Automatic Adaptation Rule Optimization via Large Language Models

TL;DR

This paper addresses the challenge of designing high-performance, robust adaptation rules for self-adaptive systems within a large design space. It proposes an LLM-driven pipeline that builds and optimizes rules inside the MAPE-K loop, supported by a knowledge base, an analyzer, and a planner, with rules emitted as executable C++. Preliminary evaluation in the SWIM simulator demonstrates that LLM-generated rules can outperform manually designed baselines and can achieve strong initial performance, though optimization via iteration is costly and exhibits fluctuations. The work highlights the potential of combining LLMs with traditional search methods to create more efficient, adaptable rule-based systems and points to runtime evolution as future work.

Abstract

Rule-based adaptation is a foundational approach to self-adaptation, characterized by its human readability and rapid response. However, building high-performance and robust adaptation rules is often a challenge because it essentially involves searching the optimal design in a complex (variables) space. In response, this paper attempt to employ large language models (LLMs) as a optimizer to construct and optimize adaptation rules, leveraging the common sense and reasoning capabilities inherent in LLMs. Preliminary experiments conducted in SWIM have validated the effectiveness and limitation of our method.

Figures (2)

• Figure 1: System Overview.
• Figure 2: Results from ten experiments.