Federated learning (FL) is a promising approach for processing on-board data in vehicular networks due to its distributed nature and its ability to accurately and efficiently handle the large amount of sensed data. However, training and transmitting the model parameters during FL process can consume a significant amount of energy and time, which is not suitable for applications with strict real-time requirements. Moreover, the dynamicity of the vehicular network, as well as the varying capabilities of each vehicle, can impact the performance of the training process, bringing to the forefront the optimization of the participants selection and their resources. In this paper, we propose VOC-FL, a Vehicular-based Offloading and Clustering framework supported by FL. The proposed scheme bypasses communication bottlenecks by enabling groups of vehicles to train models simultaneously, with only the Cluster Head (CH) sending the aggregated results of each cluster to the roadside units for further processing. To form the clusters, we select a CH for each cluster based on multiple metrics, including stability, computational resources, bandwidth, and network topology. Moreover, the CH runs an offloading strategy that allows struggling nodes with limited computational resources to offload their tasks to other nodes with enough resources within the cluster, enabling efficient and effective use of resources.

With an innovative door opened for a new era of Machine Learning, Federated Learning (FL) is now revolutionizing Artificial Intelligence. It exploits both decentralized data and decentralized computation to preserve user privacy. Albeit its popularity and being the most widely used framework nowadays, FL becomes a sub-optimal solution when the convergence of the global model occurs at a slow pace, which exacerbates the communication bottlenecks. To address this challenge, we propose in this paper CISCO-FL, a Clustered FL with Intelligent Selection and Computation Offloading. First, we partition the clients into different groups, where sub-aggregations of the clients models are performed at each cluster before the global aggregation. Second, we study the computing resources of the clients, and we embed in the proposed approach an intelligent offloading model, where the clients with high computational resources can assist and optimize the model of those struggling with limited resources. As such, both communication cost and computation resources are reduced and optimized. Finally, thorough experimental results are presented to support our findings and validate our model.

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.

Internet of Things (IoT), Digital Twin (DT), and Federated Learning (FL) are redefining the future vision of globalization. While IoT is about sensing data from physical devices, DTs reflect their digital representation and enable optimized decision-making by tightly integrating Artificial Intelligence (AI). Although swiftly growing, DTs are raising new challenges in privacy concerns, which are nowadays addressed by FL. However, the limited IoT resources, the communication overhead, and the lack of trust among clients are major obstacles that hinder the effectiveness of learning systems. In this paper, we design a new IoT-based architecture empowered by DT to improve the efficiencies of limited-resources devices. On top of this architecture, we leverage FL to construct the DT models. We further propose CISCO-FL, a Clustered FL with Intelligent Selection and Computation Offloading. Particularly, we study the computing resources of the clients and the quality of their models, and we embed in the proposed approach an intelligent offloading model, where the clients with high computational resources can assist and optimize the model of those struggling with limited resources. As such, both communication cost and computation resources are reduced and optimized. Finally, thorough experimental results are presented to support our findings and validate our model.

As an alternative centralized systems, which may prevent data to be stored in a central repository due to its privacy and/or abundance, federated learning (FL) is nowadays a game changer addressing both privacy and cooperative learning. It succeeds in keeping training data on the devices, while sharing locally computed then globally aggregated models throughout several communication rounds. The selection of clients participating in FL process is currently at complete/quasi randomness. However, the heterogeneity of the client devices within Internet-of-Things environment and their limited communication and computation resources might fail to complete the training task, which may lead to many discarded learning rounds affecting the model accuracy. In this article, we propose FedMCCS, a multicriteria-based approach for client selection in FL. All of the CPU, memory, energy, and time are considered for the clients resources to predict whether they are able to perform the FL task. Particularly, in each round, the number of clients in FedMCCS is maximized to the utmost, while considering each client resources and its capability to successfully train and send the needed updates. The conducted experiments show that FedMCCS outperforms the other approaches by: 1) reducing the number of communication rounds to reach the intended accuracy; 2) maximizing the number of clients; 3) handling the least number of discarded rounds; and 4) optimizing the network traffic.

Driven by privacy concerns and the visions of deep learning, the last four years have witnessed a paradigm shift in the applicability mechanism of machine learning (ML). An emerging model, called federated learning (FL), is rising above both centralized systems and on-site analysis, to be a new fashioned design for ML implementation. It is a privacy-preserving decentralized approach, which keeps raw data on devices and involves local ML training while eliminating data communication overhead. A federation of the learned and shared models is then performed on a central server to aggregate and share the built knowledge among participants. This article starts by examining and comparing different ML-based deployment architectures, followed by in-depth and in-breadth investigation on FL. Compared to the existing reviews in the field, we provide in this survey a new classification of FL topics and research fields based on thorough analysis of the main technical challenges and current related work. In this context, we elaborate comprehensive taxonomies covering various challenging aspects, contributions, and trends in the literature, including core system models and designs, application areas, privacy and security, and resource management. Furthermore, we discuss important challenges and open research directions toward more robust FL systems.