Challenges faced in 5G wireless technologies arise from the lack of radio spectrum bandwidth for envisioned data services. Spectrum reframing and millimeter wave technology are two proposed solutions to meet the growing demand for wider bandwidth and higher data rates. Spectrum reframing, which involves reallocating spectrum licenses, is very complex, while millimeter wave technology is still not fully developed. The potential for shared spectrum in the <6 GHz bands represents a significant opportunity for more efficient spectrum use to meet the increasing need for mobile connectivity.
- Retrieve the RadioML dataset and perform a brief exploratory data analysis to become familiar with the dataset’s content
- Train and evaluate the performance of a Convolutional Neural Network (CNN) for signal modulation classification.
- Investigate the effect of the model’s receptive field (filter size) on the classification accuracy.
- Develop and optimize a deep learning processor for the CNN model, and generate an implementation and timing report for the design.
A simple MATLAB-based application was developed to demonstrate the model's effectiveness in signal modulation classification at different SNR values.