Locations are normally estimated with a certain level of uncertainty, referred to as accuracy. This is true regardless of the technique used to determine that location. Higher accuracy means lower uncertainty.

The key to obtaining highly accurate device locations using ambient radio signals is to establish a strong correlation between locations of interest, and the set of ambient radio signals detected at each location. Locations of interest are the areas where people normally walk or dwell. The correlation process is referred to as fingerprinting process and is described inside the Setup section of this user guide (available in the left sidebar).

In this page you can find information regarding all the factors that influence the level of accuracy provided by Accuware Indoor Navigation. Being aware of these factors will allow you to understand what to expect from each installation and how to bring out the most from Accuware Indoor Navigation.

1 – Environment

1.1 – Amount of obstacles

The best results with Accuware Indoor Navigation are obtained when there is a high attenuation of the radio signals available in the environment. The Accuware Indoor Navigation engine determines the devices positions by comparing the radio signals detected in real-time with a database of radio signals created during the fingerprinting phase. This means that, as strange as it might sound, it works better inside building with walls because of the stronger attenuation of the radio signals. Open space environments are characterized by a weaker attenuation, so everything being equal, the accuracy will be lower.

  • You should expect a higher accuracy when the system is used inside shopping malls, hospitals, school campus, airports, hotels, train stations, subways, museums, corporate buildings (with offices), jails and in general inside areas with walls.
  • You should expect a lower accuracy when the system is used inside a large warehouses, trade shows, industrial plants, stadiums, retail stores etc.. and in general inside area with only few walls.

1.2 – Types of obstacles

Although the Accuware Indoor Navigation engine provides excellent results when used in areas with walls, it must be said that different types of obstacles made of different materials affect the accuracy of  the system:

  • Thick walls – even if Accuware Indoor Tracking works better inside building with walls, if the walls are too thick, then the absorption of the walls is too high and during the fingerprinting process only few radio signals are seen all together in each location and this affects negatively the accuracy.
  • People – the human body is a source of interference because its high percentage of water attenuates the radio signals. This means that if the fingerprinting is done in an environment without people and later on the system is used in a crowded environment, then there could be some differences in the radio signals measured with a possibility of reduced accuracy. The negative effects just described will be more evident in open spaces where there is a weak attenuation of the radio signals and the presence of people will produce more noticeable effects on the values measured by the Accuware Indoor Navigation engine. If possible, we recommend to perform the fingerprinting in conditions comparable to the conditions that will be found during regular system usage. For example, if Accuware Indoor Navigation is used to provide navigation to devices in a shopping mall during business hours, the fingerprinting should be done during business hours as well.
  • Metal objects and mirrors – big metal objects and/or mirrors affect negatively the level of accuracy because of the reflection of these kind of objects on the radio signals propagation.

1.3 – Size of the area

It must be said that the accuracy perceived is also a function of the dimension of the areas in which the system is used. Using the system in an open space that is smaller than 10 x 10 meters may not make much sense considering that the average accuracy in this kind of environment is of about 1 meter.

2 – Fingerprinting process

The fingerprinting process is the one with the biggest impact on the overall performances of the Accuware Indoor Navigation engine. In particular these are the factors with a major effect on the accuracy of the system:

  • the routes chosen for the fingerprinting – the fingerprinting must be performed along ALL the routes that will contain the devices that will take advantage of the Accuware Indoor Navigation engine.
  • the number of passes on the routes – during the fingerprinting each one of the chosen routes should be walked 2 times: once in each direction.
  • the technical specification of the device used for the fingerprinting – different devices behave differently when the Accuware Indoors App is used during the fingerprinting phase for the collection of the radio signals. For example there are some devices that can not be used to collect radio signals and there are also some devices that can collect more radio signals than others. As we will see inside the page about the fingerprinting process, all these variables have a big impact on the accuracy provided by the system.
  • the presence/absence of portable WiFi hotspots during the fingerprinting – recent devices can share the data connection over Wi-Fi by creating hotspots. If, during the fingerprinting process, you are collecting fingerprints related to Wi-Fi signal then you need to make sure that there are no hotpots moving around. Otherwise the indoor location results will be affected negatively. Please look at the paragraph below Move – existing radio signals.

3 – Radio signals

Even if it is not always possible to change the amount, the deployment model and the physical locations of the radio signals in the environment, you need to be aware of how these variables have an impact on the accuracy of the system.

3.1 – Amount

The accuracy provided by the Accuware Indoor Navigation engine is a function of the amount of radio signals detected during the fingerprinting phase. A higher amount of radio signals always increases the system accuracy. Even if the accuracy is affected by multiple factors (and not just by the spacing), here is a rough estimate of how the level of accuracy varies with the number of radio signals detected per surface (square meters):

    \[Radio Signals = (\frac{\sqrt{Area}}{Radio Signal Spacing}+1)^{2}\]

Here is a rough estimate of how the level of accuracy varies with the RADIO SIGNALS SPACING:

~12 meters of RADIO SIGNALS SPACING -> leads an AVERAGE ACCURACY of ~1 meter 
~18 meters of RADIO SIGNALS SPACING -> leads an AVERAGE ACCURACY of ~2 meters

3.2 – Transmission power

The accuracy provided by the Accuware Indoor Navigation engine is a function of the transmission power of the radio signals detected during the fingerprinting phase. By setting the radio signals at their maximum transmission power the accuracy provided by the Accuware Indoor Navigation engine will be higher.

3.3 – Deployment model

The best results are obtained when the sources of radio signals are distributed uniformly (ideally in a in a grid pattern). Please keep in mind also these notions:

  • having the sources of radio signals close to each other inside a small area won’t improve the accuracy in that area because the fingerprinting process won’t lead to a strong correlation between locations and the set of ambient radio signals detected at each location.
  • if you do not need the same level of accuracy everywhere, you can use distribute the radio signals using a different spacing for different areas.

3.4 – Physical deployment

The physical location of the radio signal sources affects the system accuracy. If you have the freedom to install the radio signal sources in your preferred locations then we suggest to install them in these physical locations (in order of preference):

  1. on the ground.
  2. on the lower part of a wall/column.
  3. an any other item that is very close to the ground or to the ceiling. Please make sure the signal source is attached to an object that won’t be moved.
  4. on the ceiling.
  5. inside a false-ceiling
  6. on the upper part of a wall/column.

IMPORTANT SUGGESTIONS:

  • DO NOT install the radio signal sources halfway between the ceiling and the floor(e.g. on desks).
  • DO NOT mix the physical locations (e.g. some radio signal sources close to the ground and some radio signal sources close to the ceiling)
  • print your floor plan and write down the location of each radio signal installed and the value of the tuple (Major, Minor).

3.5 – Changes over time

Over time, it could be that some sources of radio signals will be added or turned off or removed etc.. All the changes described below might affect the results provided by Accuware Indoor Tracking. These “dynamic” environments (subject to changes over time) should be monitored periodically to gauge a possible decay of the accuracy.

Add new radio signals

Add new radio signals does not have a negative impact on the accuracy . On the contrary, it could be possible to take advantage of the presence of new radio signals, increasing thus the accuracy provided by the Accuware Wearabouts engine. Basically it is necessary to perform a NEW fingerprinting session in the area where the NEW radio signal sources have been installed using the Accuware Indoors App in order to add the NEW radio signals to the database of fingerprints of your site. There is no need to delete the database of fingerprints!

Turn off/remove existing radio signals

Turn off or remove existing radio signals might affect the accuracy negatively depending on the number of radio signals turned off or removed. Basically once some radio signals are removed or simply turned off, the radio signals collected by the device during the positioning process will be lower compared to the amount of radio signals seen by the device during the fingerprinting process. This means that the amount of data available for the computation of the location will be lower affecting thus the accuracy. If you see a significant decrease of the accuracy, you can compensate this problem by:

  1. adding NEW radio signals.
  2. performing a new fingerprinting session using the Accuware Indoors App.

There is no need to delete the database of fingerprints!

Move or change the transmission power of existing radio signals

Move or change the transmission power of existing radio signals has a drastic effect on the accuracy. This because, once some radio signal sources are moved or the transmission power is changed, the radio signal strengths that will be detected in real-time in a specific location will lead to a computation of a location that won’t be the correct one because the radio signals and/or the signal strengths that should be detected in that specific location (from the database of fingerprints) are different. If one or more radio signal sources are moved the only way to fix this problem is to follow these 2 steps:

  1. Delete completely the database of ALL the fingerprints (mandatory!)
  2. Perform a new fingerprinting from scratch using the Accuware Indoors App

Change the identifier of existing radio signals

Change the identifier of the radio signals (MAC address of Wi-Fi access points and/or combination of UUID/Minor/Major of iBeacons) might affects the accuracy negatively. This because, once the identifier of some radio signal sources are changed, the radio signal sources that will be detected in real-time in a specific location will be different compared to the radio signal sources that should be detected in that specific location (collected during the fingerprinting process). The only way to fix this problem is to follow these 2 steps:

  1. Delete completely the database of ALL the fingerprints (mandatory!)
  2. Perform a new fingerprinting from scratch using the Accuware Indoors App

4 – Usage conditions

4.1 – Position with respect to the trained routes

Another factor that affects the accuracy is the relative location of the devices compared to the area covered during the fingerprinting phase. For this reason you need to be aware that the most accurate results are obtained when the devices are located along the routes where the fingerprints have been collected. The system is less accurate when devices are physically in areas where there are few or no fingerprints.