Software Classification and Detection of Communication Signals Using Artificial Neural Networks

Spectrum distribution and channel detection have long been seen as an impending addition to intelligent radios for wireless communications systems with permit-free groups. Standard approaches have been put forth to handle periodic scanning as a signal characterization technique for applications where carrier frequencies and transmission speeds are unclear, despite the fact that it is computationally complex and requires a considerable amount of realization time to implement satisfactorily. Only in situations where the baseband signal is accessible have Artificial Neural Networks (ANN) been used for signal categorization. By combining these processes, a more reliable and efficient classifier might be produced, reducing the need for web-based computation in situations when a large amount of preparation is done separately. In order to test and classify mixed signals of QPSK and MSK modulation under noise, we use a new check-out signal classification method in this study that combines FFT spectral analysis with neural networks. The ANN methodology describes how to provide this method. The findings indicated that the LM function achieved the most favorable results, yielding optimal probability values of Pd at 0.991 and Pfa at 0.005, when using 10 neurons in the hidden layer.