Interwoven Systems: Self-Improving Systems Integration

Abstract

Current trends in information and communication technology show that systems are increasingly influencing each other -which is seldom completely anticipated at design-time. As a result, mastering system integration with traditional methods becomes infeasible due to the resulting complexity. In this paper we argue that self-improving system integration is the most promising solution to counter the resulting challenges. Thereby, we highlight the different aspects of such a process with special attention to the optimisation question and discuss how approaches from the domain of self-organising systems -in particular Organic and Autonomic Computing -will be beneficial when researching possible solutions. I. MOTIVATIONInformation and communication technology (ICT) pervades every aspect of our daily lives. This inclusion changes our communities and all of our human interactions. It also presents a significant set of challenges in correctly designing and integrating our resulting technical systems. For instance, the embedding of ICT functionality in more and more devices (such as household appliances or a power grid's thermostats) leads to novel interconnections and a changing structure of the overall system. Not only technical systems are increasingly coupled, a variety of previously isolated natural and human systems have consolidated into a kind of overall system of systems -an interwoven system structure.In this context, the term "interwoven" [12] refers to the several aspects of coupling and mutual influences between component systems: heterogeneous elements interact, indirect influences of behaviour can be observed, and the context, cooperation and composition of systems are uncertain during runtime. For instance, legacy systems have to cooperate with novel solutions or different versions of the same system have to coexist and cooperate. This leads to novel challenges for system engineers and administrators.The ongoing integration of interwoven systems depends on the valid and timely knowledge of each of the participating systems on dynamically changing goals or priorities, on the state of its available resources and on the details of its operational context. It is infeasible to have such timely and intimate dynamically-changing knowledge of the participating systems come from an external master controller for the overall interwoven system. Hence, the most promising approach to master such systems relies on increasing their self-organisation capabilities. Within the last decade, initiatives like Autonomic Computing (AC) [6] and Organic Computing(OC) [11] have started to investigate and build systems based on the fundamentals of self-organisation: Concepts from natural organisms have been transferred to technical systems in order to achieve "life-like" characteristics such as adaptivity, robustness, and flexibility. Architectural patterns, customised machine learning techniques and collaboration mechanisms have been developed. However, these approaches and mechanisms have been developed with the focus on individual...

Self-improving system integration: Mastering continuous change

“…b) Problem classes: The application scenarios underlying the various theoretical considerations range from autonomous vehicles [42] over vehicular traffic control [45] and route guidance [46], robotics and simulation [40], emergency response [47], to intelligent surveillance systems [44] and energy systems [48]. From these contributions, together with those that take a more generalist system-of-systems stance [6,43,37] or that involve specific interaction models, e.g., [41], we can extract the following core problem classes:…”

“…This situation is further complicated by the lack of shared knowledge and/or pooled authority when considering the control of all involved systems [3]. Thus the self-improving system integration community [4,5] has emphasised the consideration of what are known as Interwoven Systems [6,7,8] -that is, systems that integrate themselves and adapt as necessary, in response to their own behaviour and each other, within open and unpredictable environments.…”

Verification and Uncertainties in Self-integrating System

“…In order to operate efficiently these systems need to integrate themselves properly and improve over time. This can be achieved by optimising their individual activities but also by combining their efforts in achieving their individual and common goals [1]. Along with their own adaptation, the environment may change and adapt accordingly, further increasing the challenges of the integration and improvement process.…”

Coopetitive Soft Gating Ensemble

“…Recently, an ensemble method called Coopetitive Soft Gating Ensemble or CSGE was proposed for wind power forecasts. However, CSGE also offers a basic technology to make the functionality of self-improving system integration (SISSY) more robust against interference and to better avoid uncertainty by its weighting and optimization scheme [7].…”