For battery temperature sensors on every floor of an office building, why is ordinary Wi-Fi usually the wrong radio, and which links fit better?
answer
- power budget before data rate
- a few bytes, years on a coin cell
- sleep almost always, wake cheaply
- hub or border router to the LAN
basics
~20 sA coin-cell sensor sending a few bytes a minute needs a radio that sleeps almost always; classic Wi-Fi spends energy on rates it never uses. Bluetooth LE, Zigbee, Thread or Z-Wave fit, reaching IP via a hub, gateway or border router.
solid answer
~50 sThe deciding property for a battery sensor is its **energy budget**, not its data rate: it sends a few bytes every few minutes and must last years. A building's Wi-Fi is designed for throughput, and a client that stays associated to an access point and wakes to keep up with it spends far more energy per useful byte than the reading is worth. Low-power links fit better: **Bluetooth LE** (2.4 GHz, a star to a phone or gateway), **Zigbee** and **Thread** (low-power mesh radios), and **Z-Wave** (a sub-GHz mesh). Most of them do not put the sensor straight onto the building LAN, so a hub, gateway or Thread border router joins them to the IP network. Wi-Fi has narrowed the gap, with Target Wake Time in Wi-Fi 6 and the sub-1 GHz `802.11ah` (HaLow), but it stays the natural choice for mains-powered, high-rate devices such as cameras and displays.
go deeper
Recall that a battery sensor's first constraint is energy, not speed, and name the low-power links: Bluetooth LE, Zigbee, Thread, Z-Wave. Know that a hub or border router joins them to the IP network.
Explain why Wi-Fi costs a sleeping device energy: staying associated, waking to listen, rejoining and a heavier stack. Contrast that with links built to sleep most of the time and wake briefly.
Show you know where Wi-Fi has moved: Target Wake Time since Wi-Fi 6 and HaLow below 1 GHz. Judge which building devices still belong on ordinary Wi-Fi and which should move to a low-power link.
Frame the choice as battery-replacement labour against infrastructure: a low-power link means hubs or border routers to run and patch, Wi-Fi means more battery changes. Decide which cost the facilities team should own.
## The question a battery sensor asks A temperature sensor on an office floor has a very small job. It measures, sends a reading of a few bytes, and goes back to sleep, perhaps once a minute or once every few minutes. It runs on a coin cell or a small battery and nobody wants to climb a ladder to change it for years. That makes its **power budget** the first property in the choice of radio, ahead of data rate, ahead of range and ahead of which standard the IT team already knows. Such a device is usually **constrained** in the sense of RFC 7228, which sorts constrained nodes into classes: a Class 1 device has roughly 10 KiB of RAM and 100 KiB of code space, a Class 2 device roughly 50 KiB and 250 KiB. A sensor at that size can run a lean protocol stack, but every millisecond its radio is awake costs battery. ## Why classic Wi-Fi spends the battery Wi-Fi is built to move a lot of data to laptops, phones and screens. For a sensor that sends almost nothing, that design works against it: - **The radio is tuned for throughput.** Wi-Fi spends its effort on high rates, wide channels and many clients, none of which a few-byte reading needs. - **Staying connected costs energy.** A client has to stay associated to an access point, which means waking to listen for it; a radio that could sleep for minutes is woken far more often. - **Joining costs energy too.** A device that fully disconnects to save power pays for association and key exchange again each time it wakes. - **The stack is heavier.** A Wi-Fi client carries a full IP stack and Wi-Fi security, a heavy load for a Class 1 device. None of this makes Wi-Fi bad. It makes Wi-Fi the wrong tool for a coin-cell device whose whole traffic is a few bytes. ## The low-power candidates | Link | Band | Topology | How it reaches the IP network | |---|---|---|---| | **Bluetooth LE** | 2.4 GHz ISM (2400-2483.5 MHz) | star to a central device; broadcast | a gateway, or IPv6 over Bluetooth LE (RFC 7668) through a border router | | **Zigbee** | 2.4 GHz and sub-GHz bands | self-healing mesh | a hub that translates | | **Thread** | IEEE 802.15.4 radio | IPv6 mesh | a border router that routes IPv6 | | **Z-Wave** | sub-GHz (ITU-T G.9959) | mesh; Long Range is a star | a controller that translates | | **Wi-Fi HaLow** (`802.11ah`) | sub-1 GHz | star around an access point | native IP | Each of these radios is built to sleep most of the time and wake briefly to send. The mesh links (Zigbee, Thread, Z-Wave) also let mains-powered devices, such as light fittings, relay for the battery ones, so a sensor in a far corner of a floor still has a path to its hub. ## Where Wi-Fi does fit Wi-Fi has moved towards low power in two ways, and an answer that ignores them is out of date: 1. **Target Wake Time**, listed among the key features of Wi-Fi 6 (2018), lets a client and an access point agree when the client will next wake, which the Wi-Fi Alliance credits with better device battery life. 2. **Wi-Fi HaLow** (`802.11ah`, certified by the Wi-Fi Alliance from 2017) runs below 1 GHz with a range of about 1 km, native IP support and, in the Alliance's words, months to years of coin-cell battery life. So the honest rule is not "Wi-Fi can never run on a battery". It is that a building's ordinary 2.4/5/6 GHz Wi-Fi is sized for throughput, and the low-power links above usually give a coin-cell sensor more years for less effort. Wi-Fi remains the right link for mains-powered devices that move real data: cameras, displays, access-control panels with screens. ## What a hub, gateway or border router is for Most low-power links do not put the sensor directly on the building LAN. Something sits between the two networks: 1. A **hub or gateway** for Zigbee or Z-Wave runs the radio network, keeps track of its devices and translates their readings into messages an IP application can use. 2. A **Thread border router** routes IPv6 packets between the Thread mesh and the LAN; the sensor already speaks IPv6 and nothing has to translate it. 3. A **Bluetooth LE gateway** collects readings from nearby devices; with IPv6 over Bluetooth LE, the central device usually acts as a 6LoWPAN border router. ## A first-pass rule - Battery device, small and infrequent data, indoors: a low-power link (Bluetooth LE, Zigbee, Thread, Z-Wave). - Mains-powered device with real data: Wi-Fi. - Battery device that must be native IP and reach far: consider HaLow. - Then ask how each link reaches IP, because that decides what infrastructure the building must run.
- Does Target Wake Time in Wi-Fi 6 make ordinary Wi-Fi a good choice for coin-cell sensors?It helps: a client and access point agree when the client next wakes, so the radio sleeps longer. But the client still needs a full Wi-Fi and IP stack and an access point that supports the feature, and the radio is still sized for throughput. For a few bytes a minute, a low-power link usually still wins on battery years; Wi-Fi with Target Wake Time suits devices that need Wi-Fi anyway.
- Why do low-power mesh links still need a hub or border router in the building?The mesh is a separate radio network. Zigbee and Z-Wave do not speak IP to the LAN, so a hub or controller terminates the radio network and translates readings into IP messages. Thread devices do speak IPv6, but a border router is still needed to route between the 802.15.4 mesh and the Ethernet or Wi-Fi LAN.
- Wi-Fi is always better because it is faster and already installed
- Data rate is the first thing to compare for a sensor
- Wi-Fi can never run on a battery for months
- Zigbee and Z-Wave sensors get their own address on the building LAN
- Zigbee and Z-Wave sensors reach the LAN with no hub at all
Assumes Wi-Fi 7 as the current generation; Target Wake Time, the power-saving feature cited, is listed among Wi-Fi 6's key features (2018).