A Survey of Anomaly Detection for Connected Vehicle Cybersecurity and Safety

Rajbahadur, Gopi Krishnan; Malton, Andrew J.; Walenstein, Andrew; Hassan, Ahmed E.

doi:10.1109/ivs.2018.8500383

Cited by 52 publications

(18 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Regarding the analysis of multiple vehicles, the focus of scientific research lies in the field of misbehavior detection in VANETs based on V2V communication and less on centralized fleet monitoring. As indicated by recent surveys [82], [83], the VANET solutions, in general, are mostly designed for use in Road-Side Units or a Central Authority and not in a backend system. To give an example, Ghaleb et al [84] introduce context-awareness in their solution.…”

Section: A Fleet Securitymentioning

confidence: 99%

See 1 more Smart Citation

Context-Aware Security for Vehicles and Fleets: A Survey

2021

View full text Add to dashboard Cite

Vehicles are becoming increasingly intelligent and connected. Interfaces for communication with the vehicle, such as WiFi and 5G, enable seamless integration into the user's life, but also cyber attacks on the vehicle. Therefore, research is working on in-vehicle countermeasures such as authentication, access controls, or intrusion detection. Recently, legal regulations have also become effective that require automobile manufacturers to set up a monitoring system for fleet-wide security analysis. The growing amount of software, networking, and the automation of driving create new challenges for security. Context-awareness, situational understanding, adaptive security, and threat intelligence are necessary to cope with these everincreasing risks. In-vehicle security should be adaptive to secure the car in an infinite number of (driving) situations. For fleet-wide analysis and alert triage, knowledge and understanding of the circumstances are required. Context-awareness, nonetheless, has been sparsely considered in the field of vehicle security. This work aims to be a precursor to context-aware, adaptive and intelligent security for vehicles and fleets. To this end, we provide a comprehensive literature review that analyzes the vehicular as well as related domains. Our survey is mainly characterized by the detailed analysis of the context information that is relevant for vehicle security in the future.

show abstract

Section: A Fleet Securitymentioning

confidence: 99%

“…Special attention in the development of context-adaptive security measures will therefore have to be paid to the simulation of such data sets. For example, simulations are usually used to investigate security measures for VANETs [82], since almost the same problem arises here.…”

Section: G Simulation and Evaluationmentioning

confidence: 99%

Context-Aware Security for Vehicles and Fleets: A Survey

2021

View full text Add to dashboard Cite

show abstract

“…Detection. To detect unusual values, we adopt statistically anomaly detection technique that is widely used in safety-critical systems [32]. Note that we perform value-level fault detection rather than conventional bit-level mismatch detection due to the observation that many bit-flips may eventually get masked or only result in small value deviations which won't impact final performance.…”

Section: Inference: Range-based Anomaly Detectionmentioning

confidence: 99%

Analyzing and Improving Fault Tolerance of Learning-Based Navigation Systems

Wan¹,

Anwar²,

Hsiao³

et al. 2021

Preprint

View full text Add to dashboard Cite

Learning-based navigation systems are widely used in autonomous applications, such as robotics, unmanned vehicles and drones. Specialized hardware accelerators have been proposed for highperformance and energy-efficiency for such navigational tasks. However, transient and permanent faults are increasing in hardware systems and can catastrophically violate tasks safety. Meanwhile, traditional redundancy-based protection methods are challenging to deploy on resource-constrained edge applications. In this paper, we experimentally evaluate the resilience of navigation systems with respect to algorithms, fault models and data types from both RL training and inference. We further propose two efficient fault mitigation techniques that achieve 2× success rate and 39% qualityof-flight improvement in learning-based navigation systems.

show abstract

“…The objective is to detect rare or corrupted data, that are different from what we consider has normal data. This research topic has multiple practical applications, such as risk management [71], safety [72] or automatic inspection and non destructive control [73]. Anomalies can also be linked to the knowledge uncertainty [66] of the DNNs.…”

Section: Introductionmentioning

confidence: 99%

One Versus All for Deep Neural Network for Uncertainty (OVNNI) Quantification

Franchi¹,

Bursuc²,

Aldea³

et al. 2022

IEEE Access

View full text Add to dashboard Cite

Deep neural networks (DNNs) are powerful learning models yet their results are not always reliable. This drawback results from the fact that modern DNNs are usually overconfident, and consequently their epistemic uncertainty cannot be straightforwardly characterized. In this work, we propose a new technique to quantify the epistemic uncertainty of data easily. This method consists in mixing the predictions of an ensemble of DNNs trained to classify One class vs. All the other classes (OVA) with predictions from a standard DNN trained to perform All vs. All (AVA) classification. First of all, the adjustment provided by the AVA DNN to the score of the base classifiers allows for a more fine-grained inter-class separation. Moreover, the two types of classifiers enforce mutually their detection of out-of-distribution (OOD) samples, circumventing entirely the requirement of using such samples during training. The additional cost involved by the construction of the ensemble is offset by the ease of use of our proposed strategy and by its enhanced generalization potential, as it does not bind its performance in a given context to specific OOD datasets. The extensive experiments confirm the wide applicability of our approach, and our method achieves state of the art performance in quantifying OOD data across multiple datasets and architectures while requiring little hyper-parameter tuning.INDEX TERMS Uncertainty estimation, DNN ensembles, one vs all classification, all vs all classification.

show abstract

A Survey of Anomaly Detection for Connected Vehicle Cybersecurity and Safety

Cited by 52 publications

References 59 publications

Context-Aware Security for Vehicles and Fleets: A Survey

Context-Aware Security for Vehicles and Fleets: A Survey

Analyzing and Improving Fault Tolerance of Learning-Based Navigation Systems

One Versus All for Deep Neural Network for Uncertainty (OVNNI) Quantification

Contact Info

Product

Resources

About