Engineering Resilient Systems (EReS)

The Linnaeus University Centre for Data Intensive Sciences and Applications (DISA) focuses its efforts on open questions in collection, analysis and utilization of large data sets. With its core in computer science, it takes a multidisciplinary approach and collaborates with researchers from all faculties at the university.

Considering both the cyber and physical factors in the design of CPS raises many challenges as they stress the known methods, processes, and tools to their extreme, and introduce new difficulties to deal with. While on the one hand, CPSs provide new opportunities, on the other hand, the integration of cyber and physical components increases the complexity of the overall system, as it seriously complicates the design of systems delivering high-quality aspects – i.e., performance, safety, security, sustainability (i.e., near-zero power consumption), scalability, usability.

In this context, EReS is part of the DISA Smart Industry Group (SIG), an interdisciplinary research initiative in collaboration with the Department of mechanical engineering. SIG investigates data-driven approaches for (i) inferring implicit CPS models from data (data-driven discovery), and (ii) estimating and predicting CPS quality using models and real-time data (data-driven analysis).

The overall objective is to leverage data-driven approaches for improving production processes and products. For example, a production system can automatically adjust to desired production level with optimal logistics and flow at lowest cost, whereas a product can automatically reconfigure with respect to operation and human factors.

Data-driven discovery

Models (usually expressed in terms of laws and equations) facilitate the understanding of complex phenomena and allow for systems analysis. However, CPSs have proved resistant to explicit modeling due to their intrinsic complexity arising from the combination of cyber and physical components and the interaction between them.

To this end, the main objective of SIG is to investigate data-driven methodologies for facilitating the data-to-discovery process of implicit CPS Models. Specifically, data-driven methodologies would allow us to learn and infer implicit models from the observation of raw data. These methodologies have the advantage of (i) testing correlations between different variables and observations, and (ii) learning unforeseen patterns in systems and discovering new or hidden implicit behaviors.

Data-driven analysis

Model-based analysis facilitates the evaluation of complex systems and allows for estimating and predicting systems quality. However, it is well known that the computational complexity of model-based analysis techniques is one of the key challenges in complex system validation and evaluation.

To this end, SIG aims at applying data-driven analysis (e.g., machine learning) to enable online model-based estimation and prediction of quality aspects, e.g., performance, security, maintenance.

Smarter Systems is a complete knowledge environment focusing on the systems and challenges of the future. It addresses challenges together with its partners through education and research initiatives in collaboration.

“Perpetual change” changes everything. Smarter systems are complex and unstable. Instability leads to loss of control, which in turn makes the system untrustworthy. To prevent this cause-and-effect chain, smarter systems shall be provided with proper mechanisms to be resilient and return in balance whenever they face instability. That is, software systems shall be able to confront instability, mitigate uncertainty, and accommodate changes to preserve their goals.

To deal with perpetual change, EReS investigates continuous (model-based) system engineering approaches enabling smarter systems to be resilient and able to accommodate the change. Continuous refers to the fact that the engineering approaches shall be automated and never-ending, whereas model-based system engineering refers to the fact that the engineering approaches shall be model-based, holistic and interdisciplinary. Indeed, able to consider at the same time all the system constituents, the possible affecting factors, and the properties of interest.

However, model-based systems engineering approaches have several difficulties, including (but not limited to): the integration of different modeling formalisms and analysis techniques, ensuring consistency and completeness, integrating human behaviours, as well as considering uncertainty as an integrated part of the approach. To this extent, EReS aims at tackling the following grand challenges:

1. Modeling and analysis for resilience (KO1^[1])

• How to specify resilience?
  • E.g., goals, requirements, qualities, uncertainty ...
• How to model resilience?
  • E.g., structure, behavior, uncertainty
• How to assess resilience?
  • E.g., quality models, metrics, techniques (e.g., testing, simulation, formal methods, …)

2. Develop for resilience (KO2^[2])

• How to achieve resilience?
  • Design (e.g., principles, practices, patterns)
  • Development (e.g., languages, tool, platforms)
  • Operate (e.g., infrastructures, technologies)
  • Decision-making process (e.g., human-in-the-loop, autonomous, trustable, explainable)

3. Resilient applications (KO1, KO2)

• How to exploit resilience (industry/society level)?
  • Smart Cities, Green Economy, Industry 4.0

^[1] KO1: To assure product capabilities in uncertain conditions.

^[2] KO2: To devise engineering processes for perpetual adaptation and evolution.

EReS Research Lab is responsible for the following set of courses:

• 1DT305, Introduction to Applied Internet of Things, 7.5 credits 
• 1DV517, Formal Languages and Logic, 7.5 credits
• 2ME302, Research challenges in Media Technology, 7.5 credits
• 2DT393, Reliability in Embedded Systems (Fredrik Ahlgren)
• 2DV515, Web Intelligence, 7.5 credits
• 2DV517, Deployment Infrastructures (Diego Perez)
• 2DV611, Continuous Delivery (Diego Perez)
• 4ME307, Internet Architectures, 7.5 credits 
• 4DV650, Systems Simulation and Modeling, 5 credits 
• 4DV651, Project in Model-based development, 10 credits 
• 4DV701, Formal Methods, 5 credits