Sep 2026· IEEE Transactions on Mobile Computing· Vol 25, pp. 13814-13828· 0 citations· 44 references
Abstract
The integration of Internet of Things (IoT) and Federated Learning (FL) marks a significant step towards pervasiveness, supported by distributed computational resources and enhanced by 5G advancements. However, the efficient and sustainable operation of IoT-enabled FL systems faces critical challenges, particularly in selecting devices that balance energy consumption and system performance. To address this, we propose the Battery-Aware Dynamic and Automated Device Selection (DADS) framework, a novel solution specifically designed for IoT-enabled FL environments. DADS introduces an innovative adaptive mechanism that dynamically adjusts optimization parameters, such as inertia weights and crossover/mutation rates, ensuring a seamless balance between exploration and exploitation. Unlike conventional optimization approaches, DADS employs uniquely developed Adaptive Particle Swarm Optimization (APSO) and Adaptive Genetic Algorithm (AGA) within a cohesive framework. This design enables DADS to respond to dynamic IoT network conditions, such as fluctuating battery levels and device capabilities, ensuring energy-efficient device selection while preserving robust performance. Through extensive performance analysis, DADS demonstrates its novelty by achieving a 20% reduction in energy consumption and a 30% improvement in FL training time compared to state-of-the-art methods. Moreover, DADS significantly enhances battery lifespan, reducing degradation by more than 50%, and optimizes communication efficiency, extending the operational sustainability of IoT devices. These results position DADS as a groundbreaking framework, setting a new benchmark for energy-aware and sustainable IoT-enabled FL systems.
The rapid growth of Internet of Things (IoT) applications has led to the deployment of large-scale sensor networks in smart cities, healthcare, agriculture, industrial automation, and environmental monitoring systems. However, the limited battery capacity of sensor nodes remains a major challenge affecting network lifetime and reliability. Frequent battery replacement increases maintenance costs and limits scalability, particularly in remote and inaccessible locations. This paper proposes an Energy-Efficient IoT Sensor Network Framework that integrates intelligent energy harvesting techniques, adaptive sleep scheduling, edge computing, and Artificial Intelligence (AI)-based routing algorithms to optimize power consumption and extend network longevity. The proposed system continuously monitors residual node energy, communication quality, and environmental conditions to dynamically select optimal routing paths and operational states. Machine learning algorithms predict energy consumption patterns and network traffic conditions, enabling proactive resource management. Experimental analysis demonstrates significant improvements in network lifetime, packet delivery ratio, energy utilization efficiency, and communication reliability compared to conventional routing approaches. The proposed framework offers a sustainable and scalable solution for next-generation IoT sensor networks operating in energy-constrained environments.
Dabbeta Ganapathi Dabbeta Ganapathi, Halavath Vijaya Halavath Vijaya, P. K. P Kavitha· International Journal of Sci...· 0 citations
Growing deployment of Internet of Things (IoT) ecosystems has intensified concerns regarding long-term energy sustainability and environmental impact. Large-scale deployment of wireless sensor networks (WSNs) demands intelligent energy management strategies beyond conventional battery-based solutions However reliance on traditional batteries faces challenges such as limited lifespan, high costs of replacement in remote areas, and environmental impact of battery disposal. This paper proposes an AI-driven framework integrates hybrid energy harvesting mechanisms with Deep Reinforcement Learning (DRL) to optimize energy efficiency in IoT systems. The proposed model employs a Deep Q-Network (DQN) to dynamically regulate sensing, transmission, and sleep operations based on system states. By modeling energy management as a Markov Decision Process (MDP), the framework enables adaptive decision-making under uncertain and fluctuating harvesting conditions. Experimental results show that the proposed framework achieves up to 300% improvement in network lifetime under low-energy harvesting conditions, with an average improvement of 168% and 41.5% higher energy utilization efficiency than static policies. This work presents a scalable framework for Green IoT networks with TinyML feasibility (~2,500 parameters), though hardware validation on microcontrollers remains future work.
Elkhatim Abuelysar Elmobarak Mohammed Ali· Islamic University Journal o...· 0 citations
This work presents a multi-mode energy harvesting-assisted edge computing architecture, integrated with a joint optimization of energy consumption and communication behaviour, aimed at enhancing the sustainability, reliability and autonomy of operation in an industrial IoT context.
Dr. Deepa, M. Mehfooza, Padmavathy Thiruppathi Raj· Microsystem Technologies· 0 citations
The proposed intelligent energy harvesting framework provides an efficient and sustainable power solution for next-generation Internet of Things (IoT) devices by integrating multi-source ambient energy harvesting, Maximum Power Point Tracking (MPPT), hybrid energy storage, and machine learning-based energy management. The framework effectively harvests energy from solar, thermal, radio frequency (RF), vibration, and wind sources while optimizing power utilization through adaptive energy prediction and intelligent task scheduling. Experimental evaluation demonstrates that the proposed system achieves higher energy utilization, lower power consumption, improved communication reliability, and extended operational lifetime compared with conventional battery-powered IoT systems. Furthermore, the integration of cloud and edge computing enables real-time monitoring, predictive analytics, and scalable deployment across diverse IoT applications. Overall, the proposed framework offers a reliable, cost-effective, and environmentally sustainable solution for smart cities, healthcare, industrial automation, environmental monitoring, and precision agriculture, while providing a strong foundation for future research on AI-driven energy optimization and next-generation wireless-enabled self-powered IoT networks.
B. Vaishnavi, Kalasani Siddhartha, Dasarinki Ramprasad· International Journal of Cre...· 0 citations
The advent of smart urban networks based on 6G computing requires low-latency and intelligent edge computing solutions to support the massive distributed data generation. Federated Learning (FL) is an attractive concept of facilitating privacy-conscious distributed intelligence but the traditional FL models frequently ignore the important limitations like battery capacity, wireless communication energy, and variable participation of scale and heterogeneity in massive urban settings. The energy-aware Federated Edge Intelligence (EA-FEI) framework suggested in this paper aims to optimize the energy consumption of the computation and communication processes, latency, and model quality in 6G smart city scenarios simultaneously. The suggested scheme incorporates the energyconscious client selection, battery-sensitive local training, and channel-conscious compression schemes into a multi-objective optimization scheme. EA-FEI can control unnecessary energy consumption through a dynamic adaptive deployment of participation and communication policies depending on battery level, uplink rate and data drift indicators and ensures a strong convergence. Through experimental assessment, the offered framework is found to offer faster convergence and performs better in classification and reduces the per-round energy consumption by about 2728% relative to traditional FedAvg. The findings support the idea that the introduction of energyawareness into federated learning pipelines is a fundamental requirement towards the realization of scalable and sustainable intelligence in future 6G enabled smart urban networks.
Vamsi Krishna Manam, P. L. Devi, Akhila Akula et al.· 2026 5th OPJU International...· 0 citations
Wireless Sensor Networks (WSNs) use resource-constrained sensor nodes to continuously monitor ambient conditions and serve as data sinks for various Internet of Things (IoT) applications. However, maximizing Energy Efficiency (EE) while maintaining reliable data delivery remains a significant challenge. Existing clustering and routing algorithms struggle with issues such as uneven energy consumption, premature node failures, and poor network performance. Additionally, many existing metaheuristic schemes are not adaptable to dynamic network environments, which leads to ineffective energy management. To address the above issues and enhance the energy efficiency of IoT-based WSN, this research introduced a novel Energy-Aware Cluster-Optimized Intelligent Routing (EACO-IR) protocol. The EACO-IR protocol performs in three stages: stable cluster formation, Cluster Head (CH) selection, and energy-aware route finding. At first, stable clusters are formed using the Hybrid Fuzzy–Density Adaptive Kronecker Clustering (HF-DAC) algorithm. Subsequently, CHs are optimally selected using the Mutation-Enhanced Armadillo–Devil Optimization (MEADO) based on a multi-objective function for the Base Station (BS). Finally, the Multi-level Energy-Aware Attention Transformer- Based Reinforcement Learning is introduced to create intra and inter-cluster data travel ways to minimize communication overhead from SNs to the BS. Experimental results show that the proposed protocol yields an average throughput of 4 Mbps, an average Packet Delivery Ratio (PDR) of 98.71%, and an end-to-end latency of 0.06 seconds, outperforming current state-of-the-art clustering and routing algorithms. Ultimately, this framework establishes a highly adaptable template for deploying self-optimizing, long-lasting IoT architectures capable of supporting real-time data streaming without premature network degradation.
P. Kumbhar, A. Naik· International Journal of Ele...· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduAug 18, 2026
A new method for surgically removing training examples from a model reveals that as datasets grow, the link between what a model learns and what it produces dissolves.
