12 March 2018
The Long and the Short of Transmission Performance Evaluation
Comparing performance measurement techniques of multiplexed signal transmissions reveals channel correlation driven differences that depend on distance and become indistinguishable at transoceanic-scale transmission lengths
SAN DIEGO —
If you trust standard assessment methods of broadband transmission performance in real systems, your outlook might be unnecessarily pessimistic. You might also be overestimating how much benefit digital signal processing (DSP) can offer to nonlinearity compensation (NLC) in the transmission channels. By testing transmitter measurement schemes, engineers from TE SubCom, New Jersey, USA, will demonstrate that differences between performance measurements depends on the transmission distance and choice of DSP used. They will present their results at OFC 2018
, 11-15 March, San Diego, California, USA.
Whether by taking measurements in a laboratory, designed to mimic a real transmission system, or in the field to reflect daily performance scenarios, the reality of what is practical limits how thoroughly transmission channels can be tested. According to Jin-xing Cai, who will present the team’s work, ideal tests of a system’s potential capacity require full loading of the optical bandwidth, but limitations typically result in using a single measurement channel or a few odd/even measurement channels. The remaining optical bandwidth is either loaded with modulated channels or equalized amplified spontaneous emission (ASE).
In order to accurately interpret the implications of these single- or few-channel measurements on the performance of the whole system, correlations between channels -- induced by the transmitters used -- must be accounted for accurately. Recent theoretical and experimental studies showed that insufficient decorrelation among transmitted wavelength-division-multiplexing (WDM) channels could be producing misleading assessments. This revelation inspired Cai and his colleagues to further investigate the differences between different methods and the impact of channel correlation.
“In real transmission system[s], there is [a] minimum channel correlation since all user data are random,” said Cai. “Because of the limited number of transmitters used in laboratory setups, transmission experiments require careful consideration of channel correlation effects. This kind of channel correlation could enhance or diminish the nonlinearity induced penalty, hence lead to results not representative of real transmission system with all random data.”
For channels in the “C” and “L” transmission bands, the team looked at two measurement schemes. One served as a reference to compare to the other in which they had reduced channel correlations by using four independent channels. They also looked at the impact of NLC, and if compensating for the nonlinear effects would impact the measurement differences. To mimic the long distances data travels in real systems, their testbed consisted of 12 52.8-kilometer fiber spans.
The two schemes demonstrated a number of differences in their performance assessments. After 634 kilometers of transmission, the performances as a function of transmitter pre-emphasis, which is the technique commonly used to vary the signal’s power, demonstrated different responses to NLC. When employed, the performance measurements of the two systems were closer to each other and the benefit from the NLC use was always greater for the more highly correlated reference scheme.
Cai and his colleagues also looked at how the assessment differences trended with transmission distance up to 10,000 kilometers. With no pre-emphasis, higher channel correlation scheme degrades performance at short distances, but beyond 1,900 kilometers the performance measurements are the same despite the different correlation levels. They attribute this behavior to the fact that dispersion that accumulates with the distance de-correlates the measurement channel and its neighbors.
The difference in NLC benefits also disappears at the longer distances, meaning all of the modulation formats the group investigated showed negligible performance dependence on the channel correlation for transoceanic distances. Cai notes the implication that more attention may be needed for shorter-distance effects and plans to use other transmission schemes to investigate them further.
“This work will benefit all experimental work, whether it is lab measurement or field trial. The paper shows extra attention are needed for short transmission distance (<2000km),” said Cai. “We will use other transmitter loading schemes for more accurate performance measurements in short distance transmission experiments.”
Hear from the research team:
“On the Effects of Transmitter Induced Channel Correlation in Broadband WDM Transmission," J. -X. Cai, Y. Hu, A. Turukhin, M. V. Mazurczyk, M. Paskov, H. G. Batshon, C. R. Davidson, M. Bolshtyansky, and D. G. Foursa, will take place at 9:00 am PT, March 15, 2018, in San Diego Convention Center, California.
Media/analyst registration for OFC 2018 can be accessed online
. Further information is available on the event website at OFC
, including travel details.
The Optical Fiber Conference and Exposition (OFC
) is the largest global conference and exposition for optical communications and networking professionals. For more than 40 years, OFC has drawn attendees from all corners of the globe to meet and greet, teach and learn, make connections and move business forward.
OFC includes dynamic business programming, an exposition of more than 700 companies, and high impact peer-reviewed research that, combined, showcase the trends and pulse of the entire optical networking and communications industry. OFC is managed by The Optical Society (OSA) and co-sponsored by OSA, the IEEE Communications Society (IEEE/ComSoc), and the IEEE Photonics Society. OFC 2018 will be held from 11-15 March 2018 at the San Diego Convention Center, California, USA. Follow @OFCConference
, learn more OFC Conference LinkedIn
, and watch highlights OFC YouTube