面向智慧交通系统的语义通信框架  

The Semantic Communication Framework for Intelligent Transportation System

作　　者：傅宇舟 程文驰[1] 王婧青 FU Yuzhou;CHENG Wenchi;WANG Jingqing(School of Telecommunications Engineering,Xidian University,Xi'an 710071,China)

机构地区：[1]西安电子科技大学通信工程学院,陕西西安710071

出　　处：《移动通信》2024年第2期41-46,共6页Mobile Communications

基　　金：国家重点研发计划“灾害复杂现场空中探测载荷自组网和高效信号传输技术装备研制”(2021YFC3002102);陕西省重点研发计划“大规模集群无人机多模态组网及网络化抗干扰技术”(2022ZDLGY05-09);国家自然科学基金“立体空间甚高频阵列波束构造与容量优化方法”(62341132)。

摘　　要：智慧交通系统是物联网的重要应用领域之一,其依靠先进的信息处理和无线通信技术实时回传道路图像数据,以实现各类机器任务辅助城市交通的智能化管理。面向任务的语义通信可将机器任务需求融合进信息处理中,有望成为物联网的新通信范式,并有效地支持智慧交通系统的无线通信需求。然而,现有的语义通信框架以图像的高质量重建为目标,不能紧密地耦合下游机器任务,以实现高效的面向任务通信。因此,提出了一种面向任务的语义通信框架,该框架联合基于神经网络的语义编码和传统的数字化信道编码,以实现数字化的语义传输。此外,所提的语义通信框架基于自适应编码控制器,可根据下游任务需求和当前信道条件进行自适应联合语义-信道编码,从而实现面向任务的高效语义传输。仿真分析表明,相比传统无线通信框架,所提的语义通信框架在不同的信噪比条件下,均能更好地完成智慧交通系统中的下游机器任务。Intelligent Transportation System(ITS)is one of the important applications of the Internet of Things.It relies on advanced information processing and wireless communication technology to transmit road image data in real time to realize intelligent traffic management assisted by various machine tasks.Task-oriented semantic communication can integrate machine task requirements into information processing,and it is expected to become a new communication paradigm for the Internet of Things(loT),thus supporting the requirements of wireless communications in ITS.However,the existing semantic communication frameworks aim at high-quality image reconstruction and thus fail to tightly couple downstream machine tasks to achieve efficient task-oriented communications.Therefore,a task-oriented semantic communication framework is proposed that combines neural network-based semantic coding and traditional digital channel coding,thus achieving digital semantic transmission.In addition,based on the adaptive coding controller,the proposed semantic communication framework performs joint semantic-channel coding according to the downstream task requirements and current channel conditions,thus achieving efficient task-oriented semantic transmission.Simulation analysis shows that the proposed semantic communication framework can better complete various downstream tasks in ITS than traditional wireless communication framework under different signal-to-noise ratio conditions.

关 键 词：语义通信 智慧交通系统 神经网络 

分 类 号：TN929.5[电子电信—通信与信息系统]