Scholarlink

<p>The increase in connected renewables across metro grids of smart cities has driven a need for coordination that is secure, adaptive, sustainable, and flexible. This need is not met by centralized control systems, which are struggling under scaling challenges, cyber threats, and a lack of granularity in the spatial and temporal correlations of energy use. In this work, we address these challenges by proposing a graph learning approach augmented with gradient boosting techniques to improve demand forecasting, and a distributed ledger framework to ensure trusted, low-latency coordination of diverse IoT devices. The system uses PBFT for consensus and zk-STARKs to ensure privacy-preserving validation of energy transactions. The MATLAB/Simulink and Hyperledger Fabric co-simulation illustrates that the proposed design reaches 95.3% operational efficiency, 2.1% forecasting error (MAPE), 980 transactions per second throughput, and scales to 25,000 nodes maintaining an average latency of 1.1&#xa0;s. Looking at centralized baselines existing methods, our representations have an improvement of 1.1% in accuracy, a 9.4% in efficiency, and—more importantly—more than double the transaction rate. These results show the promise of the framework to enable resilient and sustainable energy infrastructures for the cities of tomorrow.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Energy-efficient graph learning with blockchain–IoT integration for secure smart city energy management

  • 期刊论文

  • 发表日期：

    2026年1月13日

查看全文
  • Varsha H. Bodade,
  • Sushila Shelke,
  • Neeta Maitre,
  • Satish Samptaro Salunkhe,
  • Sandip Shingade

摘要

The increase in connected renewables across metro grids of smart cities has driven a need for coordination that is secure, adaptive, sustainable, and flexible. This need is not met by centralized control systems, which are struggling under scaling challenges, cyber threats, and a lack of granularity in the spatial and temporal correlations of energy use. In this work, we address these challenges by proposing a graph learning approach augmented with gradient boosting techniques to improve demand forecasting, and a distributed ledger framework to ensure trusted, low-latency coordination of diverse IoT devices. The system uses PBFT for consensus and zk-STARKs to ensure privacy-preserving validation of energy transactions. The MATLAB/Simulink and Hyperledger Fabric co-simulation illustrates that the proposed design reaches 95.3% operational efficiency, 2.1% forecasting error (MAPE), 980 transactions per second throughput, and scales to 25,000 nodes maintaining an average latency of 1.1 s. Looking at centralized baselines existing methods, our representations have an improvement of 1.1% in accuracy, a 9.4% in efficiency, and—more importantly—more than double the transaction rate. These results show the promise of the framework to enable resilient and sustainable energy infrastructures for the cities of tomorrow.