Given the fast growth of intelligent devices, it is expected that a large number of high-stakes artificial intelligence (AI) applications, e. g., drones, autonomous cars, and tac?tile robots, will be deployed at the e...Given the fast growth of intelligent devices, it is expected that a large number of high-stakes artificial intelligence (AI) applications, e. g., drones, autonomous cars, and tac?tile robots, will be deployed at the edge of wireless networks in the near future. Therefore, the intelligent communication networks will be designed to leverage advanced wireless tech?niques and edge computing technologies to support AI-enabled applications at various end devices with limited communication, computation, hardware and energy resources. In this article, we present the principles of efficient deployment of model inference at network edge to provide low-latency and energy-efficient AI services. This includes the wireless distribut?ed computing framework for low-latency device distributed model inference as well as the wireless cooperative transmission strategy for energy-efficient edge cooperative model infer?ence. The communication efficiency of edge inference systems is further improved by build?ing up a smart radio propagation environment via intelligent reflecting surface.展开更多
Recent breakthroughs in artificial intelligence(AI) give rise to a plethora of intelligent applications and services based on machine learning algorithms such as deep neural networks(DNNs). With the proliferation of I...Recent breakthroughs in artificial intelligence(AI) give rise to a plethora of intelligent applications and services based on machine learning algorithms such as deep neural networks(DNNs). With the proliferation of Internet of things(IoT) and mobile edge computing, these applications are being pushed to the network edge, thus enabling a new paradigm termed as edge intelligence. This provokes the demand for decentralized implementation of learning algorithms over edge networks to distill the intelligence from distributed data, and also calls for new communication-efficient designs in air interfaces to improve the privacy by avoiding raw data exchange. This paper provides a comprehensive overview on edge intelligence, by particularly focusing on two paradigms named edge learning and edge inference, as well as the corresponding communication-efficient solutions for their implementations in wireless systems. Several insightful theoretical results and design guidelines are also provided.展开更多
With the rapid advancement of Artificial Intelligence(AI),the Radio Access Network(RAN)is poised to undergo a transformative evolution toward the convergence of AI and RAN functionalities,referred to as the AI-RAN par...With the rapid advancement of Artificial Intelligence(AI),the Radio Access Network(RAN)is poised to undergo a transformative evolution toward the convergence of AI and RAN functionalities,referred to as the AI-RAN paradigm.AI-RAN integrates high-performance computing resources into RAN infrastructures,thereby enabling the execution of both AI and RAN workloads on the same infrastructure.As a result,it improves resource utilization,reduces energy consumption,and promotes swift AI-related responses.In this paper,we provide a comprehensive overview of AI-RAN,whereby we broadly categorize the discussion into three aspects:AI and RAN,AI for RAN,and AI on RAN.In particular,we begin with AI and RAN,which encompass the hardware architecture,software stack,as well as orchestration of computational and communication resources.We subsequently elaborate on the AI on RAN,examining various approaches to leveraging AI methods to enhance RAN performance.For the topic of AI on RAN,we conduct an in-depth investigation into the schemes that take RAN as a platform to facilitate AI services,where we review distributed learning for multi-cell and multi-vendor RANs,including federated and multi-agent reinforcement learning,highlighting issues of data heterogeneity,control-plane overhead,convergence under mobility,privacy,and adversarial robustness in the RAN ecosystems.We also demonstrate several use cases pertaining to the AI-RAN framework.We conclude by outlining key open issues and research directions.展开更多
To improve the accuracy and efficiency of coal-rock interface recognition,this study proposes a model built on the real-time detection algorithm,you only look once(YOLO),and the lightweight bilateral segmentation netw...To improve the accuracy and efficiency of coal-rock interface recognition,this study proposes a model built on the real-time detection algorithm,you only look once(YOLO),and the lightweight bilateral segmentation network.Simultaneously,the regional similarity transformation function and dragonfly algorithm are introduced to enhance the quality of coal-rock images.The comparison with three other models demonstrates the superior edge inference performance of the proposed model,achieving a mean Average Precision(mAP)of 90.2 at the Intersection over Union(IoU)threshold of 0.50(mAP50)and 81.4 across a range of IoU thresholds from 0.50 to 0.95(mAP[50,95]).Furthermore,to maintain high accuracy and real-time recognition capabilities,the proposed model is optimized using the open visual inference and neural network optimization toolkit,resulting in a 144.97%increase in the mean frames per second.Experimental results on four actual coal faces confirm the efficacy of the proposed model,showing a better balance between accuracy and efficiency in coal-rock image recognition,which supports further advancements in coal mining intelligence.展开更多
文摘Given the fast growth of intelligent devices, it is expected that a large number of high-stakes artificial intelligence (AI) applications, e. g., drones, autonomous cars, and tac?tile robots, will be deployed at the edge of wireless networks in the near future. Therefore, the intelligent communication networks will be designed to leverage advanced wireless tech?niques and edge computing technologies to support AI-enabled applications at various end devices with limited communication, computation, hardware and energy resources. In this article, we present the principles of efficient deployment of model inference at network edge to provide low-latency and energy-efficient AI services. This includes the wireless distribut?ed computing framework for low-latency device distributed model inference as well as the wireless cooperative transmission strategy for energy-efficient edge cooperative model infer?ence. The communication efficiency of edge inference systems is further improved by build?ing up a smart radio propagation environment via intelligent reflecting surface.
文摘Recent breakthroughs in artificial intelligence(AI) give rise to a plethora of intelligent applications and services based on machine learning algorithms such as deep neural networks(DNNs). With the proliferation of Internet of things(IoT) and mobile edge computing, these applications are being pushed to the network edge, thus enabling a new paradigm termed as edge intelligence. This provokes the demand for decentralized implementation of learning algorithms over edge networks to distill the intelligence from distributed data, and also calls for new communication-efficient designs in air interfaces to improve the privacy by avoiding raw data exchange. This paper provides a comprehensive overview on edge intelligence, by particularly focusing on two paradigms named edge learning and edge inference, as well as the corresponding communication-efficient solutions for their implementations in wireless systems. Several insightful theoretical results and design guidelines are also provided.
基金supported in part by Start-up Funding for the 2025 National High-End Talent Frontier Project(01)(70012501A01)in part by National Natural Science Foundation of China(62571056,62301222,62201504,62571191,and 62301328)+2 种基金in part by Major Science and Technology Projects of Jiangsu(BG2025038)in part by the Xiaomi Young Talents Programin part by Zhejiang Provincial Natural Science Foundation of China(LDT23F02023F02).
文摘With the rapid advancement of Artificial Intelligence(AI),the Radio Access Network(RAN)is poised to undergo a transformative evolution toward the convergence of AI and RAN functionalities,referred to as the AI-RAN paradigm.AI-RAN integrates high-performance computing resources into RAN infrastructures,thereby enabling the execution of both AI and RAN workloads on the same infrastructure.As a result,it improves resource utilization,reduces energy consumption,and promotes swift AI-related responses.In this paper,we provide a comprehensive overview of AI-RAN,whereby we broadly categorize the discussion into three aspects:AI and RAN,AI for RAN,and AI on RAN.In particular,we begin with AI and RAN,which encompass the hardware architecture,software stack,as well as orchestration of computational and communication resources.We subsequently elaborate on the AI on RAN,examining various approaches to leveraging AI methods to enhance RAN performance.For the topic of AI on RAN,we conduct an in-depth investigation into the schemes that take RAN as a platform to facilitate AI services,where we review distributed learning for multi-cell and multi-vendor RANs,including federated and multi-agent reinforcement learning,highlighting issues of data heterogeneity,control-plane overhead,convergence under mobility,privacy,and adversarial robustness in the RAN ecosystems.We also demonstrate several use cases pertaining to the AI-RAN framework.We conclude by outlining key open issues and research directions.
基金funded by the National Natural Science Foundation of China(Grant Nos.U21A20153 and 52074258)the Key Research and Development Project of Hubei Province,China(Grant No.2021BCA133)+1 种基金the Outstanding Youth Fund Program of the Natural Science Foundation of Hubei Province,China(Grant No.2022CFA084)the Wuhan Knowledge Innovation Supporting project(Grant No.2022010801010162).
文摘To improve the accuracy and efficiency of coal-rock interface recognition,this study proposes a model built on the real-time detection algorithm,you only look once(YOLO),and the lightweight bilateral segmentation network.Simultaneously,the regional similarity transformation function and dragonfly algorithm are introduced to enhance the quality of coal-rock images.The comparison with three other models demonstrates the superior edge inference performance of the proposed model,achieving a mean Average Precision(mAP)of 90.2 at the Intersection over Union(IoU)threshold of 0.50(mAP50)and 81.4 across a range of IoU thresholds from 0.50 to 0.95(mAP[50,95]).Furthermore,to maintain high accuracy and real-time recognition capabilities,the proposed model is optimized using the open visual inference and neural network optimization toolkit,resulting in a 144.97%increase in the mean frames per second.Experimental results on four actual coal faces confirm the efficacy of the proposed model,showing a better balance between accuracy and efficiency in coal-rock image recognition,which supports further advancements in coal mining intelligence.
