Multitask Routing Design for Load-Balancing Satellite Networks

The benefit of low latency has led to growing interests in Space-Air-Ground (SAG) integrated networks. To provide better service, we need to consider how to schedule paths for many communication tasks concurrently, thus resulting in a complex optimization problem. Considering the potential benefits of evolutionary multitasking, a new routing framework is designed to draw upon excellent evolutionary experience and enhance the routing performance. In the proposed framework, the population of each task is divided into better-performing and worse-performing sub-populations. A random exchange strategy is designed for the former, while an inter-task knowledge transfer strategy is proposed for the latter with the aim for improving the population diversity. A series of results demonstrate that the proposed framework is an effective routing solution strategy. Moreover, it significantly outperforms baseline solvers and the shortest path strategy.