2.4 Megabit Cellular Modem

lew writes: "Ars has a review of a cellular modem that provides 2.4 megabits / second downsteam and 153 kilobits / second upsteam... and it works! Check it out" How much for unmetered service on such a system? :)

Comments

Re:that's PER CELL

[tswinzig]

What that article doesn't mention, and what people usually don't know when discussing 3G mobile is that the data rates quoted are PER CELL not PER USER (unless only one user per cell is active at a given moment).

Try reading the entire article. Page 3, near the bottom, does a nice job of explaining this, and why it's not such a big deal:

Which brings us to the next point: that 2.4 Mbps is shared among all users on a cell sector, just like cable bandwidth is shared by everyone in a neighborhood. What's a sector, then? Cell sites are generally divided into three sectors that each cover different parts of the surrounding area, so each site can have up to 7.2 Mbps of bandwidth to play with. In contrast to cable, bandwidth in 1xEV is intelligently scheduled to maximize throughput for everyone. The modems actively monitor signal strength and request the highest data rate they can handle without dropping too many packets. If the packet error rate gets too high, the system switches to a more reliable transmission scheme and the data rate is throttled down. The cell site uses a sophisticated scheduling algorithm that tracks the modem's average receive signal strength from millisecond to millisecond and takes advantage of local peaks in the signal conditions to send packets when they are most likely to get through. That way, bandwidth is not wasted on packets that will likely have to be retransmitted anyway, and one user with a bad connection can't cause a storm of retransmits that slows down service for everyone. Of course, if everybody on your sector is doing large downloads at the same time, the bandwidth will be divvied up among them, factoring in signal conditions. Of some consolation is that fact that your typical usage scenario is rather more sporadic: you download a web page for maybe 10 seconds, then stare at it for a minute, and so on. When you aren't actually downloading, the airwaves are free for someone else to surf. The likelihood of everyone clicking at once is very low, and the average response as seen by any particular user is pretty good; that's the miracle of statistical multiplexing.

Re:Slow transmissions.

[joshuac]

Why do all the new broadband technologies limit the upload to a very slow speed? 2.4Mbps is nice and all, but for it to be useful beyond surfing the web 153Kbps doesn't leave for much of anything else.

Collisions. Same reason your upstream is often capped on a cable modem. On shared media you will get a lot of collisions from the individuals on the network as they choose to transmit at random times.

From the downstream perspective this is simple to control; you have one broadcast point, you simply queue things to be sent, and there are no collisions. On the upstream side, you need to know when someone else will be transmitting, and this is harder.

I imagine one way of doing this is to assign time slices to groups of people; you do not transmit unless it is your turn, and you compete with far fewer people (the others in your group). If you have 2.4Mbps available and you, say, divide this by 16 groups, you get a ~153Kbps window to transmit in (plus 9.6Kbps left over on the spectrum possibly for out of band housekeeping duties).

This is what is probably happening here.

Another options (and a long shot), but perhaps they are just plain mean (or not confident in their ability to control who uses their service) and want to discourage people from using the system to host anything. "Hey, our security is lousy, we know people will start stealing our wireless service to host copyrighted material/launch dos attacks from, maybe if we lock the bandwidth down at the tower this will not be attractive and the phreaks will go elsewhere".