This document discusses how investments in communications technologies can boost a nation's competitiveness globally. It argues that enhancing communications for business workers, such as through integrated access networks, can significantly increase productivity and competitiveness at the national level. The technologies needed to provide advanced communications services to remote business sites, such as small branch offices, are analyzed. These include asynchronous transfer mode (ATM) networks, which can transport data, voice and video in a unified way over a variety of physical layer technologies to create a flexible, non-hierarchical system.

1. Building a Network Infrastructurefor Global Competitiveness
Joseph L. Lias, Jr.
Premisys Communications, Inc.
email: jlias@premisys.com
Abstract
This paper provides a strategic impact analysis of
communications technology on a nations' global
competitiveness. Global competitiveness and its
definition is described, along with communications
aivchitectures,technologies and economics that have a
direct impact on national competitiveness. It is shown
that enhancing the communications needs of business
workers canfundamentally shft the global competitive
position of a nation. The technologies and
architectures required to enable the satisfaction of
these communications needs are analyzed.
Technology and architectural evolution required to
provide a business worker oriented service offering
from PTTs and communications Service Providers is
alsopresented.
I, Competitiveness via Communications
The changing global marketplace is forcing countries
to be more competitive at the national level, since
countriesi are strugg(iing wlti'[ t'he emerging g oIlla
economy. The key decision point for a nations' policy
makers is the definition of competitiveness at the
national level, as shown in figure 1.
- no nationmeatsthis t
-India & Mexico
figure 1 National Competitiveness
Possibilities
Some would argue a large positive balance of trade is
the key to global competitiveness. However there are
several countries that do not have a large positive
balance of trade (e.g. the U.S.A.), and is still
considered very competitive at the national level.
Some would argue low labor costs is a good indicator
of national competitiveness. Clear examples contrary
to this argument are India and Mexico. Some would
argue a competitive exchange rate for pricing in
international markets is a good indicator of national
competitiveness. Contrary examples to this argument
are Germany and Japan. Some would argue that if
every company or industry is competitive, then the
nation is competitive. Clearly, no nation meets this
test.
Therefore, the only meaningful concept of
competitiveness at the national level is "worker
305

2. productivity". The reason is a nation's standard of
living depends on the capacity of its businesses to
achieve high levels of productivity. Nations also
depend on enterprises to increase productivity over
time inorder to remain competitive. Worker
productivity, then, is the root cause of national per
capita income and global competitiveness. Sustained
national productivity growth, then, requires a nations'
key industries continually increase worker efficiency
and effectiveness. The majority of key industries
within any nation is increasingly based on knowledge
and information. Therefore, the intemational basis of
competition has shifted to the creation and assimilation
of knowledge, which forms the basis for continuous
investment in knowledge based technology such as
computers and communications.
Therefore, the deployment of new communications
network technology would not only impact the local
Service Provider (SP) positively, but would also have a
profound impact at the national level. In the last ten
years, profound changes in the structure of the
telecommunications industry has affected the
evaluation and justification of new technology
deployment. Because of its strategic nature, the
industry has become increasingly global and
competitive with loosening regulatory control.
The introduction of new network technology will
largely depend on the services requested by users.
Based on the service requirements, communications
users can typically be categorized into two large
groups: business(corporations) and residential.
Although corporations represent the smaller number of
customers, they generally account for the bulk of the
SPs' revenue streams. A corporations' remote branch
offices are often among the most important locations,
because they are the sites where a corporations'
customers actually get served. However, all too often,
inadequate communications facilities and network
access equipment are used to link branches back to the
corporate headquarters or network backbone. With the
deployment of new business applications based on
remote application servers, file servers or multimedia,
the requirements in remote offices will become even
more formidable.
11. CorporateCommunicationsNetwork
Business communications traffic is currently
undergoing a drastic shift from voice to data, mainly
due to rapid advancement and growth in business
information networking. Each business has its own
traffic characteristics and financial restrictions with
regards to communications networking. The ultimate
goal for SPs is to satisfy these diverse requirements
with a single access network infrastructure technology.
Integrated Access Server (IAS) technology offers SPs
the potential for access network infrastructures that
handle the diverse requirements of business services.
These applications also typically have a need for
dialed access to packet and cell based networks, and
bandwidth grooming and concentration at the SPs'
network entry points.
Current corporate networks link the larger sized
physical sites with high speed lines such as leased
CEPTs, DSls, and DS3s. The aggregate bandwidth is
managed as a network by intelligent multiplexers using
proprietary signaling and connection management. It
306

3. is a closed sub-network allowing no interoperability
with other sub-network elements, or with SP services
or sub-networks. The services provided by these
"private" transport networks are used in part to
prrovision private voice networks, data networks, video
teleconferencing networks, image networks, and even
LAN inter-networks.
Typically, the larger sized corporate sites generate
about half of the corporations' communications
network traffic, but constitute a very small fraction of
the total number of physical sites. The large majority
of physical sites are the remote branch offices.
Increasingly, large corporations are also moving to
bring their vendors and strategic business partners onto
their communications networks as Reach Through
Departments (RTDs) to improve business efficiency.
As large corporations bring vendors and partners "on-
line," they become virtual departments. In turn, the
vlendors and partners become virtually
indistinguishable from intemal departments of the
corporation itself. Currently, the only significant
drfference between internal departments and RTDs is
communications response time. Consequently, a more
meaningful measure of distance between physical sites
(both internal and external) may be seconds of
communications response time rather than kilometers.
Fast communications response times in remote branch
offices is still only a dream, since the usual means of
access to the corporate backbone network from these
locations is via dial-up, point-to-point, or polled-access
via low-speed leased lines or modems. Therefore, in
most countries a reasonable case could be made that
smaller blranch offices have the greatest unmet
communications needs.
In fact, the hottest market segment in the U.S.A. is
remotelbranch office networking. This has not gone
unnoticed by communications equipment vendors and
SPs. Most equipment vendors are augmenting their
portfolios with branch office devices that support
multiple data protocols and voice, and SPs are pricing
frame re1a.y services to attract traffic from even the
smallest remote locations.
To provide economical services, it is important to
simplify communications network configurations and
to integrate network resources. Network simplification
can be realized by a integrated, non-hierarchical access
network equipment. The major issue of a non-
hierarchical network is the appropriate level of
interconnect between the SP and business customer.
Alternatives exist all the way from the lowest (e.g.
physical) level to the highest (e.g. application) level.
When different networks and protocols are involved,
the interconnection usually involves a conversion
process. The complexity of both the corporations' and
SPs' communications network is the type of access
network equipment chosen.
111. Remote Branch Access Architectures
Generally speaking, the traffic density of access
networks is low because the access network is close to
individual users. This is especially true in multi-tenant
business buildings, which contain the majority of
remote branch offices. However, SPs can continue to
employ the Distribution Area (DA) and Carrier
307

4. Serving Area (CSA) as fundamental outside plant
planning units, even in multi-tenant downtownlurban
areas. Fiber cable and electronics can be deployed in
the feeder route, being terminated at the carrier service
area's (CSA's) business building site. At this point, the
traffic is distributed to the DA's (i.e. floors) throughout
the building as shown in figure 2.
Customer Located Narrowband voice
Drop Cable
figure 2 Business Building Application
Planning
A Bandwidth Management System (BMS) can route
narrowband services to the Narrowband switch and
also allow communications between existing
narrowband and future broadband networks. Traffic at
individual remote branch offices is bursty, and the
voice traffic is fragmented, and typically can not
justify dedicated SP transport facilities. SPs can not
afford a network of nearly empty pipes, and vastly
underutilized switch ports. Therefore, a BMS
"Concentrator" product is also a fundamentally more
efficient and economical network technology.
Grooming and concentration of the traffic from
multiple individual remote end users, allow the SPs to
economically combine the traffic transport of voice,
video, and data.
IV. Broadband Access Evolution
The long-term view is pretty clear: in another decade,
a large fraction of the communications traffic in the
developed world will be carried by ATM technology.
ATM is the best available compromise between the
conflicting requirements of data, voice, and video.
The real question is what is the best way to phase
ATM into the network. Different types of ATM-based
services, such as end-to-end leased circuit services,
end-to-end switched circuit services, and center-to-end
services, are expected to be popular in different time
frames.
Interconnection across private or public networks has
been discussed in several publications. Most proposals
interconnect networks at the network layer. It is
generally desirable to devise a LANIMAN
interconnection platform that is independent of the
higher-layer protocols running in the various
interconnected environments (TCPAP, OSI, etc.).
If end-to-end Synchronous Optical Network (SONET)
and Synchronous Digital Hierarchy (SDH) was
available to corporate communications network sites,
there would be an opportunity to manage all the
corporate sub-network resources as a single
homogeneous network. However, end-to-end SONET
in a global enterprise network is highly unlikely in the
near future for four primary reasons:
other public services will just be more cost effective;
the technology may be unavailable in some areas;
some locations will preserve existing investment in
older technology.
308

5. SlONET'siSDH's initial benefits are expected to come
in the form of reduced provisioning costs for inter-
exchange SPs, made possible by SONET'siSDH's
ability to economically drop channels from high speed
trunks. Later, SONETiSDH will move into the realm
of corporate networks in much the same manner as T1
and T3. Some researchers predict that SONETiSDH
transmission will grow gradually to 15 percent of fiber
access lines by 1996. The most likely implementations
of SONETiSDH by enterprise networks will be in
multiple isolated applications, such as building
backbones, campuses, and Metropolitan Area
Networks (MANS),providing high-speed transport for
various traffic types. These installations will
continually be interconnected with DS3s, or multiple
CEPTs and DSls.
However, the lack of end-to-end SONETiSDH based
networks providing logically non-hierarchical network
support can still be alleviated via ATM. ATM
provides a non hierarchical structure in which cells
from different connections are transported commonly,
independent of the bit rate or burstiness of the
connections. This inherent flexibility makes ATM the
technology of choice for the final B-ISDN. However,
before B-ISDN is available to everyone, some decades
will pass.
Recently, a solution which has gained some interest is
the use of scini-permanent virtual paths (VPs) to
interconncct the gateways across the ATM network A
virtual path can Re viewed as a "pipe" which can
multiplex several VCs (virtual circuits) Thus,
connectionless data units are channeled through the
appropriate VP to the desired destination gateway.
Figure 3 illustrates the relationship among virtual
channels, virtual path, physical link, and physical
laycr. Unlike the SONET "physical" path concept (i.e.
each STS or VT path has a fixed capacity), the
capacity of a VP may range from zero (e.g. backup
VP) to t:he line rate; thus the hierarchical path
multiplexing structure of SONET is not needed in the
VP-based network.
figure 3 VPNC and Physical Layer
Relationships
Two levels of ATM multiplexer stages increase the
flexibility ATM VP and VC network planning. An
ATM Service Multiplexer that is located in the
customer premises, provide the function of statistically
multiplexing calls belonging to different services. An
ATM Network Multiplexer can provide concentration
of traffic originated by different subscribers on the
same or different floors.
The physical layer of the broadband access network
can consist of not only SONET, but older technologies
(e.g. DS1, DS3, and CEPT rates) as well. However,

6. the unifying layers in this architecture are the ATM
and AAL layers, as shown in figure 4.
tHigher Layers
I
figure 4 Network model with ATM transport
Just as the SONET standard conserved the public
network's investment in wideband technology, ATM
can conserve the corporations' network in existing
technology while facilitating migration from the
current corporate low-capacity implementations to a
high-capacity infrastructure. This can be
accomplished by allowing ATM cells to be carried
over imbedded metropolitan, campus, and building
networks until upgrading to SONET. A broadband
non-hierarchical network can only be offered if the
access infrastructure is appropriate. An adequate
introduction strategy has to take into account the
existing infrastructure (local loop, in-office cabling,
terminals, etc.) and future planned investments.
Fiber's dark side is the high cost of installing optical
connectors. Optical fiber cable installations are about
five times as expensive as twisted-pair wiring.
Additionally, engineers had tremendous success in
increasing the capacity of twisted-pair wiring. Data-
grade (Category 5) unshiclded twisted pair will
probably suffice for the runs between individual
workstations and wiring closets. Optoelectronics is
still too expensive for an individual subscriber to bear.
User needs for data rates up to 100 Mbps can be met
by the UTP Category 5 wiring used for horizontal
drops.
Some would argue ATM at the DS1 rate is very
inefficient. However, the major concern in enterprise
networking today is network management. Therefore,
corporations are more concerned with controlling and
managing facilities (e.g. DSl), instead of using them in
the most efficient manner. Corporations are willing to
sacrifice some communications network efficiency for
manageability of this strategic asset.
V. Economics of Broadband Facilities
While the market for data applications is growing, use
of ATM services is still primarily a cost tradeoff
against frame relay and private line services, since end
users can use various bandwidth techniques to
maximize efficient use of facilities. Three examples of
end user bandwidth management agents are
compression systems, statistical multiplexing systems,
and queuing systems. Today's price structure for
subscribers offers b-times-the-bandwidth for a-times-
the-cost, where b is much greater than a.
In fact, a brief look at historical bandwidth costs will
give an indication of future bandwidthldeployment
cost ratios. Figure 5 is an illustration of transmission
facilities deployment costs per megabit-mile. It is a
ratio that indicates the effectiveness of deploying a
technology as a function of deliverable bandwidth.
310

7. I Year
figure 5 HistoricalCosts of Fiber and Copper I Year
figure7 Technology vs. Bandwidth Costs
As a sanity check of figure 5, one can compare the
actual North American deployment of inter-wirecenter
fiber facilities as illustrated in figure 6.
Year
1
figure 6 Inter-Wirecenter Fiber Deployment
Note the 1983 escalation in deployment of fiber in the
inter-office. This coincides with the derived cross-
over point of fiber vs. copper inter-office plant
facilities. Overlapping this information with the
introduction dates of key transmission technology
shows the non linearity of cost structures, as illustrated
in Figure 7.
VI. Conclusion
In many nations, governments are struggling with
positioning their countries as being competitive in the
global marketplace. This paper has shown that
strategic investments in computers and
communications can significantly benefit enterprise
workers, which will have the immediate effect of
making the countrymore competitive.
In addition, SPs initially focusing on their corporations
communications needs with an integrated access
network communications infrastructure will further
enable Corporations to manage their end products in a
more effective and efficient manner.
References
(1) A. Tantaway, and M. Zitterbart, "A scheme for High-
Performance LAN Interconnection Across Public MANS",
IEEE Joum. on Selected Areas in Comm., Vol. 11, No. 8,
Oct. 1993,PI".1133-1144.
(2) G. Clapp, "LAN interconnection across SMDS," IEEE
Netw., Sept. 1991.

8. (3) W. M. Seifert, "Bidges and routers," IEEE Netw. Mag.,
vol. 2, no. 1,Jan. 1988.
(4) T. Van Landegem and R. Peschi, "Managing a
connectionless virtual overlay netowrk on top of an ATM
network," in Proc. IEEE Int. Conf: Commun.,Denver, CO,
June 1991,paper 31.5.
( 5 ) M. Gerla, T. C. Tai, and G. Gallassi, "Internetting LAN's
and MAN'S to B-ISDN's for Connectionless Traffic
Support", IEEE Joum. on Selected Areas in Comm.,Vol. 11,
NO.8,Oct. 1993,pp. 1145-1159.
(6) D. Irving, "The Role of Customer-Premises Bandwidth
Management," IEEE Netw. Mag., vol 8, no. 3, MayiJune
1994.
(7) D. P. Malley and Ozan K. Tonguz, "Fiber in the Loop:
Where and When is it Feasible?", IEEE Journ. on Selected
Areas in Comm.,Vol 10,No 9, Dec. 1992,pp. 1523-1544.
(8)Members of the Technical Staff AT&TBell Laboratories,
"Engineering and Operations in the Bell System", Issue 2,
Chapter 14.
(9) B. Goch, P. Julien, and J. Lias, "Intelligent Network
Element Software Procurement and Delivery," in Proc. ICC
'93,pp. 1189-1196, 1993.
(10) S. Kheradpir, W. Stinson, J. Vuceti, and A. Gersht,
"Real-Time Management of Telephone Operating Company
Networks: Issues and Approaches," IEEE Joum. on Selected
Areas in Comm., Vol. 11,No. 9,Dec. 1993,pp. 1385-1403.
(11) C. Ersoy and S. Panwar, "Topological Design of
Interconnected LAN/MAN Networks," IEEE Journ. on
Selected Areas in Comm., Vol. 11, No. 8, Oct. 1993, pp.
1172-1182.
(12) P.D.Lattner, R.L. Fike and G.A. Nelson, "Business and
residential services for the evolving subscriber loop,"IEEE
Communications Magazine, Vo. 29, No. 3, Mar. 1991,pp.
109-114.
(13) H. Uno, Y . Maeda and Y. Kanayama, "OAM
Mechansims for BISDN UN1 and Access Networks,"
International Journal of Digital and Analog
CommunicationsSystems,Vol. 4 1991,pp. 161-168.
(14) A. Malik, "Network management and control systems
and strategic issues," IEEE CommunicationsMagazine, Mar.
1990,pp. 26-29.
312