Skip to content

And we’re back

Hello! Just a quick post this time to say that I’m now back from my week away, so there should be some more posts coming up this week.

Both the R6300s and the Wiced boards have now arrived from Broadcom, so there should be lots of new stuff to talk about over the coming few days. As I type the R6300s are in the anechoic chamber, so we can take a measurement of their antenna patterns and then pass that information back to Broadcom. Once that’s done we’ll be able to start using them for link testing.

In addition to the R6300s I also mentioned about these: http://www.broadcom.com/products/wiced/wifi/ which are the Wiced boards. Apparently they’re primarily aimed at washing machines, but the hope is that they may provide a useful alternative to using the microchip boards we’re currently looking at. I’m really looking forward to having a play with them.

Anyway, there is a humming sound coming from a motor under my desk which can mean only thing – another measurement has just finished. Thus it is time to go tend to the chamber! See you later…

Anechoic chamber time!

Ok, a really exciting post this time! Over the last few days I’ve been working in the anechoic chamber – soo… time for some cool pictures, and then I’ll explain what it is that I’ve been up to so far in there, and what it all means…

This is inside the chamber. The cones you can see stop the radiation from the antenna being reflected and scattered.

This is inside the chamber. The cones you can see stop the radiation from the antenna being reflected and scattered.

This is the antenna that is used to measure the antennas under test. It measures the amount of radiation produced by the antenna being tested while it is moved through various orientations.

This is the antenna that is used to measure the antennas under test. It measures the amount of radiation produced by the antenna being tested while it is moved through various orientations.

This is where antennas being tested are placed. This motorised platform allows us to rotate the antenna around so we can observe which direction radiation is being emitted in.

This is where antennas being tested are placed. This motorised platform allows us to rotate the antenna around so we can observe which direction radiation is being emitted in.

So, to explain. The anechoic chamber is used to measure the performance of antennas. The reason that this has to be done in a special chamber is that in a normal room the radiation produced by an antenna would simply bounce around the room, reflecting and scattering and generally making it very hard to determine what radiation is being emitted directly from the antenna, and what is simply reflections coming in from other angles.

In order to obtain the measurement of the antenna the antenna is mounted on a motorised platform and spun around. In this way we can measure the pattern. The test is carried out by a piece of equipment called a network vector analyser (which looks something like a large oscilloscope). This provides the test signal for the antenna under test, and then takes in the signal from the reference antenna and then sends the resulting data back to the computer. This process is repeated over and over for different orientations, and the end result looks something like this:

This plot is the radiation patter produced by a simple patch antenna when placed in the chamber.

This plot is the radiation patter produced by a simple patch antenna when placed in the chamber.

This diagram simply shows the radiation pattern. The further away from the origin of the graph a particular point is the more the antenna radiates in that direction. From this graph we can see that this antenna radiates primarily in the z direction, but very little round the back (-z) of the antenna. Other antennas radiate more uniformly. We call this property the directivity of the antenna.

And that’s it really! Using MATLAB I am able to determine other metrics based on the data collected (for example antenna efficiency and polarisation), and using this information I will be able to look for antennas which may be better suited for use either with the gloves, or on the access points we’re going to use.

The requirements for the two antennas are really quite different. The glove antennas must be as omni-directional as possible, and small and easy to incorporate into the gloves. The router antennas, however, must be reasonably directional to try and block out interference from the audience, but can be moderately bulky if required.

Anyway, that’s it for now, back to looking for the right antenna for the job!

Fixed it…

After some more tinkering with the MATLAB model yesterday I finally got results that look like a model of network traffic!

n vs g mk 2

 

I’m really pleased with this graph, as if you look in text books this is the line you see used all the time for packet loss in a network with increasing traffic load, which is in effect what I’m simulating here.

You can see from the graph that the 802.11g client remains able to send all its traffic no problem, while the 802.11n client sheds more and more packets. The is because it is simply never able to get hold of the channel. I’m going to put together a video in the next couple of days to explain why this is.

The next step with this is going to be putting realistic wifi parameters into the model (at the moment the amount of attempts the client makes to send a packet before giving up isn’t very realistic). Hopefully I’ll be able to expand this model to investigate other wifi parameters too.

Anyway, that’s all for now. I’m working on getting some anechoic chamber measurements today, but first I need to convince my laptop to talk to it!

Project update – Books and Matlab

Well it’s been a few days since my last blog post, so I thought it was high time for another to explain what I’ve been up to since my last.

Sadly thus far we haven’t got the R6300s, which has limited what I’ve been able to get on with as quite a lot of the work hinges on those. Hopefully we’ll be getting them soon, but in the mean time I’ve been using the opportunity to do some swatting up on the 802.11 standard. I’m hoping to get a video out soon explaining some of the key concepts of how it works in a user friendly way (watch this space!).

Also, based on the reading I have been doing, I’ve been working on a MATLAB model to try and predict what impact the presence of traffic using the (older) 802.11g standard has on the throughput of 802.11n traffic. It seems to be generally agreed that there is a performance impact on running routers in mixed mode (i.e where 802.11g and 802.11n are both accepted by the same access point), but there are a couple of ideas floating around as to why this might be. For those reading this blog that already know about the details of wifi one suggestion is that the DIF length for the two standards is different, which then of course means that clients using the different standards do not in effect compete fairly. The alternative hypothesis is that it has something to do with the use of different frame headers – but I’m still trying to find a resource that is able to tell me what the differences are and what impact it has!

The MATLAB model is proving interesting… it is showing the expected drop in performance in 802.11n with different DIF times, but it’s a more linear graph than I was expecting when plotted against varying traffic load – I’m still tinkering with the model to see if I can get something that looks a little more believable…

Not really looking quite right at the moment - you'd normally expect a rather more exponential pattern for varying traffic load really.

Not really looking quite right at the moment – you’d normally expect a rather more exponential pattern for varying traffic load really.

If you don’t understand the guts of wifi – don’t panic! The promised explain-802.11 video will make all clear. It’s not as complicated as it sounds – promise!

Just briefly I also want to mention that I have been in touch with Priyanshi, a very clever second year electronics and communications student who is also doing a wifi based project with the department over the summer. Hopefully we’ll be able to share a lot of what we find out and help each other out! She’s also going to be writing a blog about her project, which I’ll put up as soon as I have it.

Anyway, that’s it for now. Keep your eyes peeled – hopefully there will be lots about anechoic chamber testing tomorrow!

 

X-Osc and iperf

A couple of interesting things to talk about today. First of all I paid a visit to Seb in his (incredible) office – I say office it’s very much a glorified work bench, there’s a lot of cool gear there! We had a long chat about the X-osc, and he’s given me one to use for the project as well.

One of the key things on this project is going to be getting the best performance possible out of the X-osc, as this is what will be transmitting the wifi from the glove. The x-osc uses a Microchip MRF24WG0MA to handle all its wifi operations. If you look carefully at this image of the X-Osc you’ll see that the MRF24 has a printed antenna on the PCB (it’s the squiggly bit of copper at the end of the board).

A picture of the X-osc board

The X-osc board

If the data sheet is to be believed this has some reasonably nasty pattern nulls, pointing along the board. This means that the chip will have much poorer wifi reception to and from any device that is that part of the antenna pattern. It will be interesting to see what we see when we put the board into the universities anechoic chamber and measure the radiation pattern ourselves.

Simulated radation pattern from the MRF24WG0MA data sheet - hopefully we'll get to see if it mathches the real anechoic chamber measurements  soon!

Simulated radation pattern from the MRF24WG0MA data sheet – hopefully we’ll get to see if it mathches the real anechoic chamber measurements soon!

One potential way of improving the reliability of the wifi link will be to replace this on board antenna with some form of external antenna. To this end Seb is producing a version of the X-Osc with an external antenna port so we can plug different antennas in. For this project a wearable antenna may prove to be very useful.

In other news – I also went to see Denys, who is the department’s expert at getting the Netgear R6300 to do traffic measurement. As well as giving me some of his own software to work with which I can use to effectively cause the routers to produce lots of (empty) packets in order to allow me to simulate running the system in a very busy wifi environment he also pointed me in the direction of a piece of software called iperf, which you can use to test the maximum throughput achievable of a particular network link. It’s free to download and well worth a play with, although as far as I can see where you download it from seems to rather depend on your OS…

Anyway, that’s enough for today I think – hopefully over the next couple of days I’ll get round to explaining Professor Wifi’s wireless N/G hypothesis which I’ll be testing, and also keep your eyes out for some anechoic chamber action – probably on Tuesday.

Meeting with Andy and Seb

Just spotted Seb’s blog post about the meeting I had with him and Professor Andy Nix yesterday. Seb is the director of x-io and is working with the glove team to help with sensors and networking, so I’ll be working very closely with him over the next few months. Andy or “professor wifi” as Seb has dubbed him will be supervising my project. He is an expert in everything wireless, and has already given me lots of tips on how to get started!

Simon Rankine and Andy Nix

A picture from Seb’s blog of our meeting. You can’t see Seb though; he’s taking the picture…

The Project Begins…

Hello, and welcome to the blog of my summer project. This is where I’ll be putting lots of stuff about the project I’m doing for the University of Bristol.

As you can probably guess from the title of my blog the project is all about wifi. Specifically I’ll be working with the team developing these: http://theglovesproject.com/xosc-gloves/ in order to find a way to make the wifi link to the gloves as reliable as possible.

As part of the project I’ll be looking at using different antennas and putting them in the universities anechoic chamber to see how they perform, evaluating different boards which can be used on the gloves and exploring different firmware settings on the Netgear R6300.

Lots to do, so keep watching this space for more. I’ll try and get some exciting pictures up once I get some hardware to play with.