Optimization Scheme for Vehicle Communication Resource Allocation Based on NOMA

At present, most of China adopts Train to Ground (T2G) communication, which requires a large number of trackside equipment to maintain and has the disadvantage of high transmission delay. Train to Train (T2T) communication can achieve lower delay and autonomous safety control of trains. This article proposes a vehicle communication related algorithm based on Non Orthogonal Multiple Access (NOMA) technology, using Convolutional Neural Network (CNN) Long Short Term Memory (LSTM). The CNN is used to simulate the channel state, extract channel features and power allocation factors, simulate channels that are closer to the actual situation, and quickly calculate the optimal power allocation scheme; Use LSTM algorithm to minimize the error between actual output and ideal output. The research results indicate that the proposed scheme has better spectral efficiency than traditional Zero Forcing (ZF) algorithm and Minimum Mean Square Error (MMSE) algorithm. The more training times the CNN-NOMA-LSTM algorithm has, the better the fitting performance, and the smaller the error between the ideal output and the actual output. Moreover, it improves spectral efficiency under the lowest signal-to-noise ratio (SNR) T2T communication requirement. The research results can serve as a reference for future T2T communication.