On the Use of SVR based Machine Learning Method for Nonlinearities Mitigation in Short Range Fronthaul Links

Hadi, Muhammad Usman; Mittal, Isha

doi:10.1109/csnt51715.2021.9509717

Cited by 7 publications

(2 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For 5G applications, the digital predistortion (DPD) technique is regarded as one of the promising linearization solutions due to ordinary hardware requirements and cost. In the recent past, indirect learning architecture (ILA) based DPD has been widely employed using volterra methods 10–13 and canonical piece‐wise linearization (CPWL) methods 14 . Moreover, the concept of using machine learning (ML) for DPD that reduces the link nonlinearities is a recent addition.…”

Section: Introductionmentioning

confidence: 99%

“…In the recent past, indirect learning architecture (ILA) based DPD has been widely employed using volterra methods [10][11][12][13] and canonical piece-wise linearization (CPWL) methods. 14 Moreover, the concept of using machine learning (ML) for DPD that reduces the link nonlinearities is a recent addition. Use of K-nearest neighbor algorithms, support vector machine 15 and deep learning methods were also employed.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Enhancing5Gmulti‐bandlong haul optical fronthaul links performance by magnitude‐selective affine digital predistortion method

Hadi

Soin

Kausar

2022

Micro & Optical Tech Letters

Self Cite

View full text Add to dashboard Cite

This paper presents a novel magnitude-selective affine (MSA) based digital predistortion (DPD) method for the performance investigation of multiband 5G new radio (NR) based analog radio over fiber link. The proposed MSA-DPD method is derived from the canonical piecewise linear (CPWL) based model by employing MSA functions, which can result in reduction of the number of multiplication operations and the model complexity. The 5G NR standard at 20 GHz with 50 MHz bandwidth and flexible-waveform signal at 3 GHz with 20 MHz bandwidth is used. A dual drive Mach Zehnder modulator having two distinct RF signals modulates a 1310 nm optical carrier using distributed feedback laser for 22 km of standard single mode fiber. The proposed MSA-DPD method is compared with the CPWL and generalized memory polynomial method. The experimental results are presented in terms of adjacent channel power ratio, error vector magnitude, number of estimated coefficients and multiplications suggesting that MSA-DPD method achieves a better performance as compared to CPWL and GMP models with much lesser complexity meeting the 3GPP Release 17 requirements.

show abstract

Section: Introductionmentioning

confidence: 99%