HB-scenario
Map parameters
HB-scenario considers a model of part of the city of Vienna in 🇦🇹. Specifically, the region of interest (ROI) is near the Vienna Central Station (Wien Hbf), the railway track of the main 🚆 operator in 🇦🇹, ÖBB. It is the direction towards Meidling.
Map parameters
ROI OSM coordinate system WGS-84:
- Max. latitude = 48.1843
- Min. latitude = 48.1716
- Min. longitude = 16.3338
- Max. longitude = 16.3702
The dataset is based on the map shown below for $r$-th transmitter location $\mathbf{u} _{r}$ and a $M=8$ RRHs placed $\mathbf{b} _{m}$, where $m \in \lbrace 1, \ldots, M \rbrace$.
3D models
Below, is an example of the 3D model of HB-scenario. In addition to buildings and available objects from the OSM, this scenario contains other objects, such as two trains passing by that do change their position for every time-snapshot.
Scenario parameters
HB-scenario 3D models:
- $T$ = 200.
- $2\times$ trains
- $\approx 40 - 70$m long
- $\approx 4$m height
- $\approx 3.25$m width
3D models are provided in STL format and can be opened by any CAD tool. State of the art wireless ray-tracing tools should support this format. These files can be downloaded from our owncloud.
Download the dataset
Scenario files are all in .stl format. The dataset is compressed into a .zip file. Depending the scenario and the details of the environment, a size of each file can be 1 up to 5 MB.
Ray-traces (RT)
All path parameters values between the BS and user location are obtained using the ray-tracer available in Matlab®. The dataset contains traces for $T=200$ and a single-BS.
RT parameters
HB-scenario ray traces:
- SBR ray-tracing approach.
- Low-angular resolution with $\approx 655,362$ rays.
- Receiver sensitivity $\geq \text{-}130\text{dBm}$.
- $T$ = 200.
- $R$ = 406.
- $M = 8$
- $b_{m,3} = 20 \textrm{ m}$.
- $u_{m,3} = 1.5 \textrm{ m}$.
- $f _{c} = 3.5 \textrm{ GHz}$.
- EM properties of the scattering objects change over the time
- $\kappa \in{\text {concrete, brick, metal, wood}}$.
- Rainy periods $\mathcal{R}$ with $\mathbb{P}(\mathcal{R}) = 1/3.$
All obtained ray traces are provided in .mat file format and can be opened using Matlab or Python. The files can be from our owncloud.
Download the dataset
There is $200$ RT files in .mat format. There $81\,200$ samples. The size of this structured dataset is low, thus can be easily fit to memory of any modern computer. The total size is $\approx676\textrm{MB}$. Download the dataset in a compressed .zip file.
Channel state information (CSI)
To obtain the channel from the RT, we consider a geometric channel model. CSI is stored as $\mathbf{H} _{r}^{t} =\left[\mathbf{\widehat{h}} _{k _1}^{t}, \mathbf{\widehat{h}} _{k _2}^{t}, \ldots, \mathbf{\widehat{h}} _{k _{N _{c}’}}^{t}\right] \in \mathbb{C}^{N _{r} \times N _{c}’} .$ Dataset has channel and location pairs.
CSI parameters
HB-scenario CSI:
- $M = 8$.
- $N_{c} = 512$, $N_{c}’ = 32$ and $N_{c} \equiv N_{c}’ \quad(\bmod ; 16)$.
- $L = 4$.
- $B = 20 \textrm{ MHz}$.
- Estimation subcarriers
- $\widehat{\mathbf{h}} _{k}^{t} = \sum _{\ell=1}^{L} \eta _{\ell}^{t} e^{j{2 \pi k}{\Delta f} \tau _{\ell}^{t}} \mathbf{a}\left(\varphi _{a z,\ell}^{t}, \varphi _{e l,\ell}^{t} \right) . $
Obtained CSI are provided in .mat file format and can be opened using Matlab or Python. The files can be from our owncloud.
Download the dataset
There is only one CSI file in .mat format. The size of the structure is relatively low, thus can be easily fit to memory of any modern computer. The total size is $\approx 4.27 \textrm{GB}$. Download the dataset in a compressed .zip file.
Showcase
As the 5G mobile services start penetrating worldwide, in many B5G / 6G visions disclosed, we can find Digital-Twins, Society 5.0, etc., where wireless technologies take one significant role as the social infrastructure.
We believe that simulating the real-life conditions will be crucial for the future of wireless technologies to understand how the surrounding dynamics impact the 5G and beyond systems. A real-time virtual city-scale digital twins will be a new norm for analyzing and designing future mobile communications systems.
In addition to scientific methododology for evaluating the performance, we transfer the knowledge to more realistic environment to showcase our work for the industry partners.
MPC visualization
An example of the communication of a transmitter along the railway track is shown below. We are looking at the PDP of the transmitter along the track. Moreover, we can observe the expected driving route and estimated locations.
When our DNN is not sure of a prediciton, it associates a high uncertainty to its estimate. This is an older version of the work. Recent work has a different part of the track.
- MPCs
- Actual map
- PDP profile
- Location estimates
- Uncertainty estimates
Mobility and time-varying environment
The dataset with all 3D models is a time-varying environment. Buildings, cars, and other object change their position every time snapshot. A train moves parallel to the track considered for the transmitter. Another object moving parallel to the train is introduced to stretch the impact of multipath components.
___