IoT Sensing Systems

Sensing Systems

• Sensors are crucial for IoT applications as they monitor the physical world and provide data.
• IoT sensors are generally small, low-cost, and have limited battery life.
• These sensors can range from simple thermocouples to complex video cameras.
• Common types include smartphone-based, medical, neural, environmental/chemical, RFID, and vision systems.

Location Detection Sensors

• Smartphones are widely used in IoT solutions due to embedded sensors.
• These include accelerometers, gyroscopes, GPS, and magnetometers.

Accelerometers

• Measure motion and acceleration in 3D (X, Y, Z axes).
• Use MEMS piezoelectric technology with a central mass and spring.
• Detect deflection and produce voltage in response to movement by varying capacitance.

Gyroscopes

• Measure orientation and rotation rate.
• Use a rotating disk to detect tilt.
• Often paired with accelerometers in MEMS packages, like the Invensense MPU-6050.

Magnetometers

• Measure magnetic field strength and direction.
• Used as digital compasses and for metal detection.
• Often combined with accelerometers and gyroscopes to form an Inertial Measurement Unit (IMU).

IMUs

• Used for indoor location tracking by combining motion and orientation data.

GPS Sensors

• Used for outdoor location detection.
• Work via trilateration, using signals from at least 3 satellites in a network of around 30 orbiting Earth.

Light and Ranging Sensors

• Light sensors measure ambient light intensity.
• Applications include smart homes, security systems, and smart lighting.
• Types include photoresistors (change resistance with light intensity) and photodiodes (convert light to electrical current).
• Proximity sensors use infrared (IR) signals to detect the distance to objects.
• They trigger events when objects approach.
• LiDAR sensors measure distance by reflecting laser pulses.
• They are used in agriculture, robotics, self-driving cars, and environmental studies.
• LiDAR can analyze objects, gases, and atmospheric conditions.
• In self-driving vehicles like Google’s Waymo and Uber, LiDAR provides 360° visibility with precision up to ±2 cm.

Temperature Sensors

• Some smartphones have thermometers, barometers, and humidity sensors.
• These measure temperature, atmospheric pressure, and humidity.
• Temperature sensors are widely used in smart thermostats, cold storage logistics, refrigerators, and industrial machinery.

Thermocouples

• Do not need an excitation signal.
• Produce very small signals (e.g., microvolts).
• Suited for long-distance measurements in industrial/high-temperature environments.
• May lose accuracy over time due to aging in high-heat environments.

Resistance Temperature Detectors (RTDs)

• Require an excitation current.
• Have better accuracy than thermocouples.
• Operate within a narrower temperature range (rarely used above 600°C).

Thermistors

• Resistors whose resistance changes with temperature.
• Provide high resolution for narrow temperature ranges.
• Common in medical devices, scientific equipment, food handling, incubators, and home appliances like thermostats.

Wearable Sensors

• Measure medical parameters such as heart rate, pulse, blood pressure, body temperature, respiration rate, and blood glucose levels.
• Common devices include smartwatches, wristbands, monitoring patches, and smart textiles.

Smartwatches and Fitness Trackers

• Gained popularity with companies like Apple, Samsung, and Sony.
• Integrate features such as smartphone connectivity, accelerometers, and heart rate monitors.

Smart Patches

• Skin-applied devices that continuously monitor vital health parameters.
• They are stretchable, disposable, and affordable.
• Example: Microneedle smart insulin patches sense elevated blood glucose and release insulin.

RFID Technology

• An identification system consisting of a tag and a reader.
• Tags include a small chip with an antenna.
• The antenna transmits digital RF messages containing a unique ID and stored data.
• The reader emits a signal to the tag and receives a response with tag data.
• The Reader measures Received Signal Strength (RSSI) to estimate distance.

Active RFID Tags

• Have a built-in power source.

Passive RFID Tags

• Have no power source.
• Draw power from the reader’s electromagnetic waves.
• Cheaper with longer lifetimes.

RFID Applications in IoT

• Inventory management
• Supply chain management
• Asset tracking
• Access control
• Identity authentication
• Object tracking
• Example: RFID tags on vehicles trigger barricades to open if authorized.

Audio and Vision Systems

• Smartphones have advanced microphones and cameras for capturing audio and visual data.
• Used to detect contextual information like environment and user activities.
• Sound and vibration measurements are used to monitor equipment health and safety in IIoT and predictive maintenance.

Video Cameras and Vision Systems

• Use high-end processors, digital signal processors, FPGAs, or custom ASICs.

Charge-Coupled Devices (CCDs)

• Move charge from the sensor to the chip edge for sampling via analog-to-digital converters.
• Offer high-resolution, low-noise images.
• Have high power consumption.

Complementary Metal-Oxide-Semiconductor (CMOS)

• Sample charge at individual pixels.
• More power-efficient but more prone to noise.

CMOS Camera Components

• Integrated into a silicon die with a two-dimensional array of transistors.
• Captures images through a lens.
• Uses an image signal processor (ISP) to filter, normalize, and convert the image into a digital format.
• High data volume (e.g., 60 frames per second at 1080p resolution) makes streaming directly to the cloud impractical for real-time processing.

Sensor Fusion

• Combining readings from multiple sensors to make more accurate and informed decisions.
• Example: Integrating temperature, location, and light intensity sensors to infer high foot traffic and sunlight.
• This leads to automated adjustments in air circulation in a smart building.

Centralized Sensor Fusion

• Raw data is sent to a central service (e.g., cloud) for aggregation and fusion.

Decentralized Sensor Fusion

• Data is correlated at or near the sensor, such as at edge or fog nodes.
• Example: Autonomous vehicles use decentralized sensor fusion to understand their environment.

Autonomous Vehicles

• Sensor fusion combines readings from multiple sensors to enhance driving safety and performance.

Autonomous Vehicle Sensors

• Ultrasonic sensors detect obstacles nearby.
• GPS calculates location, speed, and course.
• Speed and angle sensors measure vehicle speed and wheel rotation.
• LiDAR identifies and maps obstacles.
• Cameras detect, classify, and determine the distance of objects.

Autonomous Vehicle Sensor Data Processing

• Sensor readings are transmitted to an onboard processing unit (e.g., in vehicles).
• Fusion algorithms like the Kalman filter are used to improve accuracy.

Kalman Filter Overview

• This is an iterative algorithm with two main steps: predict and update.
• Based on probability theory and real-time sensor measurements.
• It refines predictions by comparing expected vs. actual sensor data.

Kalman Filter Example

• Autonomous vehicle tracking a neighboring vehicle.
• Predict step: Estimates position using previous data (e.g., assuming constant velocity).
• Update step: Adjusts prediction based on actual sensor measurements and calculates Kalman gain.
• Repeats until prediction matches measurement closely.

Kalman Filter Advantages

• Can ignore unreliable sensor data.
• Continuously improves prediction accuracy over time.

Kalman Filter Limitations

• Slow response in rapidly changing environments.
• Assumes linear models, which is often unrealistic in real-world scenarios.

Alternatives for Nonlinear Systems

• Extended Kalman Filter (EKF)
• Unscented Kalman Filter (UKF)
• Particle Filter