透過功率放大器的非線性特性在多路徑通道下辨識使用者

Abstract

這篇論文中，我們利用功率放大器(power amplifier, PA)的獨立非線性(nonlinearity)特性提出一個精確的辨識系統，並設計重覆傳送的資料來得到接收訊號自相關函數(auto-correlation)的互補(complementary)特性，並利用此特性來解決多路徑通道對接收訊號的影響而無法準確辨識的問題。放大器是傳送端最關鍵的元件，不同的使用者使用的放大器有獨特的非線性特性，因此可透過此特性來達到在無線通訊系統辨識出犯罪的匿名使用者。首先，我們使用兩階段(two-box)的放大器模型，同時可以表現出非線性特性和記憶性(memory)；接著通過白高斯雜訊(additive white Gaussian noise, AWGN)通道，Rayleigh多路徑衰減通道和特定的多路徑通道: ITU-R，COS 207的通道模型，並驗證接收訊號的互補特性。藉由接收訊號自相關函數的互補特性，我們可以得到理論上和多路徑通道響應無關的辨識方法，並在多路徑通道型式下皆可達到較佳的辨識效果。隨後，使用遞迴最小平方法(recursive least square, RLS)估測出放大器的非線性參數，並使用likelihood ratio test (LRT) 解決假設檢定的問題，並達到辨識與偵測的結果。最後，除了分析提出方法的效能，也同時會和既有的論文方法做比較。

We proposed an accurate identification system based on the unique nonlinearity of power amplifiers and design the repeatedly transmitted signal to obtain complementary property of separated-by-N auto-correlation of received signal to cope with the multipath effects. The most critical components at transmitters are power amplifiers, which are distinctive for different users. Therefore, applying the individual characteristic to recognize transmitters can break criminals’ anonymity in wireless system. First, we exploit the two-box behavioral power amplifier model, which can correctly exhibit both nonlinearity and memory effects of power amplifiers. Next, the additive white Gaussian noise channel, multipath Rayleigh fading channel, and specific multipath channels: ITU-R and COS 207 channel models will be presented. Then, the complementary property of separated-by-N auto-correlation will be manifested, as well as the independence between error probability of identification and multipath effects will be verified by this property. At last, we will apply recursive least square algorithm for estimation of PA nonlinearity and likelihood ratio test for identification of transmitters. Moreover, we will illustrate the performance and compare to existed paper via simulations by MATLAB.