Power Allocation in Multi-Agent Networks via Dueling DQN Approach
Zhihao Xuan, Guiyi Wei, Zhengwei Ni · 2021 IEEE 6th International Conference on Signal and Image Processing (ICSIP) · 2021
As an emerging computing technology, edge computing transfers computing power from the cloud to the edge of the network, which greatly improves the quality of service, but different edge devices have signal interference during the communication process, which will affect the downlink rate. However, the power allocation of the edge server can effectively reduce the influence of interference. In this paper, we take channel changes in the actual scenario into account, intending to maximize the downlink rate, propose a power allocation algorithm based on Dueling DQN under the multi-cell edge computing model. We have carefully designed Dueling DQN to maximize the downlink rate, the algorithm uses centralized training and distributed execution, the executor does not need to get complete network information, so it has better robustness. Finally, we compare with five baseline algorithms including the " DQN " , "FP" and "WMMSE" algorithms in various situations. The results of the experiment show that the proposed algorithm is better than the other five baseline algorithms.