For all the experiments you will be using Linux PCs in LWSN B131 that are connected to each other and the global Internet. Specialized equipment is provided by the TA.
 

Purpose Of The Experiment

In this exercise, you will become familiar with a measuring device known as a Time Domain Reflectometer (TDR) that can be used to detect a broken cable.
 

Overview

A TDR attaches to one end of a copper cable, and measures the length of the cable. It does so by  transmitting electrical pulses of a suitable shape. Irregularities in the cable cause "reflections" of the electrical signals, which are received back at the TDR. In particular, the (unterminated) end of a broken wire always causes electrical signals to reflect in the same way a mirror reflects light.

A TDR contains an accurate timer which it uses to measure the difference in time between the original transmission and the reflection. From the time and the speed of propagation, it computes the length of the wire (i.e., distance to the fault).
 

Procedure And Details

To see a TDR in action, you will be provided with a TDR and some cables. You will use the TDR to measure the length and related properties of the provided cables. Record the various measurements. Note: some cables may be damaged, others fine. Use the WWW to research what grades (i.e., quality) of copper cable there are and the propagation speeds of electrical signals over these wires. Submit a 1/2-page write-up of what experiments you carried out, the measurements you observed, and your interpretation of the measurements. Is there a way to evaluate the accuracy of the TDR? Is there a way to infer what speed of propagation the TDR may be using when computing distances based on reflected signals?
 

Ping (25 pts):

The Ping (Packet Internet Groper) program is used to query another system and ensure a connection is active as well as determine the completion time (the query packet to reach its destination plus the response packet to arrive). It operates by sending a request to the destination machine for a reply. If the destination's IP software receives the request it issues a reply immediately. If ping is unable to reach the destination machine an error message will be shown.

Run ping as

$ ping local-machine -i 1 -c 10 -s 512

where local-machine is the hostname or IP address of another local machine in the LWSN B131 lab. The "-i" flag specifies that packets be emitted once every second, "-s" specifies the size of the packet (in bytes), and "-c" specifies the number of packets that should be emitted. Record the terminal output using script. Rerun the experiment with packet sizes "2048" and "8192". Use gnuplot to plot the average round-trip time (RTT), minimum RTT, and maximum RTT (on the y-axis) against packet size (on the x-axis). Do you observe a trend?

Rerun the previous experiments with hosts www.cs.purdue.edu, www.purdue.edu, www.iupui.edu, www.wisc.edu, www.utah.edu, www.sdsc.edu, and www.hawaii.edu (try www.hpu.edu if www.hawaii.edu does not work). Fix the payload size at 512.

Use mapquest.com (or other tool) to estimate the distance (in miles) from Purdue to the target destinations. For www.cs.purdue.edu and www.purdue.edu, try your best to locate their physical whereabouts and guess their approximate distance from LWSN B131. Use the speed-of-light to calculate the minimum latency needed for a bit to travel from here to there. How much do the calculated numbers differ from the measured numbers obtained using ping? Summarize your results in a table.

Find three Internet sites across the globe to which the ping values are especially high. The person who finds the "furthest" site with respect to delay (note that ping values vary by time of day, among other factors, so please indicate the time of day of the measurement) will receive 15 bonus points.
 

Traceroute (25 pts):

This utility tracks the route that a packet from your machine follows, over the Internet, to the destination machine.

It gives you the number of hops that your packet goes through and also displays their corresponding IP address and delay (like ping).

Run traceroute as

$ traceroute dst-machine

Rerun traceroute with the destinations from the ping experiment. Use script to record the output. How many IP hops does it take for the packet to make it to their respective destinations? Use gnuplot to draw a graph where the x-axis is the approximate distance from LWSN B131 and the y-axis is the router hop count as output by traceroute.

The traceroute destinations are approximately westerly from Purdue. Given that the traceroute packet originate from the same machine (the machine from which you are running traceroute), how common or disjoint are the traceroute discovered paths to the westerly destinations? Note that one of the reasons why there is a discrepancy in the ping latencies from mapquest distance/SOL estimations is that the actual distance traversed by packets going from router to router is likely to be different (sometimes significantly) from road distance calculated by mapquest. From the names of the intermediate router hops, estimate how many different organizations (multiple routers may belong to the same organization) a packet traverses to reach the destinations. Use gnuplot to draw a graph where the y-axis is the number of different organization hops (the x-axis remains the same). How do the two plots differ?

Go to traceroute.org and perform traceroute from SDSC back to Purdue (your local machine in the lab). Are the forward and backward paths identical, i.e., symmetric? What about the delay values? Find two additional sites on traceroute.org where one is symmetric and the other is not. If you have a broadband network connection at home (cable or DSL), perform traceroute from your home machine to the lab machine in LWSN 131. Traceroute is available for most operating systems including Windows and MAC OS. Given that your home is not too far from Purdue, what do you find in the traceroute results?