Search Results

Automobiles are getting more and more sophisticated with increased demand for more comfort and safety by customers. Due to this, the automotive Electronic Control Units (ECU) and the software applications running on these ECUs have become more complex and computationally more intensive. This has resulted in Original Equipment Manufacturers (OEMs) migrating to multicore platforms. Optimal usage of multicore platform necessitates the design of new scheduling algorithms. In the past decade, different approaches to implement hard real time scheduling in automotive domain have been proposed for single core as well as multicore architectures. We explore different scheduling techniques proposed so far which are relevant to automotive domain and also, provide a taxonomy of these scheduling algorithms, which will help the automotive design engineer to make an informed decision.

The automotive industry today follows Model Based Development (MBD) for developing modern automotive applications. This method involves creating models for a system under design and then using tools like MATLAB/Simulink to auto-generate code for target platforms. This method is popular since maintenance of MBD based applications is simple and less time consuming as compared to maintaining hand-written application code. Thus, MBD facilitates correct designs and easy maintenance of automotive applications. However, there are legacy automotive applications that are not developed using models. It is difficult to accommodate and test any changes in such application codes since it requires extensive testing. Additionally, for application code generated from models, many a times, code is changed during testing and these changes are not reflected in the model. Hence, there is a need to convert legacy automotive application codes to models.