Smart Parking: How It Works and How Cities Deploy It

Smart parkingSUse caseSmart parkingView profile tells drivers and cities which spaces are free right now, and a well-run deployment changes how a street, a car park, or a loading bay is used.

Anyone who has circled a city block looking for a space knows the problem. What is less obvious is how much of it is an information problem. Free spaces often exist a street or two away; the driver simply does not know. Meanwhile the city has little data on how its kerbside is used: how long vehicles stay, which loading bays are blocked all morning, which disabled bays are occupied by vehicles without a permit.

Smart parking puts a sensor, or a camera, on each space or group of spaces and turns occupancy into data. On top of that data you can build guidance signs, apps, enforcement routes, pricing policies, and planning reports. The sensor is the easy part. The hard parts are choosing the right detection method, keeping thousands of battery devices alive under the asphalt, and making the data useful to the people who run the streets.

This guide covers each of those, with the Spanish regulatory context and a practical rollout plan. It is written for integrators and city or car park operators preparing a first project.

What Smart Parking Is and What It Solves

A smart parking system detects whether each space is free or occupied, sends that state to a platform, and makes it available to drivers, operators, and city services.

It addresses four different problems, and it helps to know which one a project is really about, because each leads to different choices:

• Search traffic. Drivers looking for a space add traffic and emissions. Guidance signs and apps that show free spaces shorten the search.
• Rotation and enforcement. In regulated zones, the city wants vehicles to rotate. Occupancy data shows which spaces exceed the permitted time, so enforcement officers can plan routes instead of patrolling at random.
• Special bays. Loading bays, disabled bays, and electric vehicle charging bays are scarce and often misused. Monitoring them individually has a high value per sensor.
• Planning. Months of occupancy data show how the kerbside is actually used, which supports decisions on pricing, bay allocation, or turning parking into other uses.

A project aimed at guidance needs high accuracy and real-time updates on every space in an area. A project aimed at planning can work with a sample of streets and slower updates. Asking which problem comes first saves a lot of money.

Sensor Types: Magnetometer, Radar, and Camera

There are three main ways to detect a vehicle, and most deployments combine them.

In-ground sensors for streets come in two forms. Flush-mounted units sit in a hole drilled into the asphalt and survive street cleaning and snow ploughs. Surface-mounted units are glued on top: faster to install, but more exposed to damage and theft.

Cameras are attractive where there are many spaces in view, such as an open car park or a long stretch of street, and power is available. They also bring obligations. In Spain, the data protection authority's guide on video surveillance applies as soon as a camera captures identifiable people or number plates. If the project only needs occupancy, process images at the camera and send only the count.

Covered Car Parks Are a Different Problem

Indoor and underground car parks change most of the assumptions above. Power and wired networks are usually available, so sensors do not need to live on batteries. The common choice is an overhead ultrasonic or infrared sensor per bay with a coloured light that drivers can see from the aisle.

If the operator only needs free spaces per level, counting vehicles at the entrances and exits of each level is far cheaper than a sensor per bay. The count drifts over time and needs a daily reset, but for a guidance sign at the entrance it is often enough.

Radio is the other difference. Mobile coverage underground is often poor, so cellular sensors may not work at all. A LoRaWANProtocolLoRaWANOpen long-range, low-power LPWANView profile gateway installed inside the car park, connected by cable, usually solves it.

Whatever the method, accuracy in the brochure and accuracy in your street are different numbers. Ask for results under conditions like yours: angled or parallel parking, the size of vehicles in the area, buses or vans in adjacent spaces, and winter.

Connectivity and Batteries for Parking Sensors

In-ground sensors run on batteries and have to send data from under a vehicle, at street level, often between buildings. That is a demanding radio position.

Most street deployments use one of two low-power networks:

• LoRaWAN, with gateways on buildings or lamp posts that the city, an integrator, or an operator runs. One gateway covers many streets, and the same network can serve lighting, waste, and other services. Coverage at ground level under a car needs a survey; the method is in our guide to LoRaWAN range, and city architectures are covered in LoRaWAN for smart cities.
• NB-IoT, over a mobile operator's network, with no gateways to install but a connectivity fee per sensor. It works well where the operator's coverage is good at street level. See our guide to NB-IoT.

Parking sensors are event-driven. They send a message when a vehicle arrives or leaves, plus a periodic heartbeat so the platform knows they are alive. Take a busy space with 20 changes a day and an hourly heartbeat: about 44 messages a day. A quiet residential space might send half that. Battery life therefore depends on rotation as much as on the hardware, and the busiest streets run out first.

The other big factor is the radio itself. A sensor under a vehicle, far from a gateway, transmits at a high spreading factor and spends far more energy per message. The arithmetic is in our article on LoRaWAN battery life. Before signing, ask vendors for the battery estimate at your expected number of events per day and at the worst coverage in the area, not at the best case.

Plan for replacement from the start. A flush-mounted sensor with a sealed battery means opening the asphalt again. Know how many visits that implies per year before the first sensor goes in.

From Occupancy Data to Services

Occupancy data is only useful if it reaches someone who acts on it. The most common services, from simplest to most demanding:

1. Guidance signs. Panels at the entrance to a zone or a car park show free spaces per street or level. They need reliable, real-time data for the area they cover.
2. Apps and open data. The same occupancy feed goes to the city's app, to navigation apps, or to an open data portal. An API that partners can use is worth more than a single proprietary app.
3. Enforcement support. Officers see which spaces exceed the permitted time or which special bays are occupied, and plan their routes. This needs the data to be matched with permits or payments.
4. Special bay monitoring. Loading bays freed at the right times, disabled bays checked against permits, charging bays not occupied by vehicles that are not charging.
5. Planning and pricing. Occupancy by hour and by street over months supports decisions on tariffs, time limits, and changing the use of the kerbside.

Each step adds integration work: with the parking payment system, the permit register, the enforcement team's tools, or the city's open data platform. A platform with a documented API turns that into configuration; the integration patterns are in our API integration guide.

Data Rules Worth Writing Down

Occupancy data looks simple, but a few rules decide whether people trust it:

• Ignore short stops. A vehicle that stops for 30 seconds to drop someone off should not show the space as occupied. Agree on a minimum time before a change counts.
• Unknown is not free. A sensor that has missed its heartbeats should show its space as unknown. Showing it as free sends drivers to a space that may be taken.
• Mark stale data on signs. If a zone loses data, the panel should say so rather than freeze on the last number.
• Recalibrate after works. Resurfacing or new cables near a magnetometer change its baseline. Recalibration should be part of the works procedure, not discovered weeks later.

There is also a service for drivers that costs almost nothing extra: a public dashboard that shows occupancy by street. It does not replace an app, but it puts the data to work from day one.

Smart Parking in Spain: Regulation and a Real Reference

Two things make smart parking timely in Spain.

The first is regulation. The Climate Change and Energy Transition Law 7/2021 requires municipalities with more than 50,000 inhabitants, and island territories, to adopt sustainable urban mobility plans that include low emission zones. Managing access, regulated parking, and loading in those zones is much easier with data on how the kerbside is actually used. Smart parking is one of the few ways to get it street by street.

The second is that Spain has a well-documented early reference. The SmartSantander project, an EU-funded research testbed in the city of Santander, buried several hundred magnetic parking sensors under the asphalt of the city centre, each one detecting whether its space was occupied through the change in the magnetic field, and connected them through a wireless sensor network. It was a research deployment more than a commercial service, and the technology has moved on since, but it showed early what in-ground sensors can and cannot do at city scale: the importance of battery life, radio coverage at street level, and maintenance once thousands of devices are under the road.

Today, loading bays and disabled bays are a common place to start, because the value per sensor is highest there. The difficulty usually appears later, when a pilot that works on three streets has to be maintained across a whole district. For a broader view of how leading cities approach these programmes, see our review of the top smart cities of 2025.

How to Deploy Smart Parking Without Surprises

A rollout plan that avoids the usual problems:

1. Define the goal and the area. Guidance, enforcement, special bays, or planning. The goal decides the detection method, the density, and the accuracy you need.
2. Survey coverage at ground level. Measure the radio signal where the sensors will be, under a parked vehicle if possible, not at head height on the pavement.
3. Pilot and measure accuracy. Install on a few streets with different layouts. Compare sensor readings with manual counts or camera footage over at least a week, including peak hours and night. Agree on how accuracy will be measured before the contract, not after.
4. Integrate before scaling. Connect payments, permits, enforcement tools, and signs during the pilot. An occupancy map nobody uses is the most common outcome of a smart parking project.
5. Roll out by area. Complete whole zones, so guidance and statistics are reliable for that zone, instead of scattering sensors across the city.
6. Plan maintenance. Battery replacement, damaged sensors, roadworks that remove sensors, and recalibration after resurfacing. Write down who does each and how quickly.

Data ownership deserves a clause of its own. The city should be able to export its occupancy history and move to another provider without losing it. That is easier when sensors, network, and platform are not locked together.

The Platform Behind the Sensors

The platform receives occupancy events, keeps the state of each space, stores history, raises alarms for silent or faulty sensors, shows dashboards, and serves data to apps, signs, and other systems.

For integrators, two properties matter beyond the basics. The first is network independence: being able to receive from LoRaWAN and NB-IoTProtocolNB-IoT3GPP-standardized cellular LPWAN — carrier coverageView profile sensors, and from different manufacturers, in the same platform. The second is the ability to serve several clients, for example several municipalities or car park operators, each with its own data, users, and brand.

Cloud Studio IoTITermIoT (Internet of Things)The IoT (Internet of Things) is the network of physical objects with sensors, software and connectivity that collect and exchange data and act autonomously.View profile receives LoRaWAN data through external network servers, such as The Things Stack, LORIOT, or ThingPark X, and cellular sensors can send directly over MQTTProtocolMQTTThe standard pub/sub protocol of IoTView profile or HTTPS. It is multi-tenant and white-label, so an integrator can run each city or operator as a separate client under its own brand; see our article on the white-label IoT platform. Dashboards can be shared through public links, alarms can go out by email, SMS, WhatsApp, voice call, or push, and the API lets apps and signs read occupancy in real time. The smart city solution page shows how parking fits with lighting and other city services.

Conclusion

Smart parking works when it starts from a clear goal and ends in a service someone uses. The main points:

• Decide first which problem you are solving: search traffic, enforcement, special bays, or planning.
• Magnetometer plus radar is the usual choice for streets; cameras suit open areas with power, with data protection obligations.
• Battery life depends on rotation and radio coverage at ground level. Ask for estimates at your worst conditions.
• Integration with payments, permits, enforcement, and signs is what turns occupancy data into results.
• In Spain, low emission zones under Law 7/2021 make street-level data more valuable than ever.
• Roll out by complete zones, measure accuracy before scaling, and keep ownership of the data.

If you are preparing a smart parking project or tender and want to review the architecture, talk to the team.

Frequently asked questions

What is smart parking?

It is a system that detects whether each parking space is free or occupied, sends that state to a platform and makes it available to drivers, operators and city services through signs, apps, enforcement routes and planning reports.

How do smart parking sensors work?

Most street sensors use a magnetometer, which detects the change in the Earth's magnetic field caused by the vehicle's metal. Adding a short radar reading to confirm it gives much better accuracy. Covered car parks usually use an overhead ultrasonic or infrared sensor per bay, and cameras suit open areas with power.

Which network do parking sensors use?

Most street deployments use LoRaWAN, with gateways on buildings or lamp posts, or NB-IoT over a mobile operator's network. Underground, mobile coverage is often poor, and a LoRaWAN gateway installed inside the car park and connected by cable usually solves it.

How long does a parking sensor battery last?

It depends on rotation as much as on the hardware. Sensors send a message when a vehicle arrives or leaves, plus a periodic heartbeat, so a busy space with 20 changes a day sends about 44 messages a day and runs out first. Ask vendors for the estimate at your number of events and at the worst coverage in the area.

Do smart parking cameras raise data protection issues?

Yes. In Spain, the data protection authority's guide on video surveillance applies as soon as a camera captures identifiable people or number plates. If the project only needs occupancy, process the images at the camera and send only the count.

Where should a city start with smart parking?

Loading bays and disabled bays are a common place to start, because the value per sensor is highest there. Define the goal first, survey coverage at ground level, pilot on a few streets while measuring accuracy against manual counts, and roll out by complete zones.