提出去中心化智能物联网络框架，提升设备协同效率与能效。
An Internet of Intelligent Things Framework for Decentralized Heterogeneous Platforms
- 构建异构节点的去中心化自组织网络，支持分布式协作。
- 融合联邦学习与元启发式算法，实现任务分配与路径优化。
- 兼顾可靠性、能耗与延迟，适合边缘智能场景部署。
智能物联（IoIT）作为新兴领域，将物联网设备与嵌入式AI算法结合，但面临计算资源、能源供应和存储限制等挑战，尤其体现在嵌入式设备上机器学习/深度学习模型的能效部署问题。现有研究多聚焦集中式系统，存在性能瓶颈与安全风险。为此，本文提出一种异构、去中心化的传感与监控IoIT点对点网状网络系统模型。网络中节点自主协调，以可靠性、能效与低延迟为优化目标。系统采用联邦学习实现分布式模型训练，利用元启发式算法优化任务分配与路由路径，并通过多目标优化平衡相互冲突的性能指标。
原文摘要 · Abstract (English)
Internet of Intelligent Things (IoIT), an emerging field, combines the utility of Internet of Things (IoT) devices with the innovation of embedded AI algorithms. However, it does not come without challenges, and struggles regarding available computing resources, energy supply, and storage limitations. In particular, many impediments to IoIT are linked to the energy-efficient deployment of machine learning (ML)/deep learning (DL) models in embedded devices. Research has been conducted to design energy-efficient IoIT platforms, but these papers often focus on centralized systems, in which some central entity processes all the data and coordinates actions. This can be problematic, e.g., serve as bottleneck or lead to security concerns. In a decentralized system, nodes/devices would self-organize and make their own decisions. Therefore, to address such issues, we propose a heterogeneous, decentralized sensing and monitoring IoIT peer-to-peer mesh network system model. Nodes in the network will coordinate towards several optimization goals: reliability, energy efficiency, and latency. The system employs federated learning to train nodes in a distributed manner, metaheuristics to optimize task allocation and routing paths, and multi-objective optimization to balance conflicting performance goals.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。