Analysis and Evaluation of Inter-Core Communication Based on a Multi-Core Automotive Software and Hardware

The vehicles have become increasingly computerized in order to satisfy the strict safety requirements and to provide better driving experiences. Growing complexity of vehicle functionality requires higher
performance and increases complexity of electronic control units (ECUs) software.
As a result, the AUTOSAR standard was established by several major vendors and manufactures in 2003. This foundation introduced a standard with the main goal to define a uniform ECU software architecture, which reduces the software complexity.
The standardisation of software along with the increasing requirements for resources brought the single core processors to their limit of performance. Hence, new techniques had to be introduced to increase computational power. A solution to this problem is a multi-core processor.
It is expected that the main focus for the automotive industry in the next decade will be to implement the existing single-core software to a multi-core architecture. One of the key challenges is to reduce the execution time of inter-core communication in order to maintain the response time in safety critical systems such as the electronic brake system.
As the need for multi-core processors became apparent in the automotive industry, AUTOSAR introduced the initial release of multi-core support in 2012 with version 4.0. The specification provides guidelines for implementing multi-core architecture. One of the key elements is the Inter OSApplication Communication (IOC), which provides the cross-core communication between applications located in different cores on a single ECU. The current multi-core software implementation is still in development and there are performance limitations that offer scope for improvement.
This thesis addresses the challenge of multi-core software in automotive systems and proposes possible approaches for inter-core communication. In order to identify possible improvements, a number of software benchmarking experiments were designed for intra-core and inter-core communication and implemented on an universal ECU to validate the performance in term of execution time and memory consumption. The results demonstrate that there are significant overheads caused by the design of the communication functions.