即時DSP嵌入式Linux系統開發及其於腦波訊號處理之應用

Abstract

在人們的生活中，嵌入式系統應用於訊號處理上十分廣泛，如多媒體語音與影像資訊處理等。為了讓訊號得到較佳的處理，往往需要大量的數學運算，但有鑑於嵌入式系統在運算能力不足，所以本論文採用同時具有ARM和DSP微處理器之OMAP來彌補這個問題。OMAP是由TI針對嵌入式系統所開發的一個具有優秀運算能力的微處理器，使用OMAP架構可以使數學運算上更為迅速有效率，且ARM和DSP可以同步處理各自不同的行程。 本論文有三大發展主軸，分別為「OMAP系統環境」、「OMAP內部通訊」及應用於「腦波訊號處理系統」，主旨在於呈現如何在OMAP下有效運用其ARM和DSP的資源。其中OMAP運用以含有DSP模組的DSPLinux做為其系統環境，並利用DSP Gateway讓OMAP中ARM和DSP部分能相互溝通。本論文以DSP Gateway的機制將ICA演算法交由ARM和DSP分工處理，並進行系統分析與設計到落實系統設計，建構一個腦波訊號處理系統，經由腦波訊號處理系統中ICA演算法將訊號做實際的特徵擷取。此外，本論文亦分別對OMAP和ARM處理器做系統效能測試比較，由測試結果中可明白顯示出OMAP高效能的數學運算能力。

The application of embedded system in signal processing is very wide in our life, such as Multimedia information processing on speech and image, and so on. In order to get better signal processing results, it always needs a large amount of computation. Because the capability of embedded system to compute is not enough, we use a microprocessor OMAP which has both ARM and DSP microprocessors to solve this problem. OMAP is a outstanding computational ability microprocessor which TI develop on embedded system. It is more efficient to compute by using OMAP, and ARM and DSP can synchronize to handle differenct processes. In this thesis, we build DSPLinux OS on OMAP platform and utilize the resource of ARM and DSP effectively by DSP Gateway. DSP Gateway is a interface which make ARM and DSP communicate in OMAP. The thesis accomplish ICA algorithm which ARM and DSP handle cooperatively by DSP Gateway mechanism. It includes system analysis, design and realization. Then, we apply OMAP system to brain signal processing. By the ICA algorithm, we can successfully get signal feature from signals. Besides, we also proceed some tests of the system efficiency on OMAP and ARM architectures, and compare them. Finally, from the experimental results, we can find that OMAP have better computation ability than ARM.