Large machine-learning training datasets can be distilled into small collections of informative synthetic data samples. These synthetic sets support efficient model learning and reduce the communication cost of data sharing. Thus, high-fidelity distilled data can support the efficient deployment of machine learning applications in distributed network environments. A naive way to construct a synthetic set in a distributed environment is to allow each client to perform local data distillation and to merge local distillations at a central server. However, the quality of the resulting set is impaired by heterogeneity in the distributions of the local data held by clients. To overcome this challenge, we introduce the first collaborative data distillation technique, called CollabDM, which captures the global distribution of the data and requires only a single round of communication between client and server. Our method outperforms the state-of-the-art one-shot learning method on skewed data in distributed learning environments. We also show the promising practical benefits of our method when applied to attack detection in 5G networks.

Trust evaluation and management schemes have been extensively employed in a bid to alleviate diverse insider attacks. These trust values are often ascertained by taking into consideration trust parameters to determine the honesty and reliability of a vehicle and evaluating a weighted sum of the said parameters. To achieve precision and to reflect a reasonable impact of these parameters, rational weight values are imperative. Accordingly, this research primarily emphasizes on the quantification of weights associated with the contributing trust attributes by proposing a novel trust management mechanism that utilizes contextual information in addition to employing relevant impacting quantities as weights to formulate trust evaluations. Moreover, the envisaged trust management model incorporates 1) attack resilience while constituting certain parameters and 2) an adaptive and flexible threshold to mitigate malevolent behaviors. The simulation results depict that the devised parameters and the formulated trust aggregation cater to the dynamic nature of vehicular networks demonstrating the rationality of the weights’ quantification, and the introduction of the adaptive threshold for misbehavior detection aligns well with the requirements of the ever-changing vehicular networks.

The emerging yet promising notion of trust amalgamated with the paradigm of the Internet of Vehicles (IoV) has been of significant interest for researchers in both industry and academia. Trust management models are being widely employed in diverse domains seeking to mitigate multifarious insider attacks where dishonest entities counterfeit or alter the messages from legitimate users, subsequently, endangering the lives of drivers, passengers, and vulnerable pedestrians. While vehicles form opinions towards the fellow network participants, numerous influencing attributes in reference to the interactions among the vehicles are aggregated to develop a final belief on a target vehicle. The significance of these trust impacting attributes is often depicted by manually assigning a weight to each parameter. The aim of this research is to address the challenges of quantification of the influencing trust attributes and the weights associated with these parameters, subsequently, quantifying the aggregated trust. Furthermore, the time-based behavior of the aggregated trust along with the contributing parameters is also analyzed in this paper to study the behavior of each vehicle and to identify suitable trust-based patterns for safety-critical and non-safety vehicular applications. Performance analysis demonstrates that the trust values computed via the envisaged trust model are lower than those calculated via the benchmark scheme.

Recently, the convergence of the paradigm of the Internet of Things (IoT) with social networking has been widely recognized as an emerging interdisciplinary area, lately, referred to as the Social Internet of Things (SIoT). SIoT enables the object-to-object interaction and provides the platform for these objects to autonomously socialize with the other objects in the network in a bid to overcome the key challenges of IoT, i.e., data discovery and composition, network navigability, trust management, etc. Trust plays a significant role while establishing these inter-object social relationships and it is also essential to observe the trustworthiness of an object before relying on information provided by them. A number of trust evaluation models for SIoT have been proposed in the literature, nevertheless, most of these models suffer in validating and testing their respective models due to lack of appropriate datasets. To address this issue, this paper proposes a scalable plug and play trust platform referred to as SCalable and Robust Trust platform for SIoT (SCaRT-SIoT) to provide the dataset to test and analyze various SIoT-based trust models of the research community.

Over the past few decades, Vehicular Ad hoc Networks have attracted the attention of numerous researchers from both academia and industry. Today, this promising wireless communication technology plays an indispensable role as vehicles exchange low-latent safety critical messages with one another in a bid to make the road traffic more safer, efficient, and convenient. However, dissemination of malicious messages within the network not only significantly reduces the network performance but also becomes a source of threat for the passengers and vulnerable pedestrians. Accordingly, a number of trust models have been recently proposed in the literature to ensure the identification and elimination of malicious vehicles from the network. These trust models primarily rely on the aggregation of both direct and indirect observations and evict the malicious vehicles based on a particular threshold set on this composite trust value. Nevertheless, setting-up of this threshold poses a significant challenge especially owing to diverse influential factors in such a dynamic and distributed networking environment. To this end, in this manuscript, machine learning has been employed to compute the aggregate trust score for flagging and evicting of the malicious vehicles from a vehicular network. It is evident from the simulated results that the devised method is both accurate and scalable.

Considerable innovative advancements in vehicular communication have led to the emerging yet promising paradigm of Internet-of-Vehicles, wherein vehicles exchange safety-critical information with minimal delay for ensuring road safety as well as efficacious traffic flows. It is, therefore, indispensable that these safety messages are authentic and reliable, and have originated from a legitimate vehicle. This demands establishing trust among vehicles such that the malicious and dishonest vehicles (and their malicious content) could be flagged and subsequently eliminated from the network. The risk manifolds if a malicious vehicle gets elected as the cluster head of other vehicles thereby compromising the safety of vehicular passengers and pedestrians on the road. Hence, intelligent algorithms should be in place to opt for the trusted and resource efficient cluster heads which could enhance the overall security and efficiency of their respective clusters. To this end, in this paper, we have proposed a scalable hybrid trust model which takes into account a composite metric encompassing the weighted trust score and available resources of each vehicle for identification of multiple malicious vehicles in real-time, and for meeting the stringent performance requirements of vehicular safety applications. Moreover, an optimal role assignment scheme based on the Hungarian algorithm has been proposed for electing the optimal cluster head, proxy cluster head, and followers among the members of a vehicular cluster so as to maximize its overall efficiency. Preliminary simulations have been carried out and are also presented in this paper.