Orthogonal frequency division multiplexing (OFDM) delivers high data throughput in the
fixed environment, but the mobile environment poses unique challenges.
In last month’s column, I explored the putatively “disruptive” world of OFDM technology, particularly as it is applied to WiMAX systems. To date, however, WiMAX as deployed in commercial networks is limited to fixed service – a relatively benign operational environment that should be favorable for delivering the high data throughput rates that make OFDM seem like such a winner. But what will happen to those sparkling throughput numbers when WiMAX goes mobile?
More importantly, what is the outlook for 4G cellular data networks, which will rely heavily on OFDM? Will they really be able to deliver the anticipated huge increase in performance over CDMA-based 3G technologies? Lots of questions, and so far very few answers, at least based upon real-world operational experience. But let’s take a look at what we know about radio channel environments for mobile communications environments to get a few clues.
Whenever I consider radio channels in a mobile communications network, I am reminded of a comment a colleague made many years ago. Trying to explain why there were so many difficult problems that had to be overcome to make a particular digital cellular technology work, he said, “You have to understand that this is a horrible environment for a radio channel. There’s interference, all sorts of multipath and an infinite number of different path loss issues. If we could use anything other than radio we would, but we’re stuck with it, and we have to make it work.”
At the time, my colleague was talking about circuit-switched voice service. The situation for packet data is made even more difficult by the fact that the extremely changeable channel quality encountered in mobile radio networks translates to corresponding variations in throughput rates that can be delivered. To see why this is such a big deal, let’s do a little comparison.
MOBILE CHALLENGES
For circuit-switched digital voice service, the quality of the radio channel is ideally such that the provided bit error rate (BER) is just at what is needed to allow the codec to provide its best quality voice reproduction. Any improvement in channel quality above that level is essentially “wasted” in that the perceived voice performance does not get any better. For voice service, air interface designs and network engineering practices take advantage of this situation in the form of increasing capacity and coverage at the expense of interference levels that are low enough to allow for “perfect” voice codec operation but are far higher than what is tolerable to provide maximum throughput rates for packet data service.
The result is what I call the “up to” problem, referring to the way providers of various wireless data technologies tout throughput rates of “up to” a certain number. Yes, that rate can be delivered, but only under ideal channel conditions. Put the user in a location where multiple base stations all provide similarly marginal signal levels, add a bit of interference from other nearby base stations, throw in some fading, and throughput will drop substantially. Furthermore, data speeds that can be delivered in poor channel quality don’t go up with increases in the “up to” rate because Shannon’s Law dictates that for a radio channel of given quality and bandwidth there is a finite limit on data throughput.
From a marketing perspective, this factor alone could be a big problem for 4G operators. A user of a network that is advertised as delivering “up to” 500 kbps may be somewhat irritated when throughput drops to, say 100 kbps, but dissatisfaction will certainly be far greater when expected speeds are more like “up to” 5 Mbps. So the real question for using OFDM in mobile data networks isn’t so much the speeds it can deliver under optimal channel conditions but rather how well it lends itself to assuring that such optimal conditions are more the norm than the exception.
Part of the solution for increasing overall channel quality is improved network engineering. The other part is the use of technologies that mitigate specific channel impairments. (A familiar example of the latter is the use of receive diversity to counteract channel fading.) OFDM-based mobile systems impact on both, but not always in a positive direction.
The good news is that such 4G features as “multiple input multiple output” (MIMO) transceivers hold promise for significant effective reduction in both interference and fading. In addition, a lot can probably be done with adaptive subcarrier assignment and possibly smart antenna systems for base stations. Many of these technologies could be adapted to 3G as well, but they more likely can be made to work optimally with a “clean sheet” design for channel structure and air interface protocols.
The downside of OFDM for mobile systems appears to be in the area of network engineering. In particular, there is the issue of practical transmit power. Even without added interference, for optimal speeds OFDM signals have to be received at a relatively high level. To get decent coverage, including good building penetration, transmitters will need to be able to serve up some serious juice. Unfortunately, high level OFDM requires very linear transmitters, which translates to poor amplifier efficiency and could certainly affect battery life in handheld user equipment. Other issues include the need for co-channel interference management (largely a non-factor in CDMA-based technologies) and challenges in realizing optimal levels of per-subcarrier power control.
These issues are certainly being addressed in the development of 4G standards and equipment, but you should not assume, based upon announcements of “successful trials,” that they have been fully resolved. That won’t happen until real-world mobile OFDM networks begin serving real-world customers. Considering the challenges of the mobile environment, that may be some years off.
Drucker is president of Drucker Associates.
He may be contacted at edrucker@drucker-associates.com