Blockchain-Based Green Edge Computing: Optimizing Energy Efficiency with Decentralized AI Frameworks

Abstract

The deployment of Internet of Things (IoT) devices and edge computing has grown exponentially and has reinvented the world of data processing and making it possible to deliver low-latency applications in real-time settings. Notwithstanding, with this shift towards the use of distributed systems, we are faced with new challenges of ensuring there is effective management of energy consumption.  The main aim of the proposed study was to design, deploy, and test a new decentralized edge computing framework that combines blockchain technology and artificial intelligence to achieve optimized energy efficiency. To be more precise, we intended to create AI models that are able to recognize and forecast energy usage patterns at the edge in real-time. The system of 250 edge devices on a network in this study simulated the environment of the smart infrastructure, which portrays a medium-sized U.S. urban grid. All of these devices were able to record important performance and system data on an ongoing basis over more than 30 days at a resolution of 10 seconds, and provide more than 60 million data points. Prominent variables that are recorded are CPU usage (%)/memory load (MB) and energy level (Watts), which is a reflection of the device in terms of operation strain and efficiency. So that edge workloads can be classified according to their energy consumption rates and usage trends to facilitate energy-efficient scheduling. Three supervised machine learning models were chosen: Logistic Regression, Random Forest Classifier, and Support Vector Classifier (SVC). The preprocessed dataset was divided into 80:20 train and test sets to ensure that there was no data leakage, and all three models were trained on the datasets and evaluated on the test set. Based on the measurement, Random Forest had the most accurate predictions, meaning that it tended to slightly outdo the other models in this comparison. The next two models, notably logistic Regression and SVM, respectively, had the lowest accuracy of the three models. The encountered blockchain mechanism, i.e., lightweight transaction ledgers including Hyperledger Sawtooth, offered informative transparency and traceability of energy behavior in edge networks. Introducing blockchain-based green edge computing is about to change the energy management approach in smart cities and intelligent energy grids in the U.S. The introduction of IoT-powered networks in metropolitan areas such as New York City, San Francisco, and Chicago, including traffic sensors and adaptive lighting, autonomous transportation, and Wi-Fi hotspots, has also meant that the energy requirements of distributed edge networks are being placed at a serious burden. Green edge computing with blockchain has an important role in defense and the safety of the population by assuring safe, energy-saving decision-making in the field. The DOD (U.S Department of Defense) mainly depends on mobile and distributed sensor networks to perform surveillance of the theaters of operation, environmental tracking, and real-time information. The findings of the current research add value to the potential of AI-powered methods in increasing energy efficiency in edge computing solutions, especially when combined with blockchain frameworks.