Multi-DSP平台應用於DOA與Beamforming之即時模擬系統

Abstract

訊號的抽取，對於陣列天線訊號的傳輸品質來說，是很重要的一個課題。到達角度估測〈DOA〉與波束形成〈Beam forming〉演算法可以有效提高陣列天線訊號接收之訊號雜訊比，得到較佳的通訊品質，但是需要龐大的運算量。本論文以三顆數位訊號處理器為模擬平台核心，藉由分析數位訊號處理器內部運算架構、MUSIC與ESPRIT演算法資料流量分析、個人電腦與數位訊號處理器之間資料快速傳遞方法，建構出適合陣列天線訊號前端處理之即時分散式運算平台。

Signal extraction is an important topic for the transmission quality of array signal processing. Direction Of Arrival and Beamforming algorithms improve the signal to noise ratio of antenna array signal, and provide better communication quality. However, it needs huge computation. This thesis takes three DSP as simulation platform kernel. By discussing the internal architecture of DSP, analyzing data flow of both MUSIC and ESPRIT algorithm, and fast data transmission technique between PC and DSP, a real time distributed computation platform suitable for the front-end of antenna array signal processing is realized.