Anyone who has circled a city block looking for a free space understands the problem that parking sensors try to solve. A parking sensor tells a central system whether one bay is free or taken. When thousands of those reports are combined, a city or car park operator can show drivers where the empty spaces are, check how parking is really used, and set rules based on data rather than guesses.
Many of these sensors send their reports over NB-IoT, a cellular radio technology built for small, battery-powered devices. This guide explains how an NB-IoT parking sensor works, why narrowband cellular is a common choice for this job, and which questions to ask before a rollout. It describes the technology in general terms and does not recommend any vendor.
What a parking sensor actually does
A parking sensor has one main job: decide whether a vehicle is sitting in a single parking space and report any change. Wikipedia's overview of on-street parking sensing describes several ways to get this data:
- Smartphone analytics, which estimate when parking sessions start and end from phone location and motion data.
- Camera analytics, which use computer vision on street cameras to estimate occupancy.
- Lamp post radars, ultrasonic or electromagnetic rangefinders mounted on street furniture.
- On-stall road sensors, battery-powered units fixed flush with, or on top of, the surface of each bay. These use magnetic, infrared and/or ultrasound sensing.
The same source notes that San Francisco and Los Angeles have run pilot programs that collected occupancy data from sensors and smart parking meters as part of matching parking prices to demand. It also says that on-stall sensors, because of their high spatial resolution and close proximity to the vehicle, give higher accuracy than camera analytics or radars. The trade-off is that road works and wireless infrastructure make the rollout harder to manage.
An "NB-IoT parking sensor" is usually an on-stall sensor of this kind that uses NB-IoT as its radio link.
How the sensor detects a car
On-stall sensors are small sealed units. The detection methods named in public sources are:
- Magnetic sensing. A car is a large mass of metal, so it disturbs the local magnetic field when it parks over the sensor. A magnetometer measures the strength or direction of a magnetic field, so it can pick up that change.
- Infrared sensing. An infrared detector can sense a vehicle body above the unit.
- Ultrasound. An ultrasonic transducer sends sound above 20 kHz and measures the echo to estimate distance to an object, the same ranging principle used for tank levels.
Once the sensor decides that the bay state has changed, it sends a short message: in effect "space 123 is now occupied" or "space 123 is now free". That is a small amount of data, and it does not need to be sent often. That pattern is a good match for a low-power wide-area network.
Why NB-IoT fits parking
A network built for tiny, infrequent messages
A low-power wide-area network (LPWAN) is designed for long-range, low-bit-rate links between battery-powered IoT devices such as sensors. Wikipedia gives typical LPWAN data rates of 0.3 kbit/s to 50 kbit/s per channel. That would be far too slow for video, but it is plenty for a sensor that sends a few bytes when a car arrives or leaves.
NB-IoT is one LPWAN technology. It was developed by 3GPP, the body behind mobile standards such as LTE. 3GPP announced in June 2016 that NB-IoT standardization was complete and the feature was frozen in Release 13 (LTE Advanced Pro).
Indoor and below-ground coverage
According to Wikipedia's NB-IoT article, the technology focuses on indoor coverage, long battery life and high connection density. NB-IoT uses part of the LTE standard but limits bandwidth to a single narrow band of 200 kHz.
Each of those three goals matters for parking:
- Coverage. A sensor set into asphalt, or placed in a multi-storey or underground car park, is in a hard spot for radio. A technology designed for deep indoor coverage has a better chance of reaching the network from there. Coverage still varies by site, so a field test is wise.
- Battery life. A sealed in-ground unit is costly to service. Every battery swap means a person in the road and, often, a closed bay.
- Connection density. A busy district can hold thousands of bays, each with its own sensor, all in a small area.
Managed cellular infrastructure
Because NB-IoT runs on mobile operator networks, a city does not have to build and maintain its own gateways, provided a local operator offers NB-IoT service at the site. Wikipedia notes that an LPWAN can be a private network or a service offered by a third party, so sensor owners can deploy devices without investing in gateway technology. That is the main practical difference from a private LoRaWAN rollout, where the owner or a partner runs the gateways.
NB-IoT compared with the main alternatives
Parking sensors are also sold with other radios. Without naming products, the general trade-offs are:
| Option | How it connects | Points to weigh for parking |
|---|---|---|
| NB-IoT | Licensed cellular spectrum through a mobile operator | Designed for indoor coverage and long battery life; needs operator NB-IoT coverage and usually a per-device subscription |
| LTE-M | Licensed cellular spectrum through a mobile operator | Higher data rate and mobility than NB-IoT, but more bandwidth and more cost; mobility is little help for a sensor that never moves |
| LoRaWAN | Unlicensed spectrum, gateways run by the owner or a network provider | No operator needed, but someone must plan, install and maintain gateways |
| Cameras | Wired or high-bandwidth wireless | One camera can cover many spaces; Wikipedia notes lower accuracy at night and in bad weather |
On LTE-M, Wikipedia states that its advantages over NB-IoT are higher data rate, mobility and voice, and that it needs more bandwidth and costs more. It also notes that NB-IoT devices moving between cells must disconnect and re-register, while LTE-M supports seamless handover. That is why LTE-M is often chosen for trackers. A parking sensor is fixed in place, so the handover gap rarely matters.
What to check before a rollout
These are general questions, not a procurement checklist, and the answers depend on the site:
- Is there NB-IoT coverage at the actual bays? Ask the operator for coverage at the specific streets or levels, then test a few units in the worst spots (lowest basement, metal-roofed decks).
- Which detection method suits the site? Magnetic sensing may behave differently near tram rails, steel structures or buried cables. Ask the vendor how the sensor handles neighbouring bays and large vehicles.
- What battery life does the vendor claim, and on what assumptions? Battery life depends on how often the bay changes state and how often the device checks in. Wikipedia's LPWAN article notes that makers claim years to decades of battery life but that real-world tests have not confirmed this. Ask for field data.
- How are units installed and replaced? Flush-mounted sensors need drilling. Surface-mounted ones are faster to fit but sit above the road surface.
- Where does the data go? Check whether occupancy data can be exported in an open format to your own systems or a parking guidance display, not just a vendor dashboard.
- What is the ongoing cost model? Cellular sensors usually carry a connectivity fee for each device, on top of the hardware. Compare total cost over the expected life of the sensor, not just the purchase price.
- Security and updates. Ask how the device is authenticated on the network, whether firmware can be updated remotely, and how long the vendor will support it.
How parking sensor data gets used
Occupancy data from bay sensors usually feeds a parking guidance and information (PGI) system. Wikipedia describes PGI systems as combining traffic monitoring, communication, processing and variable message signs to direct drivers to car parks with free spaces. It lists Internet of Things technologies among their parts, next to ultrasound detectors, camera sensors, zone controllers and signs. The stated aim is to cut search time, which in turn reduces congestion on surrounding roads.
The same data can also support enforcement (spotting overstays), planning (seeing which streets are always full) and demand-based pricing, as in the San Francisco and Los Angeles pilots noted above. Any use that links occupancy to a particular vehicle or person raises privacy questions. Treat those as a policy decision, not just a technical one.
The bottom line
An NB-IoT parking sensor is a small, battery-powered unit in or on a parking bay. It detects a vehicle, usually magnetically and sometimes with infrared or ultrasound, and sends a tiny status message over a narrowband cellular network. NB-IoT suits this job because it was designed for indoor coverage, long battery life and many devices in a small area, and because it runs on operator infrastructure. Before buying, check real coverage at the bays, the vendor's battery assumptions, and the per-device running cost.
Sources
- Wikipedia, On-street parking sensing: https://en.wikipedia.org/wiki/On-street_parking_sensing
- Wikipedia, Parking guidance and information: https://en.wikipedia.org/wiki/Parking_guidance_and_information
- Wikipedia, Narrowband IoT: https://en.wikipedia.org/wiki/Narrowband_IoT
- Wikipedia, LTE-M: https://en.wikipedia.org/wiki/LTE-M
- Wikipedia, Low-power wide-area network: https://en.wikipedia.org/wiki/Low-power_wide-area_network
- Wikipedia, Magnetometer: https://en.wikipedia.org/wiki/Magnetometer
- Wikipedia, Ultrasonic transducer: https://en.wikipedia.org/wiki/Ultrasonic_transducer
- 3GPP, "Standardization of NB-IOT completed" (June 2016): https://www.3gpp.org/news-events/3gpp-news/nb-iot-complete