Research on large matrix transpose based on multi core DSP in real time SAR imaging

The radar echo data has a large amount of data and low efficiency during matrix transposition, and in reality, the processor still faces the problem of limited storage space. SAR real-time imaging frequently uses matrix transpose operations. In order to perfectly solve these problems, this article creatively proposes a method that can achieve fast and efficient transposition of large matrices within the limited storage space of the processor, improving the real-time performance of SAR imaging. Based on the multi-core parallel characteristics of DSP, our method preliminarily divides a large two-dimensional matrix into blocks, and then finely divides the small matrices into blocks. Invert the unit matrices on the diagonal and non diagonal lines respectively, and finally combine the transposed unit matrices to achieve fast transposition of large-scale matrices in real-time SAR imaging systems. The experimental results show that this method greatly improves the transposition efficiency of large matrices in SAR imaging and reduces the time consumption. Moreover, the larger the amount of data in the matrix, the higher the ratio of reduced time consumption, and it has great engineering application value.