الفهرس 
AbstractOnly 14 pages are availabe for public view
 |  | from | 125 |  |  |
| | | from | 125 | | |
Abstract
Modern communication systems require the existence of new tasks such as adaptive modulation. Adaptive modulation is intended to vary the modulation type according to channel state. If adaptive modulation is implemented, we are encountered with either overhead bits to introduce the modulation type to the receiver or blind modulation classification schemes to be implemented at the receiver. The first scenario is a waste of the communication resources and a burden on the communication system as redundant bits are sent with each modulation change process. The second scenario is the appropriate alternative. In this scenario, the modulation type needs to be identified at the receiver prior to demodulation. This thesis is concerned with the adaptive modulation classification task. We use AlexNet, VGG-16, VGG-19 and ResNet CNN to execute the classification process. This classifier achieves high performance for detecting the true type of the modulation and its order applied at the transmitter. This approach is applied over different fading channels. The proposed solution for automatic modulation classification depends on constellation diagram estimation. The constellation diagram images, their Radon, curvelet, wavelet, and phase congruency transform representations are considered in this thesis. The classification performances of different suggested scenarios are investigated and compared. In addition, the Optical Scanning Holography (OSH) algorithm is investigated in this thesis as a tool to randomize constellation diagrams and use a statistical thresholding strategy to detect the modulation type.