mca 13(3): e5

Research Article

Online Algorithms for Adaptive Optimization in Heterogeneous Delay Tolerant Networks

Download807 downloads
  • @ARTICLE{10.4108/mca.1.3.e5,
        author={Wissam  Chahin and Francesco De Pellegrini and Rachid El-Azouzi and Amar Prakash Azad},
        title={Online Algorithms for Adaptive Optimization in Heterogeneous Delay Tolerant Networks},
        journal={EAI Endorsed Transactions on Mobile Communications and Applications},
        volume={1},
        number={3},
        publisher={ICST},
        journal_a={MCA},
        year={2013},
        month={12},
        keywords={},
        doi={10.4108/mca.1.3.e5}
    }
    
  • Wissam Chahin
    Francesco De Pellegrini
    Rachid El-Azouzi
    Amar Prakash Azad
    Year: 2013
    Online Algorithms for Adaptive Optimization in Heterogeneous Delay Tolerant Networks
    MCA
    ICST
    DOI: 10.4108/mca.1.3.e5
Wissam Chahin1,*, Francesco De Pellegrini2, Rachid El-Azouzi1, Amar Prakash Azad3
  • 1: CERI/LIA, University of Avignon, France
  • 2: CREATE-NET, Trento, Italy
  • 3: SOE, UCSC, USA
*Contact email: wissam.chahin@etd.univ-avignon.fr

Abstract

Delay Tolerant Networks (DTNs) are an emerging type of networks which do not need a predefined infrastructure. In fact, data forwarding in DTNs relies on the contacts among nodes which may possess different features, radio range, battery consumption and radio interfaces. On the other hand, efficient message delivery under limited resources, e.g., battery or storage, requires to optimize forwarding policies. We tackle optimal forwarding control for a DTN composed of nodes of different types, forming a so-called heterogeneous network. Using our model, we characterize the optimal policies and provide a suitable framework to design a new class of multi-dimensional stochastic approximation algorithms working for heterogeneous DTNs. Crucially, our proposed algorithms drive online the source node to the optimal operating point without requiring explicit estimation of network parameters. A thorough analysis of the convergence properties and stability of our algorithms is presented.