Intelligent edge computing unloading algorithm for space-based LEO satellite network based on GA-DDPG

In LEO satellite network, the high operating speed and short operating cycle of low earth orbit satellites lead to dynamic changes of inter satellite links. In order to timely sense the topology of inter satellite links, select the optimal task offload strategy with low delay and low energy consumption and make full use of satellite computing resources, a GA-DDPG based intelligent sensing edge computing offload algorithm for space-based low earth orbit satellite networks is proposed, The algorithm is divided into two phases: inter satellite link awareness and DDPG based edge computing task offloading for space-based LEO satellite networks, and a framework for modifying inter satellite link topology is established. This method overcomes the problems of poor real-time unloading of edge computing of space-based LEO satellite network, high computational complexity of genetic algorithm and easy to fall into local optimization, as well as the problems of long credit allocation time span, scarce returns, lack of diversity exploration and fragile convergence of deep deterministic strategic gradient algorithm.