14 June 2021 12 July 2021 (extended)
Publication: March/April 2022
Systems such as production plants, logistics networks, IT service companies, and international financial companies are complex, operate in highly dynamic environments, and need to respond quickly to a variety of change drivers. The characteristic features of such systems include scale, complex interactions, knowledge of behavior limited to localized contexts, and inherent uncertainty.
Knowing how to analyze, design, implement, control, and adapt such systems is a difficult problem that lacks suitable mainstream engineering methodologies and technologies. Grand challenges such as smart cities, large-scale integration of information systems such as national medical records, and Industry 4.0 can only be achieved through the deployment and integration of software systems with existing systems.
This leads to the idea of a digital twin: a virtual high-fidelity machine-processable representation of a complex system that is amenable to solution space exploration through what-if and if-what analysis for a wide selection of application domains, including cyber-physical, business, and societal systems. Highly useful as an in-silico experimentation aid, a digital twin for software systems can go further by being used as a controller using adaptation technologies such as machine learning or model reference adaptive control to achieve or maintain the overall system objective.
Digital twins can be applied in many scenarios. Twins of physical systems can be used to provide a cost-effective way of exploring the design space for new products or optimizations. Twins of information systems can be used to achieve adaptation in complex ecosystems. Twins of populations (such as those modelled in the current COVID-19 pandemic) can be used to perform scenario playing where behavior is inherently emergent.
It is desirable to envisage a situation in which digital twins are used in various modes for complex system analysis and development:
- Analysis. A key requirement is to ascertain that a complex system is achieving its goals. A digital twin can provide a cost-effective solution through execution in a simulation environment producing traces that can be examined for occurrence of the desired (and undesired) behavioral patterns. A digital twin can also support what-if and if-what scenario playing to explore the system state space.
- Adaptation. An existing system may expose a control interface that can be used for dynamic adaptation. A digital twin can be used to address the problem of constructing the desired control inputs by running alongside the real system and producing control commands based on a comparison of the observed and desired behavior. This leads to the idea of a digital twin being used for continuous improvement of complex system behavior through a variety of classical control theory and AI-based techniques.
- Maintenance. This is the single most expensive activity in a system lifecycle and can be responsible for over 60% of the overall costs. This is largely due to the present inability to explore the solution space effectively and efficiently. A digital twin can overcome this hurdle through scenario playing to help arrive at a feasible transformation path from the as-is state to the desired to-be state in silico. Once the transformation path is validated, the necessary changes can be introduced into the real system in the right order, thus providing assurances of correctness.
- Design. A new complex system can start life as a digital twin that is used as a blueprint. The twin provides a specification of the behavior for the real system and can be integrated with existing systems in the target ecosystem by observing their outputs. The design can then use adaptation to tailor its behavior with respect to real ecosystem data.
The term digital twin is used here to cover all types of application and modes of use described above. In all cases, there is a requirement for simulation in terms of goals. Different modes of use involve different twin-architectures and can involve different lifecycles and stakeholders. Across these differences, there is a requirement to ensure that digital twins are engineered to an appropriate level of quality and use-appropriate architectures and technologies.
The importance and timeliness of applying digital twins to software and systems development is highlighted in the number of recent industry thought-leadership editorials that describe the huge breadth and potential of this approach, including Deloitte, Simio, Forbes, and Gartner.
This IEEE Software special issue solicits articles relating to any area of software engineering approaches for digital twins. This includes, but is not limited to:
- Methodologies for the construction of digital twins:
- Representing requirements and use cases for digital twins
- Articulating the business case and stakeholder benefits for digital twins
- Quality assurance techniques for digital twins:
- Ensuring that simulations produce valid predictions or control systems in the correct way
- Evaluating digital twins
- Verification techniques for digital twins and the associated data
- Security and privacy
- System architectures for digital twins:
- Model Reference Adaptive Control used as the basis of digital twin engineering
- Integrating digital twins with existing industrial approaches such as Industry 4.0
- Technologies for digital twins:
- Programming language extensions to support digital twins
- Libraries and platforms for digital twin construction
In particular, we would welcome case studies and experience reports from industry that relate to the aspects listed above. Articles describing collaborations between academic and industry are particularly welcome.
For general author guidelines: www.computer.org/publications/author-resources/peer-review/magazines
To submit an article: https://mc.manuscriptcentral.com/sw-cs
Manuscripts must not exceed 3,000 words, including figures and tables, which count for 250 words each. Submissions in excess of these limits may be rejected without refereeing. The articles we deem within the theme and scope will be peer reviewed and are subject to editing for magazine style, clarity, organization, and space. Be sure to include the name of the theme you’re submitting to. Articles should have a practical orientation and be written in a style accessible to practitioners. Overly complex, purely research-oriented or theoretical treatments aren’t appropriate. Articles should be novel. IEEE Software doesn’t republish material published previously in other venues, including other periodicals and formal conference or workshop proceedings, whether previous publication was in print or electronic form.
For more information about the focus, contact the guest editors at firstname.lastname@example.org.
- Tony Clark, Aston University, UK
- Vinay Kulkarni, TCS Research, India
- Jon Whittle, Director of CSIRO’s Data61, Australia
- Ruth Breu, University of Innsbruck, Austria