WiMax Operator's Manual
WiMax Operator's Manual
WiMax Operator's Manual
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
168 CHAPTER 7 ■ SERVICE DEPLOYMENTS OVER PUBLIC WIRELESS MANS<br />
of the broadband wireless access provider but of the owner of backbone fiber and/or Internet<br />
access points.<br />
Furthermore, several protocols exist for managing connections end to end across the<br />
backbone, including Resource Allocation Protocol (RAP) and the aforementioned RSVP.<br />
Although the signals may originate in the metro, the actual resource allocation will take place<br />
on the backbone.<br />
Making Sense of QoS Protocols<br />
QoS protocols, as previously mentioned, are often used in multiples. RSVP and DiffServ, for<br />
instance, are often used in conjunction with MPLS to support a wide variety of applications.<br />
Other protocols may be used more selectively; for example, RSTP would be used for streaming<br />
video but not for voice telephony. Given the proliferation of protocols and the complexity of<br />
their interactions, engineering an end-to-end IP network for QoS is not an easy task, and it is in<br />
fact nearly impossible where protocols are not carried consistently across the network. Unfortunately,<br />
configuring intervening switches and routers is generally out of the hands of the local<br />
access provider.<br />
With content demanding very stringent QoS such as high-definition video or Super Audio<br />
Compact Disc (SACD)–quality audio, delivery over a multimode network may simply be infeasible,<br />
and if the source is geographically remote, the local access provider may choose to utilize<br />
a satellite service to transmit the content in one hop to the local network where it can be<br />
cached on a server until needed. It can then be transmitted over a short distance where QoS is<br />
entirely under the access provider’s control.<br />
Other Methods for Supporting QoS<br />
I should not leave the subject of QoS without mentioning a long-standing argument as to the<br />
need of special protocols for supporting it over packet networks. Some in fact would contend<br />
that QoS can be achieved through other means and that special reservation and prioritization<br />
protocols are not the answer because of the difficulty of implementing them across heterogeneous<br />
networks.<br />
The simplest technique for getting around reservation requirements is to undersubscribe<br />
available bandwidth. If everyone on a network enjoys, say, a median throughput rate of<br />
50Mbps, QoS is likely to be satisfactory for almost any application absent any additional<br />
provisions for ensuring QoS. And indeed some in the industry have suggested that as network<br />
throughputs inch ever upward, the need for special QoS protocols will disappear over time.<br />
Given enough bandwidth, surely such representations are correct, but for now bandwidth<br />
is still expensive, particularly in metropolitan networks, and with some services such as lifeline<br />
telephony or backup storage, the user cannot tolerate momentary suspensions of service,<br />
however infrequent. For such services a reservation of bandwidth simply has to be in place<br />
against every contingency.<br />
Another way to attempt to get around the need for reservation is to create a sort of arti-<br />
ficial bandwidth by means of data compression technologies. Compression algorithms reduce<br />
the bandwidth required for any given application, often dramatically, and 50-to-1 reductions<br />
in the amounts of data required to transmit a signal are commonplace today. In practical terms<br />
this may approximate a 50-fold increase in bandwidth, though in reality matters are not so<br />
simple. Compression almost always introduces appreciable latency in a transmission, which