There is almost never an outlet above a store entrance or office door, yet that is where the sensor goes. So the power supply often decides whether a LoRaWAN people counter goes up in an hour or needs an electrician first. The short answer: if there is no power at the mounting point, the battery version of the Milesight VS351 is the obvious choice, rated by the manufacturer at roughly 1.6 to 1.7 years per battery set. If a USB power adapter is within reach, the USB-C version runs permanently with no battery swaps. PoE pays off when a network cable already runs to the door and you need more than a simple in-and-out count.
One distinction gets lost easily: power and data are two separate questions. A wireless people counter sends its counts over LoRaWAN to a gateway, whether it runs on batteries or on a USB cable. A PoE device, on the other hand, gets power and network over the same Ethernet cable and needs no radio network at all. Keeping them apart avoids most wrong purchases.
How does a LoRaWAN people counter work?
A LoRaWAN people counter is mounted on the ceiling or above the door frame and detects which direction people walk underneath it. The VS351 uses a thermopile sensor for this: an array of heat sensors that picks up body heat as a temperature pattern without ever capturing an image. A thermopile people counter therefore cannot recognize anyone, which suits offices or restrooms where cameras are off the table. Milesight states an accuracy of 80 to 90 percent and points out that mounting height and the height of the people passing affect the result.
The VS351 mounts between 2 and 3 meters high, and the width it covers depends directly on that height. At 3 meters it covers a 3.6-meter-wide passage, at 2.4 meters only about 2.6 meters, and at 2 meters just 1.6 meters. The sensor head tilts from 0 to 90 degrees, so it works on a ceiling and above a door frame. During setup you enter the actual height and choose between an open doorway and a door that opens and closes.
Counts travel by radio to a LoRaWAN gateway and on through the network server into your application. Each report says how many people walked in and out since the last report, optionally followed by the running totals, the temperature and, on the battery version, the battery level. The VS351 also speaks Milesight’s own D2D protocol, which lets it control other Milesight devices directly without a gateway, for example to switch on lights.
How long does a battery-powered people counter last?
The battery version of the VS351 runs on four replaceable ER26500 lithium thionyl chloride cells rated at 9,000 mAh each. Alkaline batteries are explicitly not allowed. Milesight quotes a battery life of about 1.7 years at spreading factor 7 and about 1.6 years at SF10. Both were measured in the lab with a 30-minute reporting interval, 25 °C and 1,000 people per day, explicitly as guideline values.
These conditions differ from the factory settings. Out of the box, the sensor reports every 10 minutes, three times as often as in the lab figure. If you leave the interval as is, plan for noticeably less than 1.6 years. Heavy traffic, a spot far from the gateway that forces a high spreading factor, or enabled retransmission all cost extra energy. Milesight notes that retransmission adds uplinks and shortens battery life.
The battery version lasts this long thanks to a simple trick: a radar sensor detects whether anyone is approaching and only then wakes the thermopile sensor. On top of that, you can set a hibernation period, for example overnight or on weekends, during which the device neither counts nor reports. That saves power, but nobody is counted during that window.
Plan battery replacement from day one. The battery level comes with every periodic report as a percentage, so you can set a warning threshold in your application. Milesight recommends replacing all four cells with new ones at the same time, because mixing old and new cells shortens battery life and skews the percentage reading.
When is a USB-C powered people counter the better choice?
The USB-C version of the VS351 takes 5 V at 1 A and ships with a 1.2-meter Type-C cable and a power adapter. At 70 × 68 × 32 millimeters and about 72 grams, it is much smaller than the battery version, which measures 188 millimeters wide including its battery compartment and weighs 383 grams. There are no batteries to swap, and you can set a short reporting interval without worrying about battery life.
Permanent power makes sense wherever there is already electricity close to the mounting point: above a drop ceiling, in a utility closet next to the door, or at a doorway with an electric door opener. It also pays off with several counters in one building: based on the manufacturer’s figures, ten battery units mean 40 new cells and ten ladder trips roughly every year and a half.
Two differences are worth knowing. First, according to Milesight the USB-C version uses only the thermopile sensor, without the radar that wakes the sensor in the battery version, so it is always active. Second, it depends on mains power: if the breaker trips or someone pulls the plug, it stops counting. The cable only supplies power; counts still go out by radio. So secure the adapter and cable instead of leaving them loose in a shared power strip.
PoE people counter vs LoRaWAN: when do you need a cable?
Power over Ethernet means a single network cable delivers both power and a data connection. A PoE people counter such as the Milesight VS125-P sits directly on your local network and sends data via MQTT or HTTP, optionally over LTE. It is powered via IEEE 802.3af or a 12 V DC supply and draws 5.9 watts on average and 11.1 watts at peak.
The VS125-P counts with a stereo camera and AI processing, claims an accuracy of up to 99.8 percent and works at mounting heights from 2.2 to 6 meters. It adds zone counting, dwell time, heat maps and the ability to link up to 16 devices to cover wide entrances. According to the manufacturer, no personal image data is stored or transmitted.
The table below puts the three options side by side. It shows that the two VS351 versions differ mainly in power and sensing, while the VS125-P belongs to a different class of device altogether.
| Feature | VS351 battery | VS351 USB-C | VS125-P |
|---|---|---|---|
| Power supply | 4 × ER26500, 9,000 mAh each | 5 V / 1 A via USB-C | PoE 802.3af or 12 V DC |
| Data path | LoRaWAN, Milesight D2D | LoRaWAN, Milesight D2D | Ethernet, optional LTE |
| Sensing | Thermopile and radar | Thermopile | Stereo camera with AI |
| Mounting height | 2 to 3 m | 2 to 3 m | 2.2 to 6 m |
| Local storage | 1,000 entries | 1,000 entries | up to 1 million records |
| Stated accuracy | 80 to 90% | 80 to 90% | up to 99.8% |
So you need a network cable when the counter covers a wide main entrance, hangs from a ceiling above 3 meters, sees very heavy traffic or needs analytics beyond “in and out.” Check beforehand that your switch supports 802.3af, that its power budget covers every connected device, and that the cable run stays within the 100 meters Ethernet allows. At a narrow office door, a LoRaWAN counter is usually the simpler tool.
Milesight VS351 battery vs USB-C: which version should you buy?
The battery version is the true wireless people counter: you fix the mounting plate to the ceiling or above the door frame, snap in the sensor and the battery compartment, configure it by NFC from your phone, and you are done. If you want to retrofit people counting into a leased storefront, an older building with ornate ceilings or a freshly painted hallway, this is the least invasive route.
The USB-C version fits when power is already there or when battery swaps get expensive, for example because every service visit needs a lift or because many counters are in use.
Both versions react to heat and moving air. Milesight recommends an operating range of 15 to 30 °C and advises keeping the sensor away from radiators, vents, air conditioners, windows, glass and mirrors. Avoid metal directly below it. The battery version adds a radar rule: no large, constantly moving objects, such as an oscillating fan, inside the detection area.
Otherwise both share the same features: EU868, 16 dBm transmit power, NFC configuration, storage for 1,000 entries, data retransmission, D2D and an IP30 rating for indoor use. Interestingly, the Milesight datasheet lists the 5 V / 1 A USB-C input for both versions. The rated battery life, however, applies only to battery operation, and the startup message tells your application whether the device is currently running on battery or on Type-C power.
Can you install a people counter without mains power?
Yes. The battery-powered VS351 needs neither an outlet nor a network cable at the mounting point. The LoRaWAN gateway still needs power and internet, but it can sit anywhere in the building where both are available. Indoors, one gateway often covers several rooms or floors. Reinforced concrete and fire doors weaken the signal considerably, so test the connection at the planned spot before you drill.
The counting point is still set by the passage, not by convenience. Milesight names 60 to 100 centimeters as the ideal distance between the sensor and the head of a passing person. A device that is offset to the side or mounted too low picks up the edges poorly. After installation, measure the actual height and enter it in the configuration.
Without a gateway, only the D2D protocol works. It lets the counter switch a Milesight actuator, for instance, but it cannot deliver counts to an application. For real analytics you always need a LoRaWAN network, your own gateway or public coverage at the site.
How often does a wireless people counter report counts?
From the factory, the VS351 reports every 10 minutes, and you can change the interval in the app or by downlink. You can choose whether reports arrive at fixed clock times, which with a 10-minute interval means :00, :10 and :20, or at the chosen rate starting from the moment you apply the setting. Each report contains the entries and exits since the previous report and, optionally, the totals and battery level.
Second-by-second real time is not what LoRaWAN is built for. It is designed for small packets minutes apart, EU868 limits airtime with a 1 percent duty cycle, and every transmission costs battery. To react quickly, for example to a room occupancy limit, use thresholds: when the interval count or the total reaches a set value, the counter sends an alarm message immediately.
For demand-based cleaning or a weekly comparison of visitor numbers, 30 or 60 minutes is enough. For an occupancy display at the entrance, 10 minutes is more realistic. On the battery version, a 30-minute interval matches the conditions behind the manufacturer’s battery life figure.
One detail for your analysis: the total counter resets automatically at 65,535, and you can also schedule a daily reset at a fixed time, before which the counter reports its current total once more. Calculate visitor numbers from the interval values rather than from the difference between two totals, or every reset will show up as negative foot traffic.
What happens to people counts during a power outage?
That depends on which part of the chain loses power. In the simplest case, the sensor keeps running while the gateway or internet connection drops out, and the VS351 keeps counting. With local storage enabled, it saves up to 1,000 time-stamped entries, which lasts just under 7 days at a 10-minute interval and just under 21 days at 30 minutes. With retransmission enabled, it sends the missing data after reconnecting, by default every 600 seconds. Retransmitted packets carry a timestamp and use a different format, so your decoder must handle them. For the timestamps to be correct, the device clock must be synchronized, via the app or a network server that supports time sync.
In a building-wide outage, the options diverge. The battery version keeps counting and buffers its data while the gateway is offline. The USB-C version stops, and anyone who walks through the door during the outage is lost from the statistics; no amount of storage can fix that. A PoE counter stops too once its switch goes dark. If complete numbers matter to you, put the USB adapter or PoE switch, the gateway and the router on the same UPS.
When power returns, the VS351 rejoins the network and sends an information packet that flags the restart and the current power source. If the device restarts in the middle of a retransmission, it resends all pending data after reconnecting. Milesight does not clearly document whether the total counter survives a restart. Test it before going live: note the total, cut the power, switch it back on and compare.
Where can you buy a LoRaWAN people counter?
Start with the location: measure the passage and check that a sensor 2 to 3 meters up covers its full width. Then find out whether there is power at the counting point. If not, the battery version of the VS351 is the right fit, and you plan the reporting interval and battery swaps up front. If there is an outlet, the USB-C version saves you the maintenance. If a network cable is in place and you need high accuracy, greater mounting heights or analytics such as dwell time, a PoE counter like the VS125-P is the better class of device. LoRaWAN also needs a powered gateway within range; PoE needs an 802.3af switch with enough power budget. Decide in advance how often you want reports, and enable storage and retransmission so a network outage does not leave a gap.
If you are looking for a suitable LoRaWAN people counter, you can find our models in the people counters category.