Tongxin xuebao (Dec 2020)
Energy-efficient resource allocation method in mobile edge network based on double deep Q-learning
Abstract
To improve the system energy efficiency in mobile edge networks, a resource allocation method based on double deep Q-learning(DDQL) for integration of communication, computing, storage resources was proposed for the downlink communication process under the network architecture of multiple tasks, end devices, edge gateways and edge servers.A resource allocation model was constructed, which took the minimization of average energy consumption of tasks as the optimization goal and set the constraints of task delay limits and communication, computing, and storage resource limits.According to the model characteristics, a suitable resource allocation model and method based on DDQL framework was proposed to make intelligent allocation decisions for communication and computing resources and allocate storage resources on demand.Simulation results show that the proposed DDQL-based solution can effectively solve the multi-task resource allocation problem with good converge and low time complexity, and it reduces the average energy consumption of tasks by at least 5% compared with the solving methods based on random algorithm, greedy algorithm, particle swarm optimization algorithm and deep Q-learning while ensuring the quality of service.
