Looking at the Range of Services

No value-added broadband service exists that cannot be supported by a lower microwave
broadband wireless network, but because some services are heavily bandwidth intensive
and some are not, the operator must decide on the mix of services to be made available to
subscribers and the marketing emphasis that should be brought to each.
Basic Access
Currently, basic Internet access will constitute the core service offering just as is the case
with DSL and cable. The operator may want to serve as the Internet service provider (ISP) or
may want to establish a relationship with one or more third-party providers. For an independent,
competing with a giant such as America Online or EarthLink in terms of content and programming is well nigh impossible, and, in any case, customers may specifically want to
have a high-speed link to their existing ISP. If that is the situation, working with incumbents
makes good sense.
VPN and Transparent LAN
Virtual private networks (VPNs) and transparent LANs, also known as LAN extensions, are the
most popular business-related services today and pose interesting possibilities for the network
operator inasmuch as they do not need a lot of additional bandwidth beyond that required for
simple high-speed access. They do involve additional investment in infrastructure and some
fairly hard decisions on the part of the network operators based on their perceptions as to how
the evolution of business data services is likely to proceed.
A VPN is a secure communication from a remote location back to a corporate LAN that
often involves encryption. A transparent LAN service takes the concept a step further by
providing the remote user the means to run all LAN applications with similar ease at the
remote location as at the corporate headquarters and not simply to access corporate
databases.
VPNs and LAN extension can be accomplished by various means. The older method is to
utilize tunneling protocols and special hardware for performing the necessary encryption
quickly. The preferred method today is to utilize a carrier-grade IP router or Ethernet switch
with the built-in capability to set up Ethernet or IP VPNs. Such devices will exhibit a high
degree of automation and scalability and low network overhead.
Most edge routers sold today—that is, routers designed for use in a metro network—
support IP VPNs through the agency of the Multiprotocol Label Switching (MPLS) ancillary
protocol. Manufacturers of the latter include Cisco, Juniper, Laurel Networks, Riverstone
Networks, TiMetra, and Vivace. Most of these products also include other functions such as
Ethernet switching and ATM switching, though the precise mix of functionalities varies from
one platform to another.
Enhanced Ethernet switches form another possibility, specifically those designated as
“metro Ethernet” and designed for the carrier marketplace. Atrica and Extreme Networks are
the principal manufacturers in this category. But inasmuch as 802.16a networking products
are based on IP rather than Ethernet, and because IP/MPLS has a much greater capability for
setting various service levels to accommodate different types of customers, an edge router
represents a better choice for enabling VPNs.
Voice Telephony
Another fairly obvious service offering beyond simple high-speed access is voice telephony.
Either circuit voice or packet voice may be supported on 802.16a platforms, but circuit voice is
quite bandwidth intensive, requiring the reservation of a 56 kilobit per second (Kbps) channel
for each voice transmission. In contrast, IP telephony vocoders (digital speech compression
devices) can operate at rates as low as 2Kbps with good fidelity. Fifty-six kilobits may not sound
like much, but it takes only 24 voice circuits to eat up 1.5Mbps of throughput. If the total
throughput of the available band is only 100Mbps, then a couple of thousand voice circuits
could take up all the resources of the network.
Another factor to consider when choosing between circuit voice and IP voice is the cost
of the infrastructure. Circuit telephony is delivered within the metro via class 5 telephone switches costing millions of dollars. IP telephony, on the other hand, is supported by softswitches,
which are software emulations of circuit switching run over packet networks and
utilizing large, hardened, general-purpose workstations such as the Sun Netra platform. Total
cost is in the tens of thousands of dollars. Among the newer, specialized manufacturers, Sonus,
Tacqua, Syndeo, and Oresis are prominent, and old-line equipment manufacturers that have
added softswitches to their product lines include, most visibly, Nortel, Telcordia, and Marconi.
Incidentally, hybrid softswitch/circuit switches exist, of which Nortel’s product has been the
most successful. It should be noted that within the last two years VoIP architectures have been
developed that can dispense with softswitches but only in cases where a pure IP network has
been established between the two calling points. In cases where the call must traverse the
public switched telephone network (PSTN), a softswitch will still be required.
IP telephony may rightly be considered an emergent technology, one characterized by
warring standards bodies and a lack of consistent, generally accepted approaches to network
engineering. Some systems, for instance, use devices called gatekeepers that perform access
control and address translation while others dispense with them. Conversely, some incorporate
the switching and PSTN interface functions in a single box, and others favor a distributed
platform. Another problem has been lack of real functional equivalence between traditional
class 5 switches and IP softswitches. IP softswitches to date have not supported the entire
myriad of features programmed into circuit switches, which in sum represent millions of lines
of code.
Another problem with IP telephony is the requirement for special customer premises
equipment that will interface with standard desksets. If the subscriber is a business and already
has an IP private branch exchange (PBX), that is not a problem, but otherwise voice will have to
be recovered from the packet stream coming into the subscriber terminal and subsequently
sent to the desksets via ordinary copper telephone lines. That often involves a separate black
box, though some broadband wireless access equipment does support IP telephony today.
As matters stand today, the wireless broadband operator is unlikely to be able to compete
with the wireline incumbent as primary providers of voice services whether they offer circuit or
packet voice services. Still, if the wireless operator is in a position to provide a subscriber expeditiously
with extra phone lines, and the DSL or cable operator in the area cannot do so (or, for
that matter, the telco incumbent), then the wireless operator enjoys an advantage. And, if the
opposite is true—that is, if the competitive broadband access providers are offering secondline
telephony—then it is incumbent upon the wireless service provider to match those service
offerings.
In the interest of broadening the service portfolio, the wireless operator should definitely
consider offering long-distance voice service as well as local service. In this instance the operator
will inevitably be functioning as a reseller, and the long-distance service itself may be a
loss leader since the operator can compete only on the basis of price and can offer no unique
calling features. One may question the wisdom of offering a service that is not in itself highly
profitable, but studies have shown that operators providing the greatest number of converged
services are likely to be the least affected by “churn,” the tendency for subscribers to switch
service providers impulsively.
Long-distance telephone services may be either traditional circuit, generally carried over
ATM, or IP. If IP is chosen, make sure that specified peering arrangements are in place within
end-to-end connections between both calling parties. Such peering arrangements involve the
carriers providing long-distance data transport, and in the case of voice they must ensure low latency and jitter and minimum transit times. Ordinary best-effort service through the public
Internet will not do. Chapter 6 covers this topic in more detail.
Mobile Voice
Finally you must consider the issue of broadband operators moving into the mobile telecommunications
space and competing with cellular and Personal Communications Service (PCS)
operators. Several hardware manufacturers, including IP Wireless, Flarion, MeshNetworks,
and Navini, already support mobility, and each has suggested that mobile voice services are
well within the scope of a broadband wireless operator.
Such services could enhance the attractiveness of the broadband wireless network to
some customers especially if a single terminal, presumably some sort of smart phone, could be
made to serve as both a phone and a high-speed access device—which is really the whole
notion behind third-generation cellular. But before this kind of far-reaching convergence can
take place and win the acceptance of network operators, several developments will have to
happen.
First, the 802.20 mobile broadband standard will have to be completed (or, alternately, the
less comprehensive 802.16e amendment to the 802.16 standard will have to be completed). To
what extent this will be complementary to 802.16a remains to be determined, but it will most
likely apply to the same bands currently covered by 802.16. Second, the handset manufacturers
will have to produce phones that conform to the new standard, because without phones,
mobile telephony over broadband will remain conjectural. Third, some sort of modus vivendi
will have to be fashioned by the broadband wireless service providers and the mobile telecom
incumbents that themselves are moving toward fourth-generation systems, which will likely
have much in common with the 802.20 networks. Already mobile phone operators are
discussing using unlicensed frequencies in the 2.4GHz band to augment the mobile bands
during large downloads and are anticipating further allocations of licensed bandwidth to
support true high-speed services.
I do not expect 802.20 to be finalized until 2006 at the earliest, and there is no telling when
802.16e will emerge. A further span of time will be required for chipmakers to embody the standards
in silicon, and yet more time must elapse before any kind 802.20- or 802.16e-compliant
terminals become available. Gaining the support of handset manufacturers becomes essentially
a chicken-and-egg conundrum since the manufacturers are not interested in undertaking
a small production run. Still, Nextel (now part of Sprint), with its proprietary Integrated
Digital Enhanced Network (IDEN) system, managed to obtain handsets for its few million
subscribers, so 802.20- or 802.16e-enabled phones are not inconceivable within two or three
years, though prices are unlikely to be comparable to those of cellular units.
In the meantime, the cellular industry will proceed slowly but inexorably with the development
of fourth-generation systems. Fourth-generation will share in common with 802.20
the use of packet transmissions to convey voice messages and an elimination of circuits altogether
except to serve legacy customers. It will also provide much higher throughputs for data,
purportedly in the megabits per second, at least in burst mode.
The question then becomes, how will the two services differentiate themselves in the
marketplace, or will they? Similar questions were asked concerning PCS when it emerged in the
early 1990s, the supposition at the time being that PCS would fill a different market niche. It did
not, and ultimately fourth-generation cellular and broadband wireless may not fill different
niches either. Although mobile voice may be a desirable service offering for broadband wireless operators
in that it may be something to entice customers onto the network, it may also be a loss
leader. Mobile voice is already a price-eroded business, and the arrival of new service providers
and one-number portability cannot improve that situation. For now, fixed voice is the proper
concern of broadband wireless operators, however, and they will have plenty of tasks with
which to occupy themselves as they attempt to work through the intricacies of implementing
IP voice on a grand scale. Currently, only IP Wireless manufactures broadband wireless equipment
for mobile applications. Figure 3-3 shows a Compaq personal digital assistant (PDA). Conferencing
Video- and audioconferencing are well within the capabilities of 802.16a systems and are
services the network operator should strongly consider promoting, especially videoconferencing.
In the past, very expensive proprietary hardware/software platforms were necessary
to set up a videoconference with acceptable image quality, but today a multitude of highperformance
IP videoconferencing software products are available and can be run over
ordinary edge routers. Videoconferencing, while never experiencing the explosive growth
predicted for it in the past, must be considered a successful application and is increasingly
utilized by enterprises for training purposes and for controlling travel expenses. Furthermore,
it has always been a high-value application for which businesses have been willing to pay
premium prices. Finally—and this is good news for the broadband wireless service providers
striving to differentiate themselves from other broadband providers in the local marketplace—
it has seldom been offered by anyone other than incumbent telcos or large long-distance
service providers. Telemetry
Wireless telemetry, discussed in brief in the previous chapter, is a business application for
which there is a considerable sales potential and which, at the same time, makes small
demands upon bandwidth. To date, few broadband service providers have gone after
the telemetry or remote monitoring market, and, in many cases, telemetry is done over
private networks, often using special narrow telemetry bands in the UHF/VHF region. Wireless
networks are particularly well suited to telemetry functions, and the wireless broadband operator
can often pursue telemetry accounts without worrying about competition from the
wireline broadband providers. Marketing telemetry services is not easy, however. There is no
consistent profile for the users of such services and no established venue for advertising them.
Probably the best strategy in selling such services is to bundle them in with other data services
and let prospective business customers know that the wireless network can support telemetry.
Backhaul
The lower microwave network operator can also function as a carrier’s carrier by offering
backhaul to other types of network operators such as mobile telephone operators or Wi-Fi
hotspot operators. This can be a fairly lucrative business and is one that may enable the wireless
broadband carrier to avoid paying for tower space by utilizing the facilities of the carrier
being served.
A backhaul link is quite distinct in terms of network architecture from the sort of pointto-
multipoint topology that tends to characterize last-mile access networks. Instead of
apportioning the available spectrum to a number of simultaneous users and employing
medium-beam-width antennas to illuminate a designated sector, the operator uses a single,
narrow-beam point-to-point airlink often described as a wireless bridge. This will normally use
the entire bandwidth allotted to the network. With 5.8GHz spectrum, which is often employed
for the purposes of backhaul, 100Mbps throughputs are easily attainable over distances of
severalmiles.
Storage Area Networks
Yet another business application that conceivably could be supported by wireless is storage
area networking, also briefly touched upon in Chapter 2. My assumption is that the higher
microwave bands encompassed by 802.16 are better suited for this purpose, however.
Data storage is a network application where vital information is off-loaded to remote
storage facilities and invoked thereafter as it is needed. Storage equipment sales, including
those for networking gear, have remained brisk when other areas of business computing and
data networking have been plagued with declining demand, but thus far service providers have
had indifferent success in selling storage networking services. Most storage still takes place
within a LAN environment and involves special network protocols such as Fibre Channel and
ESCON. Fairly recently IP and Ethernet storage solutions have come on the market, and many
industry analysts think these will displace the old closed storage standards eventually. To my
knowledge, storage services have never been offered over a broadband wireless network, and
no 802.16 equipment is currently architected to provide such services. However, storage
services may be offered by wireless operators in the future and constitute a potential market
worthy of examination.