In this paper, the deployment of federated learning (FL) is investigated in an energy harvesting wireless network in which the base station (BS) employs massive multiple-input multiple-output (MIMO) to serve a set of users powered by independent energy harvesting sources. Since a certain number of users may not be able to participate in FL due to the interference and energy constraints, a joint energy management and user scheduling problem in FL over wireless systems is formulated. This problem is formulated as an optimization problem whose goal is to minimize the FL training loss via optimizing user scheduling. To find how the factors such as transmit power and number of scheduled users affect the training loss, the convergence rate of the FL algorithm is first analyzed. Given this analytical result, the user scheduling and energy management optimization problem can be decomposed, simplified, and solved. Further, the system model is extended by considering multiple BSs. Hence, a joint user association and scheduling problem in FL over wireless systems is studied. The optimal user association problem is solved using the branch-and-bound technique. Simulation results show that the proposed user scheduling and user association algorithm can reduce training loss compared to a standard FL algorithm.
翻译:在本文中,在能源收获无线网络中,对联合学习(FL)的部署进行了调查,基础站(BS)使用大量的多投入多重产出(MIMO),为独立能源收获源驱动的一组用户提供服务;由于受到干扰和能源限制,一定数量的用户可能无法参加FL,因此在FL无线系统中形成了联合能源管理和用户排期问题;这个问题是一个优化问题,目的是通过优化用户排期,最大限度地减少FL培训损失;为了找到传输电力和预定用户数目等因素如何影响培训损失,首先分析FL算法的趋同率。鉴于这一分析结果,用户排期安排和能源管理优化问题可以被解析、简化和解决。此外,系统模型通过考虑多个BSs得到扩展。因此,对FL无线系统的联合用户联系和排期问题进行了研究。最佳用户联系问题通过分包技术得到解决。模拟结果显示,拟议的用户排期和用户联算法可以比标准的FL算法减少培训损失。