Deep Neural Network-Based Channel Estimation Model for Vehicular Communication System

Abstract

Wireless communication plays a pivotal role in intelligent transportation systems, with vehicular communication being a crucial application, particularly with the IEEE 802.11p standard. However, the time-varying nature of vehicular communication channels, characterized by signal propagation in multiple directions and Doppler frequency shifts, poses challenges, resulting in signal strength decline. Reliable and accurate channel estimation is crucial for achieving high system performance. To address these issues, we propose the STA- LRDNN channel estimation model, which offers enhanced reliability, improved accuracy, and remarkable adaptability to diverse channel conditions. The model that integrates the Leakey ReLU-based deep neural network (DNN) architecture with the spectral time averaging (STA) channel estimation technique is proposed for significant improvement including proper hyper- parameter tuning held to get actual Leakey points. The STA-LRDNN model improve channel estimation than conventional methods in vehicle to vehicle (VTV) in express way and urban canyon environment channel models. In high mobility scenarios, it achieves a 20dB SNR gain for 10-2 a bit error rate (BER). In same manner, the minimum value of 10-5 at 16QAM and 10- 3 at 64QAM are recorded at SNR 30dB. Notably, the STA-LRDNN model exhibits remarkable results in low mobility scenarios as well. This integration plays a crucial role in robust and efficient channel estimation for vehicular communication systems.