FlAg SwaP

Flag Swap is a project developed to optimize the aggregation placement in a hierarchical model of SDFL. The project exclusively employs Particle Swarm Intteligence (PSO) as a black-box meta-heuristic optimization algorithm to progressively optimize the placement as the FL rounds progress. Flag Swap’s simulation has been developed by Mr. Sadra Bekrani, and is available here on github. Flag Swap is also integrated into SDFLMQ source code, and can be found in the optimizers directory. A paper is written that explains the structure of the optimizer, and some early findings on the superior performance of the optimizer compared to conventional determinstic placement strategies. It will soon be publicized here.