Multi-Classification and Distributed Reinforcement Learning-Based Inspection Swarm Offloading Strategy

In meteorological and electric power Internet of Things scenarios, in order to extend the service life relevant facilities reduce cost emergency repair, intelligent inspection swarm is introduced cooperate with monitoring tasks, which collect process current scene data through a variety sensors cameras, complete tasks such as handling fault inspection. Due limitation computing resources battery patrol equipment, it will cause problems slow response long time for location. Mobile Edge Computing promising technology, can improve quality by offloading task equipment edge server nearby network. this paper, we study problem multi-devices multi-tasks multi-servers under condition dynamic network environment limited servers equipment. An effective adaptive learning strategy based on distributed reinforcement multi-classification proposed processing delay energy consumption quality. Numerical experimental results demonstrate that superior other strategies terms delay, service.