الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 185 |  |  |
| | | from | 185 | | |
Abstract
One of the primary interests of the optical communication community is
the achievement of greater data transmission capacity, which has led to the
modification of the different physical light properties including amplitude,
phase, wavelength, and polarization for data encoding and channel addressing. Multiplexing of multiple independent data channels is a common method
for enhancing the transmission capacity of optical communication systems.
The implementation of orthogonal spatially overlapping and co-propagating
spatial modes, known as Mode-Division Multiplexing (MDM), is a particular
case of Space-Division Multiplexing (SDM). Each mode can hold an independent data channel in such a scenario, and orthogonality allow multiple modes
for efficient de/multiplexing and low inter-modal crosstalk. One of the most
potential candidates for MDM systems is the Orbital Angular Momentum
(OAM) due to its ability to improve the performance of optical communication systems. Advantages arise from the circular symmetry of the OAM
modes relative to other MDM methods, making OAM modes well suited for
several optical technologies.
In this thesis, a novel chaotic-interleaver is used with Low-Density ParityCheck (LDPC) coded OAM-shift keying through Atmospheric Turbulence
(AT) channel. Moreover, a Convolution Neural Network (CNN) is used as
an adaptive demodulator to enhance the performance of the wireless optical
system. The viability of the proposed system is verified by convoying a digital image in the presence of distinctive turbulence conditions with different
codes. The proposed CNN is chosen with the optimal parameter and hyperparameter values that yield the highest accuracy, the utmost Mean Average
Precision (MAP), and the largest value of Area Under Curve (AUC) for the
different optimizers. The simulation results affirm that the proposed system can achieve better peak Signal-to-Noise-Ratios (SNRs) and lower Mean
Square Error (MSE) values in the presence of different atmospheric turbulence conditions, when the CNN classification capability is restricted. By
computing accuracy, MAP, and AUC of the proposed system, we realize that
the Stochastic Gradient Descent with Momentum (SGDM) and the Adaptive
Moment Estimation (ADAM) optimizers have better performance compared
to the Root Mean Square Propagation (RMSProp) optimizer in terms of accuracy by about 3:8% .After that, we suggest 3D chaotic interleaving for coded 3D video frames
with dissimilar spatial and temporal features transmitted via a variety of Nary Orbital Angular Momentum Shift-Keying Free-Space Optical (N-OAMSK-FSO) communication system. The LDPC-coded encrypted video frames
have the highest Peak Signal-to-Noise Ratio (PSNR) and the lowest Bit Error
Rate (BER) through N-OAM-SK-FSO model. Due to the defects of conventional OAM-SK detection mechanism, two efficient Deep Learning (DL)
techniques, namely Recurrent Neural Network (RNN) and 3D-CNN are used
to decode the OAM modes with a lower error rate in the presence of extreme
atmospheric turbulence. The simulation results imply that both techniques
have nearly the same classification and prediction performance through NOAM-SK-FSO model, but this performance is deteriorated in case of larger
dataset classes. Moreover, Graphics Processing Unit (GPU) accelerates the
classification performance by almost 67.64% and 36.93% using RNN and 3D
CNN techniques, respectively. The two applied DL techniques are approximately more efficient than other conventional classification techniques by
almost 18%.
Furthermore, this thesis presents a hybrid multi-state Orbital Angular Momentum-Multi-Pulse-Position Modulation (NOAM-MPPM) technique
over gamma-gamma Free-Space Optical (ΓΓ-FSO) channel and analyzes its
performance. Both atmospheric and Pointing Error (PE) effects are taken
into account in our analysis. In addition, approximate-tight upper bounds on
the BERs of both NOAM and NOAM-MPPM techniques are developed, considering the influences of beam divergence and PE. The ΓΓ-FSO-PE channel
parameters and the BER expressions are evaluated numerically and verified
by simulation. It turned out that the analytical results are nearly the same
as those obtained from simulation under different turbulence conditions and
OAM modes. The results demonstrate that under variable turbulence conditions, the NOAM-MPPM technique outperforms both ordinary NOAM and
MPPM techniques. Finally, different DL techniques, namely Random-Forest
(RF), CNN, and Auto-Encoder (AE), are employed to get the optimum classification accuracies on different datasets with NOAM-MPPM-ΓΓ-PE model.
Our results indicate that AE has the best performance metrics of DL compared to other models on different datasets.
The thesis also presents a novel bit-level OAM video frame encryption algorithm that is dependent on the Piecewise Linear Chaotic Maps (PWLCM)
for transmission through different turbulence conditions. Firstly, the mathematical model for the BER of OAM is derived employing the ΓΓ turbulencechannel. After that, a comparison between the theoretical results from Mathematica and the simulation results from MATLAB under different turbulence
strengths, SNRs, and propagation distances is presented to assure that there
is a perfect match between both models. The proposed OAM video cryptosystem is checked via various security key indicators such as entropy analysis,
histogram testing, attack analysis, time analysis, correlation testing, differential analysis, and other quality and security evaluation metrics. The simulation results and the performance analysis confirm that the proposed algorithm
is reliable and secure for OAM video frame encryption and communication
under different turbulence conditions in the FSO communication systems.