The increasing availability of multimodal data holds many promises for developments in millimeter-wave (mmWave) multiple-antenna systems by harnessing the potential for enhanced situational awareness. Specifically, inclusion of non-RF modalities to complement RF-only data in communications-related decisions like beam selection may speed up decision making in situations where an exhaustive search, spanning all candidate options, is required by the standard. However, to accelerate research in this topic, there is a need to collect real-world datasets in a principled manner.


Dataset: IQ samples of LTE, 5G NR, WiFi, ITS-G5, and C-V2X PC5

Thes dataset comprises IQ samples captured from ITSG-5, C-V2X PC5, WiFi, LTE, 5G NR and Noise. Six different dataset bunches are collected at sampling rates of 1, 5, 10, 15 , 20, and 25 Msps. In each dataset cluster, 7500 examples are collected from each considered technology. The dataset size at each considered sampling rate is 7500 X M, where M can be 44, 220, 440, 660, 880, and 1100 for a sampling rate of 1, 5, 10, 15 , 20, and 25 Msps,respectively.



This data set provides complex-baseband samples of vehicle-to-vehicle communication (V2V) radios in the presence of primitive jamming signals. Up to 600 vehicles are simulated in this dataset using commerical IEEE 802.11p radios. Supporting example code found on CodeOcean can be used for developing novel physical layer based jamming detection and mitigation strategies using machine learning for ad hoc vehicular networks.


The dataset was collected by performing uplink/downlink throughput measurements in Munich, Germany. The user side device was a vehicle with roof-mounted antenna (approx. 1.5 m height), and on the network side was a base station antenna mounted at the top of a building (height of the antenna with respect to the ground: 21 m). The measurements were collected at the center frequency of 3.41 GHz, with 40 MHz of bandwidth, with antenna gain of 15.5dBi (5dBi) at the base station (vehicle) side. The maximum throughput in the uplink was 40 Mbps.


The simulation code for the paper:

"AoI-Aware Resource Allocation for Platoon-Based C-V2X Networks via Multi-Agent Multi-Task Reinforcement Learning"


The overall architecture of the proposed MARL framework is shown in the figure.


Modified MADDPG: This algorithm trains two critics (different from legacy MADDPG) with the following functionalities: