Intel reconfigures antenna work

INTEL HAD A pretty in depth pre-IDF session about radios and the problems that they have working together. You may not realisee that there are problems with this because they usually are solved before you see them, but these are not trivial problems, far from it..

If you think about how many radios you have on you, with a laptop and cell phone, the number is huge. An average laptop has 2 802.11 radios (2.4 and 5GHz), bluetooth and soon Wimax. Your quad band cell phone has four for GSM, another for bluetooth, and one more for the bluetooth headset in your ear. That is somewhere between six and nine before you get into the odd devices many of us carry.

These radios tend to step on each other's toes quite a bit, from interference to antenna design to other esoterica. Designing devices that work together is tough. Designing them so it fits in a notebook instead of towering above it is tougher. Making an antenna that works in that environment is a real bitch.

Let ux start with the antenna. Each service you use works on a different frequency, if it didn't, you would only be able to use one at a time. An antenna can be made to tune in a narrow range of frequencies or a wide band. If you need many frequencies, a wide band is the way to go, right?

Yes and no. Yes, it will do the job, and do it well, but in addition to picking up the signal you want, it will also pick up all the noise you don't want. Then you have to filter it out, a task that can be more of a problem than poor reception.

Multiple antennas are also a problem, you need to fit them into a notebook. 5 different antennas won't fit, much less fit in an optimal position. Ideally, you want to have a single antenna that can tune in specific frequencies but not others.

Intel is working on a reconfigurable antenna, one that you can tune on the fly to receive specific frequencies. The guts of it are far to technical for this article, but the idea is simple, you change it to work in a narrow range where you want it to, and then retune it for the next application. If you go from Wifi @ 2.4GHz to Wimax @ 2.5GHz, your antenna changes. While it is not fully there yet, it is a near term technology. Reconfigurable antennas are the future, and they are better than any of the other solutions out there.

That brings us to the radios themselves. They are broken down into three main components, the Front End Module (FEM), a transceiver and a baseband processor. All contribute to the problems and the solution.

The FEM is the amplifiers and filters, they basically sit between the antenna and the transceiver. They filter out the non-signal from the signal, and make the job of the antenna a little more flexible. Like the antennas, you traditionally have one for each radio frequency you want to receive. In a similar fashion to the antennas, you can make an FEM that is tunable, cutting down the number needed in a multiple radio system.

There is not much going on with the transceivers, so we will move on to the baseband processor. Those, are the guts of the radio itself. Again, reconfigurability comes to the surface. You can either have a bunch of radios on a chip, or a much smaller number of radios that are reconfigurable.

This requires smarter transceivers, and of course Intel is glad to make them for you. On top of that, a lot of the components are analog designs. These don't scale down very well, as processors get smaller, the devices really don't. This is bad if your business model is based on shrinking components.

You are faced with making things mostly analog, and doing it well, with a little digital backing, or doing a passable analog part with a lot of digital backing. Since analog won't scale well, and digital does, if you chose the latter path, functionality will scale with silicon technology.

Guess which path Intel is taking? Ok that one is easy, but as you can see, they are trying to make as much of the radio as configurable and controllable as possible. Why? Interference, the thing we kicked this all off with.

As you can see, with using two radios at once, you have massive potential for problems. With three or four, things get untenable in a hurry. The more configurable you can make things, the better off you are.

On top of that, simply making radios aware of each other rather than being discrete components allows you to coordinate them. Instead of these things fighting, they can hopefully be made to dodge each other. This used to be done at the spec level, now it can be done in real time on the chip.

What you hopefully end up with is a single radio to replace multiple radios, a single antenna to replace multiple antennas, and all components working together instead of against each other. It is a far from trivial task, and it is far from done. Work continues, and each version ends up better than the last.µ