Subscriber Density

Subscriber density relates pretty directly to system capacity and more directly to how frequently
spectrum can be reused. Since frequency reuse is an absolutely key concept to
operating any metropolitan wireless network, I will devote some space to the topic in this
discussion of subscriber density.
Any given radio frequency can be occupied by only one user within a given propagation
path. Two or more users attempting to use the same frequency simultaneously will interfere
with one another. To prevent interference the network architect must either assign a single
channel to each user or assign recurring time slots to individual users within the same band
(combinations of the two approaches are possible as well). The limit of the ability of a given
slice of spectrum to carry traffic is reached when every frequency is occupied for every
wave cycle.
In practical terms, such a limit can never be reached, but it can be approached through
such modulation techniques as Code-Division Multiple Access (CDMA) and orthogonal
frequency division multiplexing (OFDM), where individual transmissions are distributed
across the entire available spectrum in complex interleavings that leave relatively little spectrum
unoccupied for any length of time during periods of heavy network traffic.
Once available spectrum is completely filled, the only way the operator can support more
traffic is to employ some means of reusing the spectrum within some restricted area. This is
accomplished by two methods: transmitting the signal in a narrow beam by means of a directional
antenna and transmitting at low power so that the signal fades to insignificance at
distances beyond the terminal for which it is intended.
Directional antennas themselves use two techniques for achieving their directional characteristics:
focusing the transmission in a parabolic dish reflector and using complex
constructive and destructive interference effects from several omnidirectional monopole
antennas to shape a beam. The second type is known as a phased array and is far more flexible.
Directional antennas ordinarily work only with fixed installations where subscriber terminals do not move in relation to the base station, and thus they impose a limit on mobility or even
much portability. They also require careful management because they must continually be
realigned as subscribers are added to or dropped from the network.
Directional antennas produce one unfortunate side effect; they extend the reach of the
transmitter considerably by concentrating energy along a narrow wave front, and they change
the attenuation characteristics of the signal. This means that spill going past the intended
subscriber terminal can interfere with distant terminals elsewhere in the network. This can be
a real problem in mature networks where the footprint is divided into a series of adjacent cells
and the intent is to reuse spectrum from cell to cell (frequencies can rarely be reused within
adjacent cells, and wireless networks ordinarily require intervening cells separating those
using identical frequencies).
Both parabolic reflector antennas and phased array antennas can be aggregated to
produce what are known as sectorized antennas—groups of directional antennas distributed
on the same vertical axis and dividing the cell defined by the base station into sectors of
roughly equal area. Such antennas will permit nearly fourfold increases in spectral efficiency
within a cell but will increase interference in adjacent cells by the square of the existing quantity.
In other words, they are no panacea, but they may provide the right solution for certain
distributions of subscribers.
In the last few years, adaptive phased antenna arrays have been developed where
computing engines continually evaluate network conditions and shape the directivity patterns
of signals emanating from the array so as to mitigate interference while permitting maximum
traffic densities. Alone among directional antennas, adaptive phased arrays can support full
mobility in the subscriber terminal.
Adaptive phased array antennas, also known as smart antennas, offer other benefits as
well, which I will cover in Chapter 4. They constitute what is truly a breakthrough technology
that can significantly extend the capabilities of the wireless network and significantly increase
both capacity and subscriber density. And yet they have been little employed to date because
of the substantial price premiums they have commanded. Prices are beginning to come down,
and more companies are entering the field, and within a two- or three-year period such devices
will most likely become commonplace. But as of this writing, choices are still limited.
The second technique for achieving high spectral efficiency (using a multitude of lowpowered
base stations defining microcells) will allow the network operator to achieve almost
any degree of subscriber density, but at a price. Base stations cost money, and leasing space on
which to situate base stations costs more money. The trick in succeeding by subdividing a
network into smaller and smaller microcells is determining beforehand how many additional
subscribers you are likely to attract. Subscriber growth rarely has a linear relationship with
infrastructure growth, and the marginal cost of gaining new customers is apt to increase. Wireless
operators seeking customers seldom face an initial situation where the number of
customers wanting to be admitted to the network exceeds the amount of network capacity to
support them. No broadband wireless operator to date has faced insatiable demand, so
network operators should proceed with the utmost caution in building excess network
capacity.
Absent adaptive antennas arrays and microcell architectures, the network operators
need to deploy fixed antennas carefully based on subscriber growth assumptions for the
network and calculate the hard limits of how many subscribers can be allocated how much spectrum. While engineering formulas and design software exist for plotting antenna deployment
and directivity for maximum utilization of spectrum within an overall cellular
architecture, performing such calculations in the face of uncertainties as to the precise distribution
of subscribers at various phases of network expansion is at best an estimate. Chapter 4
discusses such issues in further detail. Here I must emphasize that the first task facing the
network operator is to plot the probable distribution of subscribers, with breakdowns as to the
relative density of high-value business customers versus basic service subscribers in various
locales. Only after that exercise has been completed should infrastructure requirements then
be projected.