Deep learning-based channel quality indicators prediction for vehicular communication

Abstract

Vehicular communication shares essential information for safety and convenience. Vehicular communication must guarantee a high transmission rate with stable communication. In vehicular communication environments, the channel characteristic frequently varies due to the high-speed movement of the vehicles. Understanding the channel conditions is essential to maintain stable communication. We propose an optimal channel quality indicator (CQI) prediction model for expecting channel characteristics. Our prediction model defines the CQI from the received signal strength indication (RSSI) and is applied to the IEEE 802.11p wireless access in vehicular environments (WAVE) standard. The prediction part applies robust long-short term memory (LSTM) network to sequential data. The CQI prediction model is trained and evaluated using vehicular communication data collected by an IEEE 802.11p WAVE device. We compare the prediction performance of the proposed model with the auto-regressive integrated moving average, support vector regression, and multilayer perception models.

Keywords

• V2X
• Channel quality indicator (CQI)
• Deep-learning
• Long–short term memory (LSTM)
• IEEE 802.11p WAVE