面向协同感知的高效通信边缘学习网络架构设计

面向协同感知的高效通信边缘学习网络架构设计

张泽中¹，刘沛西²，朱光旭³

（1. 香港中文大学（深圳），中国 深圳 518172；2. 北京大学电子学院，中国 北京100871；3. 深圳市大数据研究院，中国 深圳 518172）

摘要：针对多场景多设备与单场景多视角两类代表性的协作感知场景，提出了针对性的基于联邦学习的协作感知学习框架，并针对无线通信场景下，通信与感知资源有限，以及信道具有随机性等挑战，提出了相对应的资源分配方案与用户调度策略。保证了提出的学习算法能够在无线网络下实现高质量且稳定的协同感知，并通过仿真验证并分析了算法的正确性与有效性。

关键词：通信感知一体化；联邦学习；资源分配；用户调度

Communication-Efficient Edge Learning Architecture Designs for Cooperative Sensing

ZHANG Zezhong¹, LIU Peixi², ZHU Guangxu³

(1.The Chinese University of Hong Kong (Shenzhen), Shenzhen 518172, China；2.School of Electronics, Peking University, Beijing 100871, China; 3.Shenzhen Institute of Big Data, Shenzhen 518172, China)

Abstract: Two typical cooperative sensing scenarios, including multiple-scene-multiple-device and single-scene-multiple-view are considered, and specific federated learning architectures for the two cooperative sensing scenarios are proposed. Moreover, under the situation where the communication and sensing resources are limited and the channel conditions are stochastic, a resource allocation scheme and a user scheduling policy for the two learning architectures are respectively proposed. It shows that the proposed algorithms can guarantee high-quality and robust cooperative sensing, and verify the effectiveness of the algorithms through simulations.

Key words: integrated sensing and communication; federated learning; resource allocation; user scheduling