Abstract
We propose a novel two-stage blind carrier phase recovery scheme tailored for probabilistically shaped QAM. Our algorithm is based on modifying the classic Viterbi&Viterbi (V&V) and maximum likelihood (ML) phase estimators by optimizing the blind decisions over the received symbols taking into account the probability distribution of the received signal. Our technique improves the standard V&V + ML in all conditions and outperforms the state-of-the-art blind phase search (BPS) algorithm for a constellation entropy ≤4. At the same time our implementation preserves the low complexity of standard V&V + ML. This results in a remarkable reduction in complexity of up to a factor thirty compared to BPS.
Highlights
P ROBABILISTIC shaping (PS) has gained increasing popularity in the last years, proving to provide superior performance for coherent transmission over an additive white Gaussian noise (AWGN) channel while enhancing at the same time the transmission rate flexibility
The theoretical gain predicted with respect to unshaped quadrature amplitude modulation (QAM) is threatened by additional penalties arising from the digital signal processing (DSP) chain, whose adaptation to the peculiar symbol probability distributions of PS constellations is still an under-investigated topic [1]
We presented a novel two-stage blind carrier phase recovery (CPR) implementation based on Viterbi and Viterbi 4th power (V&V) and maximum likelihood (ML) phase estimation tailored over probabilistically shaped QAM
Summary
P ROBABILISTIC shaping (PS) has gained increasing popularity in the last years, proving to provide superior performance for coherent transmission over an additive white Gaussian noise (AWGN) channel while enhancing at the same time the transmission rate flexibility. The practical solutions implemented at the moment are (i) to use modulation-independent heavily pilot-based DSP [2] or (ii) to apply standard blind algorithms developed for unshaped M-QAM. The former choice achieves the best flexibility but at the cost of a consistent reduction in the throughput due to the high pilot overhead (OH). Still considerably lower complexity solutions based on the simple Viterbi and Viterbi 4th power (V&V) and maximum likelihood phase estimators (ML) have been successfully implemented for unshaped modulation up to 16-QAM [8] These schemes have shown reasonable performance when applied to 64-QAM [9]. We test our algorithm with variable constellation entropy, ranging from 2.66 to 4.5, and observe that our method outperforms BPS for entropy ≤4 and standard non PS-aware V&V + ML in all conditions, with negligible additional complexity needed
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
