Experimental Study of Channel Utilization in IEEE 802.11 MAC Protocol using NS2

Task:

A major concern in the design of wireless MAC protocols is the protocol efficiency, i.e., does the protocol allow high channel utilization. Channel utilization here is measured as the percentage of time the channel is transmitting user data, i.e., a channel utilization of 80% for a 2 Mbps wireless channel would mean that the MAC protocol supports a sustained 1.6 Mbps user data rate. User data here includes all the data in a MAC PDU that is NOT MAC control data, including upper layer protocol headers such as IP etc. NS2 provides an implementation of the most widely used wireless LAN MAC protocol, IEEE 802.11. In this assignment, experimentally determine the channel utilization that IEEE 802.11 is able to achieve and how that value is impacted by different parameters/factors.

he basic idea here is to design a suite of network scenarios where one or multiple stations send data as fast as they can to a common receiver. The MAC protocol will determine who gets to transmit and whether a collision will occur. By monitoring the rate at which data packets are received at the common receiver, you can then determine the sustained data rate that the protocol supports and hence the channel utilization. Note that this description implies two things:

1. All stations are within transmission range of each other (otherwise, the MAC protocol is not really managing access to the shared medium by all stations, and multiple stations could be transmitting at the same time).

2. The rate at which a station attempts to gain access to the channel should not be slowed down by some rate control/congestion control protocol. Rather, stations should always be back-logged, i.e., they should always have data to transmit and it will be up to the MAC protocol to determine who gets access when. So you cannot use TCP to generate traffic.

We are not interested in tuning any of the protocol parameters. Rather, use IEEE 802.11 with all its default settings as implemented in NS2. However, channel utilization is impacted by a number of factors outside the control of the MAC protocol, such as:

 Number of contending stations/senders
 Packet size
 Whether the protocol uses RTS/CTS or not

Design experiments (scenarios) that determine channel utilization as a function of the first two factors. This implies, at the very least, that you need to vary the factor you are studying, using more than just two different values – ideally 5 to 10 distinct values. In designing, conducting and reporting on your experiments, please take the lessons from the papers on running simulation studies into account that are posted on the course website (under the “Working with NS2” link). In particular, you should read and apply the lessons from:

Designing Experimental Scenarios

 Simulations in Wireless Sensor and Ad Hoc Networks: Matching and Advancing Models, Metrics, and Solutions
 MANET Simulation Studies: The Incredibles
 On the Credibility of Manet Simulations

In particular, these papers stress that good experimental papers require that the scenario description is complete (contains all relevant details such as number and location of nodes, traffic generated, duration of simulation, how you measured your performance criteria, etc.). Also, since we are dealing with random access protocols, you typically need to conduct multiple simulation runs with different random number seeds. Each data point should then be shown not only with the average value but also some indication about the variability of these results (standard deviation, confidence intervals, etc.).

Formulate hypothesis about how channel utilization will vary with regard to the number of contending stations and the packet size. Describe why you believe these hypothesis to be true. Design and run the experiments, collect the data, and present your results in an appropriate form. Discuss your results. Do the results match your expectations? Can you confirm that the results you collected are valid?

Please submit a report that addresses all the issues above. The report should be kept relatively brief, not more than 5 pages or so (assuming reasonable margins and font size). I prefer you submit the report in Word or PDF format. The deadline is February 14 midnight (date/timestamp of e-mail). It should have the following main sections (in this order):

 Expected results: formulate your hypothesis and your justifications for each prediction.

 Description of experimental scenarios: ensure that the description is complete enough so that a reader could replicate your scenarios (i.e., all relevant information/parameter settings need to be listed, as well as your definition of the metric(s) you measure, and how you extract them from the simulation tracefiles)
 Experimental results: show channel utilization as a function of number of contending stations and as a function of the packet size. Show both average values and some measure of variability (ideally in the form of a 95% confidence interval).
 Discussion: do the results match your expectations? Can you explain them? Can you confirm that the results you arrived at experimentally would also be true with real hardware (which is called ‘validation’)?
Also please submit, as a softcopy, your simulation scripts, scenario files, and any scripts or programs you wrote to extract data from the NS2 tracefiles. Add a short description/README file that discusses how to run simulations and extract the data in your report.