Assignment 2

Due 19/10

  
  • Basics
    The commodity 802.11 radio platform is a popular choice for building Wireless Mesh Networks (WMNs).
    However, 802.11 MAC was not designed for multihop wireless networks, and thus scales poorly in this environment.
    In this assignment you will investigate the performance challenges of TCP in a 802.11-based WMN using the network simulator ns-2.
    You may do this assignment either in groups of two or individually.
  • Installing NS-2
    You can download the latest version of ns-2 from the web:
    http://www.isi.edu/nsnam/ns/. We recommend installing the the ns-allinone package as it is often much less of a hassle.
    Installation instructions are available in the README file included with the ns-allinone package. Any installation problems should be directed to ns-users mailing list.

  • Getting Started
    ns-2 is a widely-used simulation tool for networking research. If you have not used ns-2 before, you may wish to start with Marc Greis' tutorial available here: www.isi.edu/nsnam/ns/tutorial/index.html. For more indepth discussion on ns-2, you may refer to the ns-2 manual also available on the web:
    http://www.isi.edu/nsnam/ns/ns-documentation.html. Another good reference for simulating wireless networks in ns-2 is here:
    http://evanjones.ca/ns2.html. This also includes a set of tools you may use to parse ns-2 trace files.
  • Assignment
    For simplicity, in this assignment we will limit our performance analysis to a single TCP stream. Interactions between multiple contending streams, including a mix of TCP and UDP traffic, will be studied later in this course.

    1. A number of TCP variants have been proposed over the years. Recent ns-2 releases support TCP Reno, NewReno, and SACK, amongst others. You will evaluate the impact of hop count on the end-to-end throughput of a stream for each of these three TCP variants. Report your results by varying the hop count from 1 to 10 hops. You may use a simple chain topology with equidistant node separation and uniform wireless link rates. Identify all sources of packet losses as observed in the simulation trace.
    Describe how the observed behavior relates to our theoretical understanding of spatial reuse in a multihop wireless network.

    2. Using your choice of the TCP variants discussed above, evaluate the impact of virtual carrier sensing in eliminating collisions due to hidden terminals in a wireless chain topology, varying the hop count from 1 to 10 hops.

    3. Next, you will analyze the impact of wireless interface queue size, by experimenting with a queue size of 5, 10, and 50 packets while varying the hop count of a chain topology from 1 to 10 hops.

    4. 802.11 MAC incurs a fixed overhead per packet. You will quantify this overhead for TCP traffic by experimenting with packet sizes of 40 B and 1500 B over a chain topology, varying the hop count from 1 to 10 hops.

  • Handing In
    For this assignment you will demonstrate your results using a presentation. The presentation should not exceed 15 slides. You are expected to highlight the key inferences from your experiments and figures. Also, submit your tcl scrips and C/C++ code.

 

© 2010 Basem Shihada