多重信号分类（MUSIC）算法作为现代信号处理的重要技术,在波达方向定位（DOA）等领域有显著应用潜力,其核心是利用信号子空间与噪声子空间的正交性实现DOA等参数的高精度估计。基于国微DSP芯片,完成1024点采样、8至32天线规模的MUSIC算法设计,实现算法全流程。通过设计复矩阵乘法、共轭转置等加速算子及粗/细搜索策略,国微DSP在8天线至32天线场景下的处理耗时均显著优于TMS320C6678,验证了算法的高效性与硬件适配性。

The Multiple Signal Classification (MUSIC) algorithm, as a pivotal technology in modern signal processing, exhibits significant application potential in fields such as Direction of Arrival (DOA) estimation. Its core principle lies in leveraging the orthogonality between the signal subspace and the noise subspace to achieve high-precision estimation of parameters such as DOA. This paper presents the design of a MUSIC algorithm based on the SSMEC DSP chip, accomplishing the full algorithmic process for 1024-point sampling and antenna arrays ranging from 8 to 32 elements. By designing accelerated operators for complex matrix multiplication, conjugate transpose, and implementing coarse/fine search strategies, the SSMEC DSP demonstrates a significantly lower processing time compared to the TMS320C6678 across scenarios involving 8 to 32 antenna elements, validating the algorithm's high efficiency and hardware adaptability.