Hexagonal Convolutional Neural Network for Noma Rician Channel Estimator using Hexagonal Quadrature Amplitude Modulation

Abstract

NOMA techniques have attracted much attention due to their ability to support massive connectivity, heterogeneous data traffic, and ultra-low latency requirements, making them ideal for next-generation wireless communication networks. In this paper proposed a novel Hexa CNN for NOMA Rician channel estimator using Hexagonal Quadrature Amplitude Modulation for signal detection and channel estimation (Hexa-QAM). Conventional OFDM-NOMA with HQAM is utilized at the transmitter (tx) side as pilot symbols which inserted to OFDM-NOMA signals to employs the channel estimation (CE) and signal detection advantages of OFDM. Three pilot insertion types Comp, Block, and Hexa were used in the proposed model. The proposed Hexa-QAM can detect the symbols for all users without additional operations based on pilot responses and data signals. However, Hexa CNN is used at the receiver to accomplish a joint flexible signal detection. The Hexagonal Quadrature Amplitude Modulation (Hexa CNN) achieves much better error performance than similar detectors. The Energy Efficiency of the proposed Hexa-QAM technique is 17%, 18%, 23.2%, 23% and 28% better than existing techniques. The accuracy of the proposed technique can be as high as 99.95%, while that of traditional models like the OMA-NOMA, MRC, PD-NOMA and UR-NOMA is 84.9%, 87.58%, and 93.91%, respectively.