A Biologically-inspired Adaptation Mechanism for Autonomic Grid Networks

Abstract

Summary form only given. This poster presentation describes and empirically evaluates a biologically-inspired adaptation mechanism that allows grid network services to autonomously adapt to dynamic environment changes in the network (e.g. changes in network traffic and resource availability). Based on the observation that the natural immune system has elegantly achieved autonomous adaptation, the proposed mechanism, called the iNet artificial immune system, is designed after the mechanisms behind how the natural immune system detects antigens (e.g. viruses) and specifically reacts to them. iNet models a behavior of grid network services (e.g. migration and replication) as an antibody, and an environment condition (e.g. network traffic and resource availability) as an antigen. iNet allows each grid network service to (1) autonomously sense its surrounding environment conditions (i.e. antigens) to evaluate whether it adapts well to the current conditions, and if it does not, (2) adaptively perform a behavior (i.e. antibody) suitable for the conditions (i.e. antigens). This poster presents the iNet architecture and its algorithm design. It also shows several empirical experimental results. They show that iNet works efficiently at small memory footprint and it makes grid network services adaptive by dynamically changing their population and location against environmental changes in the network