Informer-based QoS prediction for V2X communication: A method with verification using reality field test data

Vehicle-to-everything (V2X) communication plays a critical role in connected and automated driving applications, which requires strict Quality of Service (QoS) performance in terms of delay and reliability since human safety is involved. The QoS of V2X communications is impacted by various factors, such as radio interference, mobility, and user equipment (UE) density, making accurate prediction challenging. Various methods have been exploited to predict QoS deterioration. However, they usually require high computational complexity or may not capture the complex features of V2X communication environment. In this paper, a novel Informer-based QoS prediction model with a causal convolution self-attention mechanism is presented, which achieves high prediction performance with minimal computing resources. Notably, the test data used in the evaluation of the model is collected from the National Intelligent Vehicle and Intelligent Transportation Demonstration Zone in Daxing District, Beijing. The experimental results demonstrate that the proposed method exhibits superior accuracy and stability compared to existing methods such as Back Propagation (BP), ELMAN, LSTM, and CNN.

Automated summary

Vehicle-to-everything (V2X) communication is crucial for connected and automated driving, requiring high-quality service to ensure human safety. However, accurately predicting the quality of service is challenging. This paper presents a novel Informer-based QoS prediction model that achieves high prediction performance with minimal computing resources, demonstrating superior accuracy and stability in experiments.