在接收端準確估計CR的方法.此外,我們也探討為了減少接收機中類比數位轉換器的位元數或是動態範圍所等效造成Clipping的效應,我們可將DAR與CNEC稍作修改以符合需求.另外,我們建議在處理接收端產生的Clipping Noise時,使用遞迴方式估計通道來以達到提昇DAR與CNEC的效果.最後透過模擬結果顯示以上提出之方法均能有效壓抑Clipping Noise對系統的影響以及正確估計出我們需要的參數.

Orthogonal Frequency Division Multiplexing (OFDM) signals exhibit a large peak to average power ratio (PAPR), which leads to a high amplifier power consumption. Signal clipping is known to be the simplest method to solve the problem. However, clipping will introduce clipping noise and this will adversely affect the system performance. In this thesis, we first study clipping noise mitigation methods. We consider three methods, namely, filtering oversampled clipping noise, decision-aided reconstruction (DAR), and clipping noise estimation and cancellation (CNEC). Conventionally, the clipping ratio, an important parameter for DAR and CNEC, is assumed to be known. However, this may not be possible for all applications. We then propose a new clipping ratio estimation method for DAR and CNEC. In addition to signal clipping at transmitter, we also consider the clipping in the receiver. The advantage of receiver clipping is that the number of required analog to digital conversion (ADC) bits can be reduced. We modify the DAR and CNEC algorithms such that they can be operated in this scenario. To further enhance the performance, we suggest an iterative channel estimation scheme for DAR and CNEC. Simulations show that clipping noise can be effectively reduced using proposed algorithms.