ePrints.FRI - University of Ljubljana, Faculty of Computer and Information Science

Context prediction-based prefetching in software-defined wireless networks

Nejc Bizjak (2016) Context prediction-based prefetching in software-defined wireless networks. MSc thesis.

Download (2329Kb)


    In this master thesis we focus on improving in-network caching for mobile users in a large campus WiFi network. First we pinpoint the negative effects of mobility on network conditions and user experience. We propose a method leveraging SDN technology to redirect users' requests to optimally located cache servers, resulting in improved user experience and lowered burden on the backhaul and core network links. Our contribution is a network application that controls the flows in the network via an SDN controller. The application takes user's movement traces as an input, computes the optimal location of cache servers in the network and redirects user's flows accordingly. We tested our solution in a Mininet network simulator. We devised multiple scenarios using real-world movement traces from Dartmouth Campus. We measured requests delay as the main characteristic for user experience and data traffic over core and backhaul links as an indicator of network health. Our experiments show that for mobile users our dynamic redirection approach provides noticeable improvements over traditional, static caching methods.

    Item Type: Thesis (MSc thesis)
    Keywords: Software defined networking, caching, WiFi network, user mobility
    Number of Pages: 70
    Language of Content: English
    Mentor / Comentors:
    Name and SurnameIDFunction
    doc. dr. Veljko PejovićMentor
    Link to COBISS: http://www.cobiss.si/scripts/cobiss?command=search&base=51012&select=(ID=1537242051)
    Institution: University of Ljubljana
    Department: Faculty of Computer and Information Science
    Item ID: 3639
    Date Deposited: 06 Oct 2016 14:26
    Last Modified: 26 Oct 2016 13:54
    URI: http://eprints.fri.uni-lj.si/id/eprint/3639

    Actions (login required)

    View Item