A heuristic task allocation methodology for designing the secure in-vehicle network

Advances in automobiles have led to the increased usage of electronic control units (ECUs). A modern vehicle contains 50 to 70 ECUs, each of which is responsible for automatic safety and comfort functions such as anti-lock breaking system (ABS) and adaptive cruise control (ACC). To improve the safety and comfort of the drivers, ECUs communicate with one another in an in-vehicle control network and in an inter-vehicle network. In the inter-vehicle network, ECUs can collect the real-time traffic information such as traffic congestion information and collision warnings from the neighbor vehicles. However, in process of collecting the information over the air and interacting with one another, ECUs in the in-vehicle network are exposed to a number of cyber attacks targeting the automatic safety and comfort functions. Also, as the in-vehicle control network transfers the information among distributed ECUs via a shared bus interface, the overall performance of the in-vehicle control network, whose malfunction can threaten the safety of drivers, can be affected by a single attacked ECU. In this paper, we propose a new task allocation methodology for designing the secure in-vehicle network. Compared to the typical task allocation problems in the in-vehicle control network, the proposed task allocation problem considers the varying probability of task vulnerability exploits as a new objective function, where a function consists of the logically interconnected tasks over the ECUs. To reduce high computational complexity in process of solving the problem, a heuristic repeated matching methodology is proposed. Compared to those of using the random task allocation methodology and the worst task allocation, it is shown that the proposed methodology can reduce the probability of the overall malfunction of the in-vehicle network by as much as 15% and over 60% for each.