Intelligent and Low Overhead Network Synchronization for Large-Scale Industrial IoT Systems in the 6G Era

Wang, Xianbin; Jia, Pengyi; Shen, Xuemin; Poor, H. Vincent

doi:10.1109/mnet.107.2100631

Cited by 4 publications

(2 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Apart from optimization of resource utilization, Shen et al [24] leverage the outstanding abilities of GPT in language understanding and code generation to train new models among federated edge devices. Rong et al [25] leverage LLMs to generate adaptive control algorithms for addressing the diverse, dynamic, and decentralized network conditions in 6G integrated terrestrial network (TN) and nonterrestrial network (NTN). Though LLMs have shown great potential in making intelligent decisions, especially in complex and dynamic edge computing systems, the related research is still in the early stages.…”

Section: B Llm For Optimizing Edge Computingmentioning

confidence: 99%

ENTS: An Edge-native Task Scheduling System for Collaborative Edge Computing

Zhang

Cao

Yang

et al. 2022

2022 IEEE/ACM 7th Symposium on Edge Computing (SEC)

View full text Add to dashboard Cite

Large language models (LLMs) have shown great potential in natural language processing and content generation. However, current LLMs heavily rely on cloud computing, leading to prolonged latency, high bandwidth cost, and privacy concerns. Edge computing is promising to address such concerns by deploying LLMs on edge devices, closer to data sources. Some works try to leverage model quantization to reduce the model size to fit the resource-constraint edge devices, but they lead to accuracy loss. Other works use cloud-edge collaboration, suffering from unstable network connections. In this work, we leverage collaborative edge computing to facilitate the collaboration among edge devices and cloud servers for jointly performing efficient LLM inference. We propose a general framework to partition the LLM model into shards and deploy on distributed devices. To achieve efficient LLM inference, we formulate an adaptive joint device selection and model partition problem and design an efficient dynamic programming algorithm to optimize the inference latency and throughput, respectively. Experiments of Llama2 serial models on a heterogeneous physical prototype demonstrate that EdgeShard achieves up to 50% latency reduction and 2x throughput improvement over baseline methods.

show abstract

Section: B Llm For Optimizing Edge Computingmentioning

confidence: 99%

ENTS: An Edge-native Task Scheduling System for Collaborative Edge Computing

Zhang

Cao

Yang

et al. 2022

2022 IEEE/ACM 7th Symposium on Edge Computing (SEC)

View full text Add to dashboard Cite

show abstract

“…To enhance authentication in next-generation wireless networks, such as a decentralized, dynamic, and heterogeneous AIoT network, researchers have explored the concept of lightweight flexible group authentication mechanisms for fingerprint identification [ 22 , 23 , 24 , 25 ]. A group authentication scheme was proposed in [ 23 , 24 ] to detect devices’ identities based on generated tokens for decentralized edge collaboration.…”

Section: Introductionmentioning

confidence: 99%

Channel Prediction-Based Security Authentication for Artificial Intelligence of Things

Qiu

Zhuang

et al. 2023

Sensors

View full text Add to dashboard Cite

The emerging physical-layer unclonable attribute-aided authentication (PLUA) schemes are capable of outperforming traditional isolated approaches, with the advantage of having reliable fingerprints. However, conventional PLUA methods face new challenges in artificial intelligence of things (AIoT) applications owing to their limited flexibility. These challenges arise from the distributed nature of AIoT devices and the involved information, as well as the requirement for short end-to-end latency. To address these challenges, we propose a security authentication scheme that utilizes intelligent prediction mechanisms to detect spoofing attack. Our approach is based on a dynamic authentication method using long short term memory (LSTM), where the edge computing node observes and predicts the time-varying channel information of access devices to detect clone nodes. Additionally, we introduce a Savitzky–Golay filter-assisted high order cumulant feature extraction model (SGF-HOCM) for preprocessing channel information. By utilizing future channel attributes instead of relying solely on previous channel information, our proposed approach enables authentication decisions. We have conducted extensive experiments in actual industrial environments to validate our prediction-based security strategy, which has achieved an accuracy of 97%.

show abstract