An Efficient Ensemble Deep Learning Model-based Signal Detection in 6G Wireless Communication

In sixth-generation (6G) wireless communication, non-orthogonal multiple access (NOMA), programmable meta-surfaces, and intelligent reflecting surfaces are pivotal technologies for augmenting spectrum efficacy, minimizing latency, facilitating extensive connectivity, and attaining elevated data rates. On the other hand, in wireless communication, signal detection is a fundamental aspect of communication systems, ensuring reliable transmission and reception of data. However, in 6G wireless communication, signal detection is more challenging due to the dynamic wireless environment, high mobility, increased number of devices, etc. To overcome the above limitations, this paper proposes an efficient ensemble deep learning model for signal detection in NOMA for 6G wireless communication. The proposed long short-term memory (LSTM)+1D-convolutional neural network (CNN) model combines the strengths of both LSTM and 1D-CNN models. The effectiveness of the proposed model has been compared with the conventional LSTM model, least squares, minimum mean squared error, and maximum likelihood models. Simulation findings indicated that the proposed model yields a superior symbol error rate compared to existing models and enhanced performance across differences in pilot symbols, cyclic prefixes, batch size, and learning rate.