Enhancing Channel Coding in MC-CDMA Systems Using Deep Learning for Reduced Complexity and Improved Efficiency

The necessity to provide higher data rates and the growing number of connected users have driven the advance of mobile communication systems. The MC-CDMA technology is a next-generation wireless network that is developed by mixing the best features of OFDM and CDMA by taking high spectral efficiency of OFDM and multiple access features of CDMA. However, certain forms of channel coding like the Unpunctured Turbo Trellis Coded Modulation (UTTCM) have their limitations mainly due to high computational complexity and inadequate provision for channel conditions change. Based on these challenges, this paper presents a new technique of utilizing deep learning for channel coding in MC-CDMA systems. Thus, taking into account the learning abilities of neural networks, the problem is solved by the proposed model with less BER and reduced computational complexity, thus improving the adaptability of the communication system. As shown in the simulation results, the proposed deep learning model achieves higher accuracy than UTTCM under different channel conditions, especially in the high SNR scenario. This research shows the possibility of incorporating the Deep Learning into the wireless systems, in order to achieve even higher scalability and reliability of the communications in the 6 G and further networks.