WiMax Operator's Manual

112 CHAPTER 5 ■ STRATEGIES FOR SUCCESSFUL DEPLOYMENT OF PHYSICAL INFRASTRUCTURES
neatly together, though occasionally a checkerboard pattern is substituted. Both patterns are
abstractions, and misleading ones at that.
A cell radius is always an arbitrary value. A radio signal does not abruptly cease to propagate
at so many yards from the transmitter; indeed it continues to the very edge of the cosmos,
though growing steadily weaker over distance. What this means is that not only do adjacent
cells overlap, but that every cell in the network overlaps with every other cell. Picture the propagation
of signals as ripples or wavelets spreading over the surface of a pond. Each pattern
of ripples travels everywhere, and each reflection begets new ripples. This being the case,
cells should be considered as concentrations of RF energy rather than as well-defined geographical
areas.
Nevertheless, subscriber units within a point-to-multipoint network architecture must
treat the cells as if they were well-defined entities; that is, subscriber units located at the arbitrary
boundary separating cells must communicate with only one base station even though
they are receiving signals from several, albeit at reduced levels. In other words, they must lock
onto the base station with which they are registered and reject interference from all others, and
for that to happen transmit power levels must be strictly controlled throughout the network.
This, as it happens, has important implications for cell splitting. Because the average radii
of all the cells in the network decrease with cell splitting, so perforce does the transmit distance
from the subscriber terminal to the base station, and vice versa. And because of the shorter
distances involved, transmitting power must be reduced at both the base station and the subscriber
terminals to avoid interference throughout the network.
Because in a fixed broadband network (excepting the mesh variety, which really does not
have cells as such) subscriber units are normally assigned to a specific cell (an assignment that
is ultimately based on the strength of the signal in either direction), redetermining the optimal
transmit power levels becomes extremely important during cell splitting.
Cell splitting may also necessitate the reassignment of channels within each cell in the
network since the new cells will be establishing new channel relationships with surrounding
cells. Altogether, it is not a process to be undertaken lightly and without a thorough reexamination
of the entire network.
As indicated earlier, it is advisable to plot out the location and capacity of every base station
that the network will ever need at the time the network is being launched, though that may
not always be possible, and the time may come when the network operator is forced to consider
unanticipated microcells to meet demand. At that point, the network operator is faced
with the task of essentially reengineering and rearchitecting the entire network. Obviously,
there are limits to what can be done here. For practical reasons one is not going to relocate
existing base stations. But power levels and channel assignments will all have to be redone, and
the same software tools used in the initial planning process will have to be used all over again.
Line of Sight and Non–Line of Sight
I have previously alluded to NLOS many times in this book. In this section I present a fuller
explanation of how it is achieved and what it means in terms of network mapping.
First a bit of background: Radios operating in the region above 2GHz, the region that
has generally been assigned to broadband wireless services, are, as you have seen, easily
obstructed. This is a simple matter of the physics of wave propagation at such frequencies.
Since the degree to which such high-frequency transmissions are obstructed has a major bearing
on the broadband wireless operator’s ability to sign up customers, the equipment segment