在非理想環境下具有強健性的寬頻可適性波束成型技術

Abstract

直至今日,在通訊系統中,可適性天線陣列一直佔據了一個很重要的位置。而可適性天線陣列中在寬頻天線陣列的波束成型器,像是DFT架構[1]的波束成型器,在接收端對於接收寬頻信號方面提供了不錯的效能,它能抑制干擾以及雜訊的影響,使得陣列有不錯的輸出功率。但是在現實環境中,由於一些非理想的因素,會造成可適性天線陣列的效能大大的降低。在此篇論文中,我們考慮了一些非理想的因素,包括在接收端指引向量的誤差,陣列間的位置擾動,以及受到同調干擾的影響,並致力於在寬頻DFT架構[1]的波束成型器下一項一項解決此三種非理想因素所造成的影響。 在此篇論文中,對於修正接收端指引向量的誤差,陣列間的位置擾動的兩種非理想情形,我們主要用的概念是雜訊子空間投影[8]法,它包括了對於頻譜密度的特徵分解還有一套找出最佳化解的方法來疊代修正我們所想要尋求的天線陣列向量解,最後實驗模擬時,將會將其性能和使用變異負載法(Variable Loading)[5]的性能做比較。爲了對抗同調干擾的影響,首先,我們先估計出同調干擾的入射角度[12],再來使用DMV[11]以及SMV[11]兩種方法來找出最佳化的陣列權重值,最後,比較DMV以及SMV兩者的性能。

To date, adaptive antenna array has played an important role in communication systems. Broadband beamformer , such as the DFT beamformer, supplies good properties of receiving broadband signals, and it can suppress interference and noise to get a high “Output SINR ”. But, the performance of the broadband beamformer may be highly degraded if there are some non-ideal factors .The mismatch between the directional vectors of the signal and the steering vectors in the receiver, array position perturbations and the coherent interference effect are the non-ideal factors that we consider in the thesis .We concentrate on amending the influence for the non-ideal factors one after another with the DFT beamformer . In this thesis, we mainly use noise-subspace projection method including the eigen-decomposion of spectral density matrix and a gradient method to iteratively renew to correct the problems of the mismatch between the directional vectors of the signal and the steering vectors in the receiver as well as the array position perturbations. Finally, we compare the results by using the noise-subspace projection method and variable loading. To confront the coherent interference, first, we exploit the estimates of coherent interference incident directions, and then use the information to transfer received data to another space so as to eliminate the coherent interference. And then two methods, namely DMV and SMV, are used to find the optimal weight vectors, respectively according to LCMV criterion in the transform domain. Ultimately, we compare the performance between DMV and SMV.