Participant Selection for Hierarchical Federated Learning in Edge Clouds
Participant Selection for Hierarchical Federated Learning in Edge Clouds
复制标题
DOI：
10.1109/nas55553.2022.9925313
复制
发表时间：
2022-10
期刊：
影响因子：
--
通讯作者：
Xinliang Wei;Jiyao Liu;Xinghua Shi;Yu Wang
中科院分区：
文献类型：
--
作者：
Xinliang Wei;Jiyao Liu;Xinghua Shi;Yu Wang
联邦学习（Federated Learning，FL）是近年来出现的一种新的分布式机器学习模式。虽然FL可以通过将参与者的训练数据保存在本地设备上来保护参与者的数据隐私，但最近的一些作品引起了新的隐私问题，特别是当FL的工作人员或参数服务器不值得信任或恶意时。解决这个问题的一种有效方法是使用分层联邦学习（HFL），其中使用一些中间层聚合器（或称为组领导者）来聚合来自工作者的本地模型更新并将组模型更新发送到参数服务器。本文研究了边缘云环境下多个FL模型的HFL参与者选择问题，其中每个模型需要从边缘服务器中选择一个参数服务器、几个组长和一定数量的工作者来共同执行HFL.我们首先将这个问题表示为一个非线性整数规划问题，目标是在满足边资源约束的情况下，最小化所有模型的总学习成本。然后，我们设计了一个三阶段的算法，将原问题分解为三个子问题，并迭代求解。通过对真实数据集和FL模型的仿真实验表明，与现有方法相比，该算法能够有效地降低边缘云的平均总学习代价.
Federated learning (FL) has been emerging as a new distributed machine learning paradigm recently. Although FL can protect the data privacy of participants by keeping their training data on local devices, there are recent works raising new privacy concerns especially when workers or the parameter server of FL are untrustworthy or malicious. One effective way to solve the problem is using hierarchical federated learning (HFL) where a few middle-layer aggregators (or called group leaders) are used to aggregate local model updates from workers and send group model updates to the parameter server. In this paper, we consider the participant selection problem of HFL in an edge cloud with multiple FL models, where each model needs to select one parameter server, a few group leaders and a certain amount of workers from edge servers to jointly perform HFL. We first formulate this problem as a non-linear integer programming, aiming to minimize the total learning cost of all models while satisfying the constrained edge resources. We then design a three-stage algorithm by decoupling the original problem into three sub-problems and solving them iteratively. Simulations with real-world datasets and FL models confirm that our proposed algorithm can efficiently reduce the average total learning cost in edge cloud compared with existing methods.