This is the code repository for the LambdaRoad web-based map tool, which allows the user to view, add and edit stations on an interactive map, run path loss calculations from these stations, and visualize the results. The goal of this tool is to demonstrate the path loss models, and allow easy experimentation and testing. The rest of this file documents how the tool can be used, how it works internally, how it can be run offline, and partially how the source code can be understood and modified.
Two path loss models were created in the LambdaRoad project, both based on signal strength data collected in Norway: one for the mobile network (800MHz), and one for ITS G5. Both models have been implemented in the command line application that acts as a backend for the web-based tool. More information about the data collection can be found in the LambdaRoad WP1 Report.
New path loss models (or variations of the existing ones, such as mobile network models for different frequencies) can be created by implementing the IPathLossCalculator interface, then modifying the relevant config class to instantiate the new path loss model when relevant.
In general, hold your mouse over something in the tool to get a mouseover explanation. The rest of this document explains everything in further detail.
For long-running calculations with high performance requirements, it is recommended to download the model executable and run it on your own computer/server, as the current server and queue system may make calculations take a long time. See the 'Offline calculations' section further down.
Warning: The online demonstrator has primarly been tested in Mozilla Firefox and Google Chrome. It will probably work in most modern browsers, but it should not be surprising if there are issues in older browsers such as Internet Explorer.
In order to run a road network calculation, you need an API key. This is to reduce the load on the server, since road network calculations are resource intensive. Contact [email protected] to get one.
All base stations you place on the map are saved only in your web browser's local storage. This means that if you switch to another browser or another computer, you will lose your setup. You can export stations and import them in another browser or on another computer, see 'Importing/exporting data'.
If you run a single point calculation, the base station parameters as well as the coordinates of the target point is sent to the server and processed, but nothing is saved anywhere.
If you run a road network calculation, the parameters of all selected base stations are sent to the server, and the relevant parameters are stored as part of the result files. Calculations and results started with a certain API key will only be available to users with the same API key. You can manually delete these files by deleting the calculations results in the 'Calculate road network' section of the tool.
If you do not want to send any data to our server, you can run the entire calculation offline instead. See the 'Offline calculations' section below.
In the current implementation of the tool, a base station is a single signal source. If you need to simulate a base station with multiple antennas, you can do this in one of the following ways:
- Calculate the total sum of power/gain for all antennas, and input the summed values in the Power/Gain fields on the base station.
- Zoom in as much as you can and place the base stations very close to each other.
- Use offline calculations where you can easily define multiple base stations on the same coordinate in the config file.
Option 1, while requiring some manual preparation work for the user, will result in shorter calculation times, as the total number of base stations will be reduced.
To place a new station on the map, click on the map where you want to place it, and pick "Create new station at this location" in the context menu that appears. The station will then appear as a gray circle, which will be colorized when it has been saved. Edit the base station parameters as required, and click "Save station" to finalize it.
Hold your mouse over the parameter headers in the tool for explanations.
Base stations are visualized as a circle with a surrounding signal strength indicator depending on their combined power and gain.
The center circle is colored depending on the station's state:
- Red: normal
- Gray: new/unsaved
- Green: selected for editing
- Blue: selected for calculations
The surrounding signal strength indicator will visualize the power+gain for each degree of a circle. The coloring of the signal strength indicator always moves from red to green, with no relevance for the signal strength.
The figure below shows a visualization for a station with a power of 50, and a gain of 18 to 28 around the 0 degree mark (straight East). This station will have a larger RSRP on the right hand side than on the left hand side.
Note that too large signal strengths will be capped at 128 pixels, resulting in the entire indicator being shrunk to fit this size. The visualization can therefore not be used to provide a detailed signal strength comparison, it is only meant as a convenient way of quickly visually separating different stations from each other.
To edit an existing base station, pick "Info/Edit" in menu in the top left corner, then click the station you want to edit. When you are done, remember to click "Save changes", or "Cancel" if you want to undo your current edits.
To remove a base station, pick "Info/Edit" in menu in the top left corner, click the station you want to edit, then click "Delete station" below the editor. This cannot be undone.
To import base stations into the tool, you can use the "List" item in the menu in the upper left corner, and click "Import CSV". This allows you to import a simple CSV format with the parameters used in this tool. When clicking "Import CSV" you will be presented with an example of how the CSV should look.
To export base stations as CSV, click "Export CSV" at the bottom of the station list. This exports data in the same CSV format as above.
In addition to your base stations and calculation results, you can also toggle layers showing infrastructure (from infrastructure.org) or the road network used in the road network calculations using the layer picker in the upper right corner. Click a layer to turn it on or off. For all layers, you can hold your mouse over an element on the map to get further information.
To run a path loss calculation from a base station to a point on the map, you need to select a base station first. Pick the "Info/Edit" section in the menu in the top left corner, then click the station you want to calculate from. When a station is selected (green center circle), you can click anywhere on the map and select "Calculate path loss from (station name)".
For performance reasons, a maximum of 1000 points are visualized in the plot. If the distance is greater than 1000 meters, a suitable number of points are skipped between each visualized point, so that the number of points does not exceed 1000 (it will, for example, skip every other point if the distance is about 2000 meters). In these cases, the distance between each visualized point will therefore be more than 1 meter. For more detail, the complete results can be exported as CSV.
Note that results are reported as RSRP, but if this is an ITS G5 station, it corresponds to a simplified RSSI (the same calculation as for RSRP, but without the resource block term). See the 'RSRP vs RSSI in the results' for more about this. You can also see the path loss directly if you export the results as CSV.
A road network calculation is a calculation where the received signal strength from one or more base stations is calculated along all roads within the base station's range.
Before starting a road network calculation, you need to input an API key, and consider the calculation parameters, as shown below. You will also need to select which base stations you want to include in this calculation by clicking them on the map (stations selected for this calculation gets a blue center circle).
The road network calculations consists roughly of five steps:
Step 1: All road links within the maximum radius (defined on each base station) of each base station is read into memory, and connected so that each base station has a list of all relevant links. One road link may be in the list of multiple base stations.
Step 2: Then each base station processes all its links once with an algorithm for finding the lowest possible path loss (assuming a perfectly flat terrain) to each point in this link. If the RSRP based on the lowest possible path loss is lower than the minimum RSRP (defined in the calculation parameters), this link will not be processed at all, as it is guaranteed to have too low RSRP, and therefore removed after the calculation anyway (step 4). Removing such links before actually performing the calculation speeds up the process greatly.
Step 3: The next step is the actual path loss calculation. Each base station runs through all its links once more, this time retrieving detailed terrain profiles between the base station and each point on the road network, calculating path loss and RSRP for each point. If the link calculation point frequency (defined in the calculation parameters) is 1, this will happen for every meter along the road link. If it is 10, it will happen for every 10th meter, making the calculation much faster, but also less detailed.
Step 4: To save storage space, all links with no points with an RSRP above the minimum RSRP (defined in the calculation parameters) are removed.
Step 5: The results are stored in various formats to allow visualization in the map and downloads as CSV and SHP.
Since these calculations are a lot more time consuming than a single point calculation (often requiring billions of terrain lookups per base station), the calculation jobs are listed in "Calculation log", where you can see calculation progress and results when it's finished. Due to limitations on our server, only a single calculation can run at a time, which means that a job may be placed in a queue to be calculated when the current job is finished.
- A job in the queue can be deleted by clicking the trash can (it will be removed from your list and from the queue on the server).
- A job that is currently being processed can be aborted by clicking the stop symbol (it will be stopped, but the configuration and progress information will remain until you delete the job).
- A finished or failed job can be deleted by clicking the trash can (it will be removed from your list, and all results will be deleted from our server).
When a job has finished, results can be downloaded as CSV or SHP files, both containing details of the path loss from each base station to each point along each calculated road link. Results can also be visualized on the map by clicking the map symbol. A calculation result may look like shown below. Note in the rightmost image that it is possible to zoom in and click a link to split it into all the individual points it consists of. The number of points will depend on the link calculation point frequency of this calculation.
 |
 |
The RSSI, which was used when collecting data in the LambdaRoad project, contains the total received signal energy over the entire bandwidth of the signal, including the signal of interest, noise and interference. While this is a good value to use when collecting data, it is difficult to estimate since noise and interference is not a part of the model. The RSRP contains the average received signal level per resource element of the LTE signal. Therefore, the RSRP is a better parameter to use when estimating the path loss than the RSSI.
This means that the tool is focused on RSRP, but for ITS G5 it has to report a suboptimal RSSI calculation, since it is a WiFi technology without resource blocks.
As the application has to retrieve and process a one meter resolution height profile of up to tens of kilometers for every single point it has to calculate, there has been a large focus on performance. Elevation data is downloaded (either manually or on demand), then converted to an optimized file format to allow very quick file loading. Elevation data is stored in tiles of 512x512 meters, which was found in testing to be a good balance between size (less tiles must be loaded in to memory) and memory usage (less total data has to be loaded into memory).
Since multiple tiles may have to be loaded into memory to retrieve a single height profile, the road link calculation orders the links by the angle between the West-East line and the line from the base station to the center point of the road link. This means that links having similar height profiles (and therefore using the same elevation data tiles) are calculated after each other, lowering the amount of tiles that must be loaded into memory multiple times.
The road network calculation allows base stations to be run on separate threads (see Threadcount in the config file). The decision to keep the entire base station on a single thread, instead of spreading links or even points into a set of working threads is because of the angle optimization described above -- to keep that working correctly, it is easiest to keep a base station and its elevation data memory cache grouped together on a single thread.
In the current implementation, roughly 60-70% of the run time is retrieving the elevation data from arrays already loaded into memory, while 20-30% is calculating features based on the elevation profile and calculating the path loss based on these features calculations. The run time could probably be reduced further by coupling the path loss calculators tighter on the elevation tiles, to avoid the height profile array that currently has to be populated from the elevation tile, then read from the path loss calculator, but this has not been done to keep the class responsibilities separated and easier to work with.
The online map tool sends data to the server, which runs the command line application. Like the online tool, the command line application is also open source, and can be compiled with or without modifications, or downloaded as a compiled executable if you need to run offline calculations.
Offline calculations are generally more performant than the online applications, both because there is no calculation queue, and because the application for some reason is about twice as fast when compiled as a standalone executable as compared to a part of an ASP.NET REST API. Offline calculations can also be easier to set up for complex calculations, as you have full control of the config-file the application runs.
More information about the offline calculations can be found in the backend documentation.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent