Bluetooth beacons, are they useful?

Back in 2019, I got some interesting devices for testing – Bluetooth beacons. Since my work is related to the implementation of such technologies in business, I will try to describe all the pros and cons of such devices, as well as why their implementation is not observed anywhere still.

Let's start with why they are needed at all?
Are they used anywhere now and why did interest arise in this direction?
Below you will find most popular applications technologies:

• With their help you can keep track of the state of things in the room.
  For example, a keychain for a TV remote or keys that emits a signal in the “near-far” format.

• Opportunity transmit any external data.
  For example, information on electronic price tags, data from temperature or pressure sensors, and the like.

• Marketingi.e. the ability to leave some text or other marks similar to NFC tags.
  It is worth noting here that the features of the latest versions of smartphone OS have made this area practically dead in favor of device security.
  Although, to be fair, it wasn’t very lively before.

• And what passed the test with us is this the ability to track employee movements inside the retail facility in order to optimize labor costs.

APPEARANCE
The devices may be different, but most beacons are represented by key fobs or their analogues.
Inside such a device is hidden a regular passive-radiating antenna, a number of microchips and a power source, most often a CR2016 at 3V, although there are other options.

This power source is used to save battery power, since in this case a DC-DC voltage converter is not required, which means there will be no extra losses in an already small energy source.
The second factor why it is convenient to use CR2016 is its small dimensions, which fit well into the key fob form factor.

In essence, these devices are similar to TPMS tire pressure sensors.
Only in the case of TPMS sensors the data is the tire pressure, and in the case of these beacons – the distance from the smartphone to the beacon emitters.

INSTALLATION and CALIBRATION
Bluetooth beacons are placed in a closed room (in our case, a store) and then they need to be calibrated relative to the receiving device, i.e. the smartphone of any employee.

In this case, on-site calibration is mandatory process.
The fact is that Bluetooth waves do not pass well through physical objects and as a result, significant wave interference is observed.
This, in turn, creates delays in the signal arriving at the employee’s smartphone.

In our case, beacons could only be placed under shelves, which led to an “echo” effect, reducing the efficiency of signal propagation.

Tests have shown that the overlap of waves leads to errors in estimating the distance to the device, and sometimes to a complete loss of signal at the receiving device.
The employee simply “disappeared” for about 5 minutes and appeared in another place.

This is where the first disadvantage of the technology comes from:
Any device will have to be calibrated on site, which immediately leads to additional costs and an increase in the cost of the project as a whole.
And if the objects are located in the regions, then the cost of a cheap beacon will be supplemented by the cost of a very expensive flight to the object.

PRINCIPLE OF OPERATION
The devices send a signal to the receiving smartphone, and the software in the smartphone triangulates the signal, a mechanism very similar to how communication towers work, identifying the location of the subscriber.

Thus, to determine the location of an employee on the territory of a retail facility, a minimum of 3 beacons will be required, and it is better to increase this number several times to improve accuracy.

This is where the second disadvantage of the technology comes from:
Each signal requires a minimum of energy at the level of the transmitting device, but at the same time requires computational costs for that very triangulation at the level of the smartphone.
It is worth noting that the more beacons transmit a signal, the more often the receiver will have to recalculate the coordinates, which means that the energy costs for the smartphone will also increase.

In our tests, we observed a stable discharge of the smartphone battery.
Measurements showed a consumption of about 1000 mAh/hour,
i.e. a standard phone battery fails after 3-4 hours of operation.
This means that you will have to constantly charge your phone, which is very problematic when an employee works on the premises.

ACCURACY OF MEASUREMENTS
The main question that tormented us was whether the technology would be able to create a heat map of an employee’s movement around the facility and whether it would match the visual control data.

The map was successfully compiled, but its accuracy was +/- 1 meter,
which is unacceptable in the context of assessing the labor costs of an employee working with a specific pallet.
Even with the recalibration of the devices, we were unable to guarantee that the employee would move along a specific alley, as there was always a chance that he would move parallel.

The third disadvantage of the technology:
Application is possible only if the coordinate error of ~1 meter is acceptable.
The error is dictated by the operating frequency of the devices (Bluetooth), as well as interference on surfaces.

For example, for a large warehouse such an error may be acceptable, but for a small store it is unlikely.

However, such data cannot be called completely useless.
Measurement of time costs within a category (1-2 racks) – showed quite good efficiency.
In particular, it was possible to identify the most expensive categories, which, oddly enough, included the store’s warehouse.

That is, the costs of searching for goods in the warehouse accounted for a significant share of the time costs, which came as a surprise to all participants in the process.

In addition, it turned out that the efficiency of external (outgoing) employees is higherthan the efficiency of its staff.
It is worth mentioning that the effectiveness does not manifest itself immediately, but within 1-2 weeks. For some time, the employee “wanders” around the facility, which increases the time for movement and reduces the time of “effective work”.
However, after the “adaptation” period, the visiting employee becomes approximately 20% more effective than his permanent colleagues.
We assume that this is a consequence of the need to visit a number of other sites during this day, which forces the employee to look for ways to optimize his work.

Another one Smoking was a discovery.
At the level of employee perception, smoking breaks are 15-25 minutes per day.
According to actual measurements, the costs for this procedure are many times higher.
The “smoking room” was allocated as a separate zone within the framework of the measurements taken.
Thus, it can be safely stated that a smoking employee is more expensive in the case of jobs in the format of “warehouse employee” / “store worker”.

TEST CONCLUSIONS:

• High battery discharge The signal receiver is the stop factor.
  It is possible that if the phone acts only as a repeater, it will be possible to reduce the impact of this factor.
  But this will require investments that will not pay off immediately.

• Possibility of application only under conditions of low influence of measurement error.
  For example, the technology may be applicable in a warehouse, but not in a retail store.

• There is enough data to create a heat map and further analysis, which means that the measurements themselves can bring about qualitative changes in the labor costs of employees.
  This will require a quality data analyst to distinguish real dependencies from random correlations.

• Need for manual calibration leads to high wasteful expenditure.
  The possibility of application is limited by the locations of the technology providers.
  If you open a facility in a remote region, you will have to pay for the travel of an external employee to calibrate the devices.

• On “Day X” it will appear need to change batteries in all devices at once, the event is random and can lead to significant financial costs.
  You won’t be able to carry out such a replacement on your own, and will then need to recalibrate.

EPILOGUE
As of 2024, the supplier of the technology we used in our tests finished his activity.
This fact indirectly confirms the conclusions about the difficulties in launching and paying off such a technological innovation at that time.

It is worth noting that previously direct triangulation technologies were used, which reduced the accuracy of the devices.
Potentially the same approach could be used, however use neural network for analysiswhich, based on a large amount of data, can provide an advantage in accurately determining the employee’s position even in difficult conditions.
In this way, the accuracy of measurements will increase to a level acceptable for small retail.

The calibration level can be transferred from beacons to a smartphone, controlling it remotely.
In this case, there will be no need for specialists to visit the site, which means that operating costs for maintaining the system’s functionality will drop significantly.

The power supply can be replaced for 2-4 AA batteries with free access, which will give several tangible advantages at once.
The first is the ease of replacement by the facility’s employees themselves.
The signal level has dropped – a message is sent indicating which device requires a battery change.
Secondly, AA batteries are everywhere and are even used for air conditioner remote controls, unlike CR2016.
The minus in the form of the increased size of the device will be more than compensated by the ease of its use.

Reduce the frequency of requests up to 1 in 30-60 seconds.
This simple operation will reduce battery consumption on your smartphone, while providing enough information to create a map of the employee’s movements.

Use your smartphone only as a connector for the “Mayak-smartphone-server” system,
and send the entire computing load to the cloud.
This way, the battery consumption of a smartphone will be similar to that of a messenger, and all the data will be available for any business analytics.

Make a single outline for everyonecreating a boxed solution.
Since the evaluation mechanism will be the same for employees, for boxes with a beacon tracker, and for a smart robot, it makes no sense to focus on only one direction.
One personal account with an admin panel will be equally suitable in any area from retail to smart home.

Due to the fact that IoT is not our core businessthen we probably won’t launch such a product ourselves.
However, we hope that the information will be useful to the business community and will help to avoid some of the mistakes that were made by the previous technology provider.