Monday, September 25, 2017

On the Effectiveness of Source Code Metrics for Fault and Change Proneness Prediction in Industrial Automation Domain


by Lov Kumar (NIT Rourkela, India) and Ashish Sureka (ABB, India) 
Associate Editor: Sridhar Chimalakonda (@ChimalakondaSri)

Domain Specific Programmable Logic Controller (PLC) Languages

Several control systems used for factory and industrial automation of electromechanical processes are developed using Programmable Logic Controllers (PLC). PLCs are specifically adapted for the control of a manufacturing process and programmed using domain specific languages. The International Electro-technical Commission (IEC) (a non-profit international standards organization) developed a standard called as IEC 61131-3 which defines the basic programming elements, syntactic and semantic rules for text-based and graphical or visual programming languages for programming PLCs [IEC]. Structured Text (ST) is one the text-based PLC programming languages defined by the IEC 61131-3 standard and is widely used in industrial automation engineering application development. Similarly, Ladder Diagram (LD) programming language is a visual programming language and is one of the five languages defined in the IEC 61131-3 standard. ST is a domain specific language and several of its language features and programming constructs are different than that of general purpose programming languages such as Java, C++, C# and Python. ST has several characteristics which are different than that of general purpose programming languages because the primary purpose of a PLC is to control an industrial process and the ST language is defined for solving problems in a specific domain of factory and industry process automation [Roos2008].

Correlation between Source Code Metrics and Software Quality for PLC Systems

There has been a lot of work done in the area of predicting change proneness and faulty components or modules of software using source code metrics for general purpose programming languages. However, defining software metrics and investigating their impact on important software engineering prediction problems such as change proneness and faulty module identification is relatively unexplored in the domain of applications developed using IEC 61131-3 languages [Kumar2016] [Kumar2017]. We believe that one of the primary reasons of lack of empirical studies in the area of source code analysis for PLC programming languages and their impact on software maintainability and quality is the lack of publicly available data for researchers and scientists in academia. Through this blog, our objective is to share our work in this relatively unexplored but a promising research direction [Kumar2016] [Kumar2017]. In [Kumar2016], we proposed source code level metrics to measure size, vocabulary, cognitive complexity and testing complexity of a visual Programmable Logic Controller (PLC) programming language. We apply Weykur’s property to validate the metrics and evaluate the number of properties satisfied by the proposed metric [Kumar2016]. In [Kumar2017], we study the correlation between the 10 ST source code metrics and their relationship with change proneness. We built predictive models using Artificial Neural Network (ANN) based techniques to predict change proneness of the software [Kumar2017]. Similarly, we are working towards examining whether source code metrics can be used to identify defective and faulty components.

Bringing Both Relevance and Rigor through Industry Academia Collaboration

Our experimental results are encouraging and provides evidences that there is a correlation between source code metrics and software quality and maintainability in the domain of industrial automation and PLC applications. However, more research studies from industry through industry and academia collaboration is needed to further add to the body of knowledge in this topic. It is hard for companies working in the industrial automation engineering domain to make their source code available in public domain for researchers in academia which creates a natural barrier for entry for academics to conduct research in this unexplored area. Industrial research labs does not have all the competencies, resources and time within a single organization to address all the research and technical challenges in this area and we believe that university and academia collaboration is the only possible solution to advance and disseminate knowledge on this topic. Our work is also a collaboration between industry and academia and gave positive results due to the successful partnership. One possible solution to build bridges is to conduct focused workshops on software engineering issues and challenges for industrial and factory automation. Some progress has been made in this direction which is encouraging and more needs to be done.

References

[Kumar2016] Lov Kumar, Raoul Jetley, and Ashish Sureka. 2016. Source code metrics for programmable logic controller (PLC) ladder diagram (LD) visual programming language. In Proceedings of the 7th International Workshop on Emerging Trends in Software Metrics (WETSoM '16). ACM, New York, NY, USA, 15-21. DOI: http://dx.doi.org/10.1145/2897695.2897699
[Kumar2017] Lov Kumar and Ashish Sureka. 2017. Using Structured Text Source Code Metrics and Artificial Neural Networks to Predict Change Proneness at Code Tab and Program Organization Level. In Proceedings of the 10th Innovations in Software Engineering Conference (ISEC '17). ACM, New York, NY, USA, 172-180. DOI: https://doi.org/10.1145/3021460.3021481
[Roos2008] Nieke Roos. 2008. Programming PLCs using Structured Text." International Multiconference on Computer Science and Information Technology (IMCSIT’08)

[IEC] International Electrotechnical Commission, International Standards and Conformity Assessment for all electrical, electronic and related technologies  http://www.iec.ch/

No comments:

Post a Comment