Positioning convergence or Time To First Fix #
When the Situm positioning system is initialized, it starts from zero. During the first seconds, Situm has little information so the first locations may be slightly inaccurate/erratic. Little by little, the system receives sensor readings (Wifi, BLE, gyroscope, etc.) and the location estimate will converge to an stable and accurate position. The following figure shows a representation of how the system error is initially high, but decreases as it approaches the convergence.
In Situm, the convergence time, usually called the Time To First Fix (TTFF), is usually 4-15 seconds. After this convergence, Situm accuracy is usually within 1-5 meters on average.
Both the convergence time and the accuracy depend on:
- The movement of the user. If the user moves, Situm needs less time to yield a first accurate position. Further positions will also be more accurate.
- The magnetometer performance. If the smartphone has a magnetometer and it is well calibrated, the convergence time will be greatly reduced and overall accuracy increased.
- Other factors, such as sensor quality, calibration quality, etc.
Indoor positioning accuracy #
Once Situm has converged, we may measure the accuracy of the location estimate (distance from the real location). In general, Situm yields an average accuracy of 1-4 meters, although this may depend on a number of factors. The following table can be used as a reference for different environments and situations.
|Environment||Example||Typical accuracy on hand||Typical accuracy in the pocket|
|Wide corridors||5-15 meter wide corridors present in airports or shopping malls||1-4 meters on average||Reduced by an 20%-30% on average.|
|Narrow corridors||3-10 wide corridors present in hospitals or offices||1-3 meters on average||Reduced by an 20%-30% on average.|
|Open areas||Wide diaphanous areas that do not have (many) delimitations. E.g. central plaza of a shopping mall, modern airport terminal, etc.||1-5 meters on average||Reduced by an 20%-30% on average.|
|Mixed areas||Mix of open areas and corridors. These are diaphanous areas with some delimitations such as shelves. E.g. supermarket, warehouse…||1-4 meters on average. Detects correct shelve 70%-90% of time.||Reduced by an 20%-30% on average.|
In addition to the geolocation, Situm provides an estimate of the accuracy as the “area surrounding the geolocation where the user most likely is”. You can see this in the following figure: the user is within the circle centered at the arrow with a certain degree of confidence. When the confidence is high (see next figure), the circle is small and the real location should be close to the one returned by Situm. On the contrary, when the confidence is low, the circle is wide and the real location might be distant from the one returned.
Factors that influence accuracy #
In general, Situm’s accuracy will be within 1-4 meters in average. However, some factors may impact on this accuracy:
- Georeferencing and the floorplan scale. The floorplan of the venue must be correctly georeferenced and scaled. Only a meter of deviation between the floorplan and the real venue could imply on an additional error of the same measure in the geolocations that the solution computes.
- Beacon placement. Beacons must be placed according to the following guide: Using BLE beacons with Situm
- Calibration and generation of the radio-magnetic map. The calibration process must be intensive and accurate, and must be redone periodically to reflect on any significant change in the environment (new walls, readjusted WiFi APs, etc..).
- Quality of the hardware and smartphone Operation System (OP). The smartphone must have a minimum set of quality sensors, and the OS must not introduce anomalies in the treatment of data. Please check the following link for more information: Is Situm compatible with every smartphone?
- Device connected to a WiFi network (It’s not recommended). When connected to a WiFi network, some devices will not allow to retrieve WiFi scans (permanently or sporadically). In addition, some apps or isolated situations may interfere on the sensors performance of the device. For instance, some apps may usually perform a heavy data synchronization, which can overload the WiFi chipset and its ability to provide WiFi scan readings.
- Device conditions. Sometimes a device may provide anomalous sensor readings (e.g. if the device has received an impact or it has not been restarted for a while, etc.). It affects the solution accuracy negatively.
- User behaviour. The geolocation system has been designed considering that the user will carry the phone by hand or keep it into his pocket. When the user carries the smartphone by hand, it must be placed with the screen pointing upwards to the natural orientation of the user. The user must walk normally the whole time.
- Distance taken from the start. The accuracy of the geolocation computed by the system will be higher as much the walking distance increases.
- WiFi Network behaviour. Some WiFi networks may vary the transmission power (either manually or automatically). It may reduce the accuracy of the solution.
- Visitor’s affluence. In some buildings, it is common to find sporadic agglomerations that may affect the power of the received signals. It may reduce the accuracy of the solution.
- Environment topology. In general, outside environments like halls or balconies broadcast WiFi and/or BLE signals with little attenuation. The accuracy may be affected in these environments.
Orientation accuracy and convergence (Displacement to First Fix) #
If all inertial sensors are working properly (accelerometer, magnetometer, gyroscope), Situm computed the orientation of the smartphone with great accuracy (around +- 15º). On the other hand, the convergence time of the orientation (usually called the Displacement To First Fix or DTFF) can be:
- Near zero: if the magnetometer is working properly, accurate orientation can usually be produced from the start, even if the user does not move.
- Few meters (5m-20m): if the magnetometer information is not available, there is no way to tell north from south without user displacement. Nevertheless, Situm can still estimate the orientation but after the user moves for a few meters.
The following table summarizes the DTFF and accuracy of the orientation estimate depending on a few factors.
|Sensors available||Phone in…||Displacement To First Fix||Average Accuracy|
|Accelerometer, gyroscope, magnetometer||Hand||No||High (+-15º)|
|Yes (5-20m)||High (+-15º)|
|Accelerometer, magnetometer||Hand||No||Medium (+-30º)|
|Yes (5-20m)||Medium-Low (+-40º)|
|Accelerometer and gyroscope|
Accelerometer, gyroscope and magnetometer is uncalibrated or under heavy interferences
|Hand||Yes (5-20m)||Medium (+-30º)|
|Yes (5-20m)||Medium (+-30º)|
|Accelerometer only||Hand||Orientation can not be estimated|
|Orientation can not be estimated|