As data demands increase, the industry has a few options it can take to protect future revenue growth.
There has been a lot of discussion lately about how the AT&T network is showing signs of overloading, particularly in a few major cities. In the general press, the blame is usually directed at the immensely popular Apple iPhone. In terms of sheer numbers, that is no doubt correct, but the underlying cause is increased demand for data services from all network users, not just those with iPhones.
In fact, I suspect that on a per-user basis, “air card” modems may account for more loading than smartphones like the iPhone. But regardless of the sorts of user devices involved, an explosive growth in the use of wireless data will almost certainly bring with it mammoth headaches for network operators.
DÉJÀ VU, ANYONE?
Wireless industry veterans will surely get a sense of déjà vu when we start talking about rapid expansion in network traffic. Back in the days when cellular voice service was enjoying 30 percent per year growth, we wondered if capacity could continue to meet demand. The FCC helped by providing additional spectrum for wireless networks, mainly with the allocation of the 1900 MHz PCS band.
But the factors that eventually ameliorated the situation for voice service were market saturation and a limit on per-user demand. Essentially, when just about everybody has a cell phone and most use it in lieu of a landline for person-to-person communications, any increase in voice service demand has to come from people spending more time on the phone. That’s because the network data throughput for each call is limited to around 4-12 kbps, depending on the air interface technology used. In fact, I suspect that with increased use of text messaging and e-mail for personal communications, the network loading from conventional voice calls may currently be growing very little or perhaps even shrinking a bit.
The current situation and future outlook for growth in data services are radically different. We are currently just beginning to see rapid growth in wireless data. Between smartphones, data cards and new portable “slate” computers, the next few years may well resemble the big wireless voice boom era.
But much more importantly, unlike voice, there is no inherent limit in per-user demand for data. For example, streaming highdefinition video may require a throughput of several megabits per second – the loading equivalent of perhaps a thousand voice calls.
It’s easy to dismiss this example as being extreme, since streaming HDTV service is rare. But that’s today. As more and more consumers gain access to high-end broadband Internet access, the trend is toward services that exploit this capability. Even now, high data rate applications such as sharing and downloading of video files account for more than half of Internet traffic, and that figure may well top 80 percent in a few years.
THREE WAYS TO RESPOND
So, what should the industry do? We certainly don’t want to discourage adoption of data services, as they represent much of the future potential for revenue growth. That leaves us with three broad areas where we can attack the growing problem of network congestion: manage per-user demand, increase the efficiency of the spectrum we have and increase the amount of spectrum available.
Management of per-user demand is a tricky business. Many current wireless data plans offer unlimited service, but that may soon change. The problem is that any significant restrictions are sure to become a big differentiator between competitive services.
Right now the industry seems to be pinning most of its hopes on improving spectrum efficiency. For wireless network data services, we can characterize this efficiency in terms of bits per second per Hertz per square kilometer (or per square mile for the metrically challenged). For example, if a network is able to reuse its available spectrum on average every four square km and uses RF channels that on average deliver 2 bps/Hz, then the network’s spectrum efficiency is 0.5 bps/Hz per km2.
Network spectrum efficiency can be improved by increasing spectrum reuse intensity (for example, by cell splitting) and by improving modulation efficiency (i.e., providing more bps/Hz on the RF channels). These two factors are highly interrelated. As spectrum reuse becomes tighter, interference levels tend to rise. Unfortunately, as expressed by Shannon’s Law, higher interference levels reduce modulation efficiency.
This leads us to a discussion of the planned migration to 4G technologies. WiMAX and LTE offer hope for improved average network spectrum efficiency primarily with standardized tools such as MIMO systems that actively reduce effective interference levels and by allowing much higher modulation efficiencies when channel conditions allow. Just how much improvement we will see with 4G remains a very big question. Offhand, I’d say that optimistically, 4G may deliver around twice the network spectrum efficiency levels of 3G, and that figure could perhaps double again over time as MIMO and other 4G tools are perfected.
A 4X improvement in network spectrum efficiency sounds great, but if usage increases annually by 40 percent (which is modest relative to some projections), that increase will be consumed within about four years. Then where will we be? Probably scrambling to find more spectrum that can be used for wireless data networks.
But how much more spectrum can we realistically expect to get? Right now, the various bands allocated for use by licensed wireless networks in the United States amount to a total of a bit over 400 MHz. The practical frequency range for wireless networks providing ubiquitous (as opposed to purely line-of sight) coverage is much larger, maybe 500 MHz to 3.0 GHz, but virtually all of that is spoken for. Over the next decade or so, we might somehow be able to double the amount of spectrum that can be used for wireless data. More likely, the increase will be 40 percent to 50 percent, about the equivalent of a single year’s projected growth in service demand.
What all of this suggests is that the long-term viability of wireless data networks will probably require management of per-subscriber usage. This had best be done now, before users are conditioned to expect unlimited throughput at high speed. Of course, I could be wrong; we may somehow find a way to increase network capacities by a couple of orders of magnitude, at which point users will be pleased to see better service with fewer restrictions. That would be much better than the opposite.
Drucker is president of Drucker Associates. He may be contacted at email@example.com.