Learning Task Patterns to Improve Efficiency and Coordination in Decentralized Autonomic Computing Systems

We present the concept of an efficiency and coordination advisor for autonomic computing approaches for dynamic optimization problems. The problem scenarios targeted contain recurring tasks that our advisor identifies over several runs of the autonomous system thus giving it some limited way to “look into the future”. If the solutions created by the autonomous agents are much worse than the optimally possible solution, the advisor creates exception rules for the agents that make the wrong decisions for the recurring tasks. This allows them to do better decisions in the future in very specific situations while still retaining all advantages of the autonomic computing approach. Our experiments with dynamic instances of the pickup and delivery problem that have recurring tasks in it show that with our advisor approach we can improve substantially instances that result in suboptimal behavior of the autonomous agents without advisor. Our advisor approach is also successful if the recurring tasks change over time.