Bio-inspired multi-agent data harvesting in a proactive urban monitoring environment

  • Authors:

  • Uichin Lee;Eugenio Magistretti;Mario Gerla;Paolo Bellavista;Pietro Lió;Kang-Won Lee

  • Affiliations:

  • Department of Computer Science, UCLA, Los Angeles, CA 90095;Dipartimento di Elettronica, Informatica e Sistemistica, University of Bologna, Bologna, Italy 40136;Department of Computer Science, UCLA, Los Angeles, CA 90095;Dipartimento di Elettronica, Informatica e Sistemistica, University of Bologna, Bologna, Italy 40136;Computer Laboratory, University of Cambridge, England;IBM T. J. Watson Research Center, Wireless Networking Group, 19 Skyline Drive, Hawthorne, NY 10532, United States

  • Venue:
  • Ad Hoc Networks
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Vehicular sensor networks (VSNs) enable brand new and promising sensing applications, such as traffic reporting, relief to environmental monitoring, and distributed surveillance. In our past work, we have designed and implemented MobEyes, a middleware solution to support VSN-based urban monitoring, where agent vehicles (e.g., police cars) move around and harvest meta-data about sensed information from regular VSN-enabled vehicles. In urban sensing operations, multiple agents typically collaborate in harvesting and searching for key meta-data in parallel. Thus, it is critical to effectively coordinate the harvesting operations of the agents in a decentralized and lightweight way. The paper presents a bio-inspired meta-data harvesting algorithm, called datataxis, whose primary goal is to effectively cover large urban areas datataxis alternate foraging behaviors inspired by Escherichia coli chemotaxis and by Levy flights to favor agent movements towards ''information patches'' where the concentration of meta-data is high. The proposed scheme avoids harvesting duplication by preventing superfluous concentration of agents in the same region at the same time using stigmergy. We have validated datataxis via extensive simulations that demonstrate how the proposed bio-inspired behavior of harvesting agents effectively coordinates their movements, thus outperforming other decentralized strategies. Our solution was shown to be robust and to work well under a wide range of operation parameters, thus making it easily and rapidly deployable for different urban sensing operations.