Salman Karimi

Dr. Salman Karimi

orcid

Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
10 April 2021 in Silicon
Verified
Journal Article
Closed access
PDF file
AOAnis Omidniaee
Anis Omidniaee
SKSalman Karimi
Salman Karimi
AFAli Farmani
Ali Farmani

17 December 2022 in Optical and Quantum Electronics
Verified
Journal Article
Closed access
PDF file
YMYounes Mousania
Younes Mousania
SKSalman Karimi
Salman Karimi
AFAli Farmani
Ali Farmani

22 January 2025 in Plasmonics
Verified
Journal Article
Closed access
PDF file
YMYounes Mousania
Younes Mousania
SKSalman Karimi
Salman Karimi
AFAli Farmani
Ali Farmani

14 April 2023 in Optical and Quantum Electronics
Verified
Journal Article
Closed access
PDF file
MHMaryam Heidari
Maryam Heidari
SKSalman Karimi
Salman Karimi

17 May 2024 in Micromachines
Verified
Journal Article
Open access
PDF file
GMGolshan Mohamadpour
Golshan Mohamadpour
SKSalman Karimi
Salman Karimi
Saeed RoshaniSaeed Roshani
Saeed Roshani
In this paper, a 3 dB 90-degree hybrid coupler with size reduction and harmonics rejection was designed. In the proposed coupler structure, four simple low-pass filters (LPFs) were applied. An artificial neural network (ANN) was used to determine the dimensions of the applied LPFs based on EM simulation data. The applied ANN model could also provide the desired LPF parameters, including the cut-off frequency (fc), bandwidth (BW), and insertion loss (IL). Designing an applied LPF involves complex mathematical calculations and simulations to optimize parameters. However, by utilizing neural networks, the design process can be significantly streamlined and automated. Neural networks have the ability to learn complex patterns and relationships within data, making them well suited for optimizing the performance of applied components. The proposed 90-degree hybrid coupler works correctly at 1800 MHz and has a small size of 16.6 mm × 15.15 mm, which provides a 73% size reduction compared to a normal 1800 MHz coupler. The designed coupler not only decreases the circuit size but also provides a wide rejection band from 4.8 GHz to 11.2 GHz, which suppresses the second to sixth harmonics. The insertion loss parameter of this 90-degree hybrid coupler is less than 0.1 dB at the working frequency, which shows the superior performance of the proposed coupler.

1 February 2025 in ARO-THE SCIENTIFIC JOURNAL OF KOYA UNIVERSITY
Verified
Journal Article
Open access
PDF file
GMGolshan Mohamadpour
Golshan Mohamadpour
SKSalman Karimi
Salman Karimi
Saeed RoshaniSaeed Roshani
Saeed Roshani
In this paper, a planar dual-band Wilkinson power divider (DWPD) with a triangular-shaped resonator is designed. This work stands out from existing designs by addressing key limitations in conventional power dividers, i.e., physical size, harmonic suppression, and insertion loss. The proposed triangular shaped resonator has a compact size of 9.9 mm × 3.4 mm (0.26 λg × 0.09 λg ), where λg is electrical wavelength at 5.9 GHz, and provides a wide suppression band from 7.1 GHz to 20.6 GHz with a 20 dB attenuation level. In the proposed DWPD structure, two triangular shaped resonators are used in two branches. It works at 3.6 GHz and 5.5 GHz with <0.1 dB insertion loss at both operating bands. The input and output return losses and ports isolation parameters at both bands are better than 20 dB, which show good performance of the divider at operating bands. Besides the acceptable performance, the proposed DWPD provides a wide suppression band from 6.8 GHz to 20.5 GHz with more than 20dB attenuation level. In the divider design, the neural network is employed to model a triangular-shaped resonator. The proposed neural network has two outputs (S11 and S21), and two hidden layers with eight neurons at each layer. The weights of each neuron are obtained using particle swarm optimization algorithms. The proposed neural network model has accurate results, and the mean relative error of the train and test data for both outputs is <0.1 , which validates the accurate results of the proposed model.