Publication

Vehicle Electronics and Architecture (VEA) & Cyber
2020

Latent Dirichlet Allocation (LDA) for Anomaly Detection in Ground Vehicle Network Traffic

by Adam Thornton; Brandon Mieners; Donald Poole; Mark Russell

Abstract

Latent Dirichlet Allocation (LDA) and Variational Inference are applied in near real-time to detect anomalies in ground vehicle network traffic for VICTORY enabled networks. The technical approach, that utilizes the Natural Language Processing (NLP) technique to detect potential malicious attacks and network configuration issues, is described and the results of a proof of concept implementation are provided.