二次時(shí)頻分布(TFD)是一種常用的非平穩(wěn)信號(hào)分析方法，以最常用的Wigner-Ville 分布(WVD)為例，分析了離散時(shí)頻分布算法實(shí)現(xiàn)，針對(duì)該算法的復(fù)雜性和計(jì)算量大的特點(diǎn)，實(shí)現(xiàn)了離散信號(hào)時(shí)頻分布的并行算法，通過(guò)采用緩沖池和計(jì)算結(jié)果的壓縮傳輸，同時(shí)減小了并行計(jì)算中傳輸數(shù)據(jù)總量和傳輸次數(shù)，將通信開(kāi)銷(xiāo)從O(N2 )降低到O(N)，解決了并行時(shí)頻分布計(jì)算中數(shù)據(jù)傳輸“瓶頸”問(wèn)題。
關(guān)鍵詞：時(shí)頻分布；并行算法；內(nèi)存管理；
Abstract: Quadratic Time-Frequency Distribution (TFD) is common analysis method for non-stationary signal. Discrete TFD is analyzed in the example of Wigner-Ville Distribution. In view of the complexity and calculation amount of TFD algorithm, parallel WVD algorithm is realized. The buffer-pool and compressing data translation are used to decrease quantity and times of data translation. Communication spending is decreased from O(N2 ) toO(N) . The bottle-neck problem of data translation is settled by the method.
Keywords：Parallel algorithm, Time-frequency distribution, memory manage,