Fatih V. Celebi

A different approach for the computation of refractive index change in quantum-well diode lasers for different injection levels

A model is developed for the carrier induced refractive-index change in Quantum-well (QW) lasers which affects different mechanisms. The model is based on Artificial Neural Network (ANN) which provides a powerful approach for setting up a complex non-linear model. Different algorithms are tried and tested for different injection current levels. Both the training and the test results for refractive-index change are in very good agreement with the experimental findings reported elsewhere.

Computer-aided design model for a quantum-cascade laser

This study presents a simple, single and an accurate computer-aided design model for quantum-cascade laser based on multi-layer perceptrons. Each critical quantity (optical gain, differential refractive index change, linewidth enhancement factor) that is used in the model requires long mathematical calculations with a strong background knowledge. In addition to that, these quantities use different theories, estimations and assumptions of some parameter values. The model tremendously decreases the computational time in the order of microseconds and is in very good agreement with previously published results.

Optical gain modelling in type I and type II Quantum Cascade Lasers by using Adaptive Neuro-Fuzzy Inference System

In this study, one of the important characteristic of type I and type II Quantum Cascade Lasers (QCLs) that is optical gain which is a function of injection current and wavelength is implemented using Adaptive Neuro-Fuzzy Inference System (ANFIS) architecture. The proposed simple and useful models are in very good agreement when compared with the experimental findings reported in literature.

Semiconductor laser modeling with ANFIS

In this study, neuro-fuzzy (NF) models are used which are well-known robust learning systems that combine the advantages of fuzzy sets and neuro-computation theory. Particularly, a single model based on the adaptive network-based fuzzy inference system (ANFIS) is successfully developed from the amplified spontaneous emission (ASE) spectra of a semiconductor laser diode in order to obtain the critical quantities and their dependences on wavelength and currents. These critical quantities are the differential gain, the induced effective index change and the linewidth enhancement factor (¿ parameter). The comparison of single model results and the experimental measurements validate the presented approach.

Modeling quantum cascade lasers by multilayer perceptrons

This study presents the computer aided design (CAD) of type-I quantum-cascade lasers (QCLs) based on Artificial Neural Networks (ANNs). QCLs have critical quantities named modal gain, differential refractive index change and the linewidth enhancement factor (LEF, ¿ parameter). Each of these quantities requires lengthy mathematical calculations using different theories and assumptions. The single model is based Multi-layer Perceptrons (MLPs) approach which decreases the computational time with accurate values. MLPs are trained and tested with different learning algorithms and different network configurations in order to get an accurate model. The results are in very good agreement with the previously published results.

Gain and noise figure enhancements of both C and L bands double pass Erbium Doped Fiber Amplifier

In this study, designed and optimized C and L band Erbium Doped Fiber Amplifiers (EDFA) around 70 nm bandwidth, 36.53 dB average gain with an average noise figure of 5.8 dB properties are used. For this purpose, 16 channels Wavelength Division Multiplexed (WDM) system with channel spacing of 5.3 nm is taken into account. -30 dBm powers at the input signals (1530 nm-1610 nm) are applied to three port filter and then input signals are separated into C and L bands. Each band signals are separately amplified in double pass configuration. Finally, all signals are combined into a coupler and the overall configuration has wide bandwidth, high gain and low noise.

Alternative intermodulation frequency components

Volterra power series expansion is performed for four tone small signal input laser diode (LD) with optoelectronic feedback. The analysis is carried out for the IMD frequency components that are obtained from the carrier frequency (ω0) which can be alternatively used in sub-carrier systems.

Design of a high precision temperature measurement system

An experimental method is designed and proposed in order to estimate the non-linearity, test and the calibration of a thermocouple using artificial neural network (ANN) based algorithms integrated in a virtual instrument (VI). An ANN and a data acquisition board with signal conditioning unit designed are used for data optimization and to collect experimental data respectively. In both training and testing phases of the ANN, Wavetek 9100 calibration unit is used to obtain the experimental data. After the successful training completion of the ANN, it is used as a neural linearizer to calculate the temperature from the thermocouple’s output voltage.

Design of a chaotic optical communication system by using RAMAN with noise addition technique

In this study. optical carrier and data, which is coining from the 1550 am. CW laser-diode is modulated by using an external optical modulator. The noise that is obtained from the Gaussian white noise source is added to the modulated signal with the help of optical combiners. This combined signal is inputted to the Rarnan Amplifier and the data is transmitted in a 80 kin. fiber optic transmission system. The system is implemented by using chaotic communication tecirnique and the chaotic signal is obtained out of laser diode. Non-return to zero (NRZ) network system is selected in the design.

A novel hybrid method for determining the depth of anesthesia level: Combining ReliefF feature selection and random forest algorithm (ReliefF+RF)

Depth of anesthesia is a matter of great importance in surgery. Determination of depth of anesthesia is a time consuming and difficult task carried out by experts. This study aims to decide a method that can classify EEG data automatically with a high accuracy and, so will help the experts for determination process. This study consists of three stages: feature extraction of EEG signals, feature selection, and classification. In the feature extraction stage, 41 feature parameters are obtained. Feature selection stage is important to eliminate redundant attributes and improve prediction accuracy and performance in terms of computational time. Effective feature selection algorithms such as minimum redundancy maximum relevance (mRMR); ReliefF; and Sequential Forward Selection (SFS) are preferred at the feature selection stage to select a set of features which best represent EEG signals. These obtained features are used as input parameters of the classification algorithms. At the classification stage, six different classification algorithms such as random forest (RF); feed-forward neural network (FFNN); C4.5 decision tree algorithm (C4.5); support vector machines (SVM); naive bayes; and radial basis function neural network (RBF) are preferred to classify the problem. A comparison is provided between computation times and accuracy rates of these different classification algorithms. The experimental results show that better results according to other classifiers when the obtained attributes by ReliefF algorithm are used with RF classifier.