Positioning accuracy and verifying it depends on four main components:
High Level Guide for Validating Deployments
On a high level, the recommended steps to validate an IndoorAtlas deployment technology are:
Figure 1. Marking ground truth with MapCreator.
Step 1 Checking analytics should be done continuously during fingerprinting to detect issues as soon as possible and to be able to fix them while on site. Note that you can have the on-site team check the analytics while performing fingerprinting, or you can have a person at the back office checking the analytics from IndoorAtlas Web App.
If fingerprinting a larger venue, it is a good practice to test positioning in smaller areas as soon as they are finished (remember to trigger map generation after the paths have been uploaded). Such areas might be floor levels or different sections of the building, for example. This way you can detect problematic areas and possible fingerprinting errors as soon as possible. If multiple persons are mapping, they can all independently check the performance on the area they have fingerprinted.
Step 2 validates that positioning is accurate all areas and that floor changes work well. In case you find an issue on some area, you can replay the TraceId with Session Viewer to e.g. see if beacons were detected by the test device in the problematic area. In general, check the section Overview and the four main components of accurate positioning before making conclusions.
Also, in Step 2 the work can be split:
- On-site team performs test walks and stores TraceIds for those
- Back office team uses Session Viewer to validate performance and/or look for issues and reasons for issues.
Splitting tasks between on-site team and back office team is useful in several contexts, for example:
- A company uses IndoorAtlas technology and supports Customer's own team doing IndoorAtlas deployment.
- A company uses and supports outsourced workforce doing IndoorAtlas deployment.
- A company is doing a large deployment with many people on site. Someone at the back office can be in charge of the overall deployment quality.
Test Coverage / Planning Test Routes
When you plan to deploy a production app, it’s time to plan how to verify the positioning performance as well as possible. Remember to Use MapCreator's Ground Truth feature to capture the real location of the tester in all test cases.
Planning Test Routes
If you have an on-site team that you are managing, plan the test routes for them. We recommend just drawing them on the floor plan images, please see an example below:
The above is a single-floor route to validate positioning accuracy on that area of the floor as well as the outdoor-indoor transition for the venue.
Follow these high-level guidelines when planning test route coverage:
- Focus on natural aka most common routes (e.g. user starts outdoors and enters the venue)
- Walk through the main area of the floor.
- Start paths from a few different locations to test first fix is accurate enough at important areas – walk at least 30s on each test to see that the tracking gets accurate.
- Plan a test route through multiple floors that is common for the end users
- Start and walk on a floor until the tracking gets accurate (usually 10s to 30s), then take the elevator or the stairs.
- Test the route multiple times to verify floor transition performance is good
- Plan a test route from outdoors to indoors that is common for the end users
- Start well outside the venue e.g. 50m away from the main entrance
- Walk indoors and continue on the mapped area for at least 50m to see accurate tracking
- Use multiple different Android device models in your tests
- Use exactly same test routes when comparing performance of models
- Focus on device models that are most common for your app’s end users
- Repeat the test multiple times with each device
- Test evenly on iOS and Android if you plan to release your app for both platforms