Knowledge about oneself and how this influences one’s behaviour are well-established concepts from the fields of psychology and cognitive science. Over the past five years, an international team of researchers has studied how these concepts can be applied to computing systems, and the results have recently been published in a book.
“As the complexity of technical systems is steadily increasing, traditional computing systems with predefined functionality will soon reach their limits. Hence, innovative computing systems must be able to continuously assess their own state and make autonomous decisions, in order to adapt to unforeseen changes”, Bernhard Rinner (Institute of Networked and Embedded Systems) explains. He is a joint editor of a recent publication on so-called “self-aware computing systems”. Rinner expands further: “We were able to take this concept from psychology and expand it for the analysis and the design of computing systems. It works by maintaining models of the computer’s own state and its environment, and by adapting both the application (software) and the underlying platform (hardware) at runtime.”
This book is the first ever to comprehensively present computational self-awareness as a design method for computing systems and networks, and it further provides a detailed discussion of various case studies. Inspired by well-known concepts from the field of psychology and cognitive science, computational self-awareness focuses on the following research topics: First, what information is exploited for the manifestation of self-awareness; second, which level of self-awareness is realised; and third, whether the adaptation is performed in a single computer or in a network. This requires a variety of techniques and algorithms. Rinner elucidates: “Online learning methods represent a key component of these novel computing systems. With these methods, models of the own state and the environment are maintained and kept up to date.”
As a case study, the team in Klagenfurt developed a network of cameras based on computational self-awareness. The cameras pursue a common objective, however – depending on the specific level of self-awareness – they make the decision about their individual contribution to the objective autonomously. “As shown in a person tracking use case in the camera network, we were able to demonstrate that computational self-awareness leads to more resource-efficient solutions”, Jennifer Simonjan, a member of the institute’s research staff, points out. “A further advantage lies in the autonomous learning performed by the network topology. This allows the camera network to configure itself autonomously.” Computational self-awareness is not only suitable as a design method for camera networks. The international team including members from Germany, U.K., Norway and Switzerland successfully demonstrated this method in examples such as high-performance computers for financial modelling, interactive music systems, and the management of cloud systems.
Lewis, P.R., Platzner, M., Rinner, B., Torresen, J. & Yao, X. (Eds.) (2016). Self-aware Computing Systems: An Engineering Approach. Heidelberg: Springer.