IP over photonics: making it work
For some, IP over DWDM is the Holy Grail of optical networking. Before this quest can be completed, SONET?s network-management functions must be duplicated in the optical layer.
Pawan Jaggi and Larry Steinhorst
With the explosive growth of the Internet, data traffic will continue to outpace voice traffic for years to come. The resulting need for huge capacity on the network backbone has led to the deployment of dense wavelength-division multiplexing (DWDM) systems, which significantly increase the bandwidth-carrying capacity of a single optical fiber. With so much of the data traffic based on Internet protocol (IP), service providers are being tempted to maximize the efficiency of the network by eliminating layers between IP and DWDM. This goal must be achieved without sacrificing the features of current networks that provide optimal reliability and availability.
Early network architectures used time-division multiplexing (TDM) to transmit a single OC-48 channel at a rate of 2.5 Gbits/sec. The TDM rate increased to 10 Gbits/sec (OC-192) to handle the growing demand for bandwidth. This demand continued, however, resulting in the need for systems with even higher capacity.
DWDM allows multiple channels to be transmitted over a single fiber by sending them at different wavelengths (see Fig. 1). Use of the highest-rate TDM, together with DWDM, provides the most efficient use of network bandwidth. It is the combination of TDM and DWDM that provides this efficiency (see Fig. 2). Current commercial systems support as many as 32 wavelengths of OC-48 and/or OC-192, increasing the network`s capacity to up to 320 Gbits/sec. These systems can operate over conventional as well as non-zero dispersion-shifted fiber designs in both short- and long-haul networks.
Today, legacy voice circuits are time-division multiplexed to higher bit rates to be transported in such DWDM networks. Data traffic also is still multiplexed by Synchronous Optical Network (SONET) add/drop multiplexers (ADMs) before being transported by DWDM systems. A benefit of this scheme is that the SONET protocol provides many functions to simplify the management and operations of transport networks. These functions allow the network manager to ensure high-quality transport, provision the network, and repair the network when a fault occurs. The SONET protocol performs these functions by providing an embedded communications channel between network elements, orderwire channels, signal labels, remote error indications, signal traces, error detection, multiplexing, and synchronization.
However, emerging OC-48c and OC-192c high-speed interfaces could enable routers and switches to interface directly into DWDM systems without being multiplexed in the time domain by SONET ADMs. Thus, carrying IP or Asynchronous Transfer Mode (ATM) directly over photonics has become possible. But joining photonic networking with these protocols is not as simple as connecting two pieces of a puzzle. While IP/ATM over photonics is attractive to service providers as a potential means of reducing the costs of high-capacity networks, the network functions currently provided by SONET must be built into the photonic layer to reduce costs sufficiently.
Fortunately, many of the SONET management features mentioned above can be provided by DWDM systems through the use of an Optical Supervisory Channel (OSC). An OSC is carried outside the conventional erbium-doped fiber amplifier (EDFA) band and provides overhead communications among optical network elements. It typically provides an embedded data-communications channel, orderwire capability, optical multiplex section (OMS) signal label and trace, and OMS remote error indications, while DWDM systems provide multiplexing capabilities.
But some of the features of the SONET protocol are not easily duplicated within an OSC. These include error detection and fault isolation, signal trace, synchronization, and protection and restoration. Thus, other ways to provide these functions must be found.
Error detection and fault isolation
The SONET protocol includes a frame to easily identify the SONET overhead. The overhead contains an error-detection capability used for monitoring defects in the network. With simple circuitry and minimal overhead, SONET allows the condition of the signal to be checked without having to process every packet in a data stream or another payload.
The migration to optical networks will still require error detection for fault isolation and protection switching, especially at such high bit rates as OC-48 and OC-192, and for backbone networks. By using only optical parameters (for example, optical power and optical signal-to-noise ratio), degraded digital signals caused by optical nonlinearities or other problems in the network may not be detectable.
Error checking, which involves using the SONET frame within the network, can be accomplished within the transponders of the DWDM network, SONET regenerators, or the interface cards of routers and switches that use SONET frames (see Fig. 3). If other signals are carried transparently by a DWDM system, it may not be possible to do bit-error checking, which therefore limits fault detection and the furthest distance that the signal can be transported.
Signal trace is used to follow photons through the network to make sure the correct signals end up at the correct place on the other end. This capability can be provided by adding information to a wavelength and pulling it off to an electrical receiver, which detects and reports the signal trace.
Signal trace can be provided by any of the following methods:
adding an amplitude-modulated pilot tone via the signal transmitter
adding an envelope of bits to the digital signal, which slightly increases the signal bit rate, as with out-of-band forward error correction
the signal trace bytes of the SONET protocol.
In a DWDM system, transponders can be used to add a signal trace via one of the first two methods listed here to help maintain the transparency of the network.
Synchronization in SONET
Synchronization in current SONET networks supports the multiplexing of lower TDM rates into higher TDM rates and maintains the quality of the digital signal through the network to ensure it is received correctly.
DWDM networks support the multiplexing of wavelengths so that synchronization is not needed for multiple wavelength transmission. However, synchronization may still be needed to maintain the quality of the signal through the network, especially for long-haul transmission or where the source signal to be transmitted may not be of good quality. Multiple regenerations through transponders or other equipment can accumulate jitter and wander effects, which may cause bit errors in the receiver. Synchronization can clean up the signal to allow it to be transmitted correctly or over longer distances.
The current EDFA bandwidth is limited to approximately 30 nm in the C band. The maximum bandwidth can be put through this EDFA band by multiplexing lower TDM bit rates into the maximum TDM rate supported by the DWDM system. SONET ADMs or dedicated transponders that support internal multiplexing are used for the synchronization needed for reliable time-division multiplexing.
Protection and restoration
Protection and restoration can be provided by the SONET layer, photonic layer, IP/ATM layer, or some combination. Service providers currently use SONET to offer the highest reliability and availability through fast and efficient network protection and restoration. As providers migrate to data-centric DWDM architectures, they still expect to provide optimal network efficiencies and high availability.
The SONET protocol provides bytes that carry signaling for automatic protection switching (APS). It operates in near real time to synchronize protection switching in ring and linear configurations so that restoration can be accomplished in less than 50 msec in ring spans less than 1200 km.
Without the SONET APS protocol, protection and restoration can be accomplished in the photonic or IP/ATM layers. In the photonic layer, optical ADMs with wavelength-switching capability are required. They must be capable of switching based on defects in the digital signals, which may require electrical bit-error-rate monitoring or optical Q monitoring built into the switching system. Also, an APS signaling channel must be defined at the OMS or optical channel level to signal and control protection switches.
For certain architectures, restoration at the IP/ATM layer may provide adequate capabilities. Although restoration at the IP/ATM layer is expected to be slower than protection switching at the photonic layer, these slower speeds may be acceptable depending on the services being provided. IP/ATM layer restoration also typically requires additional network engineering and maintenance to sustain an adequate restoration capability.
It is important to note that multiple protection and restoration mechanisms can co-exist in the same DWDM network to support different service classes. This scenario may occur through the evolution of legacy networks to data-centric networks. The DWDM network can simultaneously support a mixture of IP and SONET protection.
At the photonic level, dynamic optical add/drop multiplexers (OADMs) and optical crossconnects (OXCs) are being developed. These technologies will enable the dynamic optical network to provide efficient optical transport as well as the foundation for optical protection and restoration.
In a ring architecture, an OADM can be used to add and drop some or all the capacity of the DWDM ring. Such OADMs can be introduced as simple back-to-back terminals or in more sophisticated configurations, such as acousto-optic tunable filter (AOTF)-based systems, which are available now.
The AOTF devices enable dynamic add and drop capabilities based on software provisioning (see Lightwave, September 1998, page 97). This, along with switching capabilities, is sufficient for network protection, wavelength routing, and wavelength selectivity.
Optical rings provide ring and span switching protection capabilities. Ring switching protection guards against cable breaks and node failures, while span switching protection guards against equipment failures and supports maintenance activities.
Optically amplified DWDM systems have revolutionized the industry by providing the capability of carrying huge amounts of capacity in the backbone network. With the demand for capacity skyrocketing, network providers are seeking more efficient ways to carry data-centric bandwidth, such as Internet traffic, over their backbones. There are various choices and possibilities that are being looked at to accomplish this task.
Traditionally, SONET-supported networks provide fundamental advantages that cannot be overlooked. Emerging IP/ATM-over-DWDM architectures must support certain key features that are currently provided by the SONET layer to be considered for deployment. These new architectures will eventually mature and co-exist with the existing SONET-supported architectures. They will become the foundation for next-generation optical networking systems. u
Pawan Jaggi is senior manager of the optical networking group at Fujitsu Network Communications Inc. (Richardson, TX), e-mail: Pawan.Jaggi@fnc.fujitsu.com. Larry Steinhorst is senior product planner of the optical networking group at Fujitsu, e-mail: Larry.Steinhorst@fnc.fujitsu.com.