1 – Type of Wi-Fi devices detectable
The nodes can detect:
- Wi-Fi devices in associated mode (e.g. smartphones, tablets, laptops, Wi-Fi printers etc..). Associated mode does NOT mean that Wi-Fi devices need to be connected to a specific Wi-Fi network.
- Wi-Fi tags in CCX (Cisco Compatible Extension) mode, like the Accuware CCX Wi-Fi Tags.
2 – Detection frequency of Wi-Fi devices
Even if the scan rate of the nodes is constant (5 seconds frequency), this does not mean that all the Wi-Fi devices within the area are seen every 5 seconds. In order to be detected, Wi-Fi devices need to transmit some Wi-Fi traffic, and when there is no transmission, there is no detection by the nodes.
So, please make sure you understand the following important and deeply different concepts:
- The scan rate of nodes is the frequency at which the nodes scan the environment looking for MAC addresses on a single 2.4 GHz Wi-Fi channel. This frequency is 1 scan every 5 seconds (it can be lowered to but it can not increased).
- The frequency at which the Wi-Fi devices generate Wi-Fi traffic is “how often” a Wi-Fi device transmits some Wi-Fi traffic. It can be that a Wi-Fi device generates Wi-Fi traffic every 5 seconds (and so is seen by the nodes every 5 seconds) but it can also be that a Wi-Fi device does not transmit constantly some Wi-Fi traffic every 5 seconds (and so is seen by the nodes less often).
Below it is possible to find all the scenarios that will help to understand what to expect from your installation and all the action that can be taken to improve the detection frequency.
2.1 – Wi-Fi devices NOT running a custom App
The detection frequency of Wi-Fi devices that are not running a custom App can vary, since there is no control on the Wi-Fi transmission frequency. So, even if the scan rate of the nodes is constant (1 scan every 5 seconds), it could be that some Wi-Fi devices are not constantly detected by the nodes because they are not transmitting any Wi-Fi traffic.
These are all the possible scenarios:
- Wi-Fi devices connected to the Open Mesh network: this is the best scenario (highest detection frequency) because all the traffic transmitted by the Wi-Fi devices will be collected by the nodes and used by the Accuware server to improve the location accuracy.
- If you can’t see the Private/Public SSIDs of the Open Mesh network (even if the SSIDs are enabled inside the Cloudtrax dashboard) please submit a support request using this form. It could be that the broadcast of the SSIDs is inhibited by us to lower the noise on the Wi-Fi channel used by the Open Mesh network. We will upload a different firmware on your nodes and you will be able to see again the SSIDs of the Open Mesh network.
- Wi-Fi devices connected to an existing Wi-Fi network operating on a single 2.4 GHz channel: this scenario can lead to the same good results of the previous scenario IF the Open Mesh network is set on the same 2.4 GHz channel of the existing Wi-Fi network.
- Wi-Fi devices connected to a Wi-Fi network operating on multiple 2.4 GHz channels (e.g. 1, 6, 11): very often a Wi-Fi networks are set to work on the three non-overlapping channels to lower the noise (1, 6, and 11). This means that Wi-Fi devices connected to these kind of Wi-Fi networks, will be detected only occasionally (when the Wi-Fi devices are connected to one of the access points that is operating on a channel that overlaps with the channel used by the Open Mesh network). In this scenario, the best thing that can be done is leaving the Open Mesh network on channel 6 in order to overlap with the highest number of channels. The accuracy of this scenario is obviously lower compared to the accuracy that can be obtained in scenarios 1 and 2, because the Wi-Fi devices will be detected less frequently.
- Wi-Fi devices NOT connected to any Wi-Fi network: in this scenario the amount of Wi-Fi traffic generated by Wi-Fi devices is unpredictable, so the location updates could be sporadic. In this scenario a Wi-Fi device can be seen only when it is seeking for Wi-Fi networks around. From some tests we carried out:
- Android devices are detected more often than Apple devices with iOS <= 8.X
- Apple devices running iOS >= 9 and not connected to a Wi-Fi network can NOT be tracked/counted because they broadcast random MAC addresses unpredictably. These random MAC addresses are detected and filtered out by default by the Accuware server.
- The MAC address randomization introduced with Android 5 is not as robust as robust as the one introduced with iOS as it can be read from multiple online articles. This means that Android devices can be detected almost always.
- The detection frequency is function of the OS version running on the devices.
- The detection frequency is function of the status of the device: Wi-Fi devices awake, in stand-by or in deep stand-by behave differently.
- The detection frequency of laptops is very high. They are detected constantly.
- Wi-Fi devices connected to a Wi-Fi network operating on 5 GHz: these Wi-Fi devices will be completely invisible because the Wi-Fi traffic detected by our nodes will be completely null since the Open Mesh network works on 2.4 GHz.
2.2 – Wi-Fi devices running an App
The detection frequency of Wi-Fi devices running an application can be very high (up to 1 update of the position every 5 seconds). Wi-Fi devices running a custom application can be tracked with the same level of accuracy of the Accuware CCX Wi-Fi tags. How? Basically it is possible to force Wi-Fi devices to generate periodic Wi-Fi traffic allowing thus a constant detection by the nodes. Please submit a support request using this form to know more about this option.
2.3 – Accuware CCX Wi-Fi tags
The Accuware Wi-Fi tags can be tracked with the highest accuracy, because they are programmed to transmit Wi-Fi traffic periodically (every 4 seconds). This ensures not only that positions are fresh but above all that are accurate. In fact the nodes are able to catch the information of the accelerometer built-in, determine if the tag is moving or not, and thus allow a motion compensation and a super high accuracy. At this link you can find more information about the Wi-Fi tags.
3 – Amount of Wi-Fi devices detected
You have to start from the general idea that it is not possible to have the same count between the number of people counted with a clicker (e.g. at the entrance of a store) and the number of Wi-Fi devices detected by Accuware Wi-Fi Location Monitor, for these reasons:
- it can be that some people are carrying smartphones with the Wi-Fi turned off.
- it can be that some people are not carrying a smartphone (or a very old smartphone without a Wi-Fi).
- it can be that some people (who are carrying a smartphone) have the smartphone in stand-by. If a smartphone is in stand-by, the network interface is turned Off an On unpredictably (to save battery) and the Wi-Fi traffic is sent out only once in a while (when the network interface is turned On to seek for known Wi-Fi networks).
- it can be that some people are using Apple devices with iOS >= 9 (with the implication described in the paragraph above “Wi-Fi devices NOT running a custom App“).
Anyway despite some Wi-Fi devices won’t be counted, our customers are extrapolating very useful insights by using the raw data provided by the Accuware CSV service or by the Accuware Wi-Fi Location Monitor Synchronous API. How? Even if the number of Wi-Fi devices detected does not reflect the exact total population of an area, there are multiplicative coefficients that can be used to correlate the total population of an area with the number of people detected by Accuware Wi-Fi Location Monitor. Once that these multiplicative coefficients are computed it is possible to derive the total population of an area over time using the data collected by Accuware Wi-Fi Location Monitor. These multiplicative coefficients can be either:
- retrieved on the Internet.
- computed empirically by counting (for a reasonable amount of time) the real number of people detected inside a specific area and comparing this number with the number of people detected by Wi-Fi Location Monitor.
Our technology, and in general, indoor tracking systems for monitoring and analysis, are made to provide data about the average movements of people in a specific environment, and to allow the analysis of this data to come out with a trend that predicts the behavior of the population. With this technology you can’t track the whole population, and it doesn’t even make sense to do so, as long as you are able to track and monitor a sample of the population which is consistent, constant, large enough and homogeneous.
If your main objective is to track 100% of the people, maybe, it’s not the right approach. Example: when statistical institutes want to understand how the political elections are going to end, they interview a sample of the population, to make projections and they don’t (and can’t) interview obviously the whole population. Same happens when a marketing department wants to know the opinion their customers have about a new product.
3.1 – Results from real-life tests
In order to better understand the real behavior of Wi-Fi enabled devices (such as smartphones and tablets) in real life, some of our customers and the Accuware engineers, performed different tests in separated and independent environments. The scope was to understand the average frequency of detection of Wi-Fi enabled devices, when they are not connected to any network, and are in “sleep mode” (scenario 4th). The tests took place in different locations, such as retail stores, supermarkets, offices and the first thing that must be said is that there are not significant differences between these environments.
The results of the tests are summarized in the following table:
|Seconds||Number of detections||Total 5~90||% of detections within 90 seconds||% of detections within 60 seconds||% of detections within 30 seconds||% of detections within 10 seconds||% of detections within 5 seconds|
|Total detections within 90 seconds||325645||97.09%|
|Total non-unique detections||335411|
Most of the devices (60.37%), even if in sleep mode, are detected within 5 seconds. Increasing the range to just 10 seconds, leads to the detection of 73.92% of the devices. And if we consider the percentage of devices detected within 30 seconds, it grows up to 87.63%. More over, the amount of devices detected in a 1 minute range, is 95.97%, which means that most of the devices are actually seen by our technology within 60 seconds, even if they are in “sleep mode” and not connected to any Wi-Fi network. This leads to the consideration that, even if it is not possible to precisely predict the detection rate of the devices, we can predict how many devices will be detected in a specific amount of time, and the fact that more than 60% of them falls in the 5 seconds range, makes this technology completely reliable for statistic purposes.
4 – Maximum number of Wi-Fi devices detected simultaneously
By detectable simultaneously, we mean these Wi-Fi devices are all active at the same time and you want to query all their locations in real time every 5 seconds. Accuware Wi-Fi Location Monitor can be used to track up to several hundred devices at the same time without problems. If you need to track more devices, we can provide more computing capacity and in that case you should be able to track tens or hundreds thousands of devices at the same time. Each node can detect up to 100 different MAC addresses in every 5 seconds time slot. This number might not seem much, but it should be kept into account that the reception radius of a node is 100-200m at most (in open space) and, even if the Wi-Fi devices are all “active” at the same time, they rarely transmit constantly, but rather they transmit in short sporadic bursts. Over a period of 1 hour, a single node can report up to 72’000 different MAC addresses (720 reports * 100 unique different MAC addresses/report). Anyway since Wi-Fi devices do not transmit simultaneously at the same time, the numbers discussed above are even larger!