POLARIZATION MODE DISPERSION: Decision feedback loop compensates at 10 Gbit/s

May 1, 2000
Upward pressure on data-communication bit rates is creating a need for improvements in the diagnosis and control of polarization mode dispersion (PMD).

Upward pressure on data-communication bit rates is creating a need for improvements in the diagnosis and control of polarization mode dispersion (PMD). At 2.5 Gbit/s with 400-ps pulses, PMD is generally not a problem. But at 10 Gbit/s with 100-ps pulses and PMD-induced pulse broadening that exceeds 20 or 30 ps, the problem becomes significant. And at 40 Gbit/s, the problem of intersymbol interference due to overlap of successive bits becomes even worse.

Researchers at Bell Labs, the R&D arm of Lucent Technologies (Holmdel, NJ), have developed an electrical solution to this problem based on a decision feedback loop (DFL) that can handle data rates up to 10 Gbit/s. In unrelated work, another group of Bell Labs researchers have developed a time-domain method of measuring PMD that provides full vector information instead of just magnitude.

Generated by birefringent effects related to factors that include cross-sectional asymmetry, stress or torsion, and temperature changes in optical fibers, PMD effectively splits a single input pulse into two pulses traveling along the fiber at slightly different speeds. Consequently, a receiver in an uncompensated system will see a broadened output pulse that consists of the two traveling pulses separated by a differential-group-delay time that typically falls in the 0- to 100-ps range. Receive errors can occur when the series of signals from one bit overlaps onto the series from the following bit.

Approaches to the problem

A team of researchers from Bell Labs, the University of Paderborn (Paderborn, Germany), and the Fraunhofer Institute for Applied Solid-State Physics (Freiburg, Germany) proposes to address this problem with a decision feedback loop in an electrical nonlinear filter that subtracts the previous bit value from the received signal when the next bit arrives (see figure on p. 67). "We think that our nonlinear filter is a good compromise between filter efficiency and filter complexity," said Bell Labs research scientist Lothar Müller, "because if your design is too complex, you cannot implement it on an integrated circuit."

Decision feedback loops have been around for the last 50 years, Müller said. His research team has implemented a DFL in a high-speed application-specific integrated circuit using aluminum gallium arsenide/gallium arsenide (AlGaAs/GaAs) high-electron-mobility-transfer technology.

An important aspect in compensating for PMD, in Müller's design as well as in others, is to accurately measure it beforehand, because PMD can vary with stress and temperature changes in the environment of the fiber. Another team of Bell Labs scientists is developing a polarization-dependent signal delay (PSD) method for measuring PMD in the time domain, as opposed to the frequency domain in which such measurements are traditionally made.

In common frequency-domain measurements, such as Jones Matrix Eigenanalysis and the Müller Matrix Method, PMD is measured by launching two different polarizations at two different frequencies into the fiber and measuring the polarization states at the receive end. Using the PSD method, however, the Bell Labs research team launched four polarizations into the fiber all at the same frequency, according to Lynn Nelson, a member of the Bell Labs team.

So while traditional methods depend on precise frequency measurements requiring an external-cavity laser at the source and a polarimeter at the receiver, the PSD method can be performed with a distributed-feedback laser at the source and a network analyzer at the receive end. The PSD method also allows measurement of chromatic dispersion independent of PMD, and provides vector measurements that agree with the traditional Müller Matrix Method.

Disadvantages of PSD include the need for four polarization launches instead of two, sensitivity to drifts in fiber length, and inability to measure the fiber rotation matrix, Nelson said. Now that an initial test has proven successful, more work will be required to refine the method and characterize it fully.

About the Author

Hassaun A. Jones-Bey | Senior Editor and Freelance Writer

Hassaun A. Jones-Bey was a senior editor and then freelance writer for Laser Focus World.

Sponsored Recommendations

Hexapod 6-DOF Active Optical Alignment Micro-Robots - Enablers for Advanced Camera Manufacturing

Dec. 18, 2024
Optics and camera manufacturing benefits from the flexibility of 6-Axis hexapod active optical alignment robots and advanced motion control software

Laser Assisted Wafer Slicing with 3DOF Motion Stages

Dec. 18, 2024
Granite-based high-performance 3-DOF air bearing nanopositioning stages provide ultra-high accuracy and reliability in semiconductor & laser processing applications.

Free Space Optical Communication

Dec. 18, 2024
Fast Steering Mirrors (FSM) provide fine steering precision to support the Future of Laser Based Communication with LEO Satellites

White Paper: Improving Photonic Alignment

Dec. 18, 2024
Discover how PI's FMPA Photonic Alignment Technology revolutionized the photonics industry, enabling faster and more economical testing at the wafer level. By reducing alignment...

Voice your opinion!

To join the conversation, and become an exclusive member of Laser Focus World, create an account today!