5G (Fifth Generation) technology represents a significant advancement in telecommunications, facilitating industry innovation and advancement for applications across various sectors. It enhances speed, minimizes latency, and optimizes connectivity. By leveraging task offloading, 5G networks maximize the efficiency of both computational and network resources, ensuring faster, more reliable data delivery and enabling the high-performance requirements of modern applications. However, offloads are complex because task migrations are highly dynamic and must balance the offloading performances. Multi-Access Edge Computing (MEC) is an approach that addresses the processing limitations of low-powered computers by transferring calculations and jobs to cloud platforms. It uses networks’ edges to move tasks to the nearest data centers, reducing latencies and increasing connection speeds. In addition to data security, MEC has restrictions when delivering information to edge nodes from current voice-only to data-driven communication networks. Framed as classification challenges in MEC, this paper uses blockchain technology to guarantee data security and privacy for task offloads. Through our proposed secure and robust dynamic task-offloading strategy Secured Adaptive Generalized Edge (SAGE), the work emphasizes on reinforcement learning (RL) for judgments on task offloads from mobile devices to edge networks considering the energy limitation and network latency. The schema efficiently derives near-optimal task-offloading decisions by leveraging contextual bandits to enhance the task-offloading paradigm in edge computing. The offloads use blockchains-based smart contracts for ensuring security and SDN-based dynamic computation offloads and migrations. Simulations under synthetic and actual circumstances are conducted to demonstrate that the suggested schema can efficiently adapt to dynamic settings in its learning process, hence enabling faster convergence and near-optimal delay performances. SAGE considers knowledge of delay in queuing, offloading cost, and trust, therefore reducing average task offloading delays by 36% when compared to other existing systems. The Dynamic Edge Server Offload (DESO) algorithm identifies harmful Edge servers (ESs) reducing the frequency of job offloading delays and failures, where results demonstrate reductions in task durations by 30%, with different arrival rates of offloaded tasks and energy consumptions compared to random and uniform offloading schemes.
11 September 2025 in IEEE Transactions on Consumer Electronics
The rapid expansion of consumer IoT devices—such as smart home hubs, smartphones, and wearables—demands secure and scalable provisioning mechanisms that can withstand emerging quantum threats. Embedded SIM (eSIM) technologies, widely adopted for device authentication and network access, remain vulnerable due to their reliance on classical cryptography. This paper presents a post-quantum secure eSIM provisioning framework for resource-constrained consumer electronics, integrating NIST-standardized Kyber and Dilithium algorithms with blockchain validation and Quantum-Enhanced Variational Neural Network (QEVNN) for efficient routing. The architecture supports secure carrier switching, dynamic profile provisioning, and tamper-resistant key management using Trusted Execution Environments. Experimental results demonstrate 97.5% security effectiveness, 46% improvement in hash performance, and a 12.3% increase in provisioning accuracy. Real-device evaluations on smartphones, wearables, and smart hubs confirm deployment feasibility with sub-second cryptographic operations on commercial hardware. The proposed solution delivers quantum-resilient, market-ready eSIM security optimized for next-generation consumer IoT environments.
In traditional e-assessments as well as in personalized learning, question bank calibration plays an important role in ensuring high-quality assessment outcomes. We propose an unsupervised learning approach that uses performance metrics derived from test-taker responses for precise calibration of question banks. We show partitioning test-takers into three groups using their scores is an effective feature extraction method. This approach, enables us to accurately capture performance variations in test-taker sub-groups thereby effectively capturing inter-item correlations. Integration of such a Machine Learning (ML) approach into the Question Bank module of a cloud-based e-assessment system enables seamless automatic question calibration. We validate the effectiveness of our approach, using a large dataset, collected from a two thousand student university-level proctored assessment.