A Proposed Systematic Software Robustness Verification Framework (SRVF) for Enhancing Critical Software Module Robustness

With the advancement of inbuilt electronics and intelligent controls, automotive and other industries are looking at efficient usage of ECUs with intelligent electronic modules and more of critical functions implemented by software. Robustness of the software involved is always critical to the ECUs health, so software engineers are tasked to ensure the same by following right development and validation life cycle. Enabling the multiple decision by single software module makes verification & validation challenging, complex and time consuming. Current method of software validation involves unit testing at module level. This may overlook some failures which occur on system level and can identify new set of requirements if not already captured during requirement gathering phase. It becomes an iterative approach of design, test and identify new requirements. Software FMEA in embedded control system [2] is used for identifying new requirements and mitigating the failures. For performing system level validation, DoE with Taguchi method [3] is used which tries to identify any gaps/failures in requirements. Combining these approaches and creating a SRVF helps in reducing these iterations and tries to capture complete requirements.

This paper summarizes a SRVF which can be followed for critical software module upfront by the verification & validation engineers to build inherent robustness in the software module. This specific methodology comprises of Failure Mode Effect Analysis (FMEA) as part of complete requirement strategy & structured experimentation such as System Behavioral Testing (SBT) to enable enhanced test coverage against dynamic responses. This paper does not supersede the typical industry standards being followed for software development life cycle but showcase a recommended practice which can be followed to enhance the robustness of the software module.

A sample case study on the deployment of SRVF on a critical software module has been presented in this paper.