A space synthetic aperture radar image processing software load balancing method and system

Abstract

This application provides a load balancing method and system for aerospace synthetic aperture radar image processing software. During parallel unwrapping processing, multiple sub-regions are divided as initial tasks and distributed to worker nodes. An iterative unwrapping algorithm is executed in the worker nodes to determine the phase residual of each pixel, extracting multiple unconverged residual regions. Each unconverged residual region is packaged into a residual sub-task and sent back to the master node. The master node dynamically reallocates tasks to idle nodes based on the load status of each worker node. The worker nodes iteratively solve the received residual sub-tasks and submit the solution back to the master node to update the global phase field. This process of determining the phase residual and updating the global phase field is repeated until the interferometric phase map meets the judgment conditions, achieving convergence of all pixels. Using this scheme, real-time load migration based on dynamic residual evolution can be achieved.

Images