A dynamic modelling framework for control-based computing system design

Abstract

This manuscript proposes a novel viewpoint on computing systems' modelling. The classical approach is to consider fully functional systems and model them, aiming at closing some external loops to optimize their behaviour. On the contrary, we only model strictly physical phenomena, and realize the rest of the system as a set of controllers. Such an approach permits rigorous assessment of the obtained behaviour in mathematical terms, which is hardly possible with the heuristic design techniques, that were mainly adopted to date. The proposed approach is shown at work with three relevant case studies, so that a significant generality can be inferred from it.

Control Strategies for Self-Adaptive Software Systems

“…The knowledge about the state variables' current values and the input values of the system allows one to formally determine the values of the output variables [Papadopoulos et al 2015]. The choice of the state variables may not be unique since an infinite number of equivalent representations can be found [Åström and Murray 2008].…”

“…This system, however, may be very complex and its dynamics may not be completely understood. Developing accurate system models over an operating range is a challenging task [Åström and Murray 2008], especially for computing systems [Hellerstein et al 2004;Maggio et al 2012;Papadopoulos et al 2015]. Even if a detailed mathematical model is available, it may be complex and making the controller design challenging and computationally expensive [Morari and Zafiriou 1989].…”

“…However, software engineers often lack the mathematical background needed for a deeper understanding of these A:2 models. Goal formalization and knob identification are additional concerns of modeling, where specific characteristics have to be taken into account to achieve a suitable controllability of the system [Papadopoulos et al 2015].…”

Software Engineering Meets Control Theory

“…Examples include the length of a queue, the percentage of frames already encoded, or the current encoding speed. Given the state variables' current values and the system's input values, one can formally determine the output variables' values [57]. The choice of the state variables is not unique since an infinite number of equivalent representations can be found [3].…”

“…Analytical abstractions of software established for quality assurance can help fill the gap between software models and dynamical models [19,27,71], but they are not sufficient for full control synthesis. In addition, goal formalization and knob identification have to be taken into account to achieve software controllability [57].…”