The proliferation of big data has fueled the success of deep learning; however, its inherent ”black box” nature poses significant challenges for its adoption in safety-critical domains. Existing interpretable machine learning methods offer partial solutions but often struggle with model fidelity, inconsistent explanations, a lack of holistic model understanding, and critically, computational inefficiency, especially when applied to models trained on large-scale datasets. To overcome these hurdles, this paper introduces Parallel Explainer (PEXP), an innovative parallel tree-based interpretation framework designed for scalability and comprehensive understanding. PEXP initiates by generating a localized sample set around a target instance through data distribution-aware perturbations. It then computes similarity scores and employs a kernel function to weight these samples effectively. Leveraging concepts from Bagging and Boosting, PEXP efficiently constructs Parallel Ensemble Trees as its core interpretable model. This model provides feature importance-based explanations and aggregates insights across all samples to achieve a global understanding of the model's behavior on the entire dataset. Experimental results demonstrate PEXP's significant advantages over mainstream interpretable methods in both runtime efficiency and the quality of explanations, particularly crucial for big data analytics. Furthermore, a case study illustrates PEXP's application in enhancing the interpretability of video anomaly detection systems within smart cities, a domain characterized by large volumes of data and offers insights for improving Transformer-based architectures.

The Internet of Vehicles (IoV), equipped with sensors, generates vast amounts of data, demanding rigorous computation and network. The cloud computing (CC) platform meets these stringent computation requirements, but it has a significant latency that fog computing (FC) effectively handles. Software defined network (SDN) has become the de facto standard for next-generation networking due to its unique and flexible features, which handle the network prerequisites in an agile manner. Task offloading (TO) is a crucial issue in software defined-IoV (SD-IoV), particularly when the vehicle’s resources are insufficient. Fog nodes schedule offloaded tasks; however, neglecting computational and network details during task scheduling can lead to longer completion times, thereby restricting the functionality of SD-IoV due to its time-sensitive nature. Therefore, this article introduces two proactive intelligent TO (ITO’ and ITO) schemes. These schemes schedule tasks onto fog nodes with enhanced computational capabilities, dynamically enable the network through the ovs-ofctl SDN utility, and consider vehicle mobility during task scheduling. Experiments with Mininet show that the proposed (ITO’ and ITO) schemes improve performance in terms of CPU availability, bandwidth, and throughput by 5.2 times, 50% and (30.7–13.9)% more than the existing scheme. They also reduce average packet loss, delay, round trip time (RTT), and offloaded node time selection by (45.35–21.3)%, (50–33.3)%, (53.9–9.9)% and (43.2–26.06)% which demonstrates the efficacy of the proposed scheme.

Synthetic data generation via generative artificial intelligence (GenAI) is essential for enhancing cybersecurity and safeguarding privacy in the Internet of Medical Things (IoMT) and healthcare. We introduce multifeature conditional generative adversarial network (GAN) (MF-CGAN), a Conditional GAN with multifeature integration, that generates realistic synthetic data using the WUSTL EHMS 2020 dataset of network traffic and health metrics. MF-CGAN’s architecture integrates conditional variables such as cyberattack types, traffic direction, and status flags, preserving their intricate interdependencies. The generator employs dense layers with batch normalization and Leaky ReLU activations to produce conditional synthetic data; the discriminator evaluates authenticity using the same conditional inputs. Experimental results show that MF-CGAN produces high-fidelity synthetic data, closely mirroring real data characteristics. In addition, we quantitatively demonstrate (MAE${=} 0.012$, RMSE${=} 0.035$, and classification accuracy up to 99%) that MF-CGAN-generated data effectively supports cybersecurity tasks, indicating its potential to significantly enhance data availability for IoMT.

The necessity to address the growing pace of industrialization and resource constraints has underscored the importance of developing innovative Information and Communication Technology solutions. The prevailing “smart industry” paradigm is distinguished by its widespread cyber infrastructure for monitoring and controlling critical industrial assets. In case sudden failure of any equipment shot down the industry and apart from financial loss, it take ample amount of time to recover. To effectively manage the multitude of continuous services while maintaining their quality of service requirements, it is crucial to concurrently utilize sensor-based real-time assessment for predicting employed equipment status and lifespan become a challenging. This research explores the feasibility of enhancing the lifespan of sensors at the fog layer by implementing intelligent sleep and wake-up duty cycles. We propose a clustered-based duty-cycling approach that optimizes service continuity with a minimal number of active sensor nodes using a genetic algorithm. The abovementioned framework simulated in iFogSim simulator and obtained results demonstrate the reduced the energy consumption by approximately 16% to 20% over the energy efficient aware and around 38% to 42% over the K-mean clustering approach with 90% duty cycle.

Social violence presents a compelling challenge to public safety, yet existing multimodal detection systems exhibit excessive reliance on RGB image semantics and opaque decision-making processes. Despite leveraging visual and auditory data, current models demonstrate RGB bias in feature prioritization, as evidenced by explainability analyzes, thereby limiting their generalization for behavioral understanding. Additionally, modality inconsistency and inefficient fusion mechanisms impair model transparency and training stability. To bridge these gaps, this study proposes modality-aligned preprocessing (VAJ) that structurally unifies visual-auditory features through conflict resolution and input optimization, explicitly suppressing color dominance while enhancing interpretable feature representations. Complementing this, we design DTVDS, an interpretable detection framework integrating knowledge distillation to transfer distilled behavioral insights from a cumbersome teacher network to an efficient student model. This dual strategy not only addresses computational overhead but also clarifies decision logic through simplified inference pathways. Evaluations on XD-Violence and UCF-Crime benchmarks demonstrate superior performance, with AP (89.64%) and AUC (88.35%) outperforming existing methods. Qualitative evaluations further validate interpretability, revealing modality-coherent attention maps and human-aligned rationale visualization. The proposed method advances violence detection by addressing persistent shortcomings in multimodal alignment and model explainability.

In an era of 5 G smart cities, precise traffic prediction remains elusive due to limited real-world data. Our paper introduces a novel approach using Generative Adversarial Networks (GANs) to create synthetic traffic data that closely mimics real-world statistics. This artificial dataset enhances our new 5GT-GAN-NET-based prediction model. The result is a significant boost in prediction accuracy, with Mean Square Error (MSE) reduced to 0.000346 and Mean Absolute Error (MAE) to 0.00685. Compared to benchmarks, our model improves MSE and MAE by up to 95.45% and 87.31%, respectively. User privacy remains a cornerstone of our approach, crucial for smart city applications. Our predictive capabilities enable more efficient resource allocation by service providers, increasing communication infrastructure reliability. Although tailored for smart cities, the approach is adaptable to other fields facing data scarcity and privacy concerns. Our research highlights the potential of GANs in generating large, accurate datasets for traffic prediction in 5 G environments while prioritizing user privacy.

As the Internet of Things (IoT) continues to advance as a technology, it has given rise to innovative and cross-domain IoT applications, particularly in smart cities. For IoT applications and services that are sensitive to latency and due resource constraints it affects the Quality of Service (QoS). To address these challenges, context-aware fog computing at the network edge requires an enhanced focus on optimizing resources for intelligent service management. Due to the dynamic change of workload at fog nodes, i.e., sudden rise in demand, an effective load balancing approach among fog nodes becomes essential. However, it’s crucial to execute load transfers, such as Virtual Machine (VM) migrations but improper migration can lead to a cascade of migrations and ultimately degrade system performance. In this paper, we introduce a resource-optimized intelligent service model (RoISM) designed to facilitate resource optimization through a forecasting technique. This technique predicts the requisite context instances and resource computation needed for efficient service delivery. The proposed hybrid approach to service management leverages context-sharing, context-migration, and live service migration strategies, all based on the forecast method. This method utilizes both current and predicted resource utilization data, as well as context availability, to fulfil service requests within the specified latency requirements for cross-domain IoT applications. To validate the effectiveness of our proposed service management algorithms, we conducted simulations using a CloudSim simulator. The results obtained from these simulations confirm the superiority of our proposed methods

The rapid progression of 5G networks has ushered in a new era of communication, marked by challenges in optimizing enhanced Mobile Broadband (eMBB) throughput and Ultra-Reliable Low Latency Communications (URLLC) latency. This research delves into Deep Reinforcement Learning (DRL) to address these challenges, with a particular emphasis on the Proximal Policy Optimization (PPO) algorithm. By leveraging a judiciously crafted environment and reward structure, our DRL agents were trained to optimize eMBB throughput and URLLC latency concurrently. Using the Colosseum O-RAN COMMAG Dataset, our agents achieved an average eMBB throughput of 0.4451 Gbps and an average URLLC latency of 0.4719 ms. These outcomes highlight the potency of DRL as a tool for 5G optimization, presenting a promising avenue for future advancements in intelligent 5G network management.

Advances in wireless charging technology give great new opportunities for extending the lifetime of a wireless sensor network (WSN) which is an important infrastructure of IoT. However, the existing greedy algorithms lacked learning from the experiences of energy dissipation trends. Unlike the existing studies, this article proposes a reinforcement learning approach, called reinforcement learning recharging (RLR), for mobile charger to learn the trends of WSNs, including the energy consumption of the sensors, the recharging cost as well as the coverage benefit, aiming to maximize the coverage contribution of the recharged WSN. The proposed RLR mainly consists of three modules, including sensor energy management (SEM), charger location update (CLP), and charger reinforcement learning (CRL) modules. In the SEM module, each sensor manages its energy and calculates its threshold for the recharging request in a distributed manner. The CLP module adopts the quorum system to ensure effective communication between sensors and the mobile charger. Meanwhile, the CRL module employs attraction rewards to reflect the coverage benefit and penalties of waiting time raised due to charger movement and recharging other sensors. As a result, the charger accumulates the learning experiences from the$Q$-Table such that it is able to execute the appropriate actions of charging or moving in a manner of state management. Performance results show that the proposed RLR outperforms the existing recharging mechanisms in terms of charging waiting time of sensors, the energy usage efficiency of the mobile charger, as well as the coverage contribution of the given sensor network.

Abstract Verifying the genuineness of official documents, such as bank checks, certificates, contract forms, bonds, etc., remains a challenging task when it comes to accuracy and robustness. Here, the genuineness is related to the degree of match of the signature contained in the documents relating to the original signatures of the authorized person. Signatures of authorized persons are considered known in advance. In this paper, a novel feature set is introduced based on quasi-straightness of boundary pixel runs for signature verification. We extract the quasi-straight line segments using elementary combinations of the directional codes from the signature boundary pixels and subsequently we obtain the feature set from various quasi-straight line classes. The quasi-straight line segments provide a blending of straightness and small curvatures resulting in a robust feature set for the verification of signatures. We have used Support Vector Machine (SVM) for classification and have shown results on standard signature datasets like CEDAR (Center of Excellence for Document Analysis and Recognition) and GPDS-100 (Grupo de Procesado Digital de la Senal). The results establish how the proposed method outperforms the existing state of the art.

With the proliferation of smart grid and deregulation of the energy market, a wide variety of peer-to-peer (P2P) energy trading systems have emerged. Common challenges for designing such systems include prosumers’ privacy and security threats. To this end, Blockchain-based solutions have gained a lot of attention, though most existing solutions have either employed permissionless blockchain, which is far from pragmatic for a P2P energy trading system with peers permitted to join or leave the network at their whim; or relatively secure yet inefficient permissioned blockchains. Hence, this article presents a flexible permissioned ascription (FPA) scheme that uses on-chain and off-chain permissioning scheme viaOrionandMetamaskwallet. It also employs contract permissioning through a JavaScript based chain code deployed over Hyperledger Besu (an Ethereum based permissioned Blockchain network) with istanbul byzantine fault tolerant (IBFT) 2.0 consensus algorithm. Additionally, the proposed framework is emulated for development of a working prototype for a P2P energy trading system. Its performance evaluation has been conducted and monitored with Grafana, Prometheus, Hyperledger Caliper, and Kibana for parameters such as latency, throughput, success rate, CPU time, block time, block behind time, memory usage, garbage collection (GC) time, and performance of the validator nodes. The latency of IBFT 2.0 was found five times lesser than that of Ethereum and two times lesser than HF RAFT and KAFKA under varying conditions. Also, the measured throughput was 1.5 times higher than RAFT and Kafka and three times higher than that of Ethereum. The average block confirmation time measured is 5–6 s. The GC usage measured very less, i.e., 0.5–0.8%, with the proposed framework. It has been observed that the proposed energy-trading framework provides an efficient performance for deploying, transferring, and querying the energy transaction to a P2P energy-trading Blockchain network when compared with other consensus mechanisms.

Software-Defined Networks ( SDNs ), with their segregated data and control planes, has proved to be capable of managing massive amounts of data by leveraging distributed information available across the network for informed decision-making at the network controller. However, with the proliferation of next-generation, real-time Internet of Things ( IoT ) applications that vary greatly in terms of data frequency and volumes, data traffic classification can substantially assist SDN controllers toward efficient routing and traffic engineering decisions. Existing works on network classification are limited by their application-centric nature, thus overlooking the key criterion for real-time IoT applications, namely, Quality of Service ( QoS ). In this article, we focus on augmenting SDN controllers’ decision-making capacity and Underwater Sensor Networks with machine learning algorithms to achieve real-time, QoS-aware, network traffic classification. Three classifiers, namely, Feed-forward Neural Network, Naïve Bayes, and Logistics Regression have been employed with a novel Artificial Neural Network and Particle Swarm Optimization hybridization scheme by carrying first- and second-order stability analysis for performance improvement of these classifiers. In short, the proposed framework exploits optimization algorithms and semi-supervised machine learning ( ML ) for precise traffic classification while keeping communication overhead between controller and switches minimal. Results obtained from real-life datasets demonstrate the efficacy of our proposed scheme.

With the proliferation of information and communication technology in every walks of the society, including healthcare services, digitization, and increased sophistication have been gaining pace, digital healthcare alternatives such as electronic healthcare record (EHR) have gained prominence with increased patients' data volume. However, traditional EHR-based systems are plagued by data loss risks, security and immutability consensus over health records, gapped communication among constituted hospitals, and inefficient clinical data retrieval systems, among others. Blockchain has been developed as a decentralized technology that holds the promise to address the aforesaid facilities in EHR-based systems. This article presents a patient-centric design of a decentralized healthcare management system with blockchain-based EHR using javascript-based smart contracts. A working prototype based on hyperledger fabric and composer technology has also been implemented which guarantees the security of the proposed model. Experiments with the hyperledger caliper benchmarking tool provide performance such as latency, throughput, resource utilization, and so on under varied scenarios and control parameters. The results affirm the efficacy of the proposed approach.

Multiple robots negotiating in a dynamic workspace may lead to collisions. To avoid such issues, multi-robot navigation and coordination becomes necessary but is computationally very challenging, particularly when there are many robots. This article addresses the problem of multi-robot navigation where individual robots require coordination. Although a few such attempts for modeling multi-robot coordination and navigation have been studied, this work proposes a game-theoretic coordination strategy, also referred to as strategic coordination. We make use of a genetic algorithm tuned fuzzy logic–based motion planner. The proposed strategic coordination strategy has been pitted against a basic potential field-based motion planner, also referred to as the heuristic method, for performance comparison. Results are compared through computer simulation with 8 to 17 robots at different rounds. From the obtained results, it was observed that the proposed coordination scheme’s efficacy is strong for a larger number of robots. In addition, the proposed strategic coordination scheme with the genetic-fuzzy-based motion planner was found to outperform other combinations as far as the quality of solutions and time to reach the goal positions. The computational complexity of different methods has also been compared and presented.

Automatic disease diagnosis is, in essence, a classification problem where the classifier has to be trained based on patients’ datasets and not entirely on doctors’ expert knowledge. In this paper, we present the design of such data-driven disease classifiers and fine-tuning classifier performance by a multi-objective evolutionary algorithm. We have used sequential minimal optimization (SMO) classifier as the base classifier and three evolutionary algorithms namely Cat Swarm Optimization (CSO), Invasive Weed Optimization (IWO) and Eagle Search based Invasive Weed Optimization (ESIWO) to diagnose disease from datasets available. In that sense, our approach is an offline data-driven approach with 18 benchmark medical datasets, and the obtained results demonstrate the superiority of the proposed diagnoses in terms of multiple objectives such as classification Prediction accuracy, Sensitivity, and Specificity. Relevant statistical tests have been carried out to substantiate the cogence of the obtained results.

With the unprecedented advancements in the field of Internet of Things (IoT), novel increasingly complex services are getting conceived and implemented with the day. A new trend in this landscape of next-generation IoT applications has been those orchestrated by the seamless integration of numerous vertical IoT applications, also known as, unified IoT services. However, such unified IoT applications are resource hungry (computation, storage, as well as network resources) and are usually characterized by real-time constraints. These render cloud and fog enactments insufficient. This article employs a context-aware computing approach that helps alleviating massive data transfers across fog nodes in real time. User requests are managed at fog nodes were depending upon their required contexts, availability of such contexts in the fog node and respective deadline, the requests are serviced either locally or at other fog nodes and Quality of Service (QoS) is fulfilled. To this end, a distributed service management algorithm is proposed that services the user requests at every fog node by either sharing contexts among context requests within a fog node, or by bringing in unavailable contexts from other fog nodes or by migrating service requests to remote fog nodes with available contexts in a deadline-aware fashion. Accordingly, this article proposes a novel smart multichannel queuing (SMCQ) model that schedules requests to virtual machines (VMs) at individual fog nodes by assigning them to specific priority groups (based on deadline vicinity) and thus minimizes service delay. The proposed algorithms for delay-tolerant service migration have been presented and simulation results carried out to demonstrate the efficacy of the proposed methodology.

Advances in wireless sensor networks and increasing Internet-of-Things devices give great opportunities for smart homecare of the elderly. Smart homecare has been a promising issue and received much attention recently. Irregularity detection is one of the most important issues in smart homecare for assessing the health condition of the elderly. However, most of the researches focused on the explicit irregularity detections which are usually based on the drastic changes of sensor data, such as falling. Existing mechanisms for detecting implicit irregularity rely on the subjective assessment of behaviors' importance by the elder and simply outputs the binary detection results. This article proposes a feature-based implicit irregularity detection mechanism (FIID), which extracts the regularity features using unsupervised learning and outputs the probability of implicit irregularity. The proposed FIID identifies the regular behaviors which satisfy the time-regular and happen-frequently properties as the regularity features of daily behaviors. These features then construct a multidimensional feature space to calculate the implicit irregularity probability of the daily health condition. Performance results show that the proposed FIID outperforms the existing implicit irregularity mechanism in terms of precision, recall as well as F-measure.

As the Internet of Things (IoT) paradigm is maturing, innovative, and novel services are being envisioned. An upcoming trend is the depiction of services enacted through seamless integration of multiple vertical IoT services, termed as cross-vertical or unified IoT services in this paper. Traditional Cloud-based centralized network architectures cannot cater to real-time responses demanded by such unified IoT applications. Moreover, introducing Fog nodes within the network architecture, though a promising alternative, cannot sustain the burden of a huge number of applications that culminates in massive data handling. In this paper, we envision employing lessons learned from context-aware computing, specifically context sharing among interdependent vertical IoT applications to address this delay requirement of such unified IoT applications by enacting context sharing among Fog nodes for minimizing system delay. The detailed network model and context sharing mechanism have been presented and the service time minimization has been framed as an optimization problem. Algorithms for context sharing and delay tolerant load balancing have been presented and simulation results carried out demonstrate the efficacy of the proposed methodology.

Cloud infrastructures are capable of leveraging massive computational as well as data processing capabilities in virtualized environments. Emerging applications on today's clouds are data intensive and this has led to the trend of employing data-parallel frameworks, like Hadoop and its myriad descendants, for handling such massive data requirements. Scheduling of jobs in such frameworks is in essence a two-step process, where the block-data distribution follows mapping of computations among those resources. Since most Hadoop-based systems make these two decisions independently, it seems a promising prospective to map computations within cloud resources based on data blocks already distributed to them. This paper proposes data partitioning and placement aware computation scheduling scheme (DPPACS), a data and computation scheduling framework that adopts the strategy of improving computation and data co-allocation within a Hadoop cloud infrastructure based on knowledge of data blocks availability. Accordingly, this paper proposes a data-partitioning algorithm, a novel partition-cum-placement algorithm and finally proposes a computational scheduling algorithm that exploits knowledge of data availability at different clusters. The proposed DPPACS has been implemented on a test bed and its comparative performance results with respect to Hadoop's default data placement strategy have been presented. Experiments conducted herein conclusively demonstrate the efficacy of the proposed DPPACS.