Deploying Minority Architectures

As indicated earlier, point-to-multipoint will be used in most instances to provide last-mile
access to the subscribers to the broadband service. The question then arises, When does using
the other topologies becomes necessary?
Point-to-point connections are generally used in three instances: to serve a single site containing
a number of high-value customers such as a business high-rise, to provide backhaul
from a base station to a central office, and to serve a single high-value user requiring extremely
high bandwidth such as a video postproduction house or a scientific research organization. In
all three cases the full spectrum available to the network operator will generally be utilized
within the single connection, and, to mitigate interference and maximize security, a very highgain,
highly directional antenna will be employed.
The tendency today is to use higher frequencies for point-to-point connections because
generally abundant bandwidth is available in the higher bands and because they lend themselves
to narrow-beam transmissions. The U-NII band at 5.8 gigahertz (GHz) is especially well
suited to point-to-point links because it can be transmitted over long distances and because it
will support throughput rates in excess of 100 megabits per second (Mbps).
Point-to-consecutive-point architectures, also known as logical rings, are chiefly applicable
with millimeter microwave equipment. Since no manufacturer of 802.16 equipment
currently makes a complete system for supporting a wireless logical ring, I will devote little attention to topology. A network operator wishing to use it will either have to jury-rig an IPbased
system by utilizing Resilient Packet Ring add-drop multiplexers or use legacy SONET or
synchronous digital hierarchy (SDH)-based point-to-point microwave equipment.
Mesh wireless equipment, in its current state of development, is best suited to relatively
small deployments serving a few dozen subscribers. To preserve bandwidth for individual
users, routes must be kept short, that is, restricted to no more than two or three hops. Hundreds
of subscribers distributed over several square miles would obviously lead to longer hop
sequences and slower throughput speeds, and where subscriber bases of such size must be
served, the network operator would have little choice but to create a number of discrete
meshes, each with its own aggregation point, or else use another topology, most probably
point-to-multipoint.