Slashdot Mirror
Vanu Replacing Cell Tower Equipment With PCs
Dwight Schwartz writes "As reported in an
article on the ScienceDaily site, researchers from
Vanu, Inc. of Cambridge, MA,
have successfully tested a system, the Vanu Software Radio(tm), that can replace a cellular tower's room full of communications hardware with a
Pentium-based computer running Linux. The system
offers the hope of making cellular technology more affordable for small, rural communities." The systems have been tested for the last several months in parts of Texas, with wider adoption planned for the near future.
Have a look at the original release from the US National Science Foundation. With some nice pictures. :-)
GSM is pretty low-bandwidth stuff (around 13kbps). Further, the line cards handle a lot of the framing and general cookery for the interconnects (whether it's a wired E1, or microwave, or whatever). So even a fairly low-end Pentium would handle a few calls. The article does say that it needs a fairly large Linux server, but an ordinary PC would work for a relatively small node. The Digitalk telephone switches we use are really just dual PII-500 machines, and they handle 120 simultaneous wired calls.
Because according to the article (as well as the writeup for this topic), they are targeted for rural communities, wehere by definition the towers will be so far apart that service will be entirely lost for that area because there will not be another tower close enough to pick up the slack....
--guru
I'm originally from Stephenville, Texas - just down the road from these two communities. There are no third-generation networks available out there. My Sprint PCS phone doesn't work there, either - not even in analog mode. But just 20 miles away I have full signal strength in Vision mode.
I still have some friends in the region using Mid-Tex, and they haven't mentioned anything odd about their service. I guess that means it's a success as far as consumers are concerned.
I'm a student of telecommunications at an university of technology in Finland, and we've had compact basestations gathering dust at a student lab facility for _years_. (one Nokia and one Siemens, if I remember right)
The unit was about two mid-tower cases of volume, had an integrated PC, integrated antennaes, the whole bunch. Everything you need for a GSM basestation. And it really is an old model. Modern models are at least more efficient (with directional tracking antennaes, etc) and more inconspicious (they can look like fake chimneys, parts of wall, etc, so that it doesn't disturb the landscape.)
Probably we're not even talking about the same things since calling a basestation unit a "tower" is ridiculous. Maybe they've replaced the switching centre with a PC? Though I doubt it, since a PC/Pentium would be severely bandlimited to handle thousands of connections. Perhaps with dedicated hardware which is merely controlled by a PC..
Well a simple i386 Linux PC that operates with a HD doesn't exactly meet the five 9's of reliability that you need to operate telcom services, but that is what Carrier Grade Linux is for. OSDL is working on, and has released specifications for CGL, a spec that quote "provides standards based, open architecture software platform for converging telecommunications/data communications systems, which require virtually zero downtime." Linux is currently making some major inroads into the telcom sector, replacing all sorts of COTS systems that are currently in use. Linux will be running on that room full of communications software before too long. So I guess those folks with the telcom PC are just getting a head start by running the system on Linux.
Anti-social? My code is just platform-specific.
Vanu is the son of Amar Bose, founder of Bose, the maker of all of those great speakers. Another MIT wizkid.
IAACT (I am a cellular tech) and I have seen at least two small scale systems - both of which failed. The first was made by an Australian company - Unilab (loony lab to its friends :)) The system used two off the shelf '386 cards and custom switching hardware to control a standalone analogue system. The problem wasn't with the processing it was in the RF paths - the transceivers were large and had to dissipate a lot of heat.
:)
The second system was a small GSM system which did fit into a rack about the size of a tower case. Sounds a lot like this one doesn't it
The problem - channel capacity and maximum power output. Even the best transmitter is not super efficient and generates a fair bit of heat. The article doesn't mention the transmitters, the maximum output power or its capacity.
The article states that the system has been installed in a Texas town. I hope it's a real small Texas town! To provide decent in building coverage and coverage on the road in and out you need a high power transmitter. You have then got to dissipate the heat from the transmitter. Thats where the airconditoners come in for GSM. At least for the equipment I have seen.
Nortel CDMA equipment can operate quite happily in a hot room without airconditoning. The way they get around it is to use one kick ass heat sink and fans as loud as jet engine that any overclocker would drool over. And once again , in my experience it's the RF circuitry not the digital processing part that fails with the heat!
Anyway bring it on - I welcome the overtime repairing another BOB (bucket of bolts)
The poster should have included this link (pdf) - much more interesting.
Reliable, Great Value Hosting: $7.95/mo 2.4G/120G
Incremental repeaters are already available. They must be low enough power to fall beneath the regulated power levels of the FCC, generally not more than 4-5 watts, I think.
aceteq.com has some good repeaters, even the 1900 mhz repeaters for US GSM for less than $800. Other systems can be $3000 or more, and can cover larger areas (I've seen some that were capable of 25,000 sq ft.)
So you would have to do a running line of small repeaters to get service to an area, similar to the 802.11 repeaters we've seen here. You might be able to convert it to line-of-sight and do it that way, too.
But the short of it is, it can be done, albeit somewhat expensively.
I'm CTO at Vanu, Inc. Here's some additional info that some posters to this thread have asked about.
- Linux version: we're using a Debian 2.4 release with the real time patches. All the signal processing code runs as an standard application level process.
- A/D and D/A: we're using an external RF front end that provides over 90 dB spurious free dynamic range. The poster who said these are big and hot was right; it's a little smaller than a PC case all by itself, with a hefty fan to dissipate the heat of the power amp. It covers 25 MHz worth of spectrum and costs a lot more than the HP server that does the signal processing.
- software features: the linux applications running on the HP server handle the complete transmit and receive chains. We go from raw digital samples on one side (exchanged with the A/D and D/A converters) to voice and data packets on the other. A separate HP server runs the Base Station Controller functions, which are the protocol logic, handover control, and similar functions.
- reliability: a huge advantage of building the GSM software on top of linux is that it's portable. Some operators want the level of reliability that comes with commercial grade servers; some want the level that comes with telco grade servers. The GSM basestation software runs on whatever they need.
It's great to see how much interest there is in the slashdot community about this.
-john chapin
The reason the US has a bad infrastructure is because of the sparse population. We have states that are bigger than some countries. And a cell network is expensive. Most companies don't want to put towers up in the middle of nowhere when they will hardly be used. So hopefully this technology will make it more feasable to cover the areas that aren't going to get much use.
Bandwidth and routing functions are not problems -- the radio function is the problem. Traditionally, decoding the modulated signal (finding the carrier and recovering likely bits) was done in dedicated hardware. Sometimes, Viterbi or other maximum-likelihood decoding is also done in hardware. WIth current PCs -- especially with SIMD instructions like the SSE family -- the demodulation function can be implemented mostly in software.
The practical effect is that instead of having a (hardware) platform that works only with certain encoding and modulation standards, you have a (software) platform that works with many. It is easier to produce software than hardware at the scales needed, and it allows for an easier upgrade path.
It's neat that Vanu is doing this on Linux, but it's not like it's a revolutionary technology breakthrough.
There's still an analog RF radio involved; all the digital processing is at the IF frequency. Digital signal processing of raw RF in the gigahertz range is still a bit out of reach. (And it will require an A/D with huge dynamic range.)
It's not clear that it's a win to do this using commodity PC hardware. Most of the crunching is in tight signal-processing loops that don't use much memory. With custom boards, you can have more CPUs on a board. Squeezing the physical size down matters in this application. If you can put the gear in a box on the pole, instead of needing a little shed, that's a big win. PCs also tend to use more power, and thus generate more heat, than DSPs per MIPS. Cooling all the gear is a constant headache in the cellular business. It typically doubles the power consumption, and the air conditioners themselves are maintenance headaches. What the industry wants is gear that doesn't require air conditioning, at least for smaller sites. Qualicomm has been shipping pole-mounted CDMA base stations since 1997.
It's also not clear that introducing a network between the radios and the processors helps reliability. If the radios are flexible enough that one can take over the job of another, it's easier to fail out a radio/processor pair and switch in another one.
None of this matters all that much because the cellular base station equipment industry is in the tank. The industry overexpanded based on forecasts of huge needs for 3G gear, and that didn't happen.