Virtual single-node processing for SAR imaging based on multi-DSP

With the development of SAR technology, the SAR imaging processing data capability and computation complexity increases greatly. According to the situation, we design a virtual single node (VSN) parallel processing method, which uses 8 TMS320C6678 DSPs as a big processor to realize the large capability distributed storage and short time delay of SAR imaging processing. This method designs a remote reading and writing time beat of range line and azimuth line in distributed DDR to improve the remote transmission bandwidth between DSPs, and a data layout optimization approach in local DDR to achieve a balanced two-dimensional matrix local storage bandwidth. Finally, this paper realizes the SAR imaging algorithm based on the VSN method using 8 C6678s.