The Cyber-physical System Security Lab (CSS Lab) conducts research addressing a wide spectrum of security problems in the cyber-physical systems space. We seek to develop mechanisms to address security, trust, and resilience of representative computer applications designed for the Internet of Things, distributed embedded systems, automotive systems, and related network-connected critical infrastructure.
Automotive Security
The interconnection of automotive in-vehicle networks with the outside world poses a significant security risk. Moreover, the modern vehicle interface exposes driving systems to cyberattacks. In addition, the communication network does not implement any security protocols. Therefore, we aim to secure both the intra-vehicular and inter-vehicular networks.
-
From Weeping to Wailing: A Transitive Stealthy Bus-off Attack [PDF]
In this project, we introduce three software-based improvements that greatly increase both the efficiency and effectiveness of the WeepingCAN bus-off attack. [Project Lead: Paul Agbaje]
-
A Framework for Consistent and Repeatable Controller Area Network IDS Evaluation [PDF]
We introduce a framework for benchmarking CAN IDS algorithms to compare them using consistent metrics, which facilitates fair and reproducible experiments. [Project Lead: Paul Agbaje]
-
In-Vehicle Network Anomaly Detection Using Extreme Gradient Boosting Machine [PDF]
We present an anomaly detection technique that uses an extreme gradient boosting machine (GBM) learning algorithm to categorize unexpected occurrences in the CAN data payload. [Project Lead: Afia Anjum]
-
Privacy-Preserving Intrusion Detection System for Internet of Vehicles using Split Learning
In this project, we implemented a split learning-based privacy-preserving IDS that deploys IDS on edge devices without sharing sensitive raw data. In addition, we implemented a regret minimization-based adaptive offloading technique that reduces the training time on resource-constrained devices. [Project Lead: Paul Agbaje]
-
Interoperability Challenges in the Internet of Vehicles [PDF]
In this project, we do an in-depth analysis of the present state of interoperability and comprehensively survey the challenges in IoV. We present a taxonomy of interoperability approaches, review solutions that prior work has proposed, and provide insights on addressing the current challenges. Finally, we identify open problems that persist and future directions for research. [Project Lead: Paul Agbaje]
Efficient and Secure Wireless Communication for Connected Devices
Efficient and secure wireless communication for connected devices is essential in today’s interconnected world. This domain focuses on developing robust, time-sensitive, and energy-efficient communication protocols to ensure reliable real-time data transmission while protecting against unauthorized access and cyber threats. By optimizing network efficiency and enhancing security measures, we aim to empower the seamless operation of IoT ecosystems with improved system reliability.
-
Noise Injection Attacks on Semantic Communication
This project aims to introduce targeted noise into the transmitted semantic information over-the-air, thereby deceiving the semantic communication. Additionally, we aim to propose an effective defense mechanism against such attacks [Project Lead: Afia Anjum]
-
Resource Allocation for NDN-enabled 5G Communication
The coexistence of diverse traffic types within 5G-NR, each with unique Quality of Service (QoS) requirements such as Enhanced Mobile Broadband and Ultra-Reliable and Low Latency Communications (URLLC), necessitates fair resource allocation strategies. Therefore, we aim to propose a novel energy-efficient resource allocation approach that satisfies the diverse QoS of various 5G-NR traffic types. [Project Lead: Afia Anjum]
-
Deadline-Aware Named Data Networking for Time-Sensitive IoT Applications [PDF]
Named Data Networking (NDN) has evolved as a networking model that can facilitate Internet of Things (IoT) applications by providing a name-based communication model, in-network caching, and inherent support for data-centric security. However, despite the benefits, the best-effort NDN cannot offer the deterministic data delivery required by safety-critical IoT applications. This project proposes a novel deadline-aware NDN protocol that utilizes a critical deadline first scheduler to prioritize traffic based on the approaching deadline. [Project Lead: Afia Anjum]
-
Deadline-Based Class Assignment for Time-Sensitive Network Frame Preemption
In this project, we present a novel approach to address priority inversion in TSN that prioritizes frames during network configuration, determines traffic paths offline with integer linear programming (ILP), and schedules transmissions online using the earliest deadline first (EDF) algorithm.
-
Towards Mitigating Blackhole Attack in NDN-Enabled IoT [PDF]
A malicious node may intentionally drop packets exploiting the ‘sleep mode’ of IoT devices, referred to as a blackhole attack. In this project, we present a reputation-based forwarding approach with a reactive reputation updating mechanism to mitigate blackhole attacks in the NDN-enabled IoT network. [Project Lead: Afia Anjum]
-
Towards Named Data Networking Technology: Emerging Applications, Use Cases, and Challenges for Secure Data Communication [PDF]
This project aims to present the concepts of NDN architecture and a comprehensive overview of emerging application use cases of the technology for secure data communications. We discuss the integration of NDN with the current Internet protocol and highlight how NDN works as a facilitator for addressing numerous concerns related to unique applications. Furthermore, we highlight the trust management and security aspects of NDN. Finally, we outline challenges relating to NDN adoption and present some proposed solutions. [Project Lead: Afia Anjum]
-
Deep Reinforcement Learning for Energy-Efficient Task Offloading in Cooperative Vehicular Edge Networks [PDF]
In this project, we propose a cooperative offloading scheme for vehicular nodes, using vehicles as mobile edge servers, which minimizes energy consumption and network delay. [Project Lead: Paul Agbaje]
Safety-critical Analysis of Vehicular Perception Systems
The research domain centers around analyzing and formulating attack vectors and developing corresponding defensive mechanisms to safeguard autonomous driving in adversarial environments and ensure road safety. These attack vectors may stem from sensor-based inputs such as cameras, model-based manipulations such as attacking the neural network perception system, or a hybrid combination of both methodologies.
-
Adversarial Analysis of Luminescent Markers in Autonomous Vehicle Perception Systems
This paper investigates the risks associated with using luminescent road markers in autonomous driving systems. We introduce novel luminescent adversarial attacks aimed at deceiving lane detection models, exploiting textural changes in the markers to cause misdetection of lanes. Through experiments, we highlight the effectiveness of these attacks, emphasizing the importance of robust defenses against adversarial manipulations in ensuring the safety of autonomous driving systems.[Project Lead: Arkajyoti Mitra]
Robust and reliable machine learning for dynamic networks
-
FedCime: An Efficient Federated Learning Approach For Clients in Mobile Edge Computing [PDF]
We present a novel tier-based FL approach that selects high-utility mobile clients likely to complete training to replace migrating clients during each round of training in a federated learning setting. [Project Lead: Paul Agbaje]
Explainability in Clinical Decision Support Systems
The clinical decision support system (CDSS) employed in healthcare relies on a variety of machine learning (ML) and deep learning (DL) algorithms to facilitate effective decision-making. However, the opaque nature of these ML and DL algorithms poses challenges to trust in the CDSS among healthcare providers. Thus, we aim to address this issue by incorporating explainability and interoperability of the CDSS decisions.
-
Explainability in Clinical Decision Support Systems for Interpretability, Trustworthiness, and Usability
-
We examine how state-of-the-art explainable artificial intelligence (XAI) algorithms can be integrated into CDSSs to enhance trust, fairness, and interpretability in healthcare decision-making processes. [Project Lead: Afia Anjum]