Dynamic pricing algorithms for task allocation in multi-agent swarms

Prithviraj Dasgupta, Matthew Hoeing

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Citations (Scopus)

Abstract

Over the past few years, emergent computing based techniques such as swarming have evolved as an attractive technique to design coordination protocols in large-scale distributed systems and massively multi-agent systems. In this paper, we consider a search-and-execute problem domain where agents have to discover tasks online and perform them in a distributed, collaborative manner. We specifically focus on the problem of distributed coordination between agents to dynamically allocate the tasks among themselves. To address this problem, we describe a novel technique that combines a market-based dynamic pricing algorithm to control the task priorities with a swarming-based coordination technique to disseminate task information across the agents. Experimental results within a simulated environment for a distributed aided target recognition application show that the dynamic pricing based task selection strategies compare favorably with other heuristic-based task selection strategies in terms of task completion times while achieving a significant reduction in communication overhead.

Original languageEnglish (US)
Title of host publicationMassively Multi-Agent Technology - AAMAS Workshops - MMAS 2006, LSMAS 2006, and CCMMS 2007, Hakodate, Japan, May 9, 2006, Honolulu, HI, USA, May 15, 2007, Selected and Revised Papers
Pages64-79
Number of pages16
DOIs
StatePublished - Oct 28 2008
Event1st International Workshop on Coordination and Control in Massively Multi-agent Systems, CCMMS 2007 - Honolulu, HI, United States
Duration: May 15 2007May 15 2007

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5043 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference1st International Workshop on Coordination and Control in Massively Multi-agent Systems, CCMMS 2007
CountryUnited States
CityHonolulu, HI
Period5/15/075/15/07

Fingerprint

Dynamic Pricing
Task Allocation
Swarm
Costs
Multi agent systems
Target Recognition
Completion Time
Large-scale Systems
Multi-agent Systems
Distributed Systems
Communication
Heuristics
Computing
Experimental Results
Strategy

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Dasgupta, P., & Hoeing, M. (2008). Dynamic pricing algorithms for task allocation in multi-agent swarms. In Massively Multi-Agent Technology - AAMAS Workshops - MMAS 2006, LSMAS 2006, and CCMMS 2007, Hakodate, Japan, May 9, 2006, Honolulu, HI, USA, May 15, 2007, Selected and Revised Papers (pp. 64-79). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5043 LNAI). https://doi.org/10.1007/978-3-540-85449-4_5

Dynamic pricing algorithms for task allocation in multi-agent swarms. / Dasgupta, Prithviraj; Hoeing, Matthew.

Massively Multi-Agent Technology - AAMAS Workshops - MMAS 2006, LSMAS 2006, and CCMMS 2007, Hakodate, Japan, May 9, 2006, Honolulu, HI, USA, May 15, 2007, Selected and Revised Papers. 2008. p. 64-79 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5043 LNAI).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Dasgupta, P & Hoeing, M 2008, Dynamic pricing algorithms for task allocation in multi-agent swarms. in Massively Multi-Agent Technology - AAMAS Workshops - MMAS 2006, LSMAS 2006, and CCMMS 2007, Hakodate, Japan, May 9, 2006, Honolulu, HI, USA, May 15, 2007, Selected and Revised Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 5043 LNAI, pp. 64-79, 1st International Workshop on Coordination and Control in Massively Multi-agent Systems, CCMMS 2007, Honolulu, HI, United States, 5/15/07. https://doi.org/10.1007/978-3-540-85449-4_5
Dasgupta P, Hoeing M. Dynamic pricing algorithms for task allocation in multi-agent swarms. In Massively Multi-Agent Technology - AAMAS Workshops - MMAS 2006, LSMAS 2006, and CCMMS 2007, Hakodate, Japan, May 9, 2006, Honolulu, HI, USA, May 15, 2007, Selected and Revised Papers. 2008. p. 64-79. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-540-85449-4_5
Dasgupta, Prithviraj ; Hoeing, Matthew. / Dynamic pricing algorithms for task allocation in multi-agent swarms. Massively Multi-Agent Technology - AAMAS Workshops - MMAS 2006, LSMAS 2006, and CCMMS 2007, Hakodate, Japan, May 9, 2006, Honolulu, HI, USA, May 15, 2007, Selected and Revised Papers. 2008. pp. 64-79 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{4acf9612c24b4848805c85a9d2d8e873,
title = "Dynamic pricing algorithms for task allocation in multi-agent swarms",
abstract = "Over the past few years, emergent computing based techniques such as swarming have evolved as an attractive technique to design coordination protocols in large-scale distributed systems and massively multi-agent systems. In this paper, we consider a search-and-execute problem domain where agents have to discover tasks online and perform them in a distributed, collaborative manner. We specifically focus on the problem of distributed coordination between agents to dynamically allocate the tasks among themselves. To address this problem, we describe a novel technique that combines a market-based dynamic pricing algorithm to control the task priorities with a swarming-based coordination technique to disseminate task information across the agents. Experimental results within a simulated environment for a distributed aided target recognition application show that the dynamic pricing based task selection strategies compare favorably with other heuristic-based task selection strategies in terms of task completion times while achieving a significant reduction in communication overhead.",
author = "Prithviraj Dasgupta and Matthew Hoeing",
year = "2008",
month = "10",
day = "28",
doi = "10.1007/978-3-540-85449-4_5",
language = "English (US)",
isbn = "3540854487",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "64--79",
booktitle = "Massively Multi-Agent Technology - AAMAS Workshops - MMAS 2006, LSMAS 2006, and CCMMS 2007, Hakodate, Japan, May 9, 2006, Honolulu, HI, USA, May 15, 2007, Selected and Revised Papers",

}

TY - GEN

T1 - Dynamic pricing algorithms for task allocation in multi-agent swarms

AU - Dasgupta, Prithviraj

AU - Hoeing, Matthew

PY - 2008/10/28

Y1 - 2008/10/28

N2 - Over the past few years, emergent computing based techniques such as swarming have evolved as an attractive technique to design coordination protocols in large-scale distributed systems and massively multi-agent systems. In this paper, we consider a search-and-execute problem domain where agents have to discover tasks online and perform them in a distributed, collaborative manner. We specifically focus on the problem of distributed coordination between agents to dynamically allocate the tasks among themselves. To address this problem, we describe a novel technique that combines a market-based dynamic pricing algorithm to control the task priorities with a swarming-based coordination technique to disseminate task information across the agents. Experimental results within a simulated environment for a distributed aided target recognition application show that the dynamic pricing based task selection strategies compare favorably with other heuristic-based task selection strategies in terms of task completion times while achieving a significant reduction in communication overhead.

AB - Over the past few years, emergent computing based techniques such as swarming have evolved as an attractive technique to design coordination protocols in large-scale distributed systems and massively multi-agent systems. In this paper, we consider a search-and-execute problem domain where agents have to discover tasks online and perform them in a distributed, collaborative manner. We specifically focus on the problem of distributed coordination between agents to dynamically allocate the tasks among themselves. To address this problem, we describe a novel technique that combines a market-based dynamic pricing algorithm to control the task priorities with a swarming-based coordination technique to disseminate task information across the agents. Experimental results within a simulated environment for a distributed aided target recognition application show that the dynamic pricing based task selection strategies compare favorably with other heuristic-based task selection strategies in terms of task completion times while achieving a significant reduction in communication overhead.

UR - http://www.scopus.com/inward/record.url?scp=54349116999&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=54349116999&partnerID=8YFLogxK

U2 - 10.1007/978-3-540-85449-4_5

DO - 10.1007/978-3-540-85449-4_5

M3 - Conference contribution

AN - SCOPUS:54349116999

SN - 3540854487

SN - 9783540854487

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 64

EP - 79

BT - Massively Multi-Agent Technology - AAMAS Workshops - MMAS 2006, LSMAS 2006, and CCMMS 2007, Hakodate, Japan, May 9, 2006, Honolulu, HI, USA, May 15, 2007, Selected and Revised Papers

ER -