Software framework for multi agent system development in embedded systems

Abstract

Multi agent systems and embedded systems both are consider as major fields in the recent research history. Both have shown the potential of developing intelligence solutions to the modern world needs. Several standards including FIPA and ACL followed by the developers and researchers when model the multi agent systems. JADE like frameworks play a major role in the field where user can design and develop multi agent system using prebuild components in the framework. Even though there are number of multi agent solutions which are run on top of the embedded systems there is no common framework to facilitate the multi agent system developments in the embedded environments. Intention is to introduce a common framework which is designed and developed to facilitate major multi agent behaviors to the researches in this field to make their work easy. In order to archive this goal completely new software framework was design, developed and evaluated for the major embedded hardware platforms available in the market. This frame work can provide communication platform for the different type of communication channels and multi agent behaviors to the host software system in embedded platforms. Since framework is fully compatible with the FIPA standard and communication using standard communication language ACL users can easily use the framework. Solution was design in such a way that users can implement their own modules and plug in easily. In additional to that this modular design allows users to port the framework to new hardware platforms with minimal changes. Evaluation of the framework was conducted by using several approaches. Completely new embedded solution was design and developed for the field of home garden watering in order to evaluate the features of the framework. Where different agent with different communication capabilities; demonstrate the usefulness of the framework. In addition to that several memory matrixes and performance matrixes were generated in order to evaluate the performance of the framework. Memory analysis shows that framework will consume 26.7% of the flash memory on average and can operate on minimum of 1KB static random access memory.