CARD: a contact-based architecture for resource discovery in wireless ad hoc networks

  • Authors:

  • Ahmed Helmy;Saurabh Garg;Nitin Nahata;Priyatham Pamu

  • Affiliations:

  • 3740 McClintock Avenue, EEB 232, Electrical Engineering Department, University of Southern California, Los Angeles, CA;3740 McClintock Avenue, EEB 232, Electrical Engineering Department, University of Southern California, Los Angeles, CA;3740 McClintock Avenue, EEB 232, Electrical Engineering Department, University of Southern California, Los Angeles, CA;Computer Science Department, University of Southern California, Los Angeles, CA

  • Venue:
  • Mobile Networks and Applications
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Traditional protocols for routing in ad hoc networks attempt toobtain optimal or shortest paths, and in doing so may incursignificant route discovery overhead. Such approaches may beappropriate for routing long-lived transfers where the initial costof route discovery may be amortized over the life of theconnection. For short-lived connections, however, such as resourcediscovery and small transfers, traditional shortest path approachesmay be quite inefficient. In this paper we propose a novelarchitecture, CARD, for resource discovery in large-scale wirelessad hoc networks. Our mechanism is suitable for resource discoveryas well as routing very small data transfers or transactions inwhich the cost of data transfer is much smaller than the cost ofroute discovery. Our architecture avoids expensive mechanisms suchas global flooding and complex hierarchy formation and does notrequire any location information. In CARD resources within thevicinity of a node, up to a limited number of hops, are discoveredusing a proactive scheme. For resources beyond the vicinity, eachnode maintains a few distant nodes called contacts. Contacts helpin creating a small world in the network and provide an efficientway to query for distant resources. Using contacts, the networkview (or reachability) of the nodes increases, reducing thediscovery overhead and increasing the success rate. On the otherhand, increasing the number of contacts also increases controloverhead. We study such trade-off in depth and present mechanismsfor contact selection and maintenance that attempt to increasereachability with reduced overhead. Our schemes adapt gracefully tonetwork dynamics and mobility using soft-state periodic mechanismsto validate and recover paths to contacts. Our simulation resultsshow that CARD is scalable and can be configured to providedesirable performance for various network sizes. Comparisons withother schemes show overhead savings reaching over 93% (vs.flooding) and 80% (vs. bordercasting or zone routing) for highquery rates in large-scale networks.