Scale your signal propagation analysis with CARTO Workflows
.png)
As wireless networks become more complex and cities get smarter, it’s more important than ever to understand how signals travel and behave. Whether you're setting up new cell towers or making sure emergency responders can communicate everywhere in a city, getting the signal patterns right is crucial. It's not just about static coverage anymore - with 5G, connected devices, and smart infrastructure, dynamic and scalable signal analysis has become essential for both telecom companies and city planners.
Traditionally, signal propagation analysis for telecommunications has been performed using legacy, dedicated tools that operate in a silo, disconnected from the broader data and analytics ecosystem deployed at the organizations. Such isolation has been not only limiting the performance of such models, but also their integration with other crucial insights for a more powerful decision-making.
In this blog post, we’ll explore how some of our latest functionality is breaking this silo and walk through an example application to show its real-world impact.
At CARTO, our mission is to break the silo that location data so often lives in, with a cloud-native and intuitive platform for turning spatial data into decision-driving intelligence.
One of the newest features of our platform is Workflow Extensions. These packages extend the capabilities of our low-code analytics tool CARTO Workflows by enabling users to create, distribute and use custom components tailored to their specific spatial analytics needs.
We have now made a new Telco Signal Propagation Models extension package available for CARTO Workflows, tailored specifically for advanced telecom signal propagation analysis. This new addition empowers users to analyze wireless communication networks effortlessly, without requiring extensive coding knowledge. Running natively on Google BigQuery (with support for other data warehouses coming soon), this new extension provides components for line-of-sight analysis and path loss estimation, complementing our existing capabilities for analytics, data processing and data enrichment for the telecommunications industry.
Let’s explore how it works!
The signal propagation extension package for Workflows brings cutting-edge analytical capabilities directly to your data warehouse - here’s a closer look at its key functionalities:
These functions allow you to perform the following:
- Path Loss Estimation: Using the Close In and Extended Hata components, users can model signal strength degradation over distance, incorporating environmental factors such as terrain elevation, urban density, and building height. This allows for precise coverage mapping and network optimization.
- Line-of-Sight (LoS) Analysis: Assess the visibility between a signal source (e.g., a cell tower) and its target locations. Using the Path Profile and Path Profile Raster components, users can identify areas with unobstructed paths to the source, helping you evaluate signal reliability and coverage in dense environments like cities;
CARTO Workflows’ signal propagation extension enhances analysis by integrating geospatial data like population density and mobility patterns, enabling more informed decision-making. Its low-code interface makes advanced telecom analysis accessible to professionals without technical expertise, streamlining the process within an intuitive platform. Built on a cloud-native architecture, the solution runs directly on your data warehouse—currently supported in Google BigQuery—ensuring fast processing and seamless scalability for projects of any size.
Let’s take a closer look at how this extension can help telecom data analysts to optimize the placement of a new 5G cell tower in a dense urban area like London through the workflow below. Note that the full workflow can be found as a workflow template in the CARTO Academy - allowing you to just download, drag and drop directly onto your Workflow canvas.
First, gather your data:
- Elevation data or a digital elevation model (DEM) to account for terrain variations. For this example, we will use the elevation data available in CARTO Spatial Features, however higher resolution data are also available for specific territories. Why not use the Google Earth Engine workflow extension to source the elevation data you need?
- Clutter footprint data to understand surface obstructions. As an example of clutter we will use the buildings profiles in London, available from Overture Maps.
The first step in the workflow is to load these datasets directly into your canvas:
For the terrain data, we had to rename the elevation column to height, as this is the naming convention that the component that computes the path profile will expect.
Before continuing with the analysis, we can visualize the data sources in this map, showing both the transmitters and receivers’ locations (right panel) and the clutter data, including buildings’ height and elevation.
The next step in the analysis is to use the Path Profile component to evaluate which areas have direct visibility to the signal source. The tool calculates whether the terrain or buildings obstruct the line of sight, ensuring you can identify zones with strong signal reliability. This component expects three mandatory inputs:
- A table with the locations (geom) of the transmitters (Tx), including its unique identifier, the height above ground (m), and a buffer representing the radius in meters (m) around the Tx that defines the area served by each transmitter;
- A table with the locations of the receivers (Rx), along with their unique identifier;
- The terrain table.
The component let the user set other additional inputs, such as:
- Clutter data, coming from buildings, vegetation, water bodies etc. These inputs are not connected as sources but rather need to be specified as fully qualified names, as shown below.
- An option to to use the First Fresnel Zone for extracting the obstructing pixels instead of the line connecting the transmitter-receiver pairs.
- When a building clutter dataset is added, an option to return the street width (in meters) at each receiver’s location. Please note that for large datasets this option requires joining large tables and might significantly increase the computing time.
- An option to extract the terrain points along the path and to specify their sampling distance in meters.
- An option to specify the operating frequency of the links in GHz (default value is 2.4 GHz).
When the input data are raster tables, rather than data identified by a geometry column, another component called Path Profile Raster is available. For the latter, all the input data (locations of Tx, Rx, clutter and terrain data) are connected directly as sources, while the names of the bands in the raster tables representing the height and type of the clutter and the terrain height are added as additional parameters.
The output of these components are two tables:
The first table (output_table) contains for each Tx-Rx pair, along with their location and height, the geometry of the line connecting the pair and of the First Fresnel Zone (if applicable), as well as other details on the pair, including:
- The street width
- Whether this link is clear of obstacles or not which can be LOS (Line-Of-Sight) or NLOS (Non-Line-Of-Sight);
- The id, type, and ratio of the obstruction of the 1st Fresnel zone of the obstacles between the transmitter and the receiver
- Thecharacteristics of the link
- A flag that specifies whether the receiver is located inside a building, when a buildings clutter is specified as input.
Using this table, we can, for example, plot for each receiver’s location, whether the link with the corresponding transmitter is line-of-sight or not, as shown in this map.
The second table (output_table_details) contains for each Tx-Rx pair the geometry of the obstacles along with its type and height. For example, the map below shows for the selected transmitter-receiver pair all the buildings obstructing the signal as well as the points used to sample the terrain data:
As we can see from the map above, in a dense city like London, tall buildings might block the signal in critical areas, prompting a review of the tower height or its placement.
Next, we use the Path Profile component to model how signal strength decreases over distance. For this analysis different models might be used, depending on the specific use case. Here, we will use the Close In model, which describes large-scale propagation path loss over distance at all relevant frequencies in a certain outdoor scenario. Users have also access to the Extended Hata model. For different or additional losses the user can write or call their own custom function using the Custom SQL Select and Call Procedure components available in CARTO Workflows.
The selected model takes as inputs the distance in meters between a transmitter and a receiver, the frequency in GHz and the type of scenario (urban macrocell, urban microcell street canyon, and urban microcell open square) and the model parameters (the pathloss exponent, the standard deviation for LOS and NLOS, and the number of samples to be drawn), with default values considered for each scenario.
The final step of the analysis involves processing the output in order to prepare the data to build an interactive dashboard to communicate the results. To do that, we first join the output of the Close In component with the input data to assign to each receiver its geometry. In this way we will be able to visualize the path loss at each receiver location. Secondly, from the Path Profile component, we also save the projected-to-the-ground Fresnel zone between a selected receiver and its corresponding transmitter as well as the details table, which reports the characteristics of each obstacle. As illustrated below, the final output is a detailed and interactive map that highlights optimal infrastructure placements.
As we can see from the interactive map, the model reveals a valley area right below the tower location, where the signal is significantly attenuated. Moreover, the map highlights how the path loss exhibits a non-isotropic pattern, being primarily influenced by the presence of buildings, which heavily impact signal propagation in urban-dense areas. These insights could be used to inform decisions to improve coverage by including additional infrastructure or adjusting the tower's power output.
As we have discussed in this blog post, traditional signal processing techniques often involve complex software with steep learning curves, high computational costs for large-scale analyses, and difficulty integrating telecom data with other spatial datasets. With CARTO you can now perform signal propagation analysis directly on your data warehouse, and with a low-code visual tool such as CARTO Workflows. If you are ready to give it a try, start your 14-day free trial today!.
