Cómo la IA hace que la analítica de geolocalización Wi-Fi de Galgus sea la más precisa del mercado

How AI makes Galgus Wi-Fi geolocation analytics the most accurate on the market

Wi-Fi geolocation analytics is one of the most valued applications by companies and public bodies seeking to know in detail and in real time the behaviour of users and citizens. However, for this data to be reliable, it is necessary to overcome certain technical challenges in which artificial intelligence has a lot to say. We tell you how we do it at Galgus to offer the most accurate solution available.

What is the main problem of Wi-Fi based analytics

This is the randomisation of MAC addresses. Modern devices (from the last 5 years to the present) continuously falsify (rotate, change) their MAC address by displaying random values, precisely to fool analytics systems. Some smartphones display 100 different MAC addresses in a minute, while others display only a couple of them.

By 2024, more than 90-95% of smartphones on the street will randomise their MAC when not connected.

So it would be like trying to do analytics on cars with a camera pointed at the number plates, and these number plates are digital displays that are continuously changing (with different cadences) to fool the system.

Therefore, it is not useful to make “rules of three” to try to “translate” MACs. This is something that other manufacturers do (by means of a calibration that usually takes weeks and requires technicians calibrating in the field), but it does not produce satisfactory results.

Moreover, this behaviour changes every few months with new releases of operating systems or new smartphone models, which would require continuous recalibration at a costly cost.

What how can Galgus and AI minimise the impact of random MACs on analytics?

As the MAC address is no longer a unique and stable identifier for each device, we look at other characteristics to create a fingerprint, i.e. a fingerprint that identifies the device as long as it stays close to our APs.

In addition, to avoid the effects of confusing the same devices, we post-process the data to statistically separate these cases of simultaneity. To do so, we use artificial intelligence algorithms.

In any case, fingerprinting alone gives a fairly good accuracy, while post-processing adds an additional 5-8% accuracy.

To the best of our knowledge, there is no other commercial Wi-Fi analytics solution that does this post-processing to spatially separate devices with the same fingerprint. An improvement that is internationally patented.

In addition, it is also necessary to take into account devices that are not connected to the Wi-Fi network, but are in its immediate vicinity.

With Galgus Fingerprint, we achieve 100% accuracy in the connected count and around 95% in the unconnected count, which is an order of magnitude between 8x and 12x higher than other WiFi analytics solutions, which are less than 10% accurate.

How AI makes Galgus Wi-Fi geolocation analytics the most accurate on the market

Beyond accuracy, what are the advantages of Galgus’ Wi-Fi Presence and Location Analytics solution over other technologies?

The first is that you don’t need to bother the user: no need to install apps, no need to grant permissions, no need to activate Bluetooth, no need to activate GPS. In most cases, you don’t even need the smartphone to have the Wi-Fi button active. And the user does not know that he or she is contributing to the analytics (all with the level of privacy and anonymity required by the specific data protection regulation).

At the infrastructure level, less hardware needs to be deployed than with a system of cameras, volumetric sensors, RFID, beacons, etc. In addition, the same network that is providing an Internet access service can be used for analytics, without any degradation of service (since data collection is passive).

Moreover, thanks to AI, anomalies in user behaviour can be detected and predicted, notifications and early warning systems can be set up, among many other options. As you can see, AI has arrived to provide enormous value when it comes to knowing the reality of various scenarios and making informed decisions accordingly. Therefore, we encourage you to learn more about the details of our Presence & Location Analytics technology by contacting our team and answering any questions you may have.

Categories