Multi-Task learning with sensor fusion for joint detection of multiple misbehaviors in Vehicular Ad Hoc Networks
|
Full Text |
Pdf
|
|
Author |
Zaid Mahmood Abdul Razzaq Hanoosh
|
|
e-ISSN |
1819-6608 |
|
On Pages
|
457-464
|
|
Volume No. |
21
|
|
Issue No. |
7
|
|
Issue Date |
June 10, 2026
|
|
DOI |
https://doi.org/10.59018/042654
|
|
Keywords |
message forgery, false location reporting, vehicular ad hoc networks (VANET), sensor fusion, machine learning, multi-task learning, misbehavior detection, network security.
|
Abstract
Vehicular Ad Hoc Networks (VANETs) are widely vulnerable to several kinds which threaten the reliability and safety of the network. Traditional diagnosis strategies sometimes concentrate on unique kinds of misbehavior and fail to consider the complicated, complex, simultaneous attacks aspect. Here, we offer a new multi-task learning architecture integrated with sensor fusion methods for diagnosing and concurrently grouping several misbehaviors in VANET areas. Through combining heterogeneous data sources-such as network parameters, behavioral features, and positional info—our model efficiently obtains interrelated features of different kinds of attacks, like denial of service, message forgery, and false location reporting. The multi-task learning framework enables knowledge sharing across tasks, developing whole diagnosis accuracy and strength. Broad simulations performed under different traffic densities show that the presented strategy performs better than conventional unique-task models in the two binary and multi-class classification scenarios. The outcomes highlight the ability of an architecture to develop security in VANETs through presenting a general and effective misbehavior diagnosis system.
Back