A machine learning based scheme for enhancing the detection of position falsification attacks in vehicular ad hoc networks

Abdelkreem, Eslam; Hussein, Sherif; Tammam, Ashraf · 2026 · DOAJ

DOI: 10.1038/s41598-026-39867-9

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the critical security vulnerability of position falsification attacks in Vehicular Ad Hoc Networks (VANETs), where malicious nodes broadcast falsified location data to disrupt safety-critical applications. The authors argue that existing Misbehavior Detection Schemes (MDSs) often lack sufficient accuracy, with some prior studies reporting performance ceilings below 87% accuracy for certain attack types. To mitigate these risks, the study proposes a machine learning-based detection scheme designed to enhance the reliability of identifying such attacks. The methodology utilizes the publicly available Vehicular Reference Misbehaviour Dataset (VeReMi), which simulates five distinct position falsification attack types: Constant Position, Constant Offset Position, Random Position, Random Offset Position, and Eventual Stop. The core innovation is the introduction of a novel feature called RSSIConf, which assesses the reliability of a sender’s claimed position by comparing the measured Received Signal Strength Indicator (RSSI) against confidence intervals corresponding to the claimed sender–receiver distance. This feature was integrated into three different feature vectors (FV1, FV2, and FV3) composed of selected and derived features from Basic Safety Messages. The researchers evaluated four machine learning algorithms using these feature sets to identify the most effective configuration. Experimental results demonstrate that the Random Forest classifier, when trained with the FV2 feature set, achieves the best overall performance. This configuration outperformed existing state-of-the-art approaches, yielding accuracy improvements ranging from 0.76% to 13.26% and F1-score improvements from 0.74% to 12.71% across different attack types. An ablation study further confirmed the quantitative contribution of the RSSIConf feature to the model’s detection capabilities. The study also benchmarked these results against previous works using the same dataset, highlighting the superior performance of the proposed scheme compared to traditional statistical methods and other ML-based detectors. The significance of this work lies in its contribution to safer and more trustworthy VANET communications. By providing a more robust detection mechanism for position spoofing, the proposed scheme helps ensure the integrity of safety-oriented applications, such as collision avoidance and lane change warnings. The findings suggest that integrating physical-layer metrics like RSSI with machine learning models offers a viable path toward overcoming the limitations of current detection schemes, thereby reducing the potential for severe accidents caused by malicious data manipulation.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-25; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• v2x connected vehicle