Speed of Light Measurement Using Ping

Thomas Colthurst writes "You've no doubt already read the story of ping, but have you ever used it to measure the speed of light?" Here's a case where all that cat5 on college campuses can actually be used for education ;)

Comments

How they do it

[Alien54]

As seen in the paper:

Here's the problem; the cable range of ethernet without a repeater is about 250 feet (that is, at most a few microseconds roundtrip in cat-5 cable) and actually all the tests described here are done in much shorter cat-5 cable (more practical for typical reuse) and coaxial cable lengths so that it can be done cheaply. A typical classroom can hold several experiments of this type, the cables being shared between pairs of computers (and thus lab groups). Since ping only returns roundtrip times as measured in microseconds the actual signal (which is the additional delay in a cable path of longer length) is below the (reported) resolution of ping.

The solution is to use noise. Although noise usually hampers one's ability to measure a signal, in this experiment, noise in the form of randomly distributed small delays (microseconds) associated with machine response actually makes the measurement of the signal (nanosecond-long cable transit delays) possible. Without the noise, the experiment we describe here would be impossible! This concept of noise-assisted sub-threshold signal detection (hereafter; stochastic resonance) is of great value because it plays a role in a great variety of systems. For a readable introduction and overview of stochastic resonance see Ref. [7] and Ref. [10] for a bibliography. For example, stochastic resonance has been used to analyze climate patterns [8] and plays a role in fundamental neuro-physiology [9] . Part of the hidden pedagogic agenda of this laboratory is to introduce the concept of stochastic resonance in a hands-on way. How well this laboratory can actually get students to ponder that depends on the approach of the instructor. Our experience with this laboratory indicates that time differences on the order of 50 nanoseconds (or about 5 % of the threshold) are reliably resolvable.

Which is damn clever of them indeed.

Re:Delays due to molecular friction?

[8bit]

Friction?! As I remember the electrions themselves move slow as molasses, but the information (current move, current not move,) travels at the speed of light. So how would any kind of friction change this?

Good as experiment.

[Bender Unit 22]

It sounds good as a experiment. You have to figure out the time the computer in the other end takes to reply and then return an answer. But don't you really need another clock than the one that comes in a standard pc. Some PC's seems to loose up to 30 secs every day. And then there is the limit to how long your cable can be. since you can't have any switches in between, can the cable be long enough so you can measure a delay with the poor accuracy of a pc? Hmm maybe counting clock cycles would be better for timetaking. oh well.

Cuckoo's Egg

[sconeu]

Cliff Stoll mentions using Kermit ack latency to measure distance in "The Cuckoo's Egg". Of course, he wasn't trying to measure c, but to figure out where his hacker was. Turns out he was pretty accurate, even though the data was ignored because it didn't fit the currently known theories...

Measuring the speed of light with a ruler

[harlows_monkeys]

One day in APh 23 (the introductory optics class) at Caltech, the professor announced it was time to measure the speed of light, and pulled a ruler out of his pocket. The class laughed.

He then turned on a laser of known wavelength, and reflected the beam off the ruler onto the chalkboard. The ruler had raised lines every 1/16th of an inch, and this made it basically act as a diffraction grating, and there was a clear diffraction pattern on the chalkboard. He marked off the pattern on the chalkboard with chalk, then took the ruler and measured the distance between the lines on the diffraction pattern. Then, still using the ruler, he measured the distance to where he had held the ruler.

A quick calculation later, and he had the speed of light.

I'm not sure that this was fully legitimate, because I can't think of a way to know the wavelength of the laser that doesn't involve already knowing the speed of light, but it was interesting nonetheless.

Speaking of interesting things to do with interference patterns, that professor did some work at Hughes on an optical weapon system. It had an array of radiators. Turn them all on, and you get a classic interference pattern, so you get a strong lobe in one direction, and not enough radiation in other directions to harm anything. The cool part was how it was aimed.

You aimed the main lobe by playing with the phase of the various radiators, so you didn't have to move things around to do fine aiming.

Here's the cool part. They used a feedback system. The modulated the phase of each radiator with a sine wave, using a different frequency for each radiator. They'd point a sensor at the target, and look for variations in the intensity of the reflection. If a particular radiator was at a phase that was contributing toward putting the max lobe on the target, there would be a weak variation in the reflection at the frequency of the sine wave they were modulating that radiator with (if the radiator is at the right phase, you are near a peak, and small variations from the modulation don't lose much). If a particular radiator's phase was way off, you'd get a strong single at the frequency of the modulation.

So, they could simply do a fourier analysis of the reflection, and see what radiators needed their phase adjusted to hit the target.

The professor had a film of a test, with a small number of radiators. They were all pointing at a black background, and you saw a kind of vague shifting light pattern. Then someone tossed a small metal model of the starship Enterprise in, and blam!, the phases were adjusted in a millisecond or so, and that thing lit up. It was very cool.

If you think that's hilarious...

[Mawbid]

I have a wireless internet connection at home. A guy came and installed a directional antenna on the roof. He had me ping their gateway and he oriented the antenna while I read ping times to him over a two-way radio.

Well, I wasn't happy with the latency, so later I adjusted the antenna myself. But I didn't have anyone to read ping times to me and I wasn't too thrilled about this method anyway, so I came up with something better.

I wrote a perl script that would ping a host, wait for a reply (or a one second timeout), play a tick sound, and repeat the process. It sounds like a Geiger counter. The more frequent and steady the ticks, the better the connection. Also, every five seconds the script calls Festival to speak the average ping time. So, I get a nice intuitive feel for the connection through the stream of ticks, and a concrete measurement too.

Speakers out the window, full blast. Me on the roof. Neighbours' quizzical faces in the windows :-)

Ping around the world

[SkewlD00d]

That means that to ping the other side of the world, through wire, takes a minimum of 318 +/- 22 ms (round-trip) best case.

round trip time=pi*diameter earth/propagation speed

diameter: 12,756.3 km = 12756300 m
pi: 3.141593
prop speed: 118000000 +/- 9000000 m/s

Re:Ping around the world

[SkewlD00d]

The absolute minimum delay would be through the earth and back, in a vacuum:

2 * diameter of earth * speed of light ~= 85 ms