A Reinforcement-Learning-Based Deployment Strategy for GPP Components
Peiwen Zhuang, Tao Xu, Shan Wang · 2021
This article presents a deployment strategy to optimally manage the computing resource in the Software Defined Radio (SDR) equipment. The proposed solution is a combination of Reinforcement Learning and the run-time resource data of the communication system. Two scenarios, based on the traditional method and Reinforcement Learning, are made through the SDR virtual platform. The simulation results show that the proposed method expands the new module for General Purpose Processor (GPP) resources management and improves the performance of Cognitive Radio (CR). The method innovatively optimizes component deployment in multi-processor platforms and adapts to the CR equipment with the new generation heterogeneous chips so that to implement more applications when resources are limited.