In this paper, a Cognitive Radio Network (CRN) based on artificial intelligence is proposed to automatically classify signal modulation more efficiently, which can further help in radio frequency modeling and pattern recognition problem solving. Three different approaches Deep Neural Network (DNN), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN) have been deployed and evaluated in the signal modulation classification. In this paper, the signals for training and validation are generated using our MATLAB based RF signal generator, which can simulate various types of modulated signal according to the configuration specification. The numerical results show that CNN network can outperform the DNN and RNN in terms of the signal modulation classification accuracy.
Proc. SPIE. 11017, Sensors and Systems for Space Applications XII
KEYWORDS: Mathematical modeling, Unmanned aerial vehicles, Data modeling, Satellites, Data transmission, Computer simulations, Telecommunications, Antennas, Satellite communications, Global Positioning System
With the explosive growth of network communication technologies, nowadays global positioning system (GPS) successfully provides the worldwide navigation service for militaries and civilians. Sailors, aviators, and car drivers rely heavily on the accuracy of navigation and position estimation provided by GPS. Instead of using conventional radio frequency (RF) crosslinks, the reliability and efficiency of information transmissions can be significantly enhanced with the aid of optical crosslinks between satellites. In this paper, we develop a toolchain-based hybrid implementation (TBHI) by designing and integrating multiple platforms and software to evaluate the performance of the next-generation global position system (GPS) with potential configuration of optical transmission crosslinks between satellites. A distributed, multi-simulation tool chain is developed in both the front-end and back-end to conduct a real-time evaluation of optical crosslinks. A comprehensive assessment is provided for the receiving or transmission chain, along with traffic loading evaluation of satellite crosslinks from both physical layer and network layer emulation. To further evaluate the effectiveness of our developed TBHI, we investigate five base-line traffic models which can cover most applications in the satellite communication, including single-time transmission, periodical transmission, regular data transmission with randomness, and small data transmission. For each model, we specify a group of parameters that can determine the statistical distribution used to generate the traffic loading. Experiments using real-world traffic traces are used to evaluate the effectiveness of our proposed TBHI framework. Our simulation validates that it can effectively and accurately visualize the GPS satellite communications.
Due to the increasing demand on bandwidth, and the demonstrated success of wireless optical communications in providing broadband applications, interest has grown into further expanding the deployment of wireless optical transmission. Lasers used in free space optical communication (FSOC) operate in optical bands that are not regulated, which shows an enormous advantage in terms of bandwidth. However, laser signals fading tends to occur due to atmospheric turbulence and many other environmental factors. Current methods for representation of laser propagation mainly focus on straightforward statistical models, where their parametrization has to be carried out from experimental data. The existing empirical models are typically obtained by using data collected by laser sensors. These sensors detect photons of light, which are capable of recording intensities of the laser beam at a certain rate. However, simple, common distributions, in some instances, cannot fully describe the dynamic of the received optical signals, especially in the battlefield scenarios that involve various terrain and weather conditions. They lack the generality and rigor of a basic physical-level formulation, i.e., a model specific for one application or scenario cannot be applied to any other case. To overcome the shortcoming of the aforementioned statistical models, a physics-based FSOC propagation is proposed to simulate and represent multipath effects properly and efficiently. In this paper, we consider a large number of factors that may affect the actual FSOC measurement including absorption, scattering, impact of weather, geometric loss, and optical turbulence, etc. The simulation results demonstrate that our proposed FSOC propagation model achieves high-fidelity prediction accuracy.