Load balancing is critical for maintaining computing systems’ stability and achieving optimal performance. Its significance is widely recognized across different computing fields, particularly in the context of heterogeneous systems. These systems comprise computing devices with varying computational capabilities and architectures, each optimized for specific workloads. This heterogeneity introduces dynamic resource constraints, architectural mismatches, and unpredictable task-device affinity, which aggravates the challenges of load balancing. This paper presents an AI-driven load balancing solution for real-time distributed heterogeneous systems. Our approach continuously monitors the system state, capturing key factors that influence performance, such as task and device characteristics. Leveraging AI-based models, it computes a dynamic load index for each device based on the collected data. Using these load estimations, the method predicts potential imbalances through a novel imbalance metric and proactively schedules incoming applications to the most suitable devices, ensuring system-wide balance. To validate our approach, we first evaluated the prediction models by comparing a variety of machine learning algorithms with device-specific deep learning models, with the latter achieving superior accuracy. We then compared our method against widely used scheduling techniques across diverse workloads. The results show that our approach achieves more balanced workload distribution, faster execution, higher throughput, improved resource utilization, and reduced energy consumption across all scenarios, showcasing its adaptability to dynamic conditions and its applicability in real-world settings.

Summary Data aggregation, in its basic form, has been widely used, and several solutions have been proposed for IoT environments. However, to calculate statistical metrics, detect anomalies, and predict future trends, we need to perform various data analysis functions on the aggregated data. Recently, multifunctional data aggregation (MFDA) has been proposed to calculate various statistical functions such as sum, mean, variance, covariance, and analyze of variance (ANOVA). The purpose of MFDA is to enable the improvement of decision making, resource allocation and system performance by providing diverse and varied statistical data. However, the existing solutions involving MFDA generate significant communication and calculation costs. Furthermore, they cannot prevent malicious aggregators from sending fake data. Recently, the Fog computing paradigm has been adopted in IoT environments to address various challenges and enhance the efficiency of data processing and storage. The blockchain technology has been integrated in various IoT applications to enhance the security, increase transparency, and facilitate decentralized data exchange and transactions. In this article, we propose BMDA, a blockchain‐based secure multifunctional data aggregation method for IoT‐Fog environments. BMDA employs an encoding function to structure the data before their transmission. Furthermore, to ensure privacy preservation, authentication, data integrity and to resist malicious aggregators, we employ Paillier homomorphic encryption, BLS signature, and blockchain technology. The security analysis demonstrates the robustness of our proposal, and the performance analysis in terms of computations and communications shows the effectiveness of BMDA compared to existing solutions.

Summary Heterogeneous systems deliver high computing performance when effectively utilized. It is crucial to execute each application on the most suitable device while maintaining system balance. However, achieving equal distribution of the computing load is challenging due to variations in computing power and device architectures within the system. Moreover, scheduling applications at real‐time further complicates this task, as prior information about the submitted applications is absent. In this context, we introduce “Equalizer,” a real‐time load balancer for heterogeneous systems. “Equalizer” leverages machine learning to continuously monitor the system's state, predicting optimal devices for application execution at runtime. It assigns applications to devices that minimize system imbalance. To quantify system imbalance, we propose a novel metric that reflects the disparity in computing loads across the system's devices. This metric is calculated using predicted execution times of applications. To validate the performance of “Equalizer,” we conducted a comparative study against widely adopted approaches, namely Round Robin and Device Suitability. The experiments were performed on a heterogeneous cluster comprising a master host and three slave servers, equipped with a total of 4 central processing units (CPUs) and 4 graphics processing units (GPUs). All approaches were deployed on the cluster and evaluated using three distinct workloads categorized by their computing intensity: medium intensity, heavy intensity, and a combination of heavy and medium intensity, simulating real‐world scenarios. Each workload consisted of a set of 80 OpenCL applications with varying input data sizes. The experimental results demonstrate that “Equalizer” effectively minimized the system's imbalance, reduced the idle time of devices, and eliminated overloads. Moreover, “Equalizer” exhibited significant improvements in workload execution time, resource utilization, throughput, and energy consumption. Across all tested scenarios, “Equalizer” consistently outperformed alternative approaches, showcasing its robustness, adaptability to dynamic environments, and applicability in real‐world practice.

Vehicular Fog Computing (VFC) is a paradigm of vehicular networks that has a set of advantages such as agility, efficiency, and reduced latency. The VFC is vulnerable to a variety of attacks, and existing security measures in traditional networks are not necessarily applicable to VFC. Among these attacks, we can find the Sybil attack that allows a vehicle to create multiple identities to perform malicious operations. In this paper, we propose a blockchain-based mechanism to detect Sybil attacks in VFC networks. The detection process consists of two levels; the first one is targeted toward the verification of the vehicle’s position by the FN using the Received Signal Strength Indicator (RSSI) technique. The FN delivers a position proof, if its position is valid, and stores it in the blockchain. At this point, the set of the obtained position proofs constitutes a trajectory. The second level is projected toward a comparison between the trajectories of the vehicles reporting an event. Two trajectories that pass through the same FNs at the same time, will be considered as Sybil trajectories. The objective of these two-level detections is to identify the Sybil attack in several attack scenarios performed by a powerful adversary. Our analysis shows that existing proposals cannot deal with such an adversary. Moreover, simulation results show the efficiency of our proposal in terms of communication, computation, and detection rate. Indeed, our system can reach a detection rate of 98% when the malicious vehicle generates several aliases simultaneously and sends position requests to the FN for each generated pseudonym.

The secure multidimensional data aggregation (MDA) has been widely investigated in smart grid for smart cities. However, previous proposals use heavy computation operations either to encrypt or to decrypt the multidimensional data. Moreover, previous fault-tolerant mechanisms lead to an important computation cost, and also a high communication cost when considering a separate identification phase. In this article, we propose an efficient and secure MDA scheme, named ESMA. Unlike existing schemes, the multidimensional data in ESMA are structured and encrypted into a single Paillier ciphertext and thereafter, the data are efficiently decrypted. For privacy preserving, the Paillier cryptosystem is adopted in a fog computing-based architecture, and to achieve efficient authentication, the batch verification technique is applied. Besides, ESMA is fault tolerant, i.e., even if some of the smart meters fail to send their data, the final aggregation result will not be affected. Furthermore, ESMA can be adapted to respond to other queries than the summation of data. The performance analysis demonstrates the cost efficiency of ESMA both in computation and communication and the scalability as well. For instance, with a 16-bits size for each data type and 500 reporting smart meters, 40 data types can be supported in a single Paillier ciphertext. ESMA also resists various security attacks and preserves the user's privacy.

The rapid growth of Internet of Things (IoT) technology offers huge opportunities and also brings many new challenges related to the authentication in (IoT) devices. Using passwords or pre-defined keys have drawbacks that limit their use for different (IoT) applications like smart hotel and smart office. In fact, they didn't provide temporary access to data in such reservation systems. Thus, authenticating users basing on password mechanism is not feasible. In this paper, we propose a new Token-Based Lightweight User Authentication (TBL UA) for (IoT) devices, which is based on token technique in order to enhance the robustness of authentication. Security analysis shows the security strength of the proposed scheme such as token security, Perfect Forward Secrecy (PFS), etc. In addition, the presented performance analysis shows that it is a strong competitor among existing ones for user authentication in (IoT) environments.

With the pervasiveness of the Internet of Things (IoT) and the rapid progress of wireless communications, Wireless Body Area Networks (WBANs) have attracted significant interest from the research community in recent years. As a promising networking paradigm, it is adopted to improve the healthcare services and create a highly reliable ubiquitous healthcare system. However, the flourish of WBANs still faces many challenges related to security and privacy preserving. In such pervasive environment where the context conditions dynamically and frequently change, context-aware solutions are needed to satisfy the users' changing needs. Therefore, it is essential to design an adaptive access control scheme that can simultaneously authorize and authenticate users while considering the dynamic context changes. In this paper, we propose a context-aware access control and anonymous authentication approach based on a secure and efficient Hybrid Certificateless Signcryption (H-CLSC) scheme. The proposed scheme combines the merits of Ciphertext-Policy Attribute-Based Signcryption (CP-ABSC) and Identity-Based Broadcast Signcryption (IBBSC) in order to satisfy the security requirements and provide an adaptive contextual privacy. From a security perspective, it achieves confidentiality, integrity, anonymity, context-aware privacy, public verifiability, and ciphertext authenticity. Moreover, the key escrow and public key certificate problems are solved through this mechanism. Performance analysis demonstrates the efficiency and the effectiveness of the proposed scheme compared to benchmark schemes in terms of functional security, storage, communication and computational cost.

Many resource-constrained applications make use of data aggregation in order to prolong the network lifetime. However, the resource-constrained devices are usually deployed in unattended environments in which providing security is of paramount importance. This leads to various attacks that can occur during data aggregation process. Internal attacks such as selective forwarding represent the most dangerous ones since they cannot be detected by existing cryptography- based protocols proposed to secure data aggregation. In this work, we propose a two-levels verification, in which data is verified using cryptography and intrusion detection techniques. Indeed, a lightweight homomorphic encryption is combined with a game-theory based technique to efficiently secure data aggregation. Our analysis and results show the applicability of the system for aggregation-based resource-constrained applications, especially those considering sensitive information (e.g. health monitoring, military) where the time-efficient detection is crucial.