Behavioral Maps : A framework to Assess and Validate Self-Adaptive Architectures at Runtime

Défense de thèse de Edilton Lima Dos Santos

A Self-adaptive System (SAS) is designed to adapt to changes in the operating environment at runtime. This adaptation can modify the system's structure, behavior, or even its adaptation mechanism. However, they can also introduce defects and architectural issues (e.g., architectural bad smells) that can cause the system to fail. Therefore, it is crucial to conduct thorough testing and architectural analysis while the system is running to ensure performance and integrity. While previous studies available in the literature have focused on analyzing architectural issues and testing during the design phase, evaluating the system at runtime is equally essential. To address this, the Behavioral Map framework was proposed.

 

The framework creates a graphical map that describes the configuration analyzed at runtime. This map allows for determining the testing boundaries and generating test cases based on the selected scope. This scope can be determined through either Feature Relationship Analysis or Architectural Bad Smells Analysis. The test case generation processes can generate tests using adaptive-random tests, evolutionary algorithms, and combinatorial test design.

 

Several studies were conducted to evaluate the approach. The first study analyzed feature interaction and ABS detection at runtime in three SASs. It was found that some ABS only appear in specific system configurations or architectures. The second study compared ABS detected at runtime to those detected at design time, revealing differences between the two. Lastly, the feasibility of the testing approach was assessed, and it was found to be feasible to select the test scope at runtime for SASs.

Contact : Isabelle Daelman - isabelle.daelman@unamur.be
Télecharger : vCal