The perception that wireless and wireline deliver comparable levels of service is a testament to the good work of wireless engineers, but maintaining high levels of satisfaction is only going to get tougher in the years ahead – and more spectrum won’t solve laws of physics.
To the average user, there are two major differences between telecommunications services provided by wireless and wireline facilities. First is the obvious advantage of mobility and convenience provided by wireless. Second is the vastly greater number of features available on wireless networks and the user devices on which they operate. But as far as basic network function goes – delivering voice and data communications – wired and wireless appear to be pretty much the same.
Of course, as any network engineer knows, there are fundamental differences in the way wireline and wireless networks work. The popular perception of similarity is really a testament to the good work that our industry has done in meeting the far greater technical challenges posed by wireless, particularly in expanding capacity within a limited amount of available radio spectrum. But maintaining high levels of user satisfaction is almost certainly about to get a whole lot tougher as demand for data services explodes over the next few years.
LAWS OF PHYSICS
The idea of a pending “spectrum crunch” is just beginning to be discussed in non-technical publications, usually with the implication that if we can free up a few more megahertz here and there, everything will be fine. But if we are going to maintain a perception of similarity between wired and wireless service, particularly if we are talking about unrestricted broadband Internet access, even hundreds of additional megahertz won’t be enough. That’s because the fundamental limitations for wireless networks have less to do with spectrum policies than with laws of physics. In particular, there are two key physical characteristics that constrain how much capacity we can cram into a given amount of spectrum. I like to define them by two metrics of my own making: signal transport efficiency and interference-to-signal ratio. They seem particularly illuminating because for most wireless networks, they look awful. In fact, the numbers are pretty mind-boggling.
Let’s consider signal transport efficiency on the forward (downlink) channel. In a typical base station sector, the transmit power amplifier may output a particular subchannel set of a 4G channel at around 10 watts, or +40 dBm. For the sake of convenience, let’s assume that all of this power actually gets radiated by the transmit antenna. At a given instant in time, those 10 watts are being transmitted to a single user device we’ll call User A. User A should be getting pretty good service on the channel at a receive signal level of around -80 dBm. (If receive signal strength is much lower than that, downlink throughput speed would be reduced because of the effects of the thermal noise floor.) The net path loss between the transmit power amp and the receiver is therefore about 140 dB. That means that, of the signal power being transmitted, only an infinitesimally small amount finds its way into the antenna of the intended receiver. How small? Here’s the number expressed as a percentage: 0.000,000,000,001%!
The problem with this seemingly appalling signal transport efficiency (the fraction of transmitted power that is captured by the intended receiver) isn’t that it saddles carriers with monster utility bills or that it contributes much to global warming. Rather, the problem is what happens to the other 99.999,999,999,999% of the transmitted power. Fortunately, it mostly gets radiated into space or absorbed by the ground or structures. But a certain amount gets into the antennas of other user devices and represents interference.
In a large urban 4G network, within the service area of one base station sector, there might at any one time be on the order of 50 active users (that is, users actively engaged in some sort of data application requiring at least periodic use of the downlink channel). If the -80 dBm receive signal level arbitrarily marks the “sector edge,” then in the above example it is reasonable to expect that that while User A is receiving its -80 dBm signal, the other 49 active users in the sector are getting interference on the same subchannel set at or above that level, making that subchannel set unusable for them. In fact, the impact of interference almost invariably extends well beyond a single sector, so that at any given instant, a particular subchannel set can be used, on average, to transmit data to a very small fraction of the active users in a given area. Another way of looking at this is that far more of the power transmitted on downlink channels in a wireless network ends up as interference than as signals input to the intended receivers. This factor is what I call the interference-to-signal ratio, and it’s often well into the hundreds.
How would a wireline broadband network compare in the areas of signal transport efficiency and interference-to-signal ratio? Well, depending on the medium – twisted pair, coax cable or optical fiber, the path loss over the “last mile” of wireline network distribution might be in the range of 10 to 40 dB, but this is mostly due to simple signal attenuation. Very, very little of the signal sent to one user ends up being received as interference by others. If anything crystallizes the fundamental differences between wireline and wireless networks, I believe it is these vast differences in signal transport efficiency and interference-to-signal ratio.
So, what we have in wireless networks are these ugly factors that constrain capacity. But that isn’t necessarily all bad, because it suggests that even modest improvements in these factors could have a large and positive impact on network capacity. For example, let’s say that interference-to-signal ratio on downlink channels in a network is an average of 200 to one. That is, for every -80 dBm of signal power input to the intended user device receiver, there are 200 times as much, or -57 dBm, of interference imposed on other active users in the network. If we could improve that ratio to 100 to one, the spectrum capacity of the network – in terms of bits per second per megahertz per square mile – would effectively be doubled.
How could that be achieved? One way would be to improve signal transport efficiency, effectively by reducing the average path loss between base station and user device. That will correspondingly reduce required transmit power and, therefore, the amount of power imposed as interference.
There are any number of ways that signal transport efficiency and interferenceto-signal ratio factors can be improved. Some, like many-element phased array antennas, seem hopelessly complex, but over time, improvements in processing power likely will make them more realistic. The point is, innovation in these areas is probably more likely to enhance network capacity in the long run than simply making more spectrum available. After all, spectrum is a finite resource; innovation is not.
Drucker is president of Drucker Associates. He may be contacted at edrucker@drucker-associates.com