On Learning in Collective Self-Adaptive Systems: State of Practice and a 3D Framework

Abstract

Collective self-adaptive systems (CSAS) are distributed and interconnected systems composed of multiple agents that can perform complex tasks such as environmental data collection, search and rescue operations, and discovery of natural resources. By providing individual agents with learning capabilities, CSAS can cope with challenges related to distributed sensing and decision-making and operate in uncertain environments. This unique characteristic of CSAS enables the collective to exhibit robust behaviour while achieving system-wide and agent-specific goals. Although learning has been explored in many CSAS applications, selecting suitable learning models and techniques remains a significant challenge that is heavily influenced by expert knowledge. We address this gap by performing a multifaceted analysis of existing CSAS with learning capabilities reported in the literature. Based on this analysis, we introduce a 3D framework that illustrates the learning aspects of CSAS considering the dimensions of autonomy, knowledge access, and behaviour, and facilitates the selection of learning techniques and models. Finally, using example applications from this analysis, we derive open challenges and highlight the need for research on collaborative, resilient and privacy-aware mechanisms for CSAS.

Learning to Learn in Collective Adaptive Systems: Mining Design Patterns for Data-driven Reasoning

“…In this section, we present how we extended our systematic literature review on learning-enabled CAS [9] and use its results as input for our data-driven methodologies. Then we Clustering allows us to capture the set of design choices applied on CAS based on past experience (i.e., the state of the art).…”

“…This paper extends the literature review of [9] by following the same search and analysis methods. The analysis of the most recent studies added 7 additional research papers.…”

“…In [9], we conducted a systematic literature review of 52 studies related to learning-based CAS. The investigated papers are classified based on their choices for the nine design dimensions envisioned for learning-enabled CAS.…”

“…This paper contributes to this direction in the following ways. (i) We extend our previous systematic literature review on learning-enabled CAS [9] and use the analysis results as data capturing relevant past experiences. (ii) By employing data-driven methodologies, we identify correlations in the collected data, based on which we present relevant design-time reasoning knowledge in the form of design guidelines.…”

On the Social Implications of Collective Adaptive Systems

“…CASs usually rely on learning and emergence to cope with change and unpredictability [50]. Both techniques (often used in conjunction) need subsequent phases of design, validation, and tuning.…”