Nippon Telegraph and Telephone
Corporation (NTT; Head Office: Chiyoda-ku, Tokyo; President:
Norio Wada), NEC Corporation (Office: Minato-ku Tokyo;
President: Akinobu Kanasugi), Fujitsu Laboratories Ltd.
(Office: Kawasaki-shi; President: Michio Fujisaki), The
Furukawa Electric Co., Ltd. (Office: Chiyoda-ku; President:
Hiroshi Ishihara), Hitachi, Ltd., (Office: Chiyoda-ku;
President, Chief Executive Officer & Director: Etsuhiko
Shoyama) and Mitsubishi Electric Corporation (President & CEO:
Tamotsu Nomakuchi) have conducted a successful demonstration
of freely controlling Gbit/s- class communication circuits
by GMPLS (Generalized MultiProtocol Label Switching) technology
in our trial photonic network consisting of multi vendor
equipment.
The photonic network, which maximizes the potential of high-speed
and large-scale optical communication technology, is attracting
attention for realizing the next-generation backbone communication
network, which must accommodate various communication services,
such as IP communication and leased line services. In this
experiment, several kinds of GMPLS-supported communication
equipment developed in each company were connected to yield
a photonic backbone network that is virtually ready for practical
introduction. This GMPLS network supports automatic failure
recovery so backup paths are activated when communication
links or equipment on the working path fail. In the normal
state, a backup path can be used to transfer low priority
traffic, which significantly boosts resource utilization
efficiency. End-to-end high-speed communication through this
network was established prior to conducting the video transmission
experiments.
The technology to be exhibited will make it possible to highly
reliable and economical high-speed and large-scale networks.
Only through these advances can the lower-cost transmission
of voice and high-resolution video like digital cinema be
made feasible.
Our technical achievements and a demonstration of high-resolution
digital video transmission will be introduced from January
26 and to 27, 2004 at the Gigabit Network Symposium 2004
(*1) in Tokyo, Japan.
[Background]
With the spread of broadband communication, services
that use the IP network to transfer voice and video,
like VoIP
and streaming, are growing every year. Accordingly, the
network capacity requirements are increasing rapidly.
Voice and video are formed into IP packets at the user
terminal, and are sent to the network. Why the conventional
Internet streaming flow demands a network capacity from
hundreds of kbit/s to several Mbit/s, the transmission
of the data signal of high-resolution video such as Hi-Vision
and digital cinema (movie contents saved in digital form)
consumes Gbit/s. In the conventional IP network, packets
are repeatedly multiplexed with packets from other users
at the metro and the backbone networks. Multiplexing
packets causes packet waiting which creates fluctuation
in the
packet arrival time (jitter) at the receiver side. Excessive
packet jitter can trigger failure of the video service
(Fig.1). The traditional solution to jitter is to set
the transmission bandwidth at up to ten times the straight
demand of the application. Therefore, it is necessary
to
enlarge network capacity much faster than required according
to the increase in Internet user number and the enhancement
of access lines (more demanding services). It is said
that increasing network capacity by adding conventional
equipment
will lead to sharp cost increases and make it hard to
maintain reasonable communication charges.
[Technical points]
MPLS(*2) and GMPLS(*3) are the key technologies to achieve
the long-sought broadband and economical network control
mechanism. MPLS is a traffic control technology for the
IP network. GMPLS is an extension of the MPLS concept to
the circuit switching network and the optical fiber network.
GMPLS enables unified control management of the network
layers (packet / TDM(*4) / wavelength / optical fiber).
The use of GMPLS unifies the network operations of disparate
networks such as the TDM network with its TDM crossconnects
and the optical wavelength network with its optical crossconects
(OXC), which will yield significant network operation cost
reductions.
GMPLS is a new control technology designed for the next-generation
photonic network. GMPLS offers the novel network service
of optical wavelength leased-lines allowing users to freely
change destination and quality.
Using the network control of GMPLS and the circuit switching
concept seen in the telephone network yields traffic flows
that can satisfy different service quality demands. The
traffic of applications sensitive to delay or jitter is
distinguished from best-effort traffic and transmitted
separately in the MPLS network. The GMPLS network goes
one step further and transmits the traffic flows sensitive
to delay or jitter through a TDM network or optical wavelength
network. Consequently, the level of packet multiplexing
is minimized because no packet multiplexing occurs in these
networks, and it becomes possible to dispense with the
excessive transmission bandwidth normally use to suppress
packet jitter (Fig. 2).
The trial network setup is shown in Figure 3. It represents
a large-scale backbone network connecting metro networks.
Metro networks 1 and 2 use MPLS technology and are constructed
around MPLS routers produced by Fujitsu Ltd. and Furukawa
(GeoStream R920 and FITELnet-G21, respectively). The
backbone network use the GMPLS technology in four kinds
of equipment;
HIKARI routers(*5) by NTT (IP and wavelength), GMPLS
routers of FITELnet-G80 by Furukawa (IP and TDM), TDM
crossconects
of SpectralWave U-Node by NEC (TDM and wavelength), optical
crossconects by Mitsubishi Electric, Hitachi, Fujitsu
Laboratories, and NTT (wavelength and optical fiber).
The bracketed expressions
indicate the function and the layer handled by each company’s
equipment. GMPLS control software that can handle at
least two layers was developed and implemented by each
company.
These GMPLS equipment are connected using control links
(to exchange control messages holding routing information)
and data links with over gigabit rate (Gigabit Ether,
OC48, and OC192).
Since a wide variety of network equipment (e.g. IP routers
and cross connects) will coexist in a GMPLS network,
it is essential to verify interoperability between multi-vendor
network equipment.
Until now, interoperability tests of GMPLS routing and
signaling protocols were conducted only at the control
software level. We have successfully realized the distribution
and collection of routing information and the setup
of paths in the multi-layer and multi-vender environment.
It is possible to use the various communication paths
provided (packet, TDM, wavelength, and a fiber) according
to the
communication quality demanded.
In addition, this GMPLS backbone network supports two
functions: seamless connection to the Metro network
and high reliability. The former is realized by MPLS
and
GMPLS interworking technology (*6). This interworking technology was first
verified to work in the multi-vender environment in October 2003; a world
first. In this experiment, practicality was verified
through the use of commercial
MPLS routers. The latter is made possible by network failure recovery technology
(restoration) that can set up detour paths automatically when failure occurs
on a certain communication link or equipment on the working path. The backbone
network is robust to network failure. This restoration technology shares
a backup path among several working paths. In addition,
this backup path is made
available for the use of lower priority traffic prior to failure of one
of the high priority working paths. Consequently, network
resources are used
very effectively (Fig. 4).
This experiment has successfully transmitted high-resolution
video like digital cinema across MPLS and GMPLS networks.
It confirms the feasibility of establishing
an economical and highly-reliable IP network that offers high-resolution
video transmission at reasonable cost.
The GMPLS equipment of NEC and Hitachi in this experiment include the
results of research sponsored by TAO (Telecommunications Advancement
Organization
of Japan).
[Future plan]
This experiment was carried out by the Photonic Internet
Lab (PIL) (*7). PIL has already conducted interoperability
tests of a standard GMPLS protocol with a number of global
companies at the interoperability site of ISOCORE at
Washington D.C, USA. PIL will promote the creation of
new control technologies that can be accepted as international
standards and the development of novel network services.
<
Glossary>
*1) Gigabit Network Symposium 2004
Symposium for publicizing the research-and-development
results for the last five years of work on the Japan
Gigabit Network (JGN), and promoting related research
the exchange of information between fields. JGN is the
research network constructed by TAO.
*2) MPLS (Multi Protocol Label Switching)
The abbreviation of Multi-protocol Label Switching. MPLS
is a packet transmission control technology for the IP
network. MPLS realizes traffic control in the IP network.
Using the circuit-switching concept seen in the telephone
network, it establishes and handles traffic flows that
satisfy different service quality demands. At present,
MPLS technology is being used to realize traffic management
in Internet service providers and to realize IP-VPN services.
*3) GMPLS(Generalized Multi-Protocol Label Switching)
GMPLS is a protocol that establishes generalized MPLS in
all layers of the IP network: layer 2, TDM (Time Division
Multiplexing), wavelength (WDM), and the fiber. MPLS
attaches fixed length labels to IP packets. GMPLS is
attracting attention for controlling the next-generation
photonic network. Standardization of GMPLS is being advanced
mainly by IETF (The Internet Engineering Task Force).
The basic function of GMPLS was released as a Proposed
Standard in February 2003, with registration number RFC
3471-3473. In order to make it a complete and truly practical
protocol, world-wide efforts are needed to elaborate
the remaining details and develop protocol code that
can be directly installed in network equipment.
*4) TDM
Transmitting technology based on time division multiplexing.
SDH/SONET is used widely in many networks.
*5) HIKARI router
A communication device with Terabit (1 trillion bits) -class
capacity. It is based on the GMPLS concept and controls
MPLS and wavelength paths dynamically and systematically
according to changes in IP traffic
*6) GMPLS and MPLS interworking technology
Permits the interworking of MPLS and GMPLS networks through
the use of a newly developed routing interworking function.
Because this approach allows the MPLS side to recognize
required IP nodes inside the GMPLS domain, GMPLS domain
resources can be efficiently managed from the MPLS domain,
such as shortest path setup and traffic control.
*7) Photonic Internet Lab
abbreviation: PIL See http//www.pilab.org
PIL, which was founded in September 2002, is promoting
research into and the development of next-generation photonic
network technologies. PIL encourages the submission of
proposals from its members to global standardization bodies,
like ITU-T, IETF, and OIF. PIL also tests the photonic
network control programs developed by PIL member companies.
This experiment directly supports these goals. At present,
PIL consists of seven companies; Oki Electric Industry
Co., Ltd. and the six above-mentioned companies. In this
experiment, Hitachi Communication Technologies Ltd., which
is a subsidiary company of Hitachi, Ltd., received the
cooperation of Hitachi, Ltd.
PIL activities are supported by the R&D support scheme
of the MPHPT (Ministry of Public Management, Home Affairs,
Posts and Telecommunications) for funding selected IT activities.
|
