BEECLUST swarm algorithm: analysis and implementation using a Markov chain model. (1st January 2013)

• Record Type: Journal Article
• Authors: Hereford, James
• Is Part Of: International journal of innovative computing and applications. Volume 5:Number 2(2013)
• Page Start: 115
• Page End: 124
• Publication Date: 2013-01-01
• Subjects: swarm algorithms -- BEECLUST -- Markov chain -- swarm robotics -- innovative applications; Evolutionary computation -- Periodicals; Neural networks (Computer science) -- Periodicals; Genetic programming (Computer science) -- Periodicals; Biologically-inspired computing -- Periodicals; Swarm intelligence -- Periodicals; Quantum computers -- Periodicals; 006.3
• Languages: English
• ISSNs: 1751-648X

Abstract

In this paper, we analyse a new swarm search algorithm based on the behaviour of social insects, specifically honey bees. The new algorithm, called BEECLUST, is unique because it does not require any bot-bot communication and does not require the bots to know their position. In this paper, we describe the BEECLUST algorithm and model the algorithm using a birth and death Markov chain. From the Markov model, we answer two questions: 1 Will the bots (eventually) congregate or cluster near the targets (maxima) in the search space? 2 How long does it take the bots to cluster? We corroborate the time-to-cluster analysis with laboratory measurements. The experimental results were done with epuck robots in two different search spaces. In general, the BEECLUST algorithm shows promise for using very simple bots in swarm search applications if conditions during the search allow for bot-bot collisions near the target.