Efficient Test Case Generation for AUTOSAR Basic Software Code Generators

Abstract

In the contemporary automotive industry, the complexity of software architectures for electronic control units (ECUs) has increased drastically. Aiming to improve the complexity management of these architectures, a worldwide partnership of car manufacturers and suppliers has created a standardized approach called AUTOSAR (AUTomotive Open System ARchitecture) (AUTOSAR Basics, 2012). At the highest abstraction level, the architecture of AUTOSAR contains three software layers which run on a Microcontroller. These three layers are Application Layer, Runtime Environment (RTE) and Basic Software (BSW) (AUTOSAR Layered software architecture, 2011). The BSW layer is further divided into multiple software modules which provide basic services such as memory management and bus communication (Mecel, 2013). These software modules can be configured to satisfy the needs of the customer. Testing these configurations requires a large amount of effort and time, especially since they are manually generated. This thesis deals with the automatic generation of these test cases, the configurations of the BSW modules. Two test case generation approaches were developed and compared. The first is random generation where elements to be added to the test case are chosen in a random manner. The second is pairwise generation where elements are added to the test case based on satisfying all pairs of element values. The experiments conducted to compare the two generation techniques ran the configurations created for three BSW modules through their SCGs (Source Code Generators) and showed that both techniques have the ability to uncover problems within a SCG. This thesis was conducted as a case study at Mecel AB in Gothenburg.