SEE-V2X: C-V2X Direct Communication Dataset:
An Application-Centric Approach

Ruoshen Mo    Bo Wu    Zhaowei Tan    Hang Qiu   

University of California, Riverside   

ACM SenSys 2025

Paper | Code | Dataset | Bibtex

SEE-V2X is a real-world C-V2X dataset, using commercial off-the-shelf standard compliant C-V2X radios. Emulating the traffic patterns of popular C-V2X applications, we investigate the gap between the demand and reality. Beyond throughput and latency, SEE-V2X contains cross-layer details, offers insight into the resource scheduling and allocation mechanism in various situations, and reveals the impact of nuanced configuration.

SEE-V2X Overview

SEE-V2X dataset are collected in two scenarios: Parameters Sweeping and Emulating Applications, with the latter further divided into three sub-scenarios: BSM-only, Perception-only and BSM&Perception. Each scenario has application layer network traces generated by the C-V2X radio, C-V2X MAC layer traces and sidelink control information (SCI) captured by USRP B210. We collected data in both indoor and outdoor environments. Additionally, for the Emulating Applications scenario, data were collected at a busy intersection, and GNSS coordinates were recorded for each experiment vehicle.
Setup
Route

Example Analysis

We present preliminary analysis results based on the collected traces, demonstrating how SEE-V2X provides insights into real-world C-V2X traffic and, more importantly, its implications for emerging futuristic C-V2X applications. Figures 1 and 2 show the throughput and packet loss rate across different packet lengths. Figures 3 and 4 illustrate latency variations under different priority levels. Figures 5 and 6 depict network jitter in two scenarios: one where the packet generation time aligns with the resource scheduling slot and another where it is misaligned.
1.Indoor
indoor throughput
2.ParkingLot
outdoor throughput
3.Indoor
indoor latency
4.ParkingLot
outdoor latency
5.Jitter Align
stable jitter
6.Jitter Misalign
latency
Moreover, beyond network-layer metrics, SEE-V2X utilizes a USRP to capture lower-layer information, specifically MAC layer scheduling data, allowing for a deeper investigation into C-V2X. The following figure were collected in an intersection scenario involving four communication participants. Each participant simultaneously transmits perception data (4,000 bytes every 20 ms) and BSM data (128 bytes every 100 ms). In the figure, the green bars represent the resources allocated for perception data, while the yellow bars indicate the resources used for BSM data.

Citation

 

  @inproceedings{see-v2x,
    title={SEE-V2X: C-V2X Direct Communication Dataset: An Application-Centric Approach},
    author={Ruoshen Mo and Bo Wu and Zhaowei Tan and Hang Qiu},
    booktitle={Proceedings of the 23rd ACM Conference on Embedded Networked Sensor Systems},
    series={SenSys '25},
    year={2025}
  }