Move Along There!

In a high performance environment, excessive dispersion on the network’s fibre infrastructure can limit performance and operation reliability of high speed transmission systems. An essential parameter that requires testing to ensure optimised performance of these systems is Polarisation Mode Dispersion (PMD).

PMD in fibre-optic links has always been a major concern for service providers to understand whether the transmission system can be upgraded to support higher bit rate signals. Therefore, to verify that a given fibre link is capable of such increase in transmission speed, it is necessary to measure the mean Differential Group Delay (DGD).

It is the random nature of PMD that poses a fundamental limit on the accuracy at which the mean DGD can be determined from a single measurement taken over a finite wavelength range. These limits become especially severe for the small values of mean DGD that are of interest when upgrading links to 2.5, 10 or 40Gbit/s (i.e. a few ps). The uncertainty can be improved by repeating the measurement over a longer time. Thus, the instrument has to be capable of long-term PMD monitoring to allow time averaging of the mean DGD.

JDSU has developed a field-deployable test instrument that employs a non-intrusive method for measuring PMD in a fibre-optic link while the link remains actively in service. The instrument analyses the polarisation states of the transmitted signals and determines the mean DGD of the fibre link from the frequency-dependence of the polarisation variations in each transmitted signal.

A test instrument of this type can not only be used for fibre link qualification, but also for troubleshooting wavelength channels that exhibit excessively high bit-error rates.

Conventional PMD Measurements

PMD of a fibre is usually measured by injecting a dedicated test signal into one end of the link and analysing the resulting polarisation transformation at the other end as a function of optical frequency. The mean DGD of the link is then determined from the average of the instantaneous DGD values measured at various optical frequencies. The most commonly used field PMD analysers use a broadband source at the transmitter end which is analysed at the receiver end.

However, to perform such a measurement, the entire fibre link must be taken out of service and the data transmission either interrupted or rerouted over an alternative link. As a result, this traditional solution is suitable only for ‘dark’ or ‘unlit’ fibre links.

Conventional test methods are difficult if not impossible to apply in modern ROADM networks, when spectral components of test signals can be routed in many directions. It is therefore mandatory to provide a non-intrusive technique for PMD measurements in a link that is in service and capable of measuring discrete DWDM channels (see Figure 1).


Figure 1: Spectral components of test signal may be routed to different locations

While the mean DGD in a fibre link is typically obtained as the average value of the instantaneous DGDs measured at various optical frequencies, it may also be determined from the time-average of the DGD variations at a fixed optical frequency, or from a combined time- and frequency-average. Moreover, as pointed out, it is not even necessary to measure the DGD directly but rather a quantity called effective DGD, or DGDeff.

Effective DGD is defined as the magnitude of the component of the PMD vector in Stokes space that is orthogonal to the launch polarisation state, or State of Polarisation (SOP) vector, of the optical signal.

In fact, DGDeff represents a fairly accurate measurement of the PMD-induced impairments in the signal. The mean DGD is related to the mean value of DGDeff, (averaged over time and/or frequency). Statistical distribution of DGDeff is well understood (Rayleigh Probability Density Function- PDF) and the Mean value is proportional to mean DGD (see Figure 2).


Figure 2: Statistical distribution of DGD Δτeff vs DGDΔτ

Therefore, the mean DGD of a fibre link may be estimated from in-situ measurements of the effective DGD in the transmitted optical signals. In contrast to conventional techniques, this method offers the advantage that the launch polarisation states of the optical signals can be arbitrary and do not have to be controlled or scanned.

Non-Intrusive Measurement

The JDSU I-PMD innovative test solution for measuring the effective DGD and extraction of PMD value in DWDM signals enables characterisation of new DWDM channels in live systems, troubleshooting of an optical path with unexpected high Bit Error Rate (BER), and the upgrade of a DWDM system to higher bit rate.

Figure 3 represents the schematic diagram of the instrument. The optical signals tapped off a fibre-optic link first pass through a scanning polarisation transformer and are subsequently separated by a polarisation beam splitter (PBS) into two orthogonal polarisation components, which we refer to as TE and TM hereafter.


Figure 3: Block diagram of the instrument

The two components are then separately mixed with the output light from a scanning local oscillator laser (LO), which is tuneable over the entire C-band at a rate of more than 100GHz/ms and with sub-GHz accuracy. The coherent beat signals are detected by a pair of balanced photodiodes and, after electrical amplification and low-pass filtering to a few hundred MHz bandwidth, fed into two RF-power detectors, thus yielding two signals, PRF-TE and PRF-TM, which are proportional to the optical powers in the two orthogonal polarisation states within a bandwidth of a few hundred MHz around the frequency of the LO laser. These two signals are then recorded while the local oscillator frequency is tuned across the spectrum of the signal under test. Measurements are repeated for various settings of the polarisation transformer.

Independent Analysis

The In-Service PMD analyser based on coherent detection technique offers sufficient spectral resolution to analyse the frequency-dependent polarisation variations in arbitrarily modulated signals at bit rates between 2.5 and 40Gbit/s.

A field trial was conducted on a 414-km long transmission link carrying a total of 19 conventional 10Gbit/s NRZ-OOK signals at various frequencies in the C-band.

All signals traversed the same fibre spans. In this investigation, the PMD analyser was connected to a monitor tap at the end of the link, where the signal powers were between -27.3 and -24.6dBm. The optical signal-to-noise ratios of the signals were between 17 and 18 dB.

The instrument measured DGDeff automatically 1680 times in each of the 19 WDM signals over a total measurement time of 191 hours, yielding 31920 samples with a mean DGDeff value of 14.8ps. The time interval between consecutive measurements varied between 5 and 30 minutes.

Hence, we estimate that the mean DGD in the fibre link is about 18.84ps, which is in excellent agreement with earlier end-to-end PMD measurements of the same link, yielding 18.56 and 18.57ps, respectively.

Figure 4 displays the variations in the frequency-averaged values over measurement time, which are approximately of the same magnitude as those of the time-averages over frequency and, hence, indicate that the PMD fluctuations in the fibre link are indeed large. The left graph shows the frequency-averaged values as a function of measurement time and the graph on the right the time-averaged data in the 19 WDM channels. The standard deviation of the frequency-averaged samples in the left diagram is 10.4% of the mean value and that of the time-averaged samples in the right diagram is 13.8%.


Figure 4: Examples of live PMD measurements at different bit rates using different modulation formats

JDSU has developed a unique non-intrusive technique for monitoring the effective DGD in modulated signals, which provides a similarly accurate estimate of the mean DGD as a conventional intrusive technique. The I-PMD instrument is based on coherent detection and offers enough sensitivity to measure individual signals tapped from a broadband monitoring port at any network access point. It has been designed for portable field testing and has been proven to address in-service field PMD measurement and monitoring requirements.