Sector explainer · Cities

The street, measured block by block.

Air quality, noise, traffic, urban heat and street flooding, with open data as a procurement credential.

Scenarios

4

Sensing layers

4

Templates

4

Open projects

6

01 · Air quality

Co-location

Low-cost sensors earn trust by co-location.

Reference analysers plus calibrated low-cost nodes; low-cost sensors alone do not meet regulatory requirements. The method is a co-location: a low-cost node sits beside a reference analyser for weeks, a correction is fitted, and the corrected node is checked on days it has not seen. The correction here is a least-squares fit, reference = a + b·raw + c·max(0, RH − 70)², which also removes the humidity swelling optical counters read as particles.

Error before correction

—

Error after correction

—

Variance explained (R²)

—

Fitted gain and offset

—

Errors are root-mean-square, in µg/m³, on the held-out days. A corrected node densifies the network; the reference analyser still carries compliance.

02 · Heat and street flooding

Microclimate

Heat plans need a node every few blocks.

Depth sensors combined with temperature, humidity and radiation nodes on poles for heat-action plans. Spacing sets the node count, n = ⌈A ÷ s²⌉, and the count sets the maintenance load: every node is a battery, a calibration and a bracket to inspect.

Nodes

—

Nodes per km²

—

Site visits a year, two per node

—

03 · Noise, traffic and mobility

Privacy by design

Count the street. Never the person.

Sound-level nodes, radar or video counters and probe sensing over low-power radio or 5G; privacy assessment required. The design rule: counters keep counts, never images or identifiers, and small counts are suppressed at the edge before anything is published.

Bins suppressed

—

Share of all passages still published

—

One counter, one day, 15-minute bins with a morning and an evening peak. The threshold is set by the city’s privacy impact assessment; the edge applies it.

04 · City-wide radio layer

Radio planning

One radio layer, shared by every sensor.

A city-wide connectivity layer as digital public infrastructure. Low-power wide-area gateways on rooftops, or private 5G small cells where bandwidth matters: either way the count follows from the planned cell radius: hexagonal cells of (3√3 ÷ 2)·r², so n = ⌈A ÷ 2.6 r²⌉. Dense-urban LoRaWAN is usually planned at 1 to 3 km; mid-band small cells at a few hundred metres.

Radio

Sites on a hexagonal grid

—

Area per site

—

05 · Proof

Evidence

What a city asks to see.

Cities buy open, accountable data. Each item below is produced by the grid as a record while it runs.

• 01Co-location calibration reports against reference monitors
• 02Data-completeness service levels
• 03An open-data API
• 04A privacy impact assessment for video and probe sensing
• 05Accessible dashboards

air quality sensor network

low-cost sensor calibration

hyperlocal air quality

urban heat island sensors

flood sensors

smart city LoRaWAN

open data portal integration

Open work in the field: Sensor.Community, OpenAQ, FloodNet, Smart Citizen Kit, Telraam, openSenseMap.