本論文首先介紹了幾種快速傅立葉轉換的演算法,包括 radix-2 , radix-4, radix-22 , split-radix2/4 ,本篇利用記憶體式架構來實現,其演算 法採用radix-2 , 並進ㄧ步的用radix-22的演算法,來增加運算的速度。 主要的研究,是利用計數器,來當作記憶體位置和多工器控制訊號的依據, 以減低控制訊號的複雜度,並用single-port memories 來取代 dual-port memories 以減少硬體的消耗, radix-2 , radix-22 ,split-radix2/4這三種演算 法中,radix-2演算法,其硬體的設計是最簡單的,但效率最差,而split-radix2/4 效率雖然比較好,不過卻會有無法產生規律性的結構的缺點, radix-22 儘管他 的控制訊號較為複雜,不過卻有極佳的效率,且具有規律性,所依我們先探討 radix-2演算法的方式,再利用類似的控制方式,推展到radix-22演算法。 此快速傅立葉轉換處理器以Verilog 硬體描述語言及ModelSim 模擬其結 果,並和Matlab 所運算的結果作比較,以驗證這種演算法的可行性。
The FFT algorithms considered here include radix-2, radix-4, radix-22 and, split-radix2/4, which is the subject of this study. We use memory-based architecture to Design and implementation, and choose the algorithm, radix-2. Then we improve it’s efficience with the radix-22 algorithm. Taking the advantages of low hardware cost of memory based FFT (MBFFT) architectures , this study improves the speed perfor- mance. The improvement can be achieved by an efficient memory retrieval scheme for reducing the control complexity and a clocksche- me. Instead of using dual-port memory for data storage and retrieval, our design uses single-port memory. The FFT processor has been implemented by using verilog and ModelSim for circuit design and simulation, respectively.