Smart cities, explained: sensing the street and its trade-offs
What a smart city really is once you drop the marketing — the domains, the sensing tech, the genuine gains, and the honest criticisms.
“Smart city” is one of those phrases that has been stretched until it means almost nothing. It gets stamped on a new traffic-light controller, a phone app for reporting potholes, and a glossy plan for a district that does not exist yet. Strip away the brochure language and a plainer idea is left: a city that instruments some of its physical systems — roads, pipes, bins, buildings, air — so it can measure what they are doing and, sometimes, adjust them.
That is the useful working definition. A smart city is not a place with a personality; it is an ordinary city that has wired up parts of itself to collect data and act on it. The interesting questions are which parts, to what end, and who decides. Those questions matter more than the technology, and they are where most of the disappointment — and most of the criticism — comes from.
What actually gets made “smart”
It helps to break the field into domains, because a city rarely does all of this at once. Most start with one or two areas where the payback is clear and the politics are gentle.
- Mobility and transport. Adaptive traffic signals, sensors that count vehicles and pedestrians, parking bays that report whether they are occupied, and real-time arrival boards for public transport. The aim is usually to move people with less delay and fewer idling engines.
- Utilities and energy. Metering that reports consumption more often than once a quarter, streetlights that dim when nobody is about, and grid equipment that flags a fault before crews are dispatched blind.
- Water and waste. Flow and pressure sensors that catch leaks in the network, and fill-level sensors in bins so collection rounds skip the containers that are still empty.
- Environment and air quality. Low-cost monitors for particulates, nitrogen dioxide, noise and temperature, spread across a city to build a finer-grained picture than a handful of official reference stations can give.
- Public safety. Cameras, gunshot or crowd-density detection, and flood or fire early-warning systems. This is the domain where the benefits and the objections collide most sharply.
- Digital public services. The unglamorous but often most valuable layer: online permits, transparent service-request tracking, and open datasets that let residents and small businesses build their own tools.
The table below is a rough map, not a ranking. A city can do any of these well or badly.
| Domain | What it typically senses | The usual goal |
|---|---|---|
| Mobility and transport | Vehicle and pedestrian counts, parking occupancy | Less congestion, better transit information |
| Utilities and energy | Consumption, streetlight state, grid faults | Lower waste, faster repairs |
| Water and waste | Flow, pressure, bin fill levels | Fewer leaks, efficient collection rounds |
| Environment | Particulates, gases, noise, temperature | A finer map of local conditions |
| Public safety | Video, acoustic events, flood and fire signals | Faster response, early warning |
| Digital services | Service requests, permits, published datasets | Convenience and accountability |
The technology underneath
Most of what makes a city “smart” is the same building block used at home or in a single building, multiplied across a metropolitan area. If you have read how the Internet of Things works, the pattern will be familiar: cheap sensors, a way to move their readings, somewhere to store and interpret the data, and — occasionally — something that acts on the result. The Internet of Things is the substrate; the city is just an unusually large deployment of it.
Three ingredients tend to define the city-scale version.
Low-power, long-range sensor networks. A parking sensor or a bin monitor cannot be trenched into mains power, and it may need to run for years on a single battery. That rules out power-hungry radios. Protocols such as LoRaWAN (Long Range Wide Area Network) exist for exactly this: they send small packets over distances measured in kilometres, using unlicensed radio spectrum, at data rates that would be useless for video but are ample for a temperature reading or a “bin 78% full” message. A handful of base stations can cover a district. The trade-off is bandwidth — these networks are deliberately slow, which is fine for the job they are given.
Open data platforms. Sensors are worth little if their output is locked in six separate vendor dashboards that cannot talk to each other. The more capable programmes publish standardised data through open interfaces, so a traffic feed and an air-quality feed can be combined, and so people outside the council can build on them. Open data is as much a policy choice as a technical one, and it is often what separates a city that learns from its instruments from one that merely owns some.
City-scale digital twins. A digital twin is a continuously updated virtual model of a physical thing, fed by live data. At city scale that might be a 3D model of the street network and buildings, wired to sensor feeds, used to simulate a road closure, a flood, or the shadow a proposed tower would cast before anyone pours concrete. Twins are genuinely useful for planning and “what if” testing. They are also easy to oversell: a model is only as honest as the data flowing into it, and a beautiful rendering can lend false confidence to a shaky assumption.
Where the value is real
None of the scepticism that follows should imply the whole idea is empty. There are concrete, unglamorous wins.
Leak detection in a water network can save enormous volumes of treated water that would otherwise vanish underground unnoticed. Adaptive signals and good transit information can shave time and emissions off everyday journeys. Dense air-quality sensing gives a neighbourhood evidence it can act on, rather than a single citywide average that hides the street canyon where children walk to school. Bin sensors cut pointless collection trips. And plain digital services — being able to track a reported fault instead of phoning and hoping — quietly improve the relationship between residents and the people meant to serve them.
The common thread is modest, measurable improvement to systems people already depend on. That is a much better test than novelty.
In short: A smart city is an ordinary city that instruments some of its systems to measure and adjust them. The technology — low-power sensors, open data, digital twins — is well understood. The hard part is governance: deciding what to sense, who sees the data, who benefits, and whether the problem was ever technical in the first place.
The criticisms, taken seriously
A guide that only listed benefits would be a brochure. The objections to smart-city projects are substantial, and the better cities treat them as design constraints rather than public-relations problems.
Cost and the maintenance tail. Buying and installing sensors is the cheap part. The expensive part arrives afterwards: batteries die, mountings corrode, firmware needs patching, calibration drifts, and the vendor who supplied the platform wants an annual fee. A network of thousands of devices is a permanent operating commitment, not a capital project you cut a ribbon on and forget. Plenty of pilots have quietly rotted because nobody budgeted for year three.
Surveillance. Much of this infrastructure watches public space by design, and the line between “counting pedestrians to time a crossing” and “tracking who walks where” is a matter of policy, not physics. The same camera can serve either. This is why the questions raised in privacy in intelligent environments apply with extra force outdoors, where nobody consented by walking down the street. Data minimisation — collecting the least that does the job, and discarding it quickly — is the difference between a sensor and a dragnet.
Equity, and who actually benefits. Instrumentation tends to follow money. The district that gets the smart lighting, the clean-air sensors and the responsive transit app is often the one that was already well served. If a programme optimises the commute of people who drive to the centre while doing nothing for the outer suburb with one bus an hour, it has made the city more efficient for the people who needed help least. Asking “better for whom?” is not a nicety; it decides whether the project is worth doing.
Vendor lock-in. Proprietary platforms, closed data formats and bespoke integrations can quietly hand a private supplier control over a public system. When the contract renews, the city may find that its data, its dashboards and its ability to change anything all live inside one company’s product. Open standards and clear rights to the data are the antidote, and they are far easier to insist on at procurement than to claw back later.
Solutionism. The deepest criticism is that technology often gets pointed at problems that are really about governance, funding or politics. Congestion is frequently a question of how road space and pricing are allocated, not of smarter signals. Illegal dumping is about enforcement and services, not only bin sensors. Dressing a political choice as a technical upgrade can make it look neutral and inevitable when it is neither. A useful habit is to ask, before buying anything, whether the underlying problem would yield to a rule, a budget line or a staffing decision instead. Sometimes the answer is a sensor. Often it is not.
How to read a smart-city claim
For a curious resident or a sceptical councillor, a few questions cut through most of the marketing. What specific problem does this solve, and for whom? What data does it collect, who can see it, and how long is it kept? Who owns and maintains it in five years, and what happens when the contract ends? Could a change in policy achieve the same thing more cheaply? Is the improvement being measured, or merely announced?
None of that requires a technical background. The engineering — a battery sensor here, an edge node there, a data platform behind it — is the settled part. The judgement is in deciding what deserves to be sensed at all, and making sure the answer serves the whole city rather than the part that already had the best of everything. A smart city, done honestly, is less a product than a set of choices anyone can hold to account.