Slashdot Mirror
Using WiMAX To Replace a Phone?
vigmeister writes "I've decided to explore the possibility of using a netbook/MID as a phone while eschewing the services of a cellphone provider. Now that Atlanta (where I live) has WiMAX from Clear, I ought to be connected to the Internet everywhere within the city (once I sign up). Theoretically, this should mean that I will be able to use my netbook as a cell phone. Of course, there are some very real issues to overcome and I am simply putting this experiment together to see if it is something that is realistically possible.
This could possibly extend to uncapped 3G connections (if they exist any more) as well. Are there any obvious problems you would foresee? Is there anything I have missed or any other questions I should attempt to answer in this 'experiment' of mine? A major issue is, of course, the fact that my pseudo-netbook has to be carried everywhere and always left on."
WiMAX was designed to handle VoIP traffic, and has specific traffic categories on the airlink for isochronous flows, like RTP and other VoIP payload streams. Unlike something like Ethernet, which is CSMA/CD (carrier sense multiple access, with collision detection), traffic is scheduled by the carrier network. For uplink data, your WiMAX card goes through a process of requesting bandwidth on what amounts to a hailing channel, and then gets a bandwidth allocation it can use. In theory, a small but constant amount of bandwidth can be allocated for VoIP at the airlink level, resulting in low jitter, low latency, and low frame loss.
There are a couple of problems with this.
The first problem is that not all WiMAX cards on the market today (in fact, quite possibly none of them) have sufficient sophistication in their device drivers and microcontrollers to send the RTP (or Skype, or whatever VoIP protocol you're using) packets on an isochronous service flow while the balance of the packets travel on a general-purpose service flow. As a result, the RTP (etc) packets have to compete with whatever else your machine is doing, either stuff you're initiating with the browser, or background things like checking email or updating the Vista weather widget, or checking for updates of one kind or another. It doesn't help matters that no operating system has a network stack that implements the service flow concept.
The second problem is that low-speed isochronous flows over the WiMAX OFDMA airlink depend upon sharing a fairly large timeslot with other users transmitting simultaneously on the uplink using a different set of carriers, at least if the system is going to be economically feasible for the carrier. Allocating an entire timeslot often enough to keep the delay below half a second or so would result in considerable wasted bandwidth, so the idea is to have users share a timeslot by have each one use only a fraction of the available carriers. Decoding the resulting burst at the base station then depends upon maintaining orthogonality between OFDM carriers, which means that exact frequency synchronization is required between multiple users. While each user's WiMAX card synchronizes its clock with the base station, doppler shift due to changes in speed or direction or a changing multipath environment can change the received frequency at the base station enough to compromise orthogonality and make the burst impossible to decode.
The result of all this is, from your perspective, is that your VoIP traffic could be jittery and have long delays and high packet loss, especially when the carrier's network is heavily loaded.