CAVIDS: Real time intrusion detection system for connected autonomous vehicles using logical analysis of data

Kumar, Anjanee; Das, Tanmoy Kanti

doi:10.1016/j.vehcom.2023.100652

Cited by 5 publications

(2 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It becomes challenging to confirm the legitimacy and integrity of the messages transferred across the network in the absence of adequate authentication procedures. Due to the lack of encryption, CAN network data transmissions are vulnerable to eavesdropping and unauthorized access (Kumar & Das, 2023;Palaniswamy et al, 2020). Due to the lack of authentication in the CAN protocol, it is possible to masquerade an ECU or replace a legitimate ECU with a malicious one using a hardware device (Boudguiga et al, 2016).…”

Section: Overview Of Can Protocol In Vehiclesmentioning

confidence: 99%

“…In order to avoid disastrous crashes and disruptive effects, real-time intrusion detection with little processing resources is required. In order to enable real-time intrusion detection in CAVs with the least amount of processing resources, Kumar and Das (2023) suggested an intrusion detection approach based on logical analysis of data. The result showed a FP rate of 0.078.…”

Section: For Intrusion Detectionmentioning

confidence: 99%

See 1 more Smart Citation

A Machine Learning Model to Predict Cyberattacks in Connected and Autonomous Vehicles

Jha,

Jaiswal

2024

JCCE

View full text Add to dashboard Cite

Connected and autonomous vehicles (CAVs) are largely at the experimental stage. Their successful deployment and field implementation require a careful consideration of their vulnerabilities to cyberattacks. The primary security vulnerability is in the controller area network (CAN) protocol, which permits communication among electronic control units in CAVs. To address this vulnerability and mitigate cyberattacks, machine learning (ML) algorithms can be developed for intrusion detection in CAV's CAN protocol. In this research, the data structure of certain experimental datasets on message injection attack from the Hacking and Countermeasure Research Lab is examined. A random forest classifier-based ML model is developed owing to its efficiency in predicting cyberattacks on CAVs consisting of over 3 million datasets. A number of procedures within the Python programming environment are employed to clean the dataset before performing the prediction. The prediction for intrusion detection is performed with a 70:30 split of the training: testing data with a random state of 11 and number of estimators as 200. The accuracy is found to be over 92% for all three scenarios in performing the prediction. The model can be deployed in real-time investigation of cyberattacks in CAVs if real-time data were available. The data cleaning method developed in this study can be applied in other ML applications consisting of large datasets, such as credit card fraud and drug discovery, to name a few.

show abstract

Section: Overview Of Can Protocol In Vehiclesmentioning

confidence: 99%

Section: For Intrusion Detectionmentioning

confidence: 99%

A Machine Learning Model to Predict Cyberattacks in Connected and Autonomous Vehicles

Jha,

Jaiswal

2024

JCCE

View full text Add to dashboard Cite

show abstract

Comparative analysis of machine learning techniques for enhanced vehicle tracking and analysis

Rani,

Dalal

2024

Transportation Engineering

View full text Add to dashboard Cite

A Novel Constructive Unceasement Conditional Random Field and Dynamic Bayesian Network Model for Attack Prediction on Internet of Vehicle

Mahendran,

Rajendran,

Pandian

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Today's Internet of Vehicles (IoV) has soared by leveraging data gathered from transportation systems, yet it grapples with security concerns stemming from network vulnerabilities, exposing it to cyber threats. This study proposes an innovative method to anticipate anomalies and exploit IoV services related to road traffic. Using the Unceasement Conditional Random Field Dynamic Bayesian Network Model (U-CRF-DDBN), this approach predicts the impact of network attacks, strategically managing vulnerable nodes and attackers. Through experimentation and comparisons with existing methods, our model demonstrates its effectiveness in mitigating IoV vulnerabilities. The U-CRF-DDBN strikes a superior balance, outperforming other approaches in intrusion detection for Internet of Vehicles systems. Evaluating its performance on the NSL-KDD dataset reveals a promising average Detection Rate of 93.512% and a low False Acceptance Rate of 0.125% for known attacks, highlighting its robustness. However, with unknown attacks, while the Detection Rate remains at 74.157%, there is an increased FAR of 16.47%, resulting in a slightly lower F1score of 0.822.

show abstract

CAVIDS: Real time intrusion detection system for connected autonomous vehicles using logical analysis of data

Cited by 5 publications

References 42 publications

A Machine Learning Model to Predict Cyberattacks in Connected and Autonomous Vehicles

A Machine Learning Model to Predict Cyberattacks in Connected and Autonomous Vehicles

Comparative analysis of machine learning techniques for enhanced vehicle tracking and analysis

A Novel Constructive Unceasement Conditional Random Field and Dynamic Bayesian Network Model for Attack Prediction on Internet of Vehicle

Contact Info

Product

Resources

About