Made in IBM Labs: Cyber Intrusion Detection System for Protecting Connected Vehicles, Part II
Co-authored by Yair Allouche.
In the Part I of this series, where we presented our idea for a prototype intrusion detection system, we highlighted the emergence of connected vehicles as a fundamental part of the connected society — and the high value they carry for stakeholders such as car owners, automakers, fleet managers and transportation service providers.
We also pointed out that along with the benefits of connected vehicles come the risks associated with security breaches and concerns over data privacy. In fact, the connected vehicle threat surface is wide and highly exposed to attacks, as various researchers have observed.
Reporting From IAA
At the International Motor Show (IAA 2015) in Frankfurt, Germany, we introduced a prototype intrusion detection system solution for securing connected vehicles based on research conducted by IBM in its Cybersecurity Center of Excellence in Israel. Our solution is based on a client/server architecture where an in-vehicle component communicates with a cloud-based server. Using a novel approach for coordinated anomaly detection, this powerful solution can identify attacks on the integrity of both individual vehicles and across vehicle networks. In doing so, this solution addresses multiple design challenges.
This year’s IAA hosted over 1,100 exhibitors, including automotive manufacturers and part suppliers from 39 countries, and welcomed tens of thousands of participants. This volume and variety provides an excellent opportunity to observe the market and develop qualified insights on trends and concepts. We want to share a few key observations and insights on the security of connected vehicles and put our solution in perspective.
A common theme of products seen at IAA can be referred to as specialized point solutions. We saw a plethora of companies working on interesting security solutions for specific purposes, such as the functionality to open doors remotely. Additionally, traditional suppliers of embedded devices or embedded software presented new solutions for securing endpoints like the single vehicle. In both types of solutions, all the effort goes into raising the security barriers against potential attackers, using techniques such as encryption and authentication based on secure embedded hardware.
A Security Assessment
While these solutions are appropriate for their specific purposes, in our perspective they remain island solutions. As such, they inherently do not provide a complete security point of view. A comprehensive end-to-end approach is necessary to cope with the plurality of attack vectors and the implications rising from a heterogeneous landscape, especially at this time since the security market for connected vehicles is still in an early stage.
We believe that original equipment manufacturers (OEMs) aiming to develop a security solution in a piecemeal approach will fail because handling multiple components from multiple suppliers and integrating them into the existing car network architecture would be a Herculean task.
Moreover, many of the solutions seen at the IAA, while interesting and sometimes even suggestive, do not provide the answer to the most important question of connected vehicle protection: How do you know if, when and how you have been hacked? While those solutions provide an acceptable level of security for a limited part of the attack surface, they remain isolated and disconnected from one another. Thus, they inherently fail to provide a complete view of the vehicle’s security.
To address this gap, a complete vehicle security solution should be designed with an end-to-end view, taking into account the wide attack surface of the car. The solution should, in particular, take into consideration the following aspects of vehicle security and privacy:
- Internal network security to protect the data on the CAN bus from tampering;
- Protecting the vehicle’s external network connectivity, including over-the-air updates;
- Application security for ECUs, infotainment apps, mobile apps, Web portal/gateway API, etc.;
- Identity governance to ensure only authorized users have access to vehicles and data;
- Mobile security for in-vehicle infotainment (which often runs a modified mobile OS, such as Apple CarPlay) and the mobile apps that the consumer uses.
Moving Forward With an Intrusion Detection System
To provide an integrated level of protection and fulfill these demanding requirements, some fundamental capabilities are required. These include efficient collecting and logging of security events from both the single car and network of cars and coordinated security analytics algorithms, which detect anomalies in, among and between cars and transportation infrastructure.
While securing connected vehicles is highly challenging, we believe that a concentrated effort from the players in the field can help our society fully benefit from the applications of connected vehicles in a secure manner. We anticipate that open architectures and standards targeting connected vehicle security will likely emerge in this space.
Lastly, we believe that the intrusion detection system we presented at IAA is the first instance of a new class of solutions for protecting connected vehicles against cyber intrusions. This system is a real-time comprehensive solution that provides the most precious asset against an invisible and unknown attacker — time to recover.