Software Challenges in Automotive

In 1978, General Motors used a Motorola 6802 CPU in the Cadillac Seville to calculate trip information (average speed\fuel usage\engine information). This is one of the first software controlled systems used within an automobile. The application calculating this information consisted of less than 10,000 lines of code. 

A few years later, the first software controlled engines were introduced. This software contained about 50,000 lines of code. Since the introduction of software solutions within automotive, the number as well as the size of these solutions grew exponentially.

These days, high end cars contain up to a hundred electronic control units running about a hundred million lines of code connected by multiple in-car computer networks. To emphasize the huge size of these solutions: the complete package of software controlling the Joint Strike Fighter (JSF) contains about 5.7 million lines of code (almost a factor 20 less)! This shift from purely mechanical to electronic solutions introduced comfort and safety increasing functions on one hand, but on the other hand it made automobiles more complex to design, test and maintain.

What about security?

The control units running this software are connected via (unsecure) computer networks, enabling them to send messages across and thereby adjusting each other’s settings. Recent research has shown the vulnerability of these networks by taking over control of the speed dial and even more important, the car's brakes. Since these interventions require a physical connection to the network, at first this doesn’t sound as an important security issue. But, additional research has shown several ways to access this network. For example by adding software to a digital music file, which while being played by the car's audio system, gets access to the car's network. This attack still requires a user of the car to connect the file to the audio system, but these days downloading mp3-files from the internet and playing them in our cars is common practice.

Another way of gaining access to the network is, through the emerging communication interfaces in a car, for example Bluetooth interfaces and internet access enabling real time traffic information. Furthermore, there exist cars that can be controlled by smartphone apps enabling users to remotely open doors and start the car’s engine to preheat it before driving. These connections can be used to gain access to the car’s network and take over control of the car. If these vulnerabilities are exploited by malicious persons, harmful situation could occur. It is interesting to note that these security issues not only apply to the automotive industry, but also to for example power plants. In June 2010 Stuxnet (a computer virus) was spread, infecting PLCs of nuclear power plants.

Protection

To be able to protect our cars from similar attacks, it is important to take security issues into account while designing the architecture of the complete electronically system of a car. Using this approach, it will be possible to identify weaknesses during design time and resolve them. For example the interfaces that somehow enable remote access to the car’s network will need to have proper authentication systems and communication needs to be encrypted. This will help solving these security issues and thereby allowing users to securely enjoy nowadays multifunctional cars.