Parallelization method of digital signal processing based on multi-core pipeline

Signal processing on multi-core Central Processing Unit (CPU) is one of the directions of software radio development. It is necessary to study the parallelization method of signal processing algorithm to make full use of multi-core CPU. Computational decomposition method is applicable to most highly serial signal processing algorithms, and its implementation method is multi-core pipeline. The real-time performance of multi-core pipeline and serial method is analyzed and compared, and a distributed adaptive thread synchronization method is studied. Combined with practical signal processing, the real-time performance of serial method and multi-core pipeline is tested, results show that throughput rate of multi-core pipeline is 2.1 times that of serial method, the processing power is greatly improved.