Measuring and Evaluating the Performance of Self-Organization Mechanisms Within Collective Adaptive Systems

Kuksa$$^{*}$$: Self-adaptive Microservices in Automotive Systems

“…We used the results of previous studies ( [8,16]) to create a metric for measuring the performance of the self-adaptive system under evaluation and comparing it with the static system architecture. The performance of software systems denotes its capabilities in its execution [8], in which two important measures include the solution quality and the time taken to achieve the solution [16]. Hence, the total performance in our designed system composed of time performance and solution quality performance [8], as follows: p(sys) = wt.tp(sys) + wq.qp(sys)…”

“…Another study [1] proposed an architecture-based self-adaptation service for cloud-native applications based on the Rainbow selfadaptation framework with support for Docker containers and Kubernetes. Current approaches mainly rely on the quantitative verification of system properties, including techniques to produce formal guarantees about the quantitative aspects of systems, such as performance [8].…”

A Metadata-Driven Approach for Testing Self-Organizing Multiagent Systems

“…In particular, there is a lack of approaches to test selforganizing MAS system [20] [21] [4]. Gardelli et al [20] provide a theoretical system-oriented approach that aims at anticipating design decisions at the early MAS design stages.…”

“…Bernon et al [21] provide a simulation-driven approach, which allows the developer to simulate different versions of the application while designing the self-organizing agents. Eberhardinger et al [4] present an approach for measuring the performance of self-organizing mechanisms at design time in order to select the best-fitting mechanism, and illustrate the proposed approach in an existing agent-based system. However, these approaches do not support the design and execution of functional tests, and fault diagnosis.…”

“…Nonetheless, there is a gap in the literature regarding the inspection of systems with these features. For example, there are few approaches to test the local interactions between agents in a self-organizing or self-adaptive MAS system and the global behavior that emerges from these interactions [4] [5]. One reason is the difficulty of specifying expected results for non-deterministic applications, especially in actual environments.…”