Runtime evolution of highly dynamic software

Abstract

Highly dynamic software systems are applications whose operations are particularly affected by changing requirements and uncertainty in their execution environments. Ideally such systems must evolve while they execute. To achieve this, highly dynamic software systems must be instrumented with self-adaptation mechanisms to monitor selected requirements and environment conditions to assess the need for evolution, plan desired changes, as well as validate and verify the resulting system. This chapter introduces fundamental concepts, methods, and techniques gleaned from self-adaptive systems engineering, as well as discusses their application to runtime evolution and their relationship with off-line software evolution theories. To illustrate the presented concepts, the chapter revisits a case study conducted as part of our research work, where self-adaptation techniques allow the engineering of a dynamic context monitoring infrastructure that is able to evolve at runtime. In other words, the monitoring infrastructure supports changes in monitoring requirements without requiring maintenance tasks performed manually by developers. The goal of this chapter is to introduce practitioners, researchers and students to the foundational elements of self-adaptive software, and their application to the continuos evolution of software systems at runtime.