Who these recommendations are for
The recommendations are addressed to the administrator of the location system and to the person responsible for issuing trackers to employees: a foreman, a tool store keeper or a security coordinator. All these roles face the same question: are the devices that come back after a shift connected to the multi-charger and actually charging?
The problem stems from a property of the hardware, not from user error. Personal devices usually do not report the charging status correctly, and the field in the data remains empty or holds the value unknown. A control policy based on that field would mark every device as uncertain.
Recommendation 1: do not base control on the charging status field
In production data, most trackers reported an unknown status despite battery levels of 95 to 100%. This means the status field carries no information and cannot serve as the basis of an alert. An alert built on that field will be either permanently red or permanently silent.
Inferring charging from the battery trend (example of 3 readings)
- Charging
- Removed from the charger
Recommendation 2: infer charging from the battery trend
The server can infer the charging state from the battery trend if the tracker is located at a multi-charger. The rule is simple: when the battery rises, the device is charging; when it does not change, the state stays as it was before; when it falls, the device has been removed from the charger.
The advantage of this approach is independence from the hardware: it is enough for the device to send its battery level in the heartbeat. The drawback is latency: the first reading after a change of state does not settle the question, and only the next one shows the direction of the trend.
- Battery rising: charging.
- Battery unchanged: previous state.
- Battery falling: removed from the charger.
Recommendation 3: name the multi-charger locations so the server recognizes them
Inference works only when the server knows that a given transmitter marks a multi-charger. This requires unambiguous naming of charger locations already at the map configuration stage. A name such as charger is readable both for a person and for the rule. The details are covered on the page about implementation notes for multi-chargers.
Recommendation 4: look at the pool, not at individual devices
The calculated charging state appears in the tracker pool view in the delivery form as a marker, without an extra database query in the heartbeat path. This way the person issuing devices sees at once which units are charging and which need attention. Checking one device after another is too slow with dozens of units.
We recommend setting an internal rule for working with the view: devices without a charging marker and with a low battery level do not go out for issue until they return to the charger.
Recommendation 5: set the battery threshold for issue
The threshold for issuing a device should be an organizational decision, not a default value. A plant where a shift lasts eight hours needs a higher threshold than a plant with short trips into the field. Write the threshold into the foreman's work instruction and into the issue procedure.
Recommendation 6: divide responsibilities and describe escalation
Someone must be responsible for returning devices to the chargers after a shift. The instruction should name the position and the time of the check, as well as who to notify when devices are missing from the pool after a shift. Without assigned responsibility, the battery trend is just a chart that nobody looks at.
Recommendation 7: verify the rule on a physical test
Once a quarter, take several devices, connect them to the charger and check whether the server marked them as charging within a few heartbeat cycles. The test takes a dozen or so minutes and detects two typical errors: a wrongly named charger location and a device whose battery does not pass on the trend.
Video
Frequently asked questions
Why can the charging status field not be trusted?
Trackers rarely report a real status, and in production data most of them reported unknown despite battery levels of 95 to 100%.
What does the server need in order to infer charging?
The device must send its battery level, and its location must be unambiguously marked as a multi-charger.
Where does the administrator see the charging state?
In the tracker pool view in the delivery form, where the calculated charging state appears as a marker.
How do you control charging of your devices?
We will show the pool view with the charging marker in a demonstration and discuss naming the charger locations.
Let's talk