Optical channel de-aggregator of 30-Gbaud QPSK and 20-Gbaud 8-PSK data using mapping onto constellation axes

Abstract

We experimentally de-aggregate the 8-PSK signal (EVM 8.7%) onto two 4-PAM signals (EVM 8.8%). QPSK signals are demultiplexed into two BPSK signals with EVMs ~11.4 %. Deaggregation performance as a function of the OSNR of the incoming signals is evaluated. The effect of phase noise is also studied. Introduction Aggregation of lower-capacity channels into a single higher-capacity channel and deaggregation of one higher-capacity channel into many lower-capacity channels are common functions in generic communication systems. This is true because often single users do not need the full bandwidth available in a highcapacity portion of a network. For a highcapacity optical network, there might be a desire to achieve aggregation and de-aggregation in the optical domain to potentially: (i) avoid inefficient optical-to-electrical conversions, (ii) enable the possibility of higher speed, and (iii) achieve linear transformations over a wide dynamic range. Given the present importance of phase-based data modulation formats, e.g., quadrature-phaseshift-keying (QPSK) or multi-level phase-shiftkeying (M-PSK), it would be valuable to demonstrate these two functions for phaseencoded data channels. In general, optical aggregation has been demonstrated for higherorder phase-based modulation formats. There have been reports of optical deaggregation of QPSK signals. However, in general, these approaches have required the use of a feedback loop to stabilize the phase within the de-aggregator. A laudable goal would be to demonstrate a technique for the optical deaggregation of QPSK and higher-order PSK into multiple data channels of lower capacity. In this paper, we demonstrate optical channel de-aggregator of 30-Gbaud QPSK and 20-Gbaud 8-PSK data using mapping onto constellation axes. In order to map the signal onto the axes, we add the signal with its conjugate coherently. To keep the coherency without a feedback loop, we implement the de-aggregation on the differentiated version of the signal. In other words, first in a nonlinear element, we generate a copy and two conjugate copies and then in a programmable phase/amplitude filter, one symbol time delay is induced on the signal and its copy. Finally, in another nonlinear element the differentiated signal could be added coherently with its conjugate which could provide the mapping concept. A 30-Gbaud QPSK signal with error vector magnitude (EVM) of 9.9% is deaggregated onto in-phase (I) and quadraturephase (Q) BPSK signals with EVMs of ~11.4 %. A 20 Gbaud 8-PSK data with EVM of 8.7% is also mapped onto Iand Q-parts (4-pulse amplitude modulation (PAM)) with EVMs of ~8.8%. We could also implement the de-aggregation of 20Gbaud QPSK signal to show the tunablity over the bit-rate. Bit error rate (BER) measurements are also shown. To study the de-aggregator performance, we change the input optical signal to noise ratio (OSNR) and measure the output OSNR and EVM. Also, we induce phase-noise on the input signal and at the output, the phasenoise could be squeezed on the mapping axes. The conceptual block diagram of optical channel de-aggregator is shown in Fig. 1. In order to de-aggregate the QPSK/8-PSK signals, we use mapping concept onto constellation axes. In other words, if we could add the signal, e , with its conjugate, i.e., e j + e -j , it would provide us the I-component of the signal. Also, to achieve the Qcomponent of the signal, we would need to map the constellation onto Q-axes, i.e., Q=e j e –j (as shown in Fig. 1(a), (b)). Figure 1(c) shows the block diagram of mapping implementation. At first, the input signal (QPSK/ 8-PSK) with two continuous wave (CW) pumps, i.e., pump-1 and pump-2, are sent through a highly nonlinear fiber (HNLF) to generate a copy and two conjugate copies of the signal in a four wave mixing