In conventional optical networks, once channels (wavelengths) are turned up, they are used for the rest of the system’s life. Capacity and reach of the channels are fixed, depending on the particular transponder card chosen. Network operators manage their system capacity in multiples of the SONET or SDH data rate, now usually 9.952 Gbps.
Recently, two new ideas have come up, which will change these fixed properties of optical channels.
One is the introduction of dynamic optical networks, where channels can be turned up on demand and might stay up only for the duration of a soccer match or until the election results have been announced. Several research projects have been carried out on the topic of dynamic optical networking, and I only mention the European DICONET project, in which I am personally involved.
Another paradigm shift concerns the data rates that need to be transported on an optical channel. So far, we were used to the SONET / SDH 2.5G, 10G, or 40G chunks of capacity. But in the datacomm world, users are already familiar with variable channel capacity. If the quality of a channel deteriorates, the data rate is reduced by throttling the transmitter. A famous example for this dynamic data rate adaptation is xDSL in the access area, where the QAM modulation format is adapted to the transmission line quality. Some users of optical networks, especially from the non-telecom world, would accept this kind of behavior also for their high-capacity optical connections. If the optical signal-to-noise ratio is better than required for the basic data rate, the data rate should be increased.
How can this data rate adaptation be realized?
The bandwidth of the electrical and electro-optical components in the transponder, like amplifiers, modulator, photodiode, are limited. They determine the symbol rate limit of the optical signal. Moreover, timing sensitive components, like the clock recovery, have a quite narrow bandwidth so that a constant symbol rate should be maintained, even though the data rate is increased. This increase in data rate can then be achieved by increasing the number of bits transported within each symbol. Quadrature amplitude modulation (QAM) of the optical carrier would be a candidate for an adaptive bit rate modulation format. The same optical I-Q modulators can be used as for instance envisaged for the QPSK modulation format, which is currently the main contender for 100Gbps long-haul transmission. The higher data rate is then achieved by modifying the control voltages going to the modulators. This can be done in a software-defined FPGA converting the digital input data to analog modulation voltages.
The Figure shows an estimate of the achievable increase in transmission distance. With sufficient OSNR margin, the transmitted bit rate can be increased threefold while maintaining a constant symbol rate. On the other hand, if the OSNR is not sufficient for 100Gbps transmission, the transponder can operate with reduced data rate to compensate for a few dB of OSNR.
The adaptation of the data rate can be done solely by adapting the program of the FPGA. Depending on the quality of the optical link, the system control plane can trigger an increase in data rate. For dynamic networks, the link quality can be predicted before the channel is turned up, and the expected data rate can be set by software control of the transponder.
What else can be done with dynamic data rate transponders? Watch Brian Teipen in his invited talk at the ICTON conference in Munich in June, where he will give some more examples of using adaptive data rates for channel protection and discuss more applications of software-defined optics.