I don’t have a good understanding of how WiFi works, so please correct me if my question is misstated. From what I’ve read, it appears that many WiFi devices use 2 or 3 frequencies, trading between frequencies when necessary.
I have looked through the Cortex microcontroller specs, but I haven’t been able to determine how many frequencies the Cortex uses (when you use the wireless keys). And is there a way to find the specs/number of frequencies on other WiFi devices, such as R/C cars or other robots? Would I need to look up the specs for the individual items?
The reason for this question is I was drafted at the last minute to supervise the Sumo Bot event for the Science Olympiad. One of the rules states that the robot must be able to operate on 3 frequencies. I understand the reasoning: they’re trying to prevent interference between multiple robots on the same field. However, according to the letter of the law, a robot with WiFi that uses only 2 frequencies doesn’t meet the requirement and should be disqualified. If it were up to me, I would exempt all WiFi bots, but I don’t get to choose. So how can I determine whether these bots are compliant? With the old crystals, it’s easy to check off teams that arrive with 3 different crystal numbers.
There’s some trickiness when it comes to talking about “frequencies” in the WiFi world. Really, when working with WiFi there’s frequencies and there’s channels. WiFi (b and g, like VEX uses) works on one channel at a time. In most parts of the world (USA included), there are 11 channels to pick from. The trouble is that some of those channels partially overlap each other. So, to avoid interference, most people use only channels 1, 6, and 11 which do not overlap each other. 2, 3, 4, and 5 all overlap channels 1 and 6 somewhat, and 7, 8, 9, and 10 overlap 6 and 11 somewhat. You can get a better understanding by looking at the Wikipedia page for 802.11g: http://en.wikipedia.org/wiki/IEEE_802.11g-2003
You can’t really compare WiFi devices like VEXnet robots to R/C cars. They’re very different technology. R/C cars (and the crystal-based VEX products) operate on one narrowband frequency at a time (something like 25kHz wide), and if anything else transmits on that frequency you will have problems with your connection. WiFi uses a much wider channel (20MHz wide, quite a bit larger) and spreads its information across that whole space. This helps it somewhat with interference (details of how and why are too complicated to post here).
The way WiFi is used in VEXnet, all robots actually share the same WiFi channel (* as far as I know). So technically all robots at an event are running on the same frequency. This works because of what I mentioned above, and plus the VEX robots are transmitting very small amounts of information compared to what the capacity of WiFi is (meaning that most of the time, they aren’t transmitting anything, so the chance of collision is low). This has been proven to work well - at the World Championship last year there were dozens of robots running at the same time on different fields in the arena.
VEXnet robots, like any WiFi device, are theoretically capable of running on any of the 11 WiFi channels. The problem, though, is that the operators of the robots have no control over which channel the robots use (as far as I have seen). It is my understanding that all VEX robots will operate on the same channel when turned on. Obviously the people at VEX know what they are doing and have determined that this works just fine. If there’s no flexibility in your rules for advances in technology like this, then I guess you’re stuck. The best thing you can probably do is appeal to whomever controls the rules and point out to them that VEXnet is designed to work this way and has been proven at many different events over the last couple of years.
The “must operate on 3 frequencies” was a poorly stated rule, which should have read “must operate on at least 3 sub-frequencies”. As I thought about it more, the crystal transmitters all operated on 75 Mhz with the crystals selecting different channels (i.e. sub-frequencies), so those technically wouldn’t have been compliant, either.
It is also important to remember that the connection on a crystal radio control link is consistant… you’re basically always using that channel.
In wi-fi, cell phones, and other more advanced, spread-spectrum technologies, you rapidly transmit a small packet of data. You send them so quickly that it appears to us slowpoke humans to be consistent, but you are sending discrete packages of information.
I haven’t worked with these details in a practical setting, but I’ve wondered how many WiFi users are you realistically looking at in an area at a tournament, including both robots and other mobile devices, and what is the probability of interference? I realize that very small, discrete packets of information use each channel for a very short amount of time, but put enough of these packets competing for space, and interference is bound to occur at some time. I noticed that at Worlds, WiFi will not be provided to phone/device users, but is there sufficient space for 400+ robots turned on at once? And what about other tournaments where WiFi is available to random devices?
Another question is, “What is the protocol when interference occurs?” If the user is simply left waiting for a fraction of a second (controls don’t respond for a few msec, then respond again), not a big problem. But if somehow failing to connect at that instant causes the user to lose the link, I can see that being a much bigger issue.
Does anyone have knowledge/experience with these issues?
I teach a graduate level networking class. The standard WiFi (802.11) wireless has a number of methods to deal with lots of devices active at one time. A combination of the different frequencies, the ability to ask for a clear channel for large data transfers, back-off and retry and reducing the data rates when the Signal/Noise ratio becomes too high means pretty reliable connections except under really adverse connections.
I’ve designed/installed wireless connectivity in a building to supply decent wireless access to 500+ concurrent users. If you go to any of the good conference facilities they are able to do the same for events. Schools / Universities can do it for classrooms full of students. So 400 robots should not be a problem.
The VEX engineers have said that VEXNet is not WiFi, you would need to get one of their engineers to talk about what they do vs. 802.11a/b/g/n networks. They have some things they could do when they pair the remotes to the Cortex that would allow them to reduce collision issues. (But, they would need to talk about that, I’m just guessing)
Web surfing has different data needs than the robots do. So Jason’s comments about small data packets is spot on.
Out of curiosity, where have you seen that? The VEXnet keys say “802.11g” on them, and it’s pretty easy to see the VEXnet signals with tools like Netstumbler.
However, they do not appear to use Access Point mode, is that what you were getting at?