Enhanced Semi-persistent scheduling (e-SPS) for Aperiodic Traffic in NR-V2X

Abstract

In cellular vehicle-to-everything (C-V2X) mode 4 and New Radio V2X (NR-V2X) mode 2 based on local observations resources are scheduled by the vehicles themselves. For resource scheduling operation third generation partnership project (3GPP) defined semi-persistent scheduling (SPS). Vehicles rely on the sensing information received in sidelink control information (SCI) over physical sidelink control channel (PSCCH). Based on the sensing information vehicle select the resources for its transmission and reserve the resources for its successive future transmissions. For periodic transmission, SPS works fine comparatively to aperiodic messages. Because aperiodic messages compelled the vehicle to select new resources for its transmission based on the latency associated with the generated packet. In turn, it results in unutilized resources which were reserved before. This would also increase resource contention. To overcome this, we have proposed the enhanced semi-persistent scheduling (e-SPS) method for resource reservation for aperiodic traffic. The proposed scheme utilizes the reinforcement learning mechanism where each vehicle act as an agent. Based on the traffic density and speed of the vehicle, the size of the sensing window is dynamically adjusted and re-evaluation mechanism is also introduced to confirm the available resources by performing the sensing again while selecting the resources. The performance of the proposed scheme is evaluated in ns-3 and compared with the naïve sensing mechanism. Results show that the e-SPS scheme outperforms the others.