Vehicular Network Security With Markov-Based Anomaly Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Modern vehicle data networks are susceptible to unauthorized and malicious access, posing risks to safety and security, including potential harm to occupants and other road users.

Innovation Solution

A Markov decision process (MDP) model is generated to monitor vehicle state transitions, identifying anomalous actions through probability thresholds, triggering responsive measures to protect the vehicle and its occupants.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional anomaly detection systems are used in vehicles, then security monitoring is provided, but the systems lack real-time responsiveness and cannot effectively prevent malicious access

Engineering Contradiction:
Improvesecurity monitoring effectivenessVSAvoidresponse time to anomalies
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by continuously learning and establishing baseline behavioral patterns of vehicle components during normal operation. This preliminary characterization of normal behavior enables the system to quickly identify deviations without requiring time-consuming analysis when anomalies occur, thus improving both reliability and reducing response time.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system implements continuous feedback by monitoring vehicle component behavior in real-time, comparing observed behavior against learned baselines, and immediately triggering alerts or protective measures when deviations are detected. This closed-loop feedback mechanism ensures rapid response to security threats while maintaining high monitoring effectiveness.

Inventive Principle:
Principle #23Feedback

2Reliability

If comprehensive monitoring of all vehicle subsystems is implemented, then security coverage is improved, but system complexity and computational requirements increase

Engineering Contradiction:
Improvesecurity coverageVSAvoidmonitoring system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system applies self-service by using unsupervised machine learning algorithms that automatically learn and adapt to each vehicle's unique operational patterns without requiring manual configuration or extensive training data. The system autonomously establishes baselines and detects anomalies, reducing the complexity of system deployment and maintenance while maintaining comprehensive security coverage across all vehicle subsystems.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP3763107B1Vehicular network security
Publication Date: 2025.09.10 BRITISH TELECOM PLC
  • EP3763107B1 patent drawingFigure 1~2
  • EP3763107B1 patent drawingFigure 3~5
  • EP3763107B1 patent drawingFigure 4

AI summary

A computer implemented security method operable with a communications network in a vehicle, the network communicatively connecting devices including sensors and actuators in the vehicle such that information provided by sensors and states of actuators are determinable by data communicated via the network, the method comprising the steps of: defining a Markov decision process model for the vehicle, the model specifying states of the vehicle and actions constituting transitions between states, wherein a state of the vehicle is indicated by information provided by one or more sensors and a state of one or more actuators, and an action corresponds to a change in the information provided by one or more sensors and/or a change to a state of one or more actuators, each action having associated a probability of occurrence; determining, by accessing data communicated via the network, a current state of the vehicle in the model; accessing data communicated via the network; responsive to the accessed data indicating an action to change the vehicle state to a new state, determining, from the model, a probability of the action; responsive to a determination that the determined probability falls below a predetermined threshold probability, generating an indication that the vehicle state transition is anomalous.