Thumbnail Image
Kord, Khaled
C-V2X , Deep-Learning , wireless communication
Autonomous vehicles (AVs) are an essential component of future intelligent transportation systems (ITS). AVs promise an increase in driving comfort, greater safety, and improved fuel consumption. These benefits are attributed to AVs ability to leverage vehicle-to-everything (V2X) communication to coordinate driving, e.g., sharing maneuver intention and acquiring updated road maps. Cellular based V2X (C-V2X) had been popularized as one of the leading technologies at the forefront of ITS. C-V2X technology is already capable of supporting a basic set of AVs’ uses cases, e.g. warning messages for collision avoidance. In contrast, support for advanced use cases, e.g. cooperative lane change, is still an obstacle. Advanced AV use cases require a network able to support up to 10ms end-to-end (E2E) latency and 99.99% packet delivery rate. These stringent service requirements pose a challenge for 4G LTE and 5G New Radio (NR). In this work, we build a simulation environment to study the feasibility of 5G NR in supporting advanced use cases in AVs and propose deep learning-based solutions to enhance its performance further. First, we propose deep leaning models to perform uplink channel state information (CSI) prediction in dynamic vehicular environments. The models are based on a combination of convolutional and recurrent neural networks (CNN-RNN), so as to suit the mobility factor in vehicular environments. Next, we propose a novel scheme based on deep learning prediction to enhance the uplink resource allocation process in 5G C-V2X. The proposed scheme enables the base station to predict vehicle maneuvers, subsequently, assign it the required resource in advance without the need for scheduling request and resource granting process. We show that this scheme improves the ability of 5G NR to support cooperative driving requirements. Moreover, we compare to traditional schemes discussing issues that arise from the introduction of prediction models and possible approaches for further enhancements in the future.
External DOI