Design of Deep Learning-Based Intelligent Blind Direct Sequence Code Division Multiple (DS-CDMA) Access Receiver

Direct Sequence Code Division Multiple Access (DS-CDMA) is a multiple access method in which multiple users can simultaneously use the same frequency band. An innovative intelligent receiver has been developed to autonomously demodulate DS-CDMA signals. This receiver utilizes deep learning techniques, specifically a convolutional neural network (CNN), operating in a regression mode. The receiver being discussed has the capability to discern the transmitted data, even in situations where the received signal encounters substantial distortion due to factors like channel noise, near-far effects, and Rayleigh fading. The CNN training process indirectly incorporates information about the channel state, effectively removing the need for traditional channel state estimation methods involving pilot signals or training sequences. The experimental findings demonstrate that the receiver we’ve introduced outperforms model-based DS-CDMA demodulators when it comes to Bit Error Rate (BER).