Automating the task of fault detection and diagnosis is essential for reducing the operational and maintenance costs of power electronic converters. Hence, this paper introduces an online optimization methodology for digital twins to diagnose multiple faults in grid connected inverters. In addition to enhancing accuracy, we channelize our efforts towards reducing computational complexity in parameter configuration under a limited set of training data. To address cybersecurity vulnerabilities during the training phase of our digital twin, we performdata sanitizationwith the aid of a quantitative association rule mining technique. For classification performance assessment, various fault cases in a virtual synchronous generator are considered to demonstrate the efficacy of our approach. Our research outcomes reveal increased accuracy and fidelity levels achieved by our digital-twin design, paving the way for reliable and trustworthy decision-making under anomalous conditions.

While substantial research has been devoted to optimizing model performance, convergence rates, and communication efficiency, the energy implications of federated learning (FL) within Artificial Intelligence of Things (AIoT) scenarios are often overlooked in the existing literature. This study examines the energy consumed during the FL process, focusing on three main energy-intensive processes: pre-processing, communication, and local learning, all contributing to the overall energy footprint. We rely on the observation that device/client selection is crucial for speeding up the convergence of model training in a distributed AIoT setting and propose two clustering-informed methods. These clustering solutions are designed to group AIoT devices with similar label distributions, resulting in clusters composed of nearly heterogeneous devices. Hence, our methods alleviate the heterogeneity often encountered in real-world distributed learning applications. Throughout extensive numerical experimentation, we demonstrate that our clustering strategies typically achieve high convergence rates while maintaining low energy consumption when compared to other recent approaches available in the literature.

One of the key concerns related to the pervasive integration of artificial intelligence and machine learning (AI/ML) models in vehicular-to-everything (V2X) communication systems pertains to adversarial attacks, which may lead trained models to exhibit undesirable behaviors. As security and user safety are tightly coupled in V2X, ensuring the resilience of AI/ML models against adversaries becomes indispensable. However, addressing adversarial attacks poses a challenging task, requiring appropriate countermeasures to elevate the trustworthiness of the targeted AI/ML models. In this paper, we propose a deep reinforcement learning (DRL)-based approach to defend against two data poisoning attacks, namely label-flipping and policy induction. Extensive evaluation with the aid of an open-source dataset demonstrates that our scheme outperforms benchmark classifiers, achieving significantly superior detection performance in the presence of label-flipping attacks. The effectiveness of our DRL-based approach is also showcased under different adversarial strategies in the policy induction attack.

Emerging vehicle-to-everything (V2X) systems call for a diverse set of novel mechanisms to address vulnerabilities and security breaches. In this context, misbehavior detection approaches aim to detect malicious behavior of rogue V2X entities and possible attacks that may originate from them. In this paper, we introduce a data-driven ensemble framework which jointly leverages clustering and reinforcement learning to detect misbehaviors in unlabeled vehicular data. A rigorous detection assessment using an open-source dataset reveals meaningful performance trends for various attacks. In particular, while the majority of attacks can be effectively detected, detection may be curtailed for certain misbehavior types due to partly inaccurate clustering and erratic activity of the attacker over time. Performance comparison against benchmark detectors reveals the robustness of our approach in the presence of potentially inconsistent or mislabeled training data. The real-time detection capabilities of our framework are also explored in an effort to evaluate its practical feasibility in mission-critical V2X scenarios.

Vehicle-to-everything (V2X) communication is contributing towards the realization of futuristic vehicular networks such as Internet-of-Vehicles (IoV). The IoV is expected to usher in a new direction of intelligence and networking to achieve the goal of intelligent transport systems, which rely on the secure exchange of messages between vehicles and infrastructure. However, the transmission of false/incorrect data by malicious vehicles may cause serious damages on road safety. Therefore, it is crucial to detect safety-threatening incorrect information and mitigate potentially detrimental effects on road users. In this paper, we propose a reinforcement learning (RL)-based misbehavior detection approach for V2X scenarios. In our method, the RL-based detection model processes V2X data broadcast by vehicles as time-series at the roadside units, and classifies incoming data as misbehaving or genuine. We evaluate the proposed RL-based approach for detection of various attack types using an open-source dataset, and compare its performance against recent work in misbehavior detection. Our scheme is able to detect all types of misbehavior with a superior recall of 0.9970 and an F1 score of 0.9845, yielding a significant improvement over the benchmarks. Our research outcomes further reveal that misbehaving vehicles can be detected with a great accuracy of 0.9882 by exploiting real-time V2X information.

The transformative Internet-of-Vehicles (IoV) paradigm comes inadvertently with challenges which involve security vulnerabilities and privacy breaches. In this context, denial-of-service (DoS) attacks may perniciously affect the normal operation of IoV systems by causing extensive periods of network unavailability where legitimate vehicles are prevented from accessing vehicular services. In this paper, we offer an in-depth vulnerability assessment of 5G-enabled IoV systems when DoS attack variants are launched at multiple network domains. We further evaluate the resilience of an IoV-tailored authentication mechanism against DoS attacks under various configurations. A data-driven detection scheme is also proposed to address DoS variants in the radio access network, which take the form of false data injection attacks on the exchanged vehicular information. Our performance assessment with the aid of an open-source dataset reveals that the proposed scheme is able to accurately detect DoS traffic originated from malicious vehicles.

Networked industrial systems capitalize on recent advancements in sensing, communications, computing and storage to improve productivity, operational and cost efficiency. The proliferation of effective techniques for knowledge extraction drive a paradigm shift in industrial environments and provide a fertile ground for enhanced process monitoring and control capabilities. In an effort to shed light on industrial data management operations, this paper presents two different approaches for dealing with information processing tasks of aggregated sensor measurements. Such tasks constitute part of an end-to-end process monitoring solution which is implemented in an open-source platform following a modular, scalable and interpretable procedure. A mapping of the industrial data processing components to the operational principles and architecture of a cyber-physical system reveals useful insights for an automated supervision of critical processes and workflows.

The reliable acquisition of monitoring information is critical for several industrial use cases relying on wireless sensor network deployments. However, missing sensor measurements are typical in industrial systems empowered by cellular connectivity due to the stochastic nature of the wireless channel. In this paper, we propose a sensor data reconstruction scheme that exploits the hidden data dynamics to accurately estimate the missing measurements. Based on an analytical framework for the network model and a closed-form expression for the outage probability, the impact on the reconstruction error performance is thoroughly explored. Considering a dataset with high spatiotemporal correlation in the sensor observations, our proposed scheme is shown to outperform two baseline data recovery methods in terms of reconstruction error for various network configurations. In addition, despite the presence of imperfect cellular connectivity, our proposed scheme exhibits high event-detection accuracy.

Selection of the most appropriate communications technology for a smart grid (SG) application is far from trivial. We propose such a feasibility assessment starting from identification of key performance indicators (KPIs) required for peer-to-peer (P2P) energy trading and grid control operations from a communications perspective. A set of cross-disciplinary KPIs, both quantitative and qualitative, are considered from communications, power, business, actor involvement, financial, and demand side management categories. They serve as a general baseline for use cases, as there have been few previous works attempting to capture the essential features of P2P SG operations. The KPIs are briefly identified along with their relations to P2P energy trading and grid control. A straightforward comparison of the quantitative and qualitative KPIs' impact on technology selection is not feasible. This paper addresses the comparison with: 1) a prioritization of the KPIs using the analytic hierarchy process; 2) a comparison of technology solutions evaluated in our previous works against the KPIs' requirements; and 3) a total feasibility evaluation of the solutions against selected KPIs. The prioritization shows latency, reliability, security, scalability, robustness, costs of information and communication technologies (ICT) devices, and costs of ICT deployment are the most important KPIs in enabling P2P energy trading and grid control. Further, the technology feasibility assessment enables identification of the most suitable candidates for an SG application.

The realization of advanced electrical grid functionalities, e.g., phasor measurement unit (PMU) information acquisition in wide-area monitoring systems (WAMS), requires a scalable and reliable underlying communication technology. Cellular networks, e.g., LTE/LTE-A systems, appear as a promising option to facilitate the smart grid evolution. However, the effect of LTE communication constraints on power system state estimation (SE) has not been thoroughly addressed in the literature. In this paper, we investigate the impact of the LTE random access channel (RACH) reliability on the transmitted PMU measurements and, consequently, on the SE accuracy. In particular, we assess the SE performance based on the achieved reliability per PMU attained with i) increasing number of contending devices in the system, i.e., smart meters and PMUs, and ii) varying cell coverage range. Numerical results demonstrate that, under different network and traffic configurations, SE accuracy can be significantly affected by the RACH reliability levels. Useful insights can thus be drawn for a reliability-aware design of a power system state estimator in LTE-based WAMS.

The reliable support of event-driven massive machine-type communication (MTC) requires radical enhancements in the standard LTE random access channel (RACH) procedure to avoid performance degradation due to a high probability of collision in the preamble transmission. In this paper, we investigate the relation between the cell size and the number of preambles generated from a single or multiple root sequences and we study their impact on the achieved reliability. Based on an analytical expression of the RACH reliability per cell, we introduce an interference- and load-aware cell-planning mechanism that efficiently allocates the root sequences among multiple cells and regulates the traffic load to guarantee reliable support of MTC. In addition, we propose a realistic traffic model that accurately captures the event-driven nature of MTC traffic. Finally, a performance evaluation of a power distribution automation scenario with MTC-overload reveals the superior performance of our proposed mechanism in terms of RACH reliability against benchmarking network-deployment schemes.

Ongoing IEC 61850 standardization activities aim at improved grid reliability through advanced monitoring and remote control services in medium- and low-voltage. However, extending energy automation beyond the substation boundaries introduces the need for timely and reliable information exchange over wide areas. LTE appears as a promising solution since it supports extensive coverage, low latency, high throughput and Quality-of-Service (QoS) differentiation. In this paper, the feasibility of implementing IEC 61850 services over public LTE infrastructure is investigated. Since standard LTE cannot meet the stringent latency requirements of such services, a new LTE QoS class is introduced along with a new LTE scheduler that prioritizes automation traffic with respect to background human-centric traffic. Two representative grid automation services are considered, a centralized (MMS) and a distributed one (GOOSE), and the achievable latency/throughput performance is evaluated on a radio system simulator platform. Simulations of realistic overload scenarios demonstrate that properly designed LTE schedulers can successfully meet the performance requirements of IEC 61850 services with negligible impact on background traffic.