Biologically-inspired adaptation of autonomic network applications†

Abstract

This paper describes and empirically evaluates a new biologically-inspired adaptation mechanism for network applications in the NetSphere architecture. The NetSphere architecture is inspired by the observation that the biological systems (e.g. bee colonies) have already developed mechanisms necessary to achieve future network requirements such as autonomy and adaptability. In the NetSphere architecture, a network application is implemented as a group of distributed and autonomous agents (analogous to a bee colony consisting of multiple bees). Each agent implements a functional service related to the application and follows simple behaviors similar to biological entities such as reproduction, replication, migration and environment sensing. Research supported by OGIS International, Inc. and Electric Power Development Co., Ltd. The proposed agent adaptation mechanism runs on the middleware platform for the NetSphere architecture, called the NetSphere platform. Designed after the mechanism behind how the immune system produces specific antibodies against an antigen invasion, the proposed adaptation mechanism allows agents to autonomously monitor their surrounding environmental conditions (e.g. traffic volume and resource availability) and adaptively perform their behaviors (e.g. reproduction and migration) suitable for the current environmental conditions. The empirical evaluation shows that the proposed mechanism achieves autonomous adaptability of network applications and the NetSphere platform is efficient and reusable to host autonomous adaptive network applications.