DeepSense 6G Machine Learning Challenge

The DeepSense 6G Machine Learning Challenge is a student competition organized by the Student and Outreach Subcommittee (SOSC) of the IEEE Information Theory Society, in collaboration with the Wireless Intelligence Lab at Arizona State University (ASU). The competition is focused on developing innovative machine learning solutions for various applications using the DeepSense 6G dataset, which comprises coexisting multi-modal sensing and communication data, such as mmWave wireless communication, camera, GPS data, LiDAR, and Radar, collected in realistic wireless environments. The competition consists of three tasks of increasing difficulty that require participants to use different machine and deep learning modalities and techniques. The scenario considered is as follows: a remote radio head (RRH) is tasked with assisting a base station (BS) in communicating toward a user equipment (UE). The RRH obtains the channel state information (CSI) of the channel between the RRH and UE and wishes to communicate this estimate to the BS through a noiseless but rate-limited channel. The BS reconstructs CSI under a given MSE distortion, while also complementing this estimate through a datastream consisting of radar, lidar, GPS, and image data. This CSI estimate is then used in downstream tasks that are not considered further.

DeepSense 6G机器学习挑战赛是由国际电气和电子工程师协会（Institute of Electrical and Electronics Engineers, IEEE）信息论学会学生与推广分委员会（Student and Outreach Subcommittee, SOSC）联合亚利桑那州立大学（Arizona State University, ASU）无线智能实验室举办的学生竞赛。本次竞赛致力于利用DeepSense 6G数据集为各类应用开发创新性机器学习解决方案，该数据集采集自真实无线环境，包含共存的多模态感知与通信数据，例如毫米波无线通信、摄像头数据、GPS数据、激光雷达（LiDAR）及雷达数据。竞赛包含三个难度逐步递增的任务，要求参赛者运用各类机器学习与深度学习范式及技术。本次竞赛设定的场景如下：远程射频头（remote radio head, RRH）负责协助基站（base station, BS）与用户设备（user equipment, UE）开展通信。RRH获取自身与UE之间信道的信道状态信息（channel state information, CSI），并希望通过无噪但速率受限的信道将该估计值传输至BS。基站需在给定均方误差（Mean Squared Error, MSE）失真约束下重构CSI，同时通过包含雷达、激光雷达、GPS及图像数据的数据流对该估计值进行补充。该CSI估计值将被用于后续未进一步展开的任务中。

搜集汇总

背景与挑战

背景概述

该数据集是'DeepSense 6G机器学习挑战赛'的竞赛数据集，包含在真实无线环境中收集的多模态传感和通信数据（如毫米波通信、摄像头、GPS和雷达数据），用于开发机器学习解决方案。数据集用于三个难度递增的任务，主要涉及信道状态信息的估计和重建，但访问受限，仅限批准参与者使用。