Towards a Priority-Based Task Distribution Strategy for an Artificial Hormone System

This paper presents a priority-based task distribution strategy as an extension to the Artificial Hormone System (AHS). The AHS is a distributed middleware based on self-organization principles. It allows to distribute tasks to processing nodes in a self-organizing way while neither having a single-point-of-failure nor requiring external user input. Node failures are detected automatically, resulting in relocation of any affected tasks to operational nodes. This provides self-healing capabilities if sufficient computational resources are available. Our extension allows tasks to have priorities and enables self-healing by gracefully degrading the system based on the task priorities if the computational resources are not sufficient to completely self-heal the system. We present our extension and analyze its worst-case time bounds for self-configuration as well as self-healing. Quickly degrading the system in overload situations requires a strategy deciding which tasks to stop in such situations. We present a simple strategy and analyze its worst- and average-case self-healing duration.