@article{discovery10190184, note = {This version is the author accepted manuscript. For information on re-use, please refer to the publisher's terms and conditions.}, pages = {1--1}, publisher = {Institute of Electrical and Electronics Engineers (IEEE)}, journal = {IEEE Network}, title = {Federated Learning Driven Sparse Code Multiple Access in V2X Communications}, year = {2024}, month = {March}, issn = {0890-8044}, keywords = {Federated learning, sparse code multiple access (SCMA), reconfigurable intelligent surfaces, vehicular communication}, author = {Chen, Zhen and Zhang, Xiu Yin and So, Daniel KC and Wong, Kai-Kit and Chae, Chan-Byoung and Wang, Jiangzhou}, url = {http://dx.doi.org/10.1109/mnet.2024.3375935}, abstract = {Sparse code multiple access (SCMA) is one of the competitive non-orthogonal multiple access techniques for the next generation multiple access systems. One of the main challenges is high computational complexity and the SCMA-aided codewords, that is, each terminal device maintains its local data and codewords, which provides no incentive for model updating to accommodate rapidly changing vehicle communication environment. Federated learning (FL) proves its effectiveness by enabling terminals to collaboratively train their local neural network models with private data while protecting the individual SCMA-aided codewords. To select reliable and trusted codewords, this article provides an overview of the salient characteristics of the application of federated learning-driven SCMA for vehicular communication and discusses its fundamental research challenges. Furthermore, we outline the advancement of federated learning-driven SCMA schemes and present a general framework with potential solutions to the challenges. Finally, several future research directions and open issues are discussed regarding federated learning-driven SCMA schemes.} }