Per-Wavelength Compensation Schemes for CD

The bandwidth increase from 10 Gbit/s to 40 Gbit/s has accommodated the demand for greater speed, yet along with this increase come a number of issues. As previously explored in this series, the main issues that arise are the increase in chromatic dispersion (CD) and the decrease in CD tolerance, which can lead to non-linear effects (e.g., four-wave mixing, self-phase and cross-phase modulation, etc.). To deal with these issues, compensation schemes that make up for CD are of the utmost importance when implementing a 40 Gbit/s network.

High-speed networks such as 40 Gbit/s networks carry dense wavelength-division multiplexing (DWDM) signals, in which each wavelength has a different CD coefficient. Therefore, compensating for an average value will result in huge and random over-compensation or, far worse, under-compensation. Since the relation between CD and CD tolerance is so close, the compensation applied must be perfect in order for the signal to reach the receiver in a useable format.

We know that CD tolerance is dependent on numerous factors such as bit rate and channel count, among others. The total all-optical distance and the fiber type dictate the accuracy level of the compensation that is required. In addition, the input power is a main contributor to non-linear effects, so that higher power must be accompanied by controlled, non-zero leftover dispersion. Modulation formats for overcoming CD were discussed in our previous article (CS-RZ, duo-binary and DPSK).

Compensation Schemes

There are two main schemes that can be considered for CD compensation at 40 Gbit/s, and both must ensure per-wavelength tunable compensation. While more complex and expensive than using the negative-fiber technique, these schemes can be tailored to the exact required value per wavelength and can be adapted if the dispersion within the fiber changes (e.g., caused by temperature changes).

Scheme 1: Many Rough Compensators and One Tunable Compensator

The simplest scheme (shown in Figure 2) consists of having a rough compensator at each regeneration site to bring the CD within reasonable values, and then placing a tunable compensator before the receiver. Since the range of accumulated CD varies, the tunable dispersion-compensating module (DCM) plays a critical role and must have a broad tenability range.

The weakness of this scenario comes with the advent of reconfigurable optical add/drop multiplexers (ROADMs). With ROADMs, wavelengths with non-optimized compensation can be re-routed onto other networks, or they can enter your network at various places/times, and with various CDs. This puts even more stress on the tunable modules at the receiving end, provided they can supply the tunable range required (since non-compensated wavelengths could enter the network).

Scheme 2: Tunable Compensators at Each Site

A more robust, yet complex and expensive scenario, is to have tunable compensators at every site, guaranteeing that the dropped and added wavelengths are always within a known CD range.

Negative Pre-Chirp 

It is well known that properly adjusted negative pre-chirp can cancel pulse spreading caused by CD. Negative pre-chirping involves red-shifting the high frequencies and blue-shifting the low ones. This allows the pulse to travel a longer distance before reaching the maximum acceptable level of CD, as shown in the figure below:

The following figure shows the gain in distance by applying negative pre-chirp, as well as over-compensation at each DCM site:

As we have seen in this article, negative pre-chirping as a compensation scheme is a helpful method, yet it does have limitations and some sort of compensation must be applied to the negative pre-chirp.

In the next edition, we will examine the PMD issues related to 40 Gbit/s transmission.