3 November 2025 in IEEE Transactions on Network Science and Engineering
In the landscape of Sixth-Generation (6G) networks, achieving efficient content delivery and caching in Unmanned Aerial Vehicle (UAV)-assisted terrestrial networks remains a critical challenge due to dynamic user demand, energy constraints, and redundant storage limitations. This paper proposes a novel framework integrating UAVs, Unmanned Ground Vehicles (UGVs), Reconfigurable Intelligent Surfaces (RISs), and Wireless Power Transfer (WPT) to eliminate the need for UAVs to return to the Base Station (BS) for energy replenishment and content reloading. A Multi-Agent Deep Reinforcement Learning (MADRL) approach is employed to optimize UAV and UGV movements, ensuring adaptive trajectory adjustments for uninterrupted connectivity and service continuity. Specifically, UAVs receive energy and data directly from the BS when a Line-of-Sight (LoS) link is available or, in the case of a Non-Line-of-Sight (NLoS) scenario, via UGV-mounted RISs that reflect signals to/from the BS, maximizing WPT efficiency and enhancing data relaying. A Softmax-based probabilistic caching strategy optimizes content placement based on request popularity while preventing redundant storage across UAVs, enabling them to autonomously update their caches and reposition according to expected demand patterns. When a requested content file is unavailable in a UAV's cache, the request is forwarded to the BS via RIS-assisted reflection or direct communication, ensuring low-latency content retrieval without requiring UAV trajectory deviations. The proposed framework significantly enhances cache efficiency and UAV mobility, enabling continuous operation. Comprehensive simulations confirm the efficacy of this method, showcasing significant gains in network sustainability, energy efficiency, and content availability over traditional UAV-assisted caching techniques.
1 December 2025 in Digital Communications and Networks
14 February 2025 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Land use land cover classification from satellite images (remote sensing) has shown many efforts from the last decade due to ecological surveillance, rapid urbanization, law enforcement, climate change, agriculture drought, and disaster recovery. The low-resolution remote sensing images impact on the accurate prediction; therefore, the high-resolution deep learning architecture is widely required. This article proposes a new deep network-level fusion approach that merges a stacked residual self-attention CNN (SRAN3) with a lightweight ViT based on 4-encoders to enhance the model performance while reducing computational costs. The SRAN3 model is proposed for extracting sophisticated prominent features, while the 4-encoder-based ViT facilitates effective learning with reduced computation time. These networks are fused using a depth concatenation approach that effectively integrates the strengths of both architectures. The fused model hyperparameters are selected through Bayesian optimization, significantly improving the learning process. The trained model is later utilized in the testing phase, extracting features from the depth-concatenation layer. The extracted features are fed to neural network classifiers and obtain the final prediction. Two publicly available datasets, EuroSAT and NWPU_RESIS45, are employed to obtain improved testing and validation accuracy. The proposed SRAN3 + WNN (Wide Neural Network) and 4-encoder ViT + WNN obtained 96.9% and 92.6% of accuracy; however, the proposed fused network + WNN achieved the highest accuracy of 98.4% on EuroSAT and 94.7% accuracy on the NWPU_RESIS45 dataset, respectively. Also, the proposed fused model interpretation is performed using the explainable artificial technique (XAI), which has shown improved land use and land cover classification.
2 June 2025 in IEEE Transactions on Consumer Electronics
Cognitive sensors and reinforcement learning are integrated into home appliances and consumer electronics to make these devices more intelligent and interactive. They create a network of connected devices that receive data from their surroundings and collectively respond to provide services based on context and user mode. The Internet of Things (IoT) is one of the paradigms established in these smart spaces, providing greater comfort and ease of use of technology. With these vast IoT sensors and reinforcement learning, the environment becomes increasingly device-centric rather than user-centric. Sensors continuously generate data and transmit it to the controlling authorities in a timely and accurate manner to allow rapid response and immediate action. A massive volume of data is generated inside the network, with numerous redundant strings of bits and repeating patterns that not only cause congestion but also degrade system performance. Frequent broadcasting on radio links for repeated strings results in idle listening and sometimes overhearing. In this research work, we have proposed a novel scheme of "Reinforcement Learning-Based Dynamic Power Management for Energy Optimization in IoT-Enabled Consumer Electronics (CIoT-DPM)" to reduce idle listening and overhearing. The proposed solution uses advanced reinforcement learning and creates a reward function to reduce network traffic by dropping redundant bits and repeating patterns. Implement different sensor states by keeping some modules in active and sleep modes to adjust dynamic power management. Using the least amount of energy while providing complete coverage, ensuring that no events are missed. In evaluation, CIoT-DPM consumes less energy with increased traffic load; BER ranges from 0.1 to 0.001 in different numbers of episodes and exhibits a 20-25% better learning strategy than traditional learning approaches. In a comparative analysis, CIoT-DPM experiences lower computational complexities by 79.33%; detection probabilities are approximately 0.10 to 0.12, respectively, and an optimal successful learning strategy of 1.00 with 50 episodes.
18 June 2025 in Ain Shams Engineering Journal
17 June 2026 in Digital Communications and Networks