Lesson 1.1Lesson 1.1 · IoT & Sensing
Sensors & What They Measure
The building's senses - temperature, humidity, air quality, occupancy, light, energy, water and more, and what each one is actually for
A building with no sensors is blind - it cannot know it is too warm, the air is stale, or a pipe is leaking. Sensors are how it opens its eyes.
Every insight in this whole course starts with a measurement. Before a digital twin can predict a failure or a BMS can hold a room comfortable, some sensor has to turn a real, physical thing - heat, moisture, movement, a flow of electrons - into a number a computer can read.
That is the humble, essential job at the base of the stack. In this lesson we tour the building's senses: what the common sensors measure, what each reading actually tells you, where the number travels next, and - the part beginners skip - why a sensor in the wrong spot or drifting out of true can quietly poison everything built on top of it.
One sensor = one quantity + one place + one decision. No decision = noise you pay to store.
A sensor is a transducer: it turns the physical into the digital
Strip away the marketing and a sensor is a transducer - a device that converts a physical quantity into an electrical signal that can be measured and, ultimately, a number. A thermistor changes resistance as it warms; a photodiode makes current in light; a microphone turns pressure waves into voltage. Everything a smart building knows about itself enters through devices like these.
Two ideas matter from the start. First, each sensor measures one quantity in one place. A temperature sensor tells you the temperature at that sensor, not the average of the room - a distinction that trips up a great many analyses. Second, the raw signal is usually analogue (a continuous voltage or resistance) and must be sampled - read at intervals and converted to a digital value by an analogue-to-digital converter. How often you sample, and how accurately you convert, are design choices that decide how much of reality you actually capture.
So when we say a building can 'sense its state', we mean it carries a population of these single-purpose transducers, each reporting one number, again and again, from one spot. The art is choosing which quantities to measure, where, and how often - because those choices define the ceiling on everything the twin can ever conclude.
Sensor = transducer. Physical quantity -> voltage -> number. One quantity, one place.
The core measurements - and what each one tells you
A working smart building leans on a fairly small, repeatable set of senses. Temperature (a thermistor or RTD, in degrees C) is the oldest and most common - it drives comfort and every heating and cooling decision. Relative humidity (a capacitive sensor, in percent) is its partner: comfort is thermal and moist, and high humidity signals condensation and mould risk. Carbon dioxide (usually an NDIR sensor, in parts per million) is the workhorse of indoor air quality - because people exhale CO2, rising ppm is a proxy for stale air and under-ventilation; broader IAQ adds particulates (PM2.5), VOCs and sometimes radon.
Occupancy is often a PIR (passive infrared) sensor that detects the heat-movement of bodies, giving a simple true/false 'is someone here'; more advanced options count people or map desks. Light is measured in lux by a photodiode, telling you daylight levels so lighting can dim to match. Energy and power come from current transformers and meters, in kW (instantaneous) and kWh (cumulative) - the single most valuable stream for cost and carbon. Water sensors catch flow and, critically, leaks - a cheap contact that has saved many a building from six-figure damage. Vibration (an accelerometer) watches rotating plant - a pump or fan whose vibration signature shifts is often failing. And door/window contacts (a magnetic reed switch) report simple open/closed state, feeding security, energy and occupancy logic.
Notice the pattern in the figure: every reading pairs a quantity with a decision it serves. CO2 exists to trigger ventilation; vibration exists to warn of a failing motor; a leak sensor exists to shut a valve. A measurement no decision depends on is just noise you are paying to store.
Beyond the basics - specialised and safety senses
The core list covers most of what a smart building watches, but real projects reach for specialised senses when a decision demands them. Particulate matter sensors (PM2.5 and PM10) measure the fine dust that most affects health, increasingly expected wherever indoor air quality is taken seriously; VOC sensors catch the volatile organic compounds off-gassed by furnishings, cleaning and cooking; together with CO2 they build a fuller air picture than any single gas. Differential pressure sensors measure the pressure drop across an air filter or between two spaces - a rising drop tells you a filter is clogging, and holding a pressure difference is what keeps a clean room clean or a stairwell free of smoke. Flow sensors quantify water, air or heat moving through a pipe or duct, turning a bare leak alarm into a measure of how much is escaping.
A second family exists purely for safety, and it deserves respect rather than reinvention. Smoke and heat detectors, carbon monoxide alarms and gas-leak detectors protect life, and they are governed by strict fire and safety codes, commissioned and certified by qualified professionals. A smart building may read their state for context, but it does not casually re-engineer them, and their sign-off belongs to specialists. The same holds for security sensing - door and window contacts, glass-break and motion detectors - which overlaps building automation but answers to its own standards.
The lesson within the lesson is that the palette of senses is broad, but you reach for each one because a specific decision or duty demands it - health, maintenance, compliance, safety - not because a catalogue offers it. Knowing that these specialised senses exist, and when they matter, is what lets you specify sensing that is complete without being indulgent.
PM2.5, VOC, differential pressure, flow - reach for a sense when a decision needs it, not because a catalogue lists it.
Where the reading goes - the journey up the stack
A number is useless sitting in a sensor. Follow it upward. The transducer's signal is read by a small on-board microcontroller, which timestamps it and formats it. It then travels over a network - a wired bus or a wireless radio - speaking a protocol such as Modbus, BACnet, KNX, Zigbee or MQTT (Module 2 is devoted to these). It usually passes through a gateway that translates and forwards it, lands in a time-series database or IoT platform where thousands of such streams are stored against time, and is finally given meaning by a data model (Brick, Haystack - Module 4) so software knows this number is 'the supply-air temperature of AHU-3 on level 2'.
Only then can analytics and the digital twin reason across it - comparing it to a setpoint, spotting a fault, predicting a trend. That whole pipeline exists to carry one honest number from a physical spot to a decision. For an architect or designer the lesson is practical: sensing is not a gadget you bolt on at the end - it implies cabling, network, storage and structure, and those are cheapest to plan early.
sensor -> MCU -> network (Modbus/BACnet/MQTT) -> gateway -> time-series DB -> model -> twin.
Placement, range and accuracy - why a reading can lie
Here is where good sensing separates from box-ticking. A sensor reports its own location, so placement is everything. A thermostat above a warm server rack reads hot and freezes the room; a CO2 sensor by an open window reads clean while the far corner suffocates; a light sensor in shadow dims lights that should brighten. Sensors also have a range (the span they can read), a resolution (the smallest change they register) and an accuracy (how close to truth) - and confusing accuracy with precision is a classic error. A sensor can be beautifully precise, returning 21.4, 21.4, 21.4 - and consistently wrong by two degrees. Precision is repeatability; accuracy is truth; you need both.
Every sensor also drifts over time and needs calibration - a whole topic we return to in Lesson 1.4, because the most expensive mistake in this field is trusting a confident number from an untrustworthy sensor. For now, carry the instinct: before you believe a reading, ask where the sensor is, what it can actually resolve, and when it was last checked. A single well-placed sensor often beats several poorly placed ones, because it reads the spot a decision actually cares about - the occupied zone, the return-air path, the coldest corner - rather than an average that belongs to nobody. And the mounting details matter more than they look: a temperature sensor in direct sun, an air sensor tucked in dead air behind a curtain, or a light sensor facing a bright wall will each report a world that does not exist. A smart building is only as smart as its worst-placed sensor.
Accuracy = truth. Precision = repeatability. A sensor can be precisely wrong.
Transducer / sensor
Converts a physical quantity to a signal
The base of the whole stack; each one measures one quantity at one point.
PIR occupancy sensor
Passive infrared motion detection
Cheap, common true/false presence; can miss still occupants, so pair with CO2 or desk sensors.
NDIR CO2 sensor
Non-dispersive infrared gas measurement
The standard proxy for ventilation adequacy; drifts and needs periodic recalibration.
Current transformer (CT) meter
Measures electrical power and energy
Turns a live circuit into kW and kWh; the single most valuable stream for cost and carbon.
Workshop — map the senses of one room
Before any hardware, train the eye. You will inventory what a real room can (and cannot) sense, and pair every sensor with the decision it serves - the exact discipline that separates a useful sensing plan from a shopping list.
None required - a notebook and a room. Optional: a cheap plug-in temperature/humidity/CO2 monitor to watch a real stream respond to opening a window.
Goal: produce a purposeful sensor map of one room Inputs: a room you use daily + a notebook or phone camera Time: ~30 minutes
- 1Pick one room and sketch its plan. Walk it and mark every sensor you can find - thermostat, smoke/CO detector, PIR, light sensor, any meter, door contacts. Note where each sits.
- 2For each existing sensor, write the quantity it measures, its likely unit, and the ONE decision it serves (comfort, safety, energy, security). If you cannot name a decision, flag it.
- 3Now find the gaps: which of the core senses (temperature, humidity, CO2, occupancy, light, energy, water, vibration, door state) is MISSING - and would any real decision improve if it existed?
- 4Question placement: is each sensor reading a representative spot, or is it near a heat source, a draught, a window, or a shadow that would bias it? Mark one you would move.
- 5Compile a one-page sensor map: existing senses, the decision each serves, the top two gaps worth filling, and one placement you would fix.
You’ll walk away with
A one-page annotated sensor map of a single room: each sensor with its quantity, unit and the decision it serves, plus two justified gaps and one placement fix.
Three altitudes on the same idea
Read the band that fits you — or all three.
Sensing is a design brief, not an afterthought. Where you put risers, ceiling voids, meter positions and structured cabling decides which senses a building can ever have. Design in space for sensors at the points that matter - each zone, each major load, wet areas, plant rooms - and specify metering granularity up front. A building conceived to be measured is a fraction of the cost to make intelligent later.
The senses that shape human experience are yours. Temperature, humidity, CO2 and light are exactly the quantities that decide whether a space feels fresh, calm and alive - or stuffy and tiring. Knowing what these sensors measure lets you argue for air-quality you can see, lighting that follows daylight, and interiors that respond to how people actually use them, discreetly and without clutter.
Learn the sensor vocabulary cold - it is the alphabet of the whole field. Being able to name what each transducer measures, its unit, and the decision it serves is the fastest way to sound (and be) fluent in a smart-building team. Buy one cheap CO2 or temperature-humidity sensor, watch its numbers move as you open a window or fill a room, and the abstractions become real.
“More sensors always means a smarter building - just cover everything in sensors.”
Do it yourself
Reason it through - no hardware needed.
- 1In one sentence, what does it mean to say a sensor is a transducer?
- 2Name five core building measurements and the decision each one serves.
- 3Why does the placement of a temperature sensor change what it reports?
- 4What is the difference between a sensor's accuracy and its precision?
- 5Trace the journey of a single CO2 reading from the sensor to the digital twin.
The one line to carry out
Peer-reviewed journals & authoritative standards
- 01Sensor — Wikipedia, 2026.
- 02Occupancy sensor — Wikipedia, 2026.
- 03Indoor air quality — Wikipedia, 2026.
- 04Internet of things — Wikipedia, 2026.
We have met the individual senses. Next we open up the device that carries them - the IoT node of sensor, compute, radio and power - and add the muscles that let a building not just sense but act.
The author
Amogh N P
Architect, interior designer, and creative polymath. Studio Matrx began in his notebooks — his vision of design made honest, useful, and open to everyone. Its Academy is written and taught in his memory, and free, forever.
More about Amogh →