Knowledge distillation has been widely used to improve the performance of small compact models for face recognition. However, selecting key knowledge and effectively transferring it from teacher to student remains a challenging problem. In this work, we propose an efficient Cluster-based Knowledge Distillation (CKD) dedicated to aligning the student model with the teacher model in terms of both sample relations and class centers. Specifically, CKD first determines the key sample relations based on the similarities between the sample features extracted by the teacher and their cluster centers generated by existing clustering algorithms. Then, CKD effectively transfers the knowledge of the above relations from the teacher to the student by designing a cluster-based relation distillation loss. Finally, CKD further improves the quality of the student's class centers by constructing a center loss between the above representative cluster centers and the student's class centers. We validate the proposed CKD on multiple face benchmarks. For example, CKD improves the baseline student performance from 91.95% to 94.20% on MegaFace and consistently outperforms recent competitive distillation methods on multiple benchmarks. These results demonstrate the effectiveness and superiority of CKD.

Fountain codes with online adaptation (OFCs) are promising for underwater acoustic sensor networks (UASNs), since they exploit limited feedback to reduce transmission over head. Yet, the performance of conventional OFCs is severely degraded in UASNs due to high error rates, long propagation delays, and sparse feedback, resulting in poor recovery efficiency and high energy cost. To address these challenges, we propose an energy-aware dual-perception OFC framework driven by deep reinforcement learning (DRL-OFC). In this design, a DRL agent jointly perceives channel dynamics and feedback sparsity to learn an optimal degree distribution that suppresses redundant coding and enhances intermediate decoding. In addition, a feedback aware transmission strategy is developed to cope with the long delay characteristics of UASNs, further reducing unnecessary retransmissions. Simulation results show that DRL-OFC achieves significant gains over existing OFC schemes in terms of decoding efficiency, transmission overhead, recovery performance, and energy consumption, confirming its suitability for resource constrained underwater communications.

Cloud computing is currently playing an essential role in supporting emerging sectors such as smart energy, intelligent transportation, and large-scale distributed systems. The scalability and adaptability features of cloud computing are efficiently enable resource utilization and continuous data exchange in dynamic and heterogeneous environments. However, the increasing demand for cloud services has raised the energy consumption of cloud data centers, which has a critical environmental impact. Therefore, sustainable resource management tactics have become crucial in today's world. Dynamic Virtual Machine (VM) consolidation is one of the major tactics for sustainable resource management in the green cloud computing environment. Herein, a novel dynamic Energy-Aware Cosine Similarity Learning Network (ECSLN) is proposed to predict the overutilized host. Further, an Impact Factor-Based VM Selection (IFBVMS) method is proposed to select VMs for migration, and an ECSLN-Packed Placement method for VM placement into hosts. The main aim of the proposed VM consolidation is to maximize energy efficiency while preserving compliance with Service Level Agreements (SLAs) for sustainable environments. The experimental evaluation using real-world traces like PlanetLab, Bitbrains, and Alibaba Cluster 2020 workload validates that the proposed VM consolidation methods significantly reduce energy consumption and SLA violation compared to the existing methods for sustainable environments. The evaluation shows that the proposed ECSLN-based consolidation framework enables green cloud infrastructure by intelligently balancing energy consumption, SLA violation, and performance of cloud data centers.

Ensuring low battery degradation in Autonomous Ground Robot (AGR) fleets operating in online environments (e.g., delivery services) is essential for enhancing their long-term sustainability. However, most existing studies either rely on offline methods—unsuitable for scenarios requiring real-time decisions—or focus solely on maximizing task allocation, resource utilization, or revenue, with limited consideration for battery health. Additionally, maximizing fleet sustainability requires bounded relative revenue losses from unassigned tasks within a user-defined acceptable limit to make it an attractive option for industry. To address these limitations, we propose an online task and charge allocation framework that jointly optimizes revenue generation and battery lifespan, while allowing users to explicitly constrain relative revenue losses. The framework includes three event-driven algorithms: BTC-M, which computes optimal decisions at each event, and two computationally efficient greedy variants, BTC-G and BTC-WG, which provide sub-optimal solutions with reduced overhead. We evaluate the performance of our approach under different task arrival distributions representative of real-world applications. Simulation results based on a real AGR, compared against multiple baselines, demonstrate that our framework can extend battery lifespan by up to 19% with minimal revenue loss.

Federated Learning (FL) is currently referred to as one of the privacy-enhancing technologies because of its service architecture. However, recent advancements in FL have high-lighted its potential not only as a new framework of privacy but also as a key enabler of sustainable computing, which is expected to minimize the impact of an individual party to further improve the capacity of the machine learning model, energy efficiency, and reliability. For the above requirement of sustainability, resource allocation and trust management in FL are very infrastructural tasks of energy efficiency and reliability. In this paper, we present a framework, called Veri-SFL, to indicate verification for resource allocation and trust measurement in FL. We use trust scores to represent the credibility of each dataset without leaking any privacy, and utilize collaborative zk-SNARKs to verify the trust scores of each local dataset. Then, after verifying the correctness of trust levels, we present a solution to verify whether workers (model owners) are training according to the required distribution ratio by using collaborative zk-SNARKs.

Air pollution monitoring systems are essential for evaluating rural and industrial environmental quality for safeguarding public health. This study presents a comprehensive IoT-based framework that uses low-cost sensors and machine learning algorithms for real-time monitoring of various pollutants, including LPG, methane, CO, alcohol, PM2.5, PM10, temperature, and humidity. The system gathers sensor data from a gateway node, which is then processed using Support Vector Machines (SVM) and Random Forest Regression (RFR) models to predict pollutant concentrations. Our approach features innovative methodologies for data validation, anomaly detection, and predictive modeling, employing Root Mean Squared Error (RMSE) as the performance metric. The model achieved a remarkably low RMSE value of 0.022, significantly improving the accuracy and reliability of air quality assessments. Experimental results highlight the system's capability to capture complex environmental patterns and predict pollutant levels with high precision. This research intends air pollution monitoring from cost-effective Internet of Things (IoT) solutions and machine learning techniques. Additionally, the user interface is designed for mobile applications, offering real-time data access, alerts, and notifications, thereby enabling personalized environmental health management and targeted pollution control strategies in industrial areas.

Convolutional Neural Network (CNN) is a powerful tool that has been extensively applied to many different applications. However, recent developments at CNN have revealed its vulnerability against adversarial example attacks. By introducing visually undetectable noise to the input image, an adversarial example attack can cause the CNN classifier to make false predictions. Multiple approaches have been proposed to defend against adversarial samples, one of which focuses on injecting noise into CNN during training. However, the existing method cannot generate noise efficiently and introduces extra time and energy overhead. In this paper, we propose an Importance-Aware undervolting training framework to improve CNN robustness. The undervolting technique is employed during training for noise generation at negligible overhead. Meanwhile, we observe that the neuron importance and bit importance in hardware can be leveraged during undervolting CNN training for controllable and flexible noise injection, which improves robustness. We design a position-aware bit mapping method in the memory unit by allocating data bits based on significance. And an importanceaware processing element (PE) mapping is also proposed at the computation unit for noise restriction. Our approach regularizes the noise injected into CNN and serves as an efficient method to defend against adversarial example attacks with significant energy savings. The proposed framework is evaluated through both FPGA implementation and software simulation. The experiment results show that our importance-aware undervolting CNN training achieves 47.8% adversarial accuracy at PGD-10 attack and 47.0% training energy savings.

Fire and rescue UAVs employ AI methods to automate navigation over fire zones to detect and extinguish fire f lames and/or rescue fire-trapped lives through low-cost and collision-free flight routes. This paper introduces a reinforcement learning framework that allows UAVs to detect and prioritise fire trapped targets based on their risk levels. It establishes three simultaneous objectives, including maximised power conservation, rescue success rate, and flight safety, to plan fire and rescue operations in complex environments where field size, the number of trapped targets, and fire source count vary. An extensive empirical evaluation is conducted to test and evaluate the performance of the proposal against three well-known benchmarks, including Double Deep Q-network, Advantage Actor-Critic, and Genetic Algorithm. The results demonstrate that the proposed solution outperforms the benchmarks in most circumstances, especially when the fire and rescue environment is large and complex.

The dynamic topology, resource constraints, and data heterogeneity inherent in the Internet of Vehicles (IoV) present fundamental challenges to the effective deployment of federated learning (FL) applications. Traditional FL strategies often suffer from suboptimal model performance and excessive communication overhead under such conditions. Existing solutions primarily focus on static resource allocation and overlook the compounded effects of high mobility, diverse device capabilities, and time-varying data quality. Moreover, current incentive mechanisms lack adaptability to dynamic resource-benefit trade-offs, making it challenging to sustain efficient participation from vehicle clients. To address these issues, we propose NBCI-FL, an incentive framework that redefines vehicular collaboration as a multi-objective coalition game. It introduces a mobility-aware coalition-formation mechanism based on heterogeneous coalition games, along with a three-dimensional evaluation-based client-selection mechanism to construct efficient and stable learning clusters. Furthermore, we propose a two-stage Nash bargaining game that dynamically balances the fairness of data contributions with the optimization of resource utilization efficiency. Theoretical analysis proves the existence and effectiveness of the Nash bargaining equilibrium under vehicular dynamic conditions. Extensive simulations on real-world datasets demonstrate that our proposed NBCI-FL substantially outperforms state-of-the-art traditional FL baselines by maintaining high global model accuracy, improving communication efficiency, enhancing network stability, and ensuring fair and rational utility allocation among all participants.

This paper presents the impact of Inverter-Based Resources (IBRs) controlled through the Grid-Forming (GFM) scheme on Low-Frequency Oscillations (LFOs) and power system dynamic behavior. These IBRs are integrated at low-inertia buses, where their effect on stability is critical. The study employs the Virtual Synchronous Generator (VSG) control-based GFM scheme for IBRs. The dynamic interaction between the power system and the active and reactive power control loops of VSG-controlled IBRs has a significant impact on the system's LFOs. This impact becomes more pronounced as the VSG control loop parameters are increased, particularly the virtual inertia constant <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$(H)$</tex-math></inline-formula>, the virtual damping coefficient <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$(D_{p})$</tex-math></inline-formula>, and the virtual voltage gain coefficient <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$(K_{q})$</tex-math></inline-formula>. These dynamic interactions can introduce new, weakly damped LFOs, negatively impacting the system's dynamic behavior. To address this challenge, a Supplementary Damping Control (SDC) scheme is proposed for IBRs. This scheme aims to improve LFO damping and mitigate power oscillations of IBRs. The SDC comprises a multi-stage mixed <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$H_{2}/H_\infty$</tex-math></inline-formula> decentralized damping controller integrated with the IBR's reactive control loop. The parameter variation uncertainty and explicit modelling of disturbance input have been considered in the design process of this SDC scheme. Further, the robustness of the proposed SDC is validated on the IEEE 39-bus system. The system is tested under various operating conditions, such as load increments, changes in network topology, and integration of renewable sources. Eigenvalue analysis is conducted using MATLAB, while dynamic simulations are performed using the Real-Time Digital Simulator (RTDS). Simulation results confirm that the proposed SDC effectively mitigates system LFOs dynamic behavior.