FedSEA

Athena – The NSF AI Institute for Edge Computing

“…To deal with contested environments, realistic threat models and security analyses for such scenarios have been introduced (e.g., Hallyburton et al., 2022) demonstrating the vulnerability of existing perception models). Further, to exploit on‐demand computation availability, we have been developing autonomous services capable of employing a combination of centralized (e.g., on an edge‐server) and decentralized data aggregation, as well as the development of secured FL models (e.g., Sun et al., 2022). Similarly, we have focused on the design of adversarially robust decision‐making policies.…”

A Node Selection Strategy in Space-Air-Ground Information Networks: A Double Deep Q-Network Based on the Federated Learning Training Method

“…To overcome challenges in the Metaverse's channel state and computing resources, a soft actor-critic-based solution has been developed for an efficient FL scheme with dynamic user selection, gradient quantization, and resource allocation [18]. Recognizing the limitations of asynchronous federated learning and semi-asynchronous federated learning methods, a new approach named FedSEA has been introduced as a semi-asynchronous FL framework tailored for extremely heterogeneous devices [19]. Additionally, in vehicular networking scenarios, studies have applied fuzzy logic for client selection, considering parameters such as the number and freshness of local samples, computational capability, and available network throughput [20].…”

“…a(t)∈A(s(t), a(t)) 10: end if 11: The edge server selects a device based on the action a(t), performs local model training on the selected device, and updates the global model. 12: Compute the instantaneous reward r(t) based on Formula(19) and update the state from s(t) to s(t+1).…”