Polarization-Switched QPSK (PS-QPSK)

Tools Used: OptSim

Increasing traffic demands have pushed boundaries of spectral efficiency by way of moving from binary intensity modulation to multi-symbol phase modulation and using both polarization of the light source. As a result, various ways of encoding information in four-dimensional (4D) constellation have emerged. A 4D constellation is generated when a standard bi-dimensional (2D) format from traditional communication systems are multiplexed over two orthogonal polarizations. Previous application note discussed one of the most popular 4D schemes, namely, polarization-multiplexed QPSK (PM-QPSK), also known as dual-polarization QPSK, or DP-QPSK.

Although not as popular as PM-QPSK, two alternative implementations of 4D constellation were proposed [1-2]. In this class of information encoding, an information bit decides which QPSK symbol gets transmitted over X-polarization or Y-polarization and hence is known as polarization-switched QPSK (PS-QPSK). Because of the hexadecachoron shape of the constellation, these implementations of PS-QPSK are also known as HEXA [3].

In this application note, we demonstrate two implementations of a PS-QPSK transmitter based on Ref [1-2]. The implementation schematics of OptSim are shown in Fig. 1. 



Figure 1. OptSim schematics of a PS-QPSK Transmitter: (a) Based on Ref. 1 (b) Based on Ref. 2

The most of the transmitter structure is the same between PM-QPSK and PS-QPSK. The difference lie in the fact that instead of 4 bits per symbol in PM-QPSK, only 3-bits per symbol are transmitted in PS-QPSK with the third data source in Fig. 1(a) controlling which polarization gets transmitted in each symbol period.

Figure 2 shows encoded constellation, optical signal and spectra at the output of the transmitter.

Figure 2. Encoded constellation (left), optical signal (center) and signal spectrum at the PS-QPSK transmitter 

The basics of the receiver remain the same for both PM-QPSK and PS-QPSK with differences in the off-line DSP. The received channel is mixed with local oscillator at 90-degree hybrid, with balanced photodetection and post-detection electronic dispersion compensation (EDC) stages.

OptSim comes pre-supplied with the receiver DSP of PM-QPSK, but does not come pre-supplied with the DSP of PS-QPSK. The post-EDC electrical signal can be processed offline in MATLAB or external tools implementing the DSP for PS-QPSK. The DSP between the two differ at the constant modulus algorithm (CMA) stage.

PS-QPSK was thought of as a fallback option for a PM-QPSK system where in case of any fault or degradation in PM-QPSK, the system can operate as PS-QPSK at 25% less capacity (because only 3 bits out the 4 of PM-QPSK gets transmitted in a PS_QPSK symbol). Although PS-QPSK can offer marginal improvement in nonlinear inter-channel crosstalk in WDM systems, in practice, it never really became a competitor of PM-QPSK since (i) the latter is 25% more efficient (i.e., can carry more data), (ii) the former needs larger electrical and electro-optical bandwidths (as much as 33% larger) (iii) the former needs faster ADC (i.e. more expenses) (iv) the former makes it hard to fit DWDM wavelengths over a typical 50GHz grid and (v) although the former has a slight sensitivity advantage, considering other factors described, cost vs. benefit hasn’t been favorable for practical deployments.


1.       Martin Sjodin, et al., “Transmission of PM-QPSK and PS-QPSK with different fiber span lengths,” Optics Express, vol. 20, no. 7, March 2012, pg. 7544-7554

2.       Magnus Karlsson and Erik Agrell, “which is the most power-efficient modulation format in optical links?” Optics Express, vol. 17, no. 13, June 2009, pg. 10814-10819

3.       P. Poggiolini, G. Bosco, A. Carena, V. Curri, and F. Forghieri, “Performance evaluation of coherent WDM PS-QPSK (HEXA) accounting for non-linear fiber propagation effects,” Optics Express, vol. 18, no. 11, May 2010, pg. 11360-11371