Case Study: Smart Greenhouse Automation System by Muheeb Ahmed

2025-10-16 14:35:31 UTC

All of the sensors mentioned will provide periodic discrete datapoints. Smart sensors have the ability to self-calibrate and process data locally. This would reduce maintenance frequency and the need for data cleaning. Sensor fusion allows data to be combined from multiple sensors such as temperature and humidity so the response is likely to be more accurate.

2025-10-16 14:55:01 UTC

Different sensing technologies:

Temperature: thermistor, thermocouple, RTD, infrared

Humidity: capacitive, resistive, thermal

Soil moisture: capacitive, resistie, TDR

Light intensity: photodiode, LDR, phototransistor

Co2 concentration: NDIR, metal oxide semiconductor, chemical

Justification of smart sensors:

Smart sensors help measure more meaninful values by combining and processing outputs from multiple physical environmental variables.

Evaluate best sensors with performance metrics and environmental suitability:

Temp: thermistors: good accuracy, fast response, robust to greenhouse temp range

Humidity: Capacitive: reliable pull RH range with a fast response, suitable for fluctuating humidity, can tolerate condensation better

Soil Moisture: Capactive: accurate, corrotion resistant and maintenance friendly, can withstand water

Light intensity: photodiode: good spectral range, durable and fast

Co2: NDIR: accurate, reliable in high humidity and needs less frequent recalibration.

2025-10-16 14:55:14 UTC

Using smart sensors would prevent wasted bandwidth and processing power spent on transmitting irrelevant data, therefore having sensors that will transmit data only if a significant change occurs or boundary conditions are exceeded would be very helpful, in the other case data transmission could be limited to a less regular interval

2025-10-16 14:58:38 UTC

Smart sensor justification:

self-calibration means drift isn't an issue that is to be accounted for

local data processing ensures only relevant data gets stored within the remote system (cloud etc.) as efficiently as possible

they can log their own health meaning the owner can be notified of a problem instead of being expected to troubleshoot

2025-10-16 15:02:19 UTC

Electrical actuators are suitable for most applications but hydraulic might be more suitable for large industrial irrigation systems. In addition to the actuators stated, heating lamps and humidifier/dehumidifiers.

2025-10-16 15:04:23 UTC

lighting would be controlled digitally, as it means you can obtain more precise control and mimic conditions more accurately

the main actuator that would require hydraulics is the irrigationa s pumping water on mass requires a lot of force and energy.

2025-10-16 15:04:54 UTC

Architecture: Sense data -> local data processing -> communication -> central control unit -> communication -> actuator. Sensor data would be cleaned and any anomalous readings would be flagged and checked. Communicate over WiFi and only take readings every 30 minutes for power saving.

2025-10-16 15:06:10 UTC

Ventilation: electrical: fast, precise, clean

Irrigation: electrical

Lighting: electrical

Co2 injectors: pneumatic: fast, economical, simple

2025-10-16 15:07:33 UTC

Beyond those stated, sensor noise and actuator wear. Make the smart sensors regularly re-calibrate and also filter noise in sensor data. With increasing number of sensors comes increased data handling and data storage. These could be addressed by making the system modular and only storing data for a necessary period.

2025-10-16 15:08:02 UTC

I would connect all the sensors to a local raspberry pi, either wired or by a local network. Then the lacal raspbery pi would have all the control parameters set on it controlling the actuators on the same network. The logic on this would be able to control the actuators based on live sensor updates.

The raspberry pi would also be controlled to a remote access point where users can control the logic and thresholds behind the control mechanisms, this would allow 24 hour operation without user input, but then a user can prompt the system to change its operation remotely whenever they want

2025-10-16 15:08:57 UTC

Controller: Raspberry Pi

Actuators: use relays or MOSFETs to control

Remote Access: Wi-Fi or Zigbee, dependent on range, probably zigbee for this application as sensors will be low power and low bandwidth

UI (cloud access): App or website

Sensors read every minute

Pre processing before being sent to the cloud, remove outliers, clean and filter the readings

Combine the data from multiple sensors for fusion to give a more accurate result within the context

Compare the data to thresholds for the actuators

Zigbee for low power, use MQTT for the data transmission, the low polling rate (every minute) is to save power