标题： 联合通信和计算设计的能效联邦边缘学习 显示英文标题

标题： Energy-Efficient Federated Edge Learning with Joint Communication and Computation Design

摘要： 本文研究了一种联邦边缘学习系统，其中边缘服务器协调一组边缘设备，基于它们各自的本地分布式数据样本训练共享的机器学习模型。 在分布式训练过程中，我们利用通信与计算联合设计以提高系统的能量效率，在其中全局机器学习参数聚合的通信资源分配和本地更新机器学习参数的计算资源分配被联合优化。 特别是，我们考虑了两种传输协议供边缘设备上传机器学习参数到边缘服务器，分别基于非正交多址接入和时分多址接入。 在两种协议下，通过联合优化边缘设备上传机器学习参数的发送功率和速率以及其中央处理单元频率用于本地更新，在特定有限的训练时间内满足给定的训练精度，从而最小化所有边缘设备的总能耗。 我们提出了高效的算法，利用凸优化技术最优地解决所提出的能耗最小化问题。 数值结果显示，与其他基准方案相比，我们提出的通信与计算联合设计显著提高了联邦边缘学习系统的能量效率，通过在通信和计算之间的能量权衡得到了合理平衡。 显示英文摘要

摘要： This paper studies a federated edge learning system, in which an edge server coordinates a set of edge devices to train a shared machine learning model based on their locally distributed data samples. During the distributed training, we exploit the joint communication and computation design for improving the system energy efficiency, in which both the communication resource allocation for global ML parameters aggregation and the computation resource allocation for locally updating MLparameters are jointly optimized. In particular, we consider two transmission protocols for edge devices to upload ML parameters to edge server, based on the non orthogonal multiple access and time division multiple access, respectively. Under both protocols, we minimize the total energy consumption at all edge devices over a particular finite training duration subject to a given training accuracy, by jointly optimizing the transmission power and rates at edge devices for uploading MLparameters and their central processing unit frequencies for local update. We propose efficient algorithms to optimally solve the formulated energy minimization problems by using the techniques from convex optimization. Numerical results show that as compared to other benchmark schemes, our proposed joint communication and computation design significantly improves the energy efficiency of the federated edge learning system, by properly balancing the energy tradeoff between communication and computation.