ABSTRACT Objectives Monitoring patient physical activities is essential for assessing health status, enabling timely clinical interventions, and supporting personalized healthcare management. Conventional cloud‐centric healthcare monitoring systems, however, suffer from high communication latency, excessive bandwidth consumption, and limited responsiveness when processing continuously generated wearable sensor data. This study proposes a Digital Twin‐enabled edge‐assisted healthcare monitoring framework that integrates wearable Internet of Things (IoT) devices, edge computing, Digital Twin technology, and deep learning‐based activity recognition to enable real‐time, scalable, and patient‐centric healthcare monitoring. Method The proposed framework employs continuously synchronized Digital Twins to maintain virtual representations of patients and wearable devices, enabling virtual–physical state alignment, contextual patient‐state assessment, healthcare decision support, and proactive alert generation. An edge‐assisted activity fusion mechanism based on Kalman filter data fusion enhances data quality while reducing communication overhead. In addition, a Bidirectional Gated Recurrent Unit (Bi‐GRU)‐based activity recognition model is utilized to capture temporal dependencies in wearable sensor data for accurate activity classification. The framework was validated using a real‐world activity dataset and the UCI‐HAR benchmark dataset, while system‐level performance was evaluated through iFogSim simulations. Results The proposed framework achieved activity recognition accuracies of 94.24% and 98.26% on the real‐world and UCI‐HAR datasets, respectively, outperforming CNN, RNN, LSTM, CNN‐LSTM, Temporal Convolutional Network (TCN), and Transformer‐based approaches. The Digital Twin layer attained 98.41% synchronization accuracy, 97.86% state consistency, 97.24% virtual–physical state alignment, and an average update latency of 0.47 s, demonstrating reliable synchronization between physical and virtual entities. Discussion The experimental findings demonstrate that the proposed Digital Twin‐enabled edge‐assisted framework effectively improves activity recognition accuracy while reducing communication latency and network overhead. The integration of edge intelligence with synchronized Digital Twins enhances resource utilization, maintains reliable virtual–physical synchronization, and supports timely healthcare decision‐making. These capabilities highlight the framework's potential for scalable, intelligent, and patient‐centric healthcare monitoring in next‐generation smart healthcare environments.
ABSTRACT Background Task scheduling in many‐core systems with static non‐uniform cache access (S‐NUCA) is complicated by spatial variability in cache latency and thermal hotspots. Prior research has considered cache asymmetry, inter‐thread distance, and locality, but these factors are often treated separately and predominantly within runtime‐based frameworks rather than being integrated into a unified compile‐time strategy. Objective To develop compile‐time topology‐aware thread mapping strategies that jointly address spatial cache locality and thermal constraints without incurring runtime migration overhead. Methods We propose three compile‐time thread mapping strategies, namely Golden‐Angle Spiral Mapping (GASM), Max Distance Anti‐Cluster Mapping (MDAC), and Latency‐Aware Minimal Mapping (LAMM). These approaches leverage geometric dispersion and Average Manhattan Distance (AMD) between cores to enhance spatial placement. Each mapping targets a distinct scenario, with LAMM designed for low‐thread compute‐bound workloads, MDAC for high‐thread thermally constrained cases, and GASM for irregular memory‐intensive workloads. These methods statically address thermal and latency constraints without runtime migration overhead, enabling low‐complexity mappings with predictable performance and thermal behavior. Results Across 8, 12, 16, and 64 thread configurations, the proposed topology‐aware mappings achieve up to 26% performance improvement and 16% energy reduction over the static baseline. Even when compared with the dynamic baseline, which benefits from runtime adaptation, they continue to deliver up to 20% higher performance and 9% lower energy consumption. Under workload‐specific thread configurations, performance gains reach 22% and energy savings 18%, with each mapping showing the benefit for its target workload behavior. Overall, these results validate the effectiveness of a complementary multi‐strategy approach for thermally constrained S‐NUCA many‐core systems.
ABSTRACT Context In the context of model‐driven engineering, the proliferation of intelligent modeling assistants (IMAs) has pushed the boundaries of automation, providing modelers with relevant recommendations based on incomplete modeling artifacts as primary input. Despite their benefits, developing IMAs has also brought a set of challenges that can compromise the quality of the recommendations, spanning from selecting the proper knowledge base to training the most suitable algorithm. In the scope of this paper, we focus on graph‐based techniques that have been proposed to enhance IMAs by exploiting the relationships among modeling artifacts. Althoughthese systems achieve notable accuracy on real‐world datasets, there is a need to understand how their performance differs with respect to different features of the input datasets, for example, the size of the graphs, the number of available artifacts, or their categories. Method We report our experiences in evaluating IMAs by comparing two state‐of‐the‐art graph‐based IMAs. First, we elicit a list of steps and issues that occur in IMAs according to our practical experience. Subsequently, we select two approaches: (i) BORA, a reuse‐oriented assistant exploiting SPARQL queries and N‐gram encoding over RDF graphs, and (ii) MORGAN, a prediction‐oriented assistant leveraging graph kernel similarity. Finally, we evaluate the accompanying tools of the two approaches by using three distinct datasets from different application domains. Results Our findings show that BORA obtains better results when heterogeneous datasets are considered, while homogeneous ones should be used to feed ML‐based techniques such as MORGAN. This finding is directly connected to the nature of the two approaches, as reuse‐oriented approaches can handle heterogeneity more than traditional machine learning approaches. Conclusion We eventually report a list of lessons learned for researchers and practitioners, highlighting the need for a careful data selection process.
ABSTRACT Objectives The Internet of vehicles development under intelligent transportation systems necessitates secure, decentralized, and privacy‐preserving authentication schemes to counter sophisticated attacks such as sybil attack, replay attack, man‐in‐the‐middle attack, and rushing attack. Current PKI and ECC‐based schemes still endure central trust schemes, high computation overheads, and weak privacy guarantees. To overcome such limitations, this paper proposes, a blockchain‐based secure routing protocol with zero knowledge handover authentication for Internet of Vehicles intertwining multiphasic security, privacy‐preserving zero‐knowledge proofs, and real‐time rushing attack detection for Internet of vehicles scenarios. Methods BSRP is built with a consortium blockchain and operates under seven mutually interconnected security phases such as secure registration, one‐time passwords assisted login, zk‐SNARK‐facilitated authentication, privacy‐preserving handover, verifiable route verification using Layer‐2 zk‐Rollups, and rushing attack detection. The protocol incorporates HMAC‐based one‐time passwords for freshness, Keccak‐256 hashing for attribute integrity, zk‐SNARKs for privacy‐resilient proofs, and AES‐256 for encrypted channel communications. Smart contracts allow for decentralized identity management, blacklist enforcement, and tamper‐evidence logs. Results A security proof with Random Oracle Model and BAN logic shows that BSRP achieves session key indistinguishability, forward secrecy, and immunity to known attack scenarios. Comparison results demonstrate that BSRP reduces computation overheads and handover latencies by orders of magnitude compared to prior blockchain‐assisted schemes. Conclusion With privacy, decentralization, and lightweight crypto construct at its viewpoint, BSRP provides a scalable and robust security architecture that can be deployed within real‐time, privacy‐conscious vehicular networks.
ABSTRACT Background The widespread expansion of IoT devices has significantly increased the scope for malicious activities on the part of cybercriminals. This has made edge computing highly vulnerable to various types of network intrusions, including DDoS floods and brute‐force attacks. However, traditional centralized Intrusion Detection Systems (IDS) do not address these concerns due to their inherent latency and lack of consideration of data privacy. To overcome these drawbacks of traditional IDS, we have developed a distributed real‐time anomaly detection system architecturally designed for deployment on edge computing hardware. Method A monitoring agent collects system and network telemetry and real‐time network flow data on a Raspberry Pi∼3 and sends it to a centralized server via a TCP socket. The server uses a dual‐engine detection system consisting of rule‐based heuristics and a DNN model. To address the concerns of data privacy and non‐IID data distribution inherent to distributed IoT networks, our system uses a collaboratively trained DNN model using three different Federated Learning (FL) techniques: Clustered Federated Learning (CFL), FedDyn, and FedNova. Results These techniques were evaluated on the CIC‐IoT‐DIAD∼2024 dataset with severe non‐IID data distribution. The results show that CFL outperforms others with an F1‐score of 0.89 and an AUC of 0.95. This work shows that production‐grade, privacy‐preserving intrusion detection is feasible in IoT networks at the edge.
