Conventional systems are usually constrained to store data in a centralized location. This restriction has either precluded sensitive data from being shared or put its privacy on the line. Alternatively, federated learning (FL) has emerged as a promising privacy-preserving paradigm for exchanging model parameters instead of private data of Internet of Things (IoT) devices known as clients. FL trains a global model by communicating local models generated by selected clients throughout many communication rounds until ensuring high learning performance. In these settings, the FL performance highly depends on selecting the best available clients. This process is strongly related to the quality of their models and their training data. Such selection-based schemes have not been explored yet, particularly regarding participating clients having high-quality data yet with limited resources. To address these challenges, we propose in this article FedAUR, a novel approach for an adaptive upgrade of clients resources in FL. We first introduce a method to measure how a locally generated model affects and improves the global model if selected for aggregation without revealing raw data. Next, based on the significance of each client parameters and the resources of their devices, we design a selection scheme that manages and distributes available resources on the server among the appropriate subset of clients. This client selection and resource allocation problem is thus formulated as an optimization problem, where the purpose is to discover and train in each round the maximum number of samples with the highest quality in order to target the desired performance. Moreover, we present a Kubernetes-based prototype that we implemented to evaluate the performance of the proposed approach.

Predicting context-aware activities using machine-learning techniques is evolving to become more readily available as a major driver of the growth of IoT applications to match the needs of the future smart autonomous environments. However, with today’s increasing security risks in the emerging cloud technologies, which share massive data capabilities and impose regulation requirements on privacy, as well as the emergence of new multiuser, multiprofile, and multidevice technologies, there is a growing need for new approaches to address the new challenges of autonomous context awareness and its fine-grained security-enforcement models. The solutions proposed in this work aim to extend our previous LCA-ABE work to provide an intelligent, dynamic creation of context-aware policies, which has been achieved through deploying smart-learning techniques. It also provides data consent, automated access control, and secure end-to-end communications by leveraging attribute-based encryption (ABE). Moreover, our policy-driven orchestration model is able to achieve an efficient, real-time enforcement of authentication and authorization (AA) as well as federation services between users, service providers, and connected devices by aggregating, modelling, and reasoning context information and then updating consent accordingly in autonomous ways. Furthermore, our framework ensures that the accuracy of our algorithms is above 90% and their precision is around 85%, which is considerably high compared to the other reviewed approaches. Finally, the solution fulfills the newly imposed privacy regulations and leverages the full power of IoT smart environments.

The rapid progress of the artificial intelligence sector has greatly impacted Vehicular Edge Components (VEC) in the Vehicular Ad-Hoc Network (VANET). Various AI applications, including automatic driving, preaccident alerts, and video broadcasting, have become essential to meet VANET’s diverse requirements. However, implementing these applications in the resource-constrained Urban Sensing Environment poses challenges. To overcome this, we propose a novel approach that partitions resource-intensive ciphertext-policy attribute-based decryption (CP-ABE) tasks based on ciphertext policy into subciphertexts using machine learning. Our technique, Ciphertext-distribution DE (CD-DE), utilizes differential evolution (DE) to decrypt CP-ABE tasks on VECs. It includes a selection algorithm that allows the data owner vehicle to choose VEC components for decryption operations. Compared to widely used techniques such as Particle Swarm Optimization (PSO) and Genetic Algorithm (GA), CD-DE offers lower overhead and provides accurate near-optimal solutions across most scenarios. Our study demonstrates the effectiveness of CD-DE in improving the efficiency and accuracy of resource-intensive CP-ABE tasks in VANETs. The proposed approach addresses the challenges posed by limited resources, bandwidth, and workload constraints in the Urban Sensing Environment. By enabling efficient implementation of AI-based applications in vehicles within urban environments, our approach holds promise for enhancing VANETs’ capabilities.

In urban environments, efficiently decrypting CP-ABE in VANETs is a significant challenge due to the dynamic and resource-constrained nature of these networks. VANETs are critical for ITS that improve traffic management, safety, and infotainment through V2V and V2I communication. However, managing computational resources for CP-ABE decryption remains difficult. To address this, we propose a hybrid RL-DE algorithm. The RL agent dynamically adjusts the DE parameters using real-time vehicular data, employing Q-learning and policy gradient methods to learn optimal policies. This integration improves task distribution and decryption efficiency. The DE algorithm, enhanced with RL-adjusted parameters, performs mutation, crossover, and fitness evaluation, ensuring continuous adaptation and optimization. Experiments in a simulated urban VANET environment show that our algorithm significantly reduces decryption time, improves resource utilization, and enhances overall efficiency compared to traditional methods, providing a robust solution for dynamic urban settings.

As the cloud moves from monolithic infrastructure to a self-isolated cloud native microservice environment, automation is becoming an important aspect for the management of the application life cycle. In this context, there are many tools available that can monitor these applications and raise alarms. However, automated orchestration is still in its early stages, and the available solutions are not capable of monitoring the whole system from application to hardware levels and performing automated operations within the system. Moreover, IP Multimedia Subsystem (IMS), which is the core part of the Telecom industry, has switched to a microservice environment. These IMS services are critical and need to be proactively monitored to provide automated orchestration operations. In this paper, we address the aforementioned problem by proposing a new scheme for monitoring the metrics from different sources and proactively and automatically performing orchestration using machine learning while improving the scalability of the cloud native Virtual IMS. Experiments carried out with a real cloud-native IMS running in a kubernetes cluster explore the relevance, efficiency and scalability of the proposed scheme.