
802.11a is the new standard for 5GHz communications. The radio band is currently broken up in three pieces: "Low band" - restricted to internal antennas and 50mW, "Middle band" - Fewer power restrictions and use of external antennas, "High band" - 1 watt output power intended for outdoor P2P and P2MP links. 802.11a's receive sensitivies are very close to the best existing 802.11b cards. This means that while your 802.11b card limps along at 1mbps, your 802.11a card will scream at an amazing 6mbps. Maximum is 54mbps.
The 108mbps mode uses two 802.11a non-overlapping channels (non-802.11a compatible) to reach its 108mbps data rate. This drops the available 8 non-overlapping channels down to 3 non-overlapping channels. A fourth channel is not possible since it doesnt fit on the end.
After about a week, the cards came in the mail. They looked similar to the orinoco, but a bit clunkier. The units had a TX and RX LED on the top. The antenna is large to the point where you are limited in card size for your second PCMCIA device. Anything with a protruding end will not fit. But thats ok, so we cant route between two wireless networks in a laptop, big deal.
Software installation was a breeze in Windows 2000 Professional. After everything was up and going, the cards started up and began looking for a signal. The software is very nice, providing every down-and-dirty option and wonderful graphing and statistics screens. We were ready to go.
The first test took place in an office building. We set two laptops side-by-side, went into ad-hoc mode on the same channel, and connected up. We turned on X2 mode to see what would happen.
"#@%$ thats fast!" We FTP'ed a 100 meg file from one laptop to the next, and it zoomed across like nothing we've ever seen on a wireless card. Of course, we were several feet apart. The first transfer gave us 3.7 megabytes a second. We turned on auto-mode (first time it was forced into 108mbps mode) and the second transfer was 4.1 megabytes a second. It appears that this amazing 108mbps mode does not hold up too well, and we also noted a variation from 108mbps to 72mbps. Putting a hand between two 108mbps forced cards will yield no signal. Glass also killed our wonderful 108mbps rate. We decided to leave on "auto" mode, which scales back the bandwidth as the signal degrades. The hard-drive was also going crazy, so we dont know if this was a limiting factor. We did not have a tool available to make a 100 megabyte ramdisk, so we decided that it would have to do for testing. Later we discovered that it takes windows 2-4 times longer to copy over from the temp file to the destination file name -- crazy!
We went around in the cube farms and around the office (which was cluddered by walls), and received some nice results. The radios kept between 24-48mbps the entire time.
After that, we turned off X2, and went into 802.11a mode. We later found out that even though X2 is really fast, 802.11a is more robust. We figure this is due to the fact that the X2 signal is twice as wide, spreading the tiny amount of 50mW very thin. The 802.11a "carrier" has much more power density, since it is smaller.
I went a couple hundred feet away with 802.11a, in a very dense office environment. We are talking multiple walls, lots of cubes, and all sorts of things that would make multipath and attenuation serious factors. Even with all that cludder, i was still crusing along at 18mbps -- nearly twice as fast as the 802.11b mode with two laptops sitting next to each other.
A week later I was able to test out how well it faired outside with another friend. We took the cards out and turned on X2, and we were actually able to go much further with the 108mbps-ish mode, perhaps around 20-40 feet. We were able to go a couple tenths of a mile with 24mbps signal. One major factor in this outdoor testing was simply the fact that the antennas were not omni-directional at all. There are big nulls on each side, making a signal strong or non-existant, on each 90 degree turn. Not all that great of an antenna.
We put ourselves between a tree after going about 100 feet or so with the cars. The tree killed the transfer 100%, but once it was put in clear sight again, the transfer started up. We can assume right now that even with good gain antennas, tree attenuation is going to be a major network killer in the 5 gig band.
Overall, these cards are very very fast. Moving back to the orinocos was sickening. Our maxiumum rate was about 440 kilobytes/sec. This is the same exact speed I got on my 10mbps half-duplex LAN at home, so it appears that a lot of this apparent overhead is also from the "half-duplexness" that one experiences on a wireless network.
Proxim charged my credit card a couple days ago, so i assume that they have FCC certification now. I guess I will have to give them back :-(
UPDATE: After having so much fun streaming full screen video, fast remote terminal sessions, crazy transfer rates, and mp3 streams, I have decided I will keep my pair of cards. Nothing like having a dual DS3 in your pocket...well, half duplex, but damn close to that!
Visit http://www.seattlewireless.net/index.cgi/ProximComments for some pictures I took of the wireless card, as well as some interesting frequency charts and (useless) "range" graphics.