Paper Info
Title
Cybertwin Assisted Wireless Asynchronous Federated Learning Mechanism for Edge Computing
Abstract
The significant advances in wireless communication together with edge intelligent (EI) technology have facilitated the decentralized edge computing paradigm for data-intensive and delay-sensitive solution on massive Internet of Things (IoT) devices. In this paper, a Cybertwin assisted asynchronous federated learning (AFL) mechanism is proposed for realizing efficient edge computing by taking full advantage of local computation capability under heterogeneous wireless environment. First, Cybertwin is introduced as intermediary communication assistant to coordinate individual model aggregation between the users and the cloud server under AFL training process. Second, for the sake of flexible and effective utilization of communication-computation resources for edge computing, Cybertwin plays the role of intelligent agent to jointly take the local computing and up-link transmission into consideration. A resource optimization problem considering the diversified computing power, varied data size, and available communication bandwidth is formulated and we leverage the block coordinate descent (BCD) method to obtain optimal resource management solution. Extensive simulations are conducted to demonstrate the effectiveness of our proposed Cybertwin assisted AFL mechanism, which can shed further light on the application of data-intensive edge computing paradigm over wireless communication network.
Year
DOI
Venue
2021
10.1109/GLOBECOM46510.2021.9685076
2021 IEEE Global Communications Conference (GLOBECOM)
Keywords
DocType
ISSN
Edge Intelligence,Asynchronous Federated Learning,Cybertwin,Resource Allocation
Conference
2334-0983
ISBN
Citations 
PageRank 
978-1-7281-8105-9
0
0.34
References 
Authors
0
5
Name
Order
Citations
PageRank
Yunting Xu152.44
Haibo Zhou220314.10
Jiacheng Chen3597.11
Ting Ma4112.51
Xuemin Shen515389928.67