A new task offloading algorithm in edge computing

Abstract In the last few years, Internet of Things (IOT), as a new disruptive technology, has gradually changed world. With prosperous development mobile and rapid growth Things, various applications continue to emerge, such payment, face recognition, wearable devices, driverless, VR/AR, etc. Although computing power terminals is getting higher traditional cloud model power, it often accompanied by latency cannot meet needs users. order reduce user delay improve experience, at same time network load certain extent, edge computing, an application IOT, came into being. view architecture after dating this paper focuses on task offloading in from migration multi-user scenarios server resource management expansion, proposes multi-agent balancing distribution based deep reinforcement learning DTOMALB, distributed allocation algorithm, can perform reasonable offload method for scenario experience balance utilization. Simulations show that algorithm adaptability compared single cell, reduces complexity centralized average response overall user. And each server, robustness scalability system.