MODULE 2

SENSORS, ACTUATORS AND EDGE DEVICES

A sensor detects (senses) changes in the ambient conditions or in the state of another device
or a system, and forwards or processes this information in a certain manner.
A device which detects or measures a physical property and records, indicates, or otherwise
responds to it.
They perform some input functions by sensing or feeling the physical changes in
characteristics of a system in response to a stimulus.
For example, heat is converted to electrical signals in a temperature sensor, or atmospheric
pressure is converted to electrical signals in a barometer.
Transducers convert or transduce energy of one kind into another.
For example, in a sound system, a microphone converts sound waves into electrical signals
and a loudspeaker converts these electrical signals back into sound waves.

Parameters Transducers Sensors Actuators

Converts electrical
Converts energy signals into various
Converts various forms of energy
Definition from one form to forms of energy,
into electrical signals.
another. typically mechanical
energy.
Can be used to
represent a sensor
Domain It is an input transducer. It is an output transducer.
as well as an
actuator.
Can work as a Used for converting
Used for quantifying
sensor or an signals into proportional
Function environmental stimuli into
actuator but not mechanical or electrical
signals.
simultaneously. outputs.
Humidity sensors, Temperature
Motors (convert
sensors, Anemometers (measures
electrical energy to rotary
flow velocity), Manometers
motion), Force heads
(measures fluid pressure),
Any sensor or (which impose a force),
Examples Accelerometers (measures the
actuator. Pumps (which convert
acceleration of a body), Gas
rotary motion of shafts
sensors (measures concentration
into either a pressure or a
of specific gas or gases), and
fluid velocity).
others.
 Fig:2.1 Outline of a simple sensing operation

Sensor Features
 It is only sensitive to the measured property.
 It is insensitive to any other property likely to be encountered in its application.
 It does not influence the measured property.
Sensor Resolution
The resolution of a sensor is the smallest change it can detect in the quantity that it is
measuring.
The more is the resolution of a sensor, the more accurate is its precision.A sensor’s accuracy
does not depend upon its resolution.
Analog Sensors
Analog Sensors produce a continuous output signal or voltage which is
generally proportional to the quantity being measured.
Digital Sensors
Digital Sensors produce discrete digital output signals that are a digital
representation of the quantity being measured.
Scalar Sensors
Scalar Sensors produce output signal proportional to the magnitude of the
quantity being measured.
Vector Sensors
Vector Sensors produce output signal proportional to the magnitude, direction,
as well as the orientation of the quantity being measured.
 Fig:2.2 sensors used for IoT-based applications

Sensors are fundamental components in modern engineering systems and industrial

environments. They are extensively used to monitor, control, and automate processes across
multiple domains such as manufacturing, energy, transportation, and logistics. In addition to
process control, sensors also support routine industrial activities like inventory management,
condition monitoring, and record keeping. Based on the nature of the physical quantity they
measure, sensors can be systematically classified into different categories. This section
focuses on thermal sensors and mechanical sensors.
Fig 2.3 Functional blocks of a typical sensor node in IoT

Figure 2.4 a typical sensor calibration graph

Sensor profile
This feature defines the expected trend and behavior of a sensor based on the input. Sensors
can be profiled as either linear or non-linear. Sensors are designed in such a manner that the
output varies proportionally with the input parameters. The input parameters can be
temperature, voltage, current, or any other characterizable parameter. If the output is directly
proportional to the input, the sensor is categorized as a linear sensor. Fig. 8.2(a) shows the
characteristics of a linear sensor. In contrast, Fig. 2 shows a non-linear sensor's
characteristics. In other words, for a fixed change in the input value, the sensor output always
changes by a fixed value, a linear sensor's behavior. In contrast, if the output changes non-
linearly, there is no fixed constant of proportionality between the output and input. The
output varies in polynomial, exponential, square, logarithmic, and others, and the sensor is
categorized as a non-linear sensor

Fig 2.5 typical sensor characteristic graph

Sensor accuracy
The accuracy of a sensor is defined as the degree of correctness or closeness with which a
sensor can quantify a physical phenomenon. For example, consider a vat of boiling water
(water boils at 100°C). A temperature sensor T1 outputs a value of 99.8°C, whereas T2
outputs 103.1°C. We say that T1 is more accurate than T2, even though they do not show the
exact correct value. If we observe closely, T1 deviates from the actual temperature of 100°C
by 0.2°C, whereas T2 deviates by 3.1°C. Therefore, T2 is prone to larger deviations from
original values as compared to T1.
Sensor resolution
The resolution of a sensor signifies the intensity or level of details a sensor can capture from
a physical phenomenon while quantifying it. For example, consider an accelerometer, which
outputs the value of acceleration (in this case, along the x-axis only) in m/s². This sensor a1,
along with another similar sensor a2 is attached to a car accelerating at 3.2 m/s². The first
sensor a1 reports the car’s acceleration as 3.225 m/s², whereas the second sensor a2 reports it
as 3.22 m/s². Here, we say that the resolution of a1 is more than a2 as it can provide
measurements of the phenomenon with more precision (up to 3 decimal places) as compared
to a2 (up to 2 decimal places).
Sensor rating
The rating of a sensor signifies the maximum and minimum values of the different conditions
(both ambient as well as input), which the sensor can safely handle without damaging itself or
distorting its readings. A non-contact infrared thermal sensor is rated to operate between
−10°C to 200°C. Operating this sensor to measure temperatures greater than the maximally
rated value may of 200°C will saturate its readings by overloading its sensing elements. This
will result in incorrect readings. Fig. 8.3 shows a typical sensor rating graph. The input and
output characteristics can be plotted against various sensor parameters. The plot shows the
variation in the sensor’s output values for variations in the sensor’s input values. It is to be
noted that the sensor output is truly active for a fixed range of sensor input values, which is
known as the sensor’s rating.

Figure 2.6: a typical sensor rating graph

Operating voltage
This feature signifies the range of operating voltages, within which the sensor can operate
without damaging its associated circuitry. It is to be noted that this feature is typically present
in sensors which require an external source of power to operate/ function. Generally,
commercial sensors available for hobbyists have an operating voltage range of 3.3 to 5 V.
Operating this sensor below the minimum value may not be enough to activate the sensor at
all. Figure shows a typical sensor output graph, plotted against varying input conditions.
Additionally, operating this sensor to supply greater than the maximum rated 5V increases
the chance of damaging the sensor.

Fig 2.7 a typical sensor output graph

Classification of sensors

I)Based on Power Requirements: The way sensors operate decides the power that must be
provided. Some sensors are provided with separate power sources for them to function,
whereas some sensors do not require any power sources. Depending on the requirements of
power, sensors can be of two types.

(i) Active: Active sensors can be classified based on: (1) power requirements, (2) sensor
output and (3) property to be measured. The active sensor requires power to operate and
provide it with power. It directly responds to the external environment and converts it into an
output signal. For example, a photodiode converts light into electrical impulse.

(ii) Passive: Passive sensors do not require an external circuitry or mechanism to power them
or respond to external stimuli from its environment, and converts it into an output signal. For
example, a thermistor's resistance can be detected by applying a current through it.

II)Based on Output:

The output of a sensor helps in deciding the additional components to be integrated

with an IoT node or system. Typically, almost all modern-day processors are digital, digital
sensors can be directly integrated to the [Link], the integration of analog
sensors to these digital processors or IoT nodes requires additional interfacing mechanisms
such as analog to digital converters (ADC), voltage level converters, and others. Sensors are
broadly divided into two types, depending on the type of output generated from these sensors,
as follows.

(i) Analog: Analog sensors generate an output signal or voltage, which is proportional
(linearly or non-linearly) to the quantity being measured and is continuous in time and
amplitude. Physical quantities such as temperature, speed, pressure, displacement, strain, and
others are all continuous and categorized as analog quantities. For example, a thermometer or
a thermocouple can be used for measuring the temperature of a liquid (e.g., in household
water heaters). These sensors continuously respond to changes in the temperature of the
liquid.

(ii) Digital: These sensors generate the output of discrete time digital representation (time, or
amplitude, or both) of a quantity being measured, in the form of output signals or voltages.
Typically, binary output signals in the form of a logic 1 or logic 0 for ON or OFF,
respectively are associated with digital sensors. The generated discrete (non-continuous)
values may be output as a single “bit” (serial transmission), eight of which combine to
produce a single “byte” output (parallel transmission) in digital sensors.

III)Measured Property: The property of the environment being measured by the sensors can
be crucial in deciding the number of sensors in an IoT implementation. Some properties to be
measured do not show high spatial variations and can be quantified only based on temporal
variations in the measured property, such as ambient temperature, atmospheric pressure, and
others. Whereas some properties to be measured show high spatial as well as temporal
variations such as sound, image, and others. Depending on the properties to be measured,
sensors can be of two types:

(i) Scalar: Scalar sensors produce an output proportional to the magnitude of the quantity
being measured. The output is in the form of a signal or voltage. Scalar physical quantities
are those where only the magnitude of the signal is sufficient for describing or characterizing
the phenomenon and information generation. Examples of such measurable physical
quantities include color, pressure, temperature, strain, and others. A thermometer or
thermocouple is an example of a scalar sensor that has the ability to detect changes in
ambient or object temperatures (depending on the sensor's configuration). Factors such as
changes in sensor orientation or direction do not affect these sensors (typically).

(ii) Vector: Vector sensors are affected by the magnitude as well as the direction and/or
orientation of the property they are measuring. Physical quantities such as velocity and
images that require additional information besides their magnitude for completely
categorizing a physical phenomenon are categorized as vector quantities. Measuring such
quantities are undertaken using vector sensors. For example, an electronic gyroscope, which
is commonly found in all modern aircraft, is used for detecting the changes in orientation of
the gyroscope with respect to the Earth's orientation along all three axes

III)Based on their underlying sensing task:

1. Thermal Sensors
Thermal sensors are used to measure temperature or thermal energy in industrial and
engineering applications. Depending on the design and application, these sensors may
provide output readings in units such as Celsius (°C), Fahrenheit (°F), Kelvin (K), or Joules
(J). It is important to clearly distinguish between heat and temperature. Heat refers to the
transfer of energy between molecules, particularly during chemical or physical processes,
whereas temperature represents the average kinetic energy of molecules in a substance and
can be directly measured.

Thermal sensors are broadly classified into three major types:

1.1 Resistive Temperature Detectors (RTDs)

Resistive Temperature Detectors (RTDs) are high-precision temperature sensors used to
measure absolute temperature. They operate on the principle that the electrical resistance of
certain pure conductive materials, such as platinum or copper, changes predictably with
temperature. Although RTDs provide excellent accuracy and stability, their sensitivity to
small temperature variations is relatively lower compared to other temperature sensors. RTDs
are passive devices and require an external voltage source to function.

Fig 2.6 thermal sensor in a industrial boiler

1.2 Thermistors
Thermistors are temperature-sensitive resistors that detect temperature changes by monitoring
variations in their electrical resistance. The material used in thermistors exhibits a significant
change in resistance with temperature, making them highly sensitive over a limited
temperature range. Like RTDs, thermistors are passive devices and require an external power
supply for operation.
1.3 Thermocouples
Thermocouples are among the most widely used temperature sensors in industrial
applications due to their simple construction, robustness, and wide operating temperature
range. A thermocouple consists of two dissimilar metals joined together to form a hot
junction and a cold junction. When a temperature difference exists between these junctions, a
voltage is generated, which is proportional to the temperature difference. This voltage is
measured and converted into a temperature reading.
2. Mechanical Sensors
Mechanical sensors are designed to detect changes in mechanical properties such as pressure,
force, motion, and acceleration. These sensors translate mechanical variations into electrical
signals that can be measured and analyzed. Mechanical sensors are extensively used in
industrial automation, safety systems, and monitoring applications.

Figure 2.9: Industrial power press

2.1 Pressure Sensors

Pressure sensors are used to measure changes in pressure within gases or liquids. Depending
on the sensing mechanism, pressure sensors may operate based on piezoresistive,
piezoelectric, capacitive, electromagnetic, strain gauge, or resonant principles. These sensors
are essential in applications such as hydraulic systems, pneumatic controls, and process
industries.

2.2 Barometer
A barometer is a type of pressure sensor specifically designed to measure atmospheric
pressure. It is typically stationary and is widely used in meteorology, weather forecasting, and
altitude estimation.
2.3 Altimeter
An altimeter operates on principles similar to a barometer but is used to measure altitude
changes relative to a fixed reference level, usually sea level. Altimeters are commonly
employed in aviation, aerospace, and navigation systems.
2.4 Liquid Flow Sensor
Liquid flow sensors measure the flow rate of liquids within a confined or bounded channel.
These sensors are crucial in industries such as chemical processing, water treatment, and oil
refining to ensure proper flow control and system efficiency.
2.5 Gas Flow Sensor
Gas flow sensors are used to measure the velocity or flow rate of gases moving through a
closed channel or pipe. They play an important role in applications such as HVAC systems,
industrial gas distribution, and environmental monitoring.
2.6 Accelerometer
An accelerometer is an electromechanical sensor used to measure the acceleration of a body.
Accelerometers are commonly based on piezoelectric or capacitive sensing techniques. They
are widely used in vibration analysis, motion detection, condition monitoring, and inertial
navigation systems.
Proximity sensor
A proximity sensor is a device used to detect the presence, absence, or distance of an object
without making physical contact with it.
How it works?
A proximity sensor emits some form of energy (such as electromagnetic fields, infrared light,
sound waves, or capacitive fields) and then detects changes in that energy when an object
comes close. Based on this change, the sensor determines whether an object is nearby.
Types of proximity sensors
 Inductive proximity sensor – Detects metal objects using electromagnetic fields.
 Capacitive proximity sensor – Detects metal and non-metal objects (plastic, wood,
liquids) by sensing changes in capacitance.
 Infrared (IR) proximity sensor – Uses reflected infrared light to detect objects.
 Ultrasonic proximity sensor – Uses sound waves and measures the echo to
determine distance.
 Magnetic proximity sensor – Detects objects with a magnetic field (e.g., reed
switches).
Applications
 Industrial automation (object detection on conveyor belts)
 Smartphones (turning off the screen during calls)
 Robotics (obstacle detection)
 Automotive systems (parking sensors)
 Security systems
Advantages
  No physical contact - less wear and tear
 Fast response time
 Reliable in harsh environments
Limitation
 Detection range is limited and depends on sensor type and object material
In short, a proximity sensor enables automatic, contactless detection of objects, making it
essential in modern industrial and electronic systems.
Flow sensor
A flow sensor is a device used to measure the rate or quantity of a fluid (liquid or gas)
flowing through a channel, pipe, or system.
Working principle:
A flow sensor detects the movement of fluid and converts it into an electrical signal
proportional to the flow rate or total flow. Different sensors use different physical principles
such as pressure difference, heat transfer, or mechanical motion.
Types of flow sensors
 Liquid flow sensor – Measures the flow rate of liquids in a bounded channel.
 Gas flow sensor – Measures the velocity or flow rate of gases.
 Mechanical (turbine/paddle) flow sensor – Uses rotating parts; rotation speed
indicates flow rate.
 Thermal (mass) flow sensor – Measures flow based on heat loss caused by moving
fluid.
 Differential pressure flow sensor – Calculates flow by measuring pressure drop
across an obstruction.
 Ultrasonic flow sensor – Uses sound waves to determine fluid velocity.

Applications
 Industrial process control
 Water and gas distribution systems
 Chemical and pharmaceutical industries
 HVAC systems
 Fuel and oil flow monitoring
 Medical devices (e.g., oxygen flow monitoring)
Advantages
 Accurate measurement of flow rate
 Helps in process optimization and safety
 Available for both liquids and gases
Limitations
 Accuracy can be affected by temperature, pressure, and fluid properties
 Some types require regular maintenance

A flow sensor monitors and measures the movement of fluids, ensuring efficient control,
safety, and performance in industrial and everyday applications.
A vibration sensor is a device used to detect, measure, and monitor vibrations or oscillatory
motion of an object, machine, or structure.
How it works:
A vibration sensor senses changes in acceleration, velocity, or displacement caused by
vibration. These mechanical movements are converted into electrical signals, which are then
analysed to assess the condition of the system.
Types of vibration sensors
 Accelerometer – Measures acceleration; the most commonly used vibration sensor.
 Velocity sensor – Measures vibration velocity, often used in rotating machinery.
 Displacement sensor – Measures displacement or position changes.
 Piezoelectric vibration sensor – Uses piezoelectric materials that generate voltage
when subjected to vibration.
 MEMS vibration sensor – Compact, low-power sensors widely used in IoT and
consumer electronics.
Applications
 Condition monitoring of industrial machines
 Predictive maintenance (detecting bearing faults, imbalance, misalignment)
 Structural health monitoring (bridges, buildings)
 Automotive systems
 Consumer electronics (smartphones, wearables)
 Earthquake and seismic monitoring
Advantages
 Enables early fault detection
 Reduces downtime and maintenance cost
 Improves safety and reliability
 Non-intrusive measurement
Limitations
 Sensitive to noise and environmental factors
 Requires proper mounting for accurate readings
A vibration sensor monitors mechanical vibrations to detect abnormalities, making it essential
for machine health monitoring, safety, and predictive maintenance in industrial and smart
systems.
Actuation
An actuator is a component of a machine or system that moves or controls the mechanism or
the system.
An actuator requires a control signal and a source of energy.

Pneumatic Actuators
Overview of Pneumatic Actuators
 Pneumatic actuators generate linear or rotary motion by utilizing the compression and
expansion of air.
 These actuators are widely used in industrial environments, especially in systems that
require reliable and fast mechanical motion.
 A common application of pneumatic actuators is in valves, where they regulate the
flow of liquids or gases in pipelines and processing units.
 Figure 2.11 Pneumatic actuators

Working Principle of Pneumatic Actuators

 Pressurized air is supplied to a sealed chamber containing a piston.

 The piston is connected to a mechanical linkage that:
o Moves a load
o Operates a valve
o Produces linear or rotary motion
 When compressed air enters the chamber:
o It exerts force on the piston
o The piston moves in the desired direction
 Once the air pressure is released:
o The piston returns to its initial position
o This return motion is typically achieved using a spring mechanism
Classification of Pneumatic Actuators

Pneumatic actuators are mainly classified based on how the piston returns to its
original position.
Single-Acting Pneumatic Actuators
 Single-acting actuators use one air inlet/outlet port.
 Pressurized air enters the chamber through this port to move the piston.
 The piston movement can occur in two ways:
o Forward movement by pushing the piston using compressed air
o Backward movement by removing air from the chamber, creating suction
  After the pressurized air is removed:
o A spring mechanism forces the piston back to its default position
 These actuators are commonly used where motion is required in only one direction
 Illustrates a typical single-acting pneumatic actuator:
o Compressed air pushes the piston forward
o The spring retracts the piston once pressure is released
Double-Acting Pneumatic Actuators
 Double-acting actuators do not use a spring for piston repositioning.
 They are equipped with two air inlet ports, one on each side of the piston.
 The direction of piston movement depends on which port receives pressurized air:
o Supplying air to the lower port moves the piston upward
o Supplying air to the upper port moves the piston downward
 These actuators allow precise and controlled motion in both directions
 They are preferred in applications requiring:
o Higher force
o Better control
o Continuous bidirectional operation
Advantages and Limitations of Pneumatic Actuators
Key Advantages:
 Pneumatic actuators have a simple design, making them easy to operate and maintain.
 They are considered safe for use in hazardous environments, particularly where fire or
explosion risks are high.
 Since air is used as the working medium:
o There is no risk of fluid leakage
o Maintenance efforts are reduced
o Surrounding machinery remains clean
Key Limitations:
 Air is highly compressible compared to liquids, which reduces the efficiency and
accuracy of pneumatic actuators.
 Continuous operation requires a constant supply of compressed air.
 The cost associated with air compressors and energy consumption makes pneumatic
actuators more expensive over long-term use.

Summary Points
 Pneumatic actuators convert air pressure into mechanical motion.
 They are widely used for valve control and industrial automation.
 Single-acting actuators use a spring for piston return, while double-acting actuators
rely on air pressure in both directions.
Their simplicity and safety make them attractive, but efficiency and operating cost are key
concerns

Hydraulic Actuators

Hydraulic actuators are devices that convert energy stored in pressurized fluids into useful
mechanical motion. In industrial systems, these actuators make use of liquids that are almost
incompressible, such as hydraulic oil. Because the fluid does not compress significantly, any
pressure applied to it is transmitted quickly and efficiently to mechanical components like
pistons or shafts, resulting in motion.

The working principle of hydraulic actuators is based on Pascal’s law. Pascal’s law states that
when pressure is applied to a confined fluid, it is distributed equally in all directions. This
means that a small input force applied over a small area can generate a much larger output
force over a larger area. This property allows hydraulic systems to produce very high forces
using relatively small input power, making them ideal for heavy-duty industrial applications.

In practical operation, hydraulic energy supplied by a pump pressurizes the fluid inside a
sealed chamber. This pressurized fluid then acts on a piston or vane, converting fluid pressure
into linear or rotational mechanical motion. Since energy losses due to compression are
minimal, hydraulic actuators are highly efficient in force transmission.

Figure 2.12 Hydraulic actuator

Hydraulic actuators are broadly classified into two main categories based on the type of
motion they generate:

1. Linear Hydraulic Actuators:

Linear hydraulic actuators generate straight-line motion when pressure is applied to the
hydraulic fluid. These actuators are widely used in lifting, pushing, pulling, and pressing
operations. Common industrial examples include:
- Single-acting cylinders, where pressure is applied in one direction and a spring or external
force returns the piston.
- Double-acting cylinders, where hydraulic pressure is used for motion in both forward and
backward directions.
- Double-rod cylinders, which provide balanced force and speed in both directions of motion.

2. Rotary Hydraulic Actuators:

Rotary hydraulic actuators convert hydraulic pressure into rotational motion. Instead of
moving a piston linearly, these actuators rotate a shaft through a certain angle or
continuously. Typical industrial rotary hydraulic actuators include:
- Rotary vane actuators, which use vanes attached to a rotor inside a chamber.
- Rack and pinion actuators, where linear piston motion is converted into rotation using a gear
mechanism.

Key Takeaways for Engineering Applications:

1. High Power-to-Weight Ratio:

Hydraulic actuators are capable of generating extremely large forces relative to their size and
weight. This makes them suitable for lifting or moving heavy loads while consuming
comparatively low power.

2. High-Speed Operation:
These actuators can operate at high speeds and can easily adapt to rapid changes in operating
speed, which is essential in many industrial automation processes.

3. Overload and Stall Resistance:

Hydraulic actuators are robust and can tolerate overload or stalling conditions without
permanent damage. Once the overload is removed, normal operation can resume
immediately.

4. Maintenance Requirements:
Despite their advantages, hydraulic actuators require careful maintenance. Regular inspection
is necessary to detect and prevent fluid leaks, as loss of hydraulic fluid reduces pressure and
system efficiency. Maintaining clean fluid and proper seals is critical for reliable operation

Electric actuators

Typically, electric motors are used to power an electric actuator by generating mechanical
torque. This generated torque is translated into the motion of a motor’s shaft or for switching
(as in relays). For example, actuating equipment such as solenoid valves control the flow of
water in pipes in response to electrical signals. This class of actuators is considered one of the
cheapest, cleanest, and speedy actuator types available. Below figure show some of the
commonly used electrical actuators.
MICROCONTROLLERS

Raspberry Pi (Single-Board Computer)

Definition: Raspberry Pi is a low-cost, credit-card sized computer that runs a full Linux OS
(e.g., Raspberry Pi OS) and is used for embedded systems, IoT, and education.
Raspberry Pi: Intermind has used the technology for performing multiple processes in the
field of HMI (Human Machine Interface), automation, control, and communications systems

Processor & memory: Uses ARM-based SoCs (e.g., Cortex-A series) with integrated CPU,
GPU, RAM, and multimedia; newer boards offer multi-core CPUs and up to several GB of
RAM.
Interfaces: Provides USB, HDMI, audio, Ethernet/Wi-Fi, camera/display connectors, and
40-pin GPIO header with UART, SPI, I²C, and PWM for hardware interfacing.

Typical use cases:

Edge computing and gateways in IoT
Robotics control with higher-level logic and vision
Teaching Linux, Python, networking, and OS concepts
Raspberry Pi is a mini-PC plus GPIO, suited for high-level applications where you need an
OS, file system, and powerful processing.

Arduino (e.g., Arduino Uno – 8-bit MCU board)

Definition: Arduino boards are open-source microcontroller platforms designed for easy
prototyping, programmed mainly in C/C++ using the Arduino IDE.
Arduino: We use the platform to perform repetitive single tasks. Arduino has been used
majorly in applications developed by Intermind, in the field of automation and control
systems
Core hardware (Uno example): ATmega328P 8-bit MCU, 16 MHz clock, 32 KB Flash, 2 KB
SRAM, 1 KB EEPROM, 14 digital I/O (6 PWM), 6 analog inputs, 5 V operation.
I/O and connectivity: Simple digital and analog pins, serial (UART) over USB, SPI and I²C
via shared pins; typically no built-in networking without shields.

Typical use cases:

Basic sensor/actuator interfacing
Introductory embedded systems and control
Low-cost, deterministic real-time tasks (no OS)
Key idea for students: Arduino is a microcontroller board (no OS) – ideal for learning
embedded basics, timing, and direct hardware control.

ESP32 (Wi-Fi + Bluetooth MCU)

Definition: ESP32 is a highly integrated Wi-Fi + Bluetooth SoC with a dual-core Tensilica
LX6 MCU, widely used for connected and IoT applications.
Core hardware: Up to 240 MHz dual-core MCU, around 520 KB SRAM, external Flash
(commonly 4 MB), ultra-low-power modes, integrated RF and power management.
Connectivity & peripherals: 2.4 GHz Wi-Fi, Bluetooth Classic and BLE, multiple
UART/SPI/I²C/I²S, ADCs, DACs, touch sensors, PWM on many GPIOs.
Software ecosystem: Programmable via Espressif IDF (C/C++), Arduino core for ESP32, or
MicroPython; supports FreeRTOS and secure features like TLS.

Typical use cases:

IoT nodes with built-in wireless
Smart home, wearables, sensor networks
Low-cost edge devices with simple RTOS
ESP32 is a networked microcontroller – combines MCU + Wi-Fi + BLE, bridging classic
Arduino-style IO with modern wireless.

STM32 (32-bit ARM Cortex-M microcontroller family)

Definition: STM32 is a large family of 32-bit ARM Cortex-M microcontrollers (and some
Cortex-A MPUs) from STMicroelectronics, spanning low-power to high-performance lines.
CPU cores: Uses Cortex-M0/M0+, M3, M4F, M7F, M33, M55, and M85 cores; series are
grouped by target: low-cost (C0, F0, G0), low-power (L0, L4, U5), high-performance (F4,
F7, H7), and wireless or secure variants.
Peripherals: Rich set including multiple timers, ADC/DAC, UART/SPI/I²C, USB, CAN,
Ethernet, and advanced timers for motor control and power electronics.

Ecosystem:
STM32CubeMX and STM32CubeIDE for code generation and configuration
Nucleo and Discovery boards with Arduino-style headers for rapid prototyping

Typical use cases:

Industrial control, robotics, motor drives
Low-power IoT nodes, medical, metering
Real-time signal processing and control

STM32 is a professional-grade MCU family, giving a path from simple lab boards to
industrial products with the same architecture

Platfor CPU/Core OS Built-in Typical use in

Category
m type support connectivity class

ARM
Full Linux Ethernet/Wi-Fi Linux, Python,
Raspberr Single-boar Cortex-A
(e.g. RPi (model-dependen IoT gateway,
y Pi d computer SoC
OS) t) vision
(various)

Embedded
8-bit No OS
Arduino MCU None onboard basics,
ATmega328 (bare-meta
Uno board (shields needed) sensors/actuato
P (16 MHz) l)
rs

Dual-core
FreeRTOS Wireless IoT
MCU with Tensilica Wi-Fi +
ESP32 / nodes, edge
wireless LX6 (up to Bluetooth/BLE
bare-metal devices
240 MHz)

ARM No OS / Depends on part Advanced

STM32 MCU Cortex-M RTOS (some with embedded,
(e.g., F4) family (M0–M7/M (FreeRTO Ethernet, industrial-style
33/…) S etc.) wireless lines) projects
Table: Different platforms and their use

What is latency
Latency is the central reason fog computing and edge devices exist. In simple terms, they
move computation closer to where data is created so decisions happen fast enough to matter.

Latency = the time delay between:

 data being generated

 data being processed

 an action or response occurring
Even delays of milliseconds can be critical in real-time systems.

Why latency is a problem in cloud-only computing

Traditional cloud computing:
 Sends data from devices → distant data centers

 Adds network delay, congestion, and jitter

 Can be unreliable if connectivity drops
For applications that need instant reactions, this is a deal-breaker.

Role of latency in edge computing

Edge devices (sensors, cameras, gateways, robots) process data at or near the source.
Latency advantages
 Ultra-low latency (microseconds to milliseconds)

 No round-trip to the cloud

 Immediate local decision-making
Why this matters
 Autonomous vehicles braking in time

 Industrial robots avoiding collisions

 Smart cameras detecting threats instantly
Edge computing exists primarily to minimize latency.

Role of latency in fog computing

Fog computing sits between edge devices and the cloud.

Think of it as a latency-aware middle layer.

How fog reduces latency

 Processes data at local or regional nodes

 Filters, aggregates, and analyzes data closer to users

 Only sends necessary data to the cloud
Latency sweet spot
 Faster than cloud

 More powerful than individual edge devices

 Handles near-real-time analytics
 Figure 2.13 Fog Computing

Edge vs Fog vs Cloud (latency perspective)

Layer Typical Latency Role
Edge Microseconds–milliseconds Instant reactions
Fog Milliseconds–tens of ms Fast local analytics
Cloud Tens–hundreds of ms Heavy processing, storage

Figure 2.14 cloud vs edge Computing

Applications where latency is critical

Latency directly determines system success in:
 Healthcare (patient monitoring, emergency alerts)

 Smart grids (fault detection, load balancing)

 Industrial IoT (predictive maintenance)
 AR/VR (motion sickness prevention)
 Smart cities (traffic control, surveillance)
In these cases, high latency = failure or danger.
Fog computing and edge devices exist because latency matters more than raw computing
power.
 Edge computing minimizes latency to near zero for instant actions

 Fog computing balances low latency with broader system coordination

 Cloud computing remains for non-urgent, large-scale processing
Local processing means data is processed on or very near the device where it’s
generated, instead of being sent to a distant cloud server. It’s the backbone of edge and fog
computing, mainly to cut latency and improve reliability.
What happens in local processing
1. Data is generated (sensor, camera, machine, wearable)
2. Processing happens locally (on-device or nearby node)
3. Immediate action is taken
4. Only useful or summarized data is sent to the cloud (optional)

Why local processing matters

1. Ultra-low latency
 No round-trip to the cloud

 Responses in milliseconds or less

 Critical for real-time systems
2. Reliability
 Works even if internet/cloud connection fails

 Essential for remote or mission-critical environments

3. Reduced bandwidth usage
 Raw data stays local

 Only filtered or aggregated data is transmitted

4. Better privacy & security
 Sensitive data processed locally

 Less exposure over networks

Local processing at different layers

On edge devices
 Microcontrollers, gateways, smart cameras

 Handles instant decisions

 Example: motion detected → alarm triggered
In fog nodes
 Local servers, routers, base stations

 Handles near-real-time analytics

 Example: traffic data optimized at city level

Examples of local processing

 Smart cameras: detect intrusions without cloud upload

 Industrial machines: stop operation if anomaly detected

 Healthcare wearables: alert instantly on abnormal vitals
 Autonomous vehicles: obstacle detection and braking

Local processing vs cloud processing

Aspect Local Processing Cloud Processing
Latency Very low High
Internet needed No / optional Yes
Privacy High Lower
Compute power Limited Very high
Best for Real-time actions Big data & training

Local processing in edge computing

 Processing happens on the edge device itself (sensor, camera, gateway).

 Used for instant, time-critical decisions.

 Eliminates network delay completely.
Example:
A smart camera detects motion and triggers an alarm locally.
Role:
Edge computing = device-level local processing for ultra-low latency.

Local processing in fog computing

 Processing happens at nearby fog nodes (routers, local servers, base stations).

 Handles data from multiple edge devices.

 Supports near-real-time analytics and coordination.
Example:
Traffic data from many cameras processed at a city fog node to optimize signals.
Role:
Fog computing = regional local processing with more resources than edge.

How local processing ties edge and fog together

1. Edge devices
o Perform immediate local processing

o Handle safety-critical tasks

2. Fog nodes
o Aggregate and analyze locally processed edge data

o Make fast system-level decisions

3. Cloud
o Receives only filtered or summarized data

o Used for storage and long-term analytics

Edge vs Fog (local processing focus)

Feature Edge Computing Fog Computing
Where processing occurs On device Near device
Latency Microseconds–ms Few–tens of ms
Scope Single device Multiple devices
Computing power Low Medium
Role of local processing Instant response Fast coordination

Local processing is fundamental to both edge and fog computing. In edge computing, data is
processed directly on edge devices to enable ultra-low-latency responses. Fog computing
extends local processing to nearby network nodes, allowing aggregation and fast analysis of
data from multiple edge devices. Together, they reduce dependence on the cloud while
supporting real-time and near-real-time applications
Latency and local processing are core to fog computing and edge devices, enabling real-time
decisions by minimizing data travel time and bandwidth needs compared to cloud-only
systems.

Role of Latency
 In Edge Devices: Processing happens directly on the device (e.g., STM32/ESP32
sensors), achieving ultra-low latency (milliseconds or less) for time-critical apps like
autonomous robotics or industrial control, where even 10-50 ms delays can fail tasks.
 In Fog Computing: Acts as an intermediate layer (e.g., Raspberry Pi gateways)
between edge devices and cloud; latency is low (tens of ms) but slightly higher than
pure edge due to local network hops, yet far better than cloud WAN delays (seconds).
 Key Benefit: Both reduce latency by 90%+ vs. cloud, supporting apps needing <100
ms responses (e.g., AR/VR, vehicle-to-vehicle comms).
Role of Local Processing
 Data Filtering and Analytics: Edge/fog nodes perform initial processing (aggregation,
filtering, ML inference), sending only summaries/alerts to cloud—cutting bandwidth
by 70-90% and enabling offline operation.
 Real-Time Enablement: Local compute (e.g., on SBCs like Pi or MCUs like STM32)
handles deterministic tasks (e.g., anomaly detection in sensors), bridging IoT scale
with cloud scalability.
 Resilience and Efficiency: Reduces cloud overload, costs, and failure risks; fog adds
geo-awareness and interoperability for distributed systems.
Edge vs Fog Comparison
Aspect Edge Devices (Local Fog Computing (Intermediate Layer)
Processing)

Latency Lowest (on-device, <10 ms) Low (local net, 10-100 ms)

Processing Directly on endpoint Fog nodes/gateways serving multiple

Location (MCU/SBC) edges

High (no transmission for raw

Bandwidth Savings data) High (filtered data to cloud)

Scalability Device-by-device Network-wide via nodes

Use Example ESP32 robot sensor fusion Pi gateway for factory sensors
This fits your microcontroller/robotics focus—e.g., STM32 on edge for low-latency control,
fog nodes for fleet coordination
Analog sensors output continuous voltage/current signals proportional to the measured
quantity, while digital sensors provide discrete binary data after internal processing.
Analog vs Digital Sensors
Aspect Analog Sensors Digital Sensors

Continuous voltage/current (e.g., 0- Discrete digital (e.g., I2C/SPI

Output 5V for temp) binary values)

Finite (bit-limited, e.g., 12-24

Resolution Theoretically infinite bits)

Noise High (EMI, interference affects

Susceptibility signal) Low (internal filtering/ADC)

Processing Needs Requires external ADC on MCU Direct digital interface

Cost/Complexity Cheaper, simpler hardware Higher but easier integration

Examples Thermocouples, potentiometers, DHT22 humidity, MPU6050 IMU

 Aspect Analog Sensors Digital Sensors

strain gauges
Sensor Calibration
 Purpose: Corrects offset, slope, and nonlinearity errors for accuracy; essential for both
analog/digital to match known references.
 Methods:
 One-point: Fixes offset (e.g., zero temp); simple for linear sensors.
 Two-point: Adjusts offset + gain using low/high references.
 Multi-point/curve fitting: For nonlinear responses (e.g., thermocouples).
 Implement on MCUs like STM32 via ADC readings or lookup tables; automate with
known standards.
Noise Reduction Techniques
 Hardware: Low-pass filters, shielding, differential signaling for analog; high-QE
sensors, low-noise amps for both.
 Software/Firmware: Averaging/oversampling (e.g., 16x on ADC), Kalman filters,
moving averages; projection methods (e.g., PCA on sensor arrays).
 MCU-Specific: STM32 DMA for burst sampling + filtering; digital sensors inherently
less noisy.
Real-Time Data Acquisition
 Process: Use MCU timers/interrupts/DMA to sample at fixed rates (e.g., 1 kHz for
vibration), process locally (filter/calibrate), and act/store/transmit.
 MCU Fit: ESP32/STM32 ADCs (up to 5 MSPS) + RTOS for deterministic timing;
FIFO buffers on smart sensors reduce CPU load.
 Edge Benefits: Local processing cuts latency (e.g., <1 ms decisions) for
robotics/control; pair with STM32 for high-speed acquisition.
These techniques are key for reliable sensor data in your robotics/IoT projects—e.g., calibrate
analog IMU on STM32, use digital for noise-prone environments.
Analog vs Digital Sensors
Analog sensors
 Output is a continuous signal (voltage/current)

 Requires ADC (Analog-to-Digital Converter)

 More sensitive to noise
Examples: temperature (thermistor), pressure sensor, microphone
Pros
 High resolution

 Simple hardware
Cons
 Noise-prone
 Needs extra processing

Digital sensors
 Output is discrete digital data (0s and 1s)

 Built-in ADC
 Communicates via I²C, SPI, UART
Examples: digital temperature sensor, IMU, digital humidity sensor
Pros
 Less noise

 Easier integration
 Better accuracy
Cons
 Slight processing delay
 Higher cost

Quick comparison
Feature Analog Sensor Digital Sensor
Output Continuous Digital
Noise High Low
ADC required Yes No
Accuracy Lower Higher
Real-time use Moderate Excellent

2. Sensor Calibration

Calibration aligns sensor output with a known reference to improve accuracy.

Why calibration is needed

 Manufacturing variations

 Temperature drift
 Aging of sensors
 Environmental effects
Calibration process
1. Apply known reference input
2. Measure sensor output
3. Calculate correction factor
 4. Store calibration values
5. Apply correction during measurement
Example:
Adjusting a temperature sensor to match a standard thermometer.

3. Noise Reduction in Sensors

Noise is unwanted disturbance in sensor signals.
Common noise sources
 Electrical interference

 Environmental vibration
 Power supply fluctuations
 Quantization error (ADC)

Noise reduction techniques

Hardware methods
 Shielded cables

 Proper grounding
 Low-pass filters
 Stable power supply
Software methods
 Moving average filter

 Kalman filter
 Median filter
 Oversampling
Key idea:
Noise reduction improves signal quality before real-time decisions are made.

4. Real-Time Data Acquisition (RTDAQ)

Real-time data acquisition means:

 Data is sampled

 Processed
 Acted upon
within strict time limits.
RTDAQ components
1. Sensor (analog/digital)
 2. Signal conditioning
3. ADC (if analog)
4. Processor / controller
5. Storage or actuator
Why it matters
 Missed deadlines = system failure

 Common in control and monitoring systems

Examples
 Industrial automation

 Medical monitoring
 Robotics
 Autonomous vehicles

How everything connects together

1. Sensors collect physical data
2. Calibration ensures accuracy
3. Noise reduction improves reliability
4. Real-time acquisition enables immediate action
Together, these steps ensure accurate, fast, and reliable data for real-time systems.

Analog sensors produce continuous signals and require ADCs, while digital sensors provide
discrete outputs with built-in conversion and lower noise. Sensor calibration corrects
measurement errors by matching outputs to known standards. Noise reduction techniques
improve signal quality using hardware and software filtering. Real-time data acquisition
ensures timely collection and processing of sensor data for immediate system response.

Embedded Programming Basics

What is embedded programming?

Embedded programming is the development of software that directly controls hardware
using microcontrollers (MCUs) or microprocessors (MPUs) under real-time constraints.
Unlike general-purpose software:
 Runs on resource-limited hardware

 Often executes continuously

 Must meet timing deadlines

Core components of an embedded system

Hardware
  MCU/MPU (CPU, RAM, Flash)
 GPIO pins
 Timers & counters
 ADC/DAC
 Communication peripherals
Software
 Firmware (C/C++)

 Device drivers
 RTOS (optional)
 Interrupt handlers

Embedded programming characteristics

Deterministic execution
 Same input → same execution time

 Critical for industrial control

Low-level hardware access
 Register-level programming

 Bit manipulation
 Memory-mapped I/O
Resource constraints
 Limited RAM and Flash

 Low processing power

 Power efficiency required

Programming concepts in embedded systems

Polling vs Interrupts
 Polling: CPU repeatedly checks device status

 Interrupts: Hardware signals CPU when action is needed (preferred)

Timers
 Generate precise delays

 Schedule tasks
 Measure time intervals
Real-Time Operating System (RTOS)
 Task scheduling
  Priority management
 Inter-task communication
 Used in complex industrial systems

2. Interfacing with Industrial Equipment

Industrial equipment includes:
 Sensors (temperature, pressure, flow)

 Actuators (motors, valves, relays)

 PLCs
 HMIs
 Industrial networks

Sensor interfacing
Analog sensors
 Output: voltage or current (0–10V, 4–20mA)

 Requires:
o Signal conditioning

o ADC

o Isolation (optocouplers)

Digital sensors
 Output: digital data

 Interfaces:
o I²C

o SPI

o UART

Industrial concern:
Noise immunity and accuracy.

Actuator interfacing
Types of actuators
 DC motors

 Stepper motors
 Servo motors
 Solenoid valves
Interfacing requirements
 Driver circuits (H-bridge, relays)

 Power amplification
 Electrical isolation
 Feedback (encoders, limit switches)
Key rule:
Embedded controller never drives high-power loads directly.

Interfacing with PLCs

Embedded systems often act as:
 Data acquisition units

 Intelligent sensors
 Communication gateways
Common PLC communication methods
 Digital I/O

 Analog I/O
 Fieldbus communication

Industrial communication protocols

Serial protocols
 RS-232: short distance

 RS-485: long distance, multi-drop (very common)

Industrial protocols
 Modbus RTU / TCP

 CAN bus
 Profibus
 Ethernet/IP
Embedded programming role:
 Packet framing

 Error detection
 Timing control
 Protocol stack implementation

Real-Time Constraints in Industrial Interfacing

Hard real-time
 Missed deadline = system failure
  Example: emergency shutdown
Soft real-time
 Performance degrades if late

 Example: monitoring systems

Embedded firmware must guarantee:
 Predictable response

 Low interrupt latency

 High reliability
Safety, Isolation, and Reliability
Electrical isolation
 Optocouplers

 Isolation amplifiers
Watchdog timers
 Reset system on software failure

Fault handling
 Sensor failure detection

 Communication timeout handling

 Safe-state transitions

Typical Industrial Embedded Workflow

1. Read sensor data
2. Apply calibration & filtering
3. Execute control algorithm
4. Drive actuators
5. Communicate with PLC/HMI
6. Log faults & status
7. Maintain real-time deadlines

Embedded programming involves developing firmware that directly controls hardware under
strict timing and resource constraints. In industrial systems, embedded controllers interface
with sensors, actuators, PLCs, and communication networks using analog, digital, and
industrial protocols. Real-time execution, reliability, electrical isolation, and safety
mechanisms are critical to ensure accurate control, fault tolerance, and continuous operation
of industrial equipment.
Embedded programming involves writing efficient C/C++ code for microcontrollers (MCUs)
to control hardware in resource-constrained environments, with industrial interfacing using
robust protocols like RS-485, CAN, and ASi for sensors/actuators/PLCs.

Embedded Programming Basics

 Languages and Tools: Primarily Embedded C (subset of C with hardware access via
registers, bit manipulation); tools include IDEs (STM32CubeIDE, Arduino IDE,
PlatformIO), compilers (GCC), debuggers (JTAG/SWD), and flash programmers.
 Core Concepts:
 Setup: Init peripherals (GPIO, clocks); main loop or superloop for tasks.
 Interrupts/Timers: Handle events asynchronously (e.g., timer for PWM,
external for buttons).
 Peripherals: GPIO (digital I/O), ADC/DAC (analog), UART/SPI/I2C (comm),
PWM (motors/servos).
 Bare-Metal vs RTOS: Start bare-metal (no OS, direct control); scale to
FreeRTOS/ChibiOS for multitasking in complex apps.
 Best Practices: Optimize for low memory/power (e.g., volatile vars, no recursion),
handle resets/watchdogs, modular code.
Example: Blink LED on STM32/ESP32—set GPIO pin high/low in loop with delay.

Industrial Interfacing Protocols

Industrial setups demand noise immunity, long-distance, and determinism; MCUs like
STM32/ESP32 interface via these:
Protocol Description Use Cases MCU Support
(STM32/ESP32)

Serial async; RS-485 Point-to-point

UART/RS-232/ multi-drop (up to 32 sensors, Modbus Native UART; RS-485
RS-485 nodes, 1.2 km) RTU via transceiver

Robust bus for real-

time control (1 Mbps, Automotive, STM32 hardware CAN;
CAN error detection) robotics, PLCs ESP32 via transceiver

2-wire master-slave Factory

ASi (Actuator (up to 62 slaves, sensors/actuators, MCU as master via
Sensor Interface) 100m) for binary I/O simple PLC IO UART modulation

Ethernet/Modbus High-speed STM32 Ethernet MAC;

TCP networked PLC gateways, HMI ESP32 WiFi/Eth

Industrial Complex Via

Profibus/Profinet fieldbus/Ethernet automation modules/transceivers

Interfacing with Industrial Equipment (Detailed Analysis)

Sensors (Input)
 Digital: Limit switches, encoders—read via GPIO/interrupts; debounce with timers.

 Analog: 4-20mA/0-10V transmitters (pressure/temp)—use MCU ADC (e.g., STM32

16-bit), calibrate/scale in firmware.
 Analysis: Noise reduction via averaging/filters; STM32 DMA for high-speed
acquisition without CPU load. ESP32 for wireless sensor nets.
Actuators (Output)
 Relays/Contactor: GPIO + opto-isolator for 24VDC isolation.

 Motors/Valves: PWM for speed, H-bridge drivers (e.g., L298N); PID loops in
firmware for control.
 VFDs/Servos: Modbus/CAN commands from MCU to drive.
 Analysis: Feedback loops (encoder + PID) for precision; safety interlocks via
watchdogs/emergency stops.

PLC Integration
 MCUs as IO expanders/gateways: Modbus RTU/TCP over RS-485/Ethernet to PLC
(e.g., Siemens S7); ASi for simple fieldbus.
 Hybrid Setup: STM32 for real-time control + ESP32 for cloud (MQTT); CAN for
vehicle/robotics fleets.
 Challenges/Solutions:

Challenge Solution

Noise/EMC Twisted pair, isolators, shielding

Distance/Power RS-485 repeaters, 24V supplies

Determinism RTOS priorities, CAN arbitration

Scalability Master-slave (ASi), mesh (ESP-NOW)

MODULE 3

IIoT Network Requirements

Reliability, Real-Time Capability, Fault Tolerance
Industrial Internet of Things (IIoT) networks are deployed in mission-critical
environments such as manufacturing plants, power grids, oil refineries, transportation
systems, and process industries. Unlike consumer IoT, failures or delays in IIoT can cause
equipment damage, safety hazards, financial loss, or production downtime. Hence, IIoT
networks must satisfy strict performance and dependability requirements.
1. Reliability
Definition
Reliability refers to the ability of an IIoT network to consistently deliver data correctly,
completely, and on time, even under harsh industrial conditions.
Why reliability is critical in IIoT
Industrial systems rely on sensor data and control commands for:
 Process automation
 Safety monitoring
 Predictive maintenance
 Closed-loop control systems
A single lost or corrupted packet can:
 Trigger incorrect actuator behavior
 Cause production errors
 Lead to unsafe operating conditions
IIoT networks demand ultra-high reliability, often exceeding 99.999% uptime, to prevent
costly downtime in manufacturing and automation. This is achieved through ruggedized
hardware tolerant to harsh environments (-40°C to +85°C, shock/vibration), redundant power
supplies (N+1 or 2N configurations), and high MTBF ratings like 224,604 hours. Network
redundancy, such as failover paths and QoS features per IEEE 802.1, minimizes data loss
from failures or spikes
Key aspects of reliability in IIoT networks
1. Guaranteed data delivery
o IIoT networks often require at-least-once or exactly-once delivery semantics.
o Protocols such as TSN, PROFINET, Ether CAT, and OPC UA ensure
deterministic delivery.
2. Low packet loss
o Industrial environments suffer from electromagnetic interference, vibrations,
dust, and temperature variations.
o Networks must maintain stable communication despite these disturbances.
3. Error detection and correction
o Use of CRC checks, acknowledgments (ACKs), retransmissions, and forward
error correction (FEC).
4. Stable connectivity
o Devices must remain connected for long durations without frequent
reconfiguration or resets.
Example
In a chemical processing plant, if pressure sensor data fails to reach the controller reliably,
incorrect pressure regulation may result in hazardous conditions.

2. Real-Time Capability
Definition
Real-time capability is the ability of an IIoT network to transmit data and control messages
within strict, predictable time limits.
In IIoT, late data is as bad as lost data.
Types of real-time requirements
1. Hard real-time
o Missing a deadline causes system failure.
o Example: Emergency shutdown systems, robotic motion control.
2. Firm real-time
o Occasional deadline misses are tolerable but degrade system performance.
o Example: Assembly line synchronization.
3. Soft real-time
o Delays reduce quality but do not cause failure.
o Example: Monitoring dashboards.
Why real-time communication is essential
 Industrial control loops operate in milliseconds or microseconds.
 Actuators must respond immediately to sensor readings.
 Delays can destabilize control systems.
Real-time performance requires deterministic latency, often in microseconds, for
synchronized control in robotics and motion systems. Time-Sensitive Networking (TSN,
IEEE 802.1 standards) enables this via time synchronization, traffic scheduling with gate
controls, and priority queues isolating critical data. Protocols like Ether CAT or
PROFINET IRT complement TSN for bounded delays, allowing IT/OT convergence
without jitter.
Key parameters
1. Deterministic latency
o Fixed and predictable end-to-end delay.
o Achieved using Time-Sensitive Networking (TSN) and industrial Ethernet.
2. Low jitter
 o Variation in delay must be minimal.
o High jitter can cause oscillations in control loops.
3. Prioritized traffic
o Critical control messages are given higher priority than monitoring or logging
data.
Example
In a robotic welding system, even a few milliseconds of delay in control commands can
misalign the welding path, leading to defective products.
3. Fault Tolerance
Definition
Fault tolerance is the ability of an IIoT network to continue operating correctly even when
components fail.
Importance in industrial environments
 Hardware failures are unavoidable due to long operating hours.
 Maintenance shutdowns are costly.
 Systems must remain operational despite partial failures.
Types of faults
 Node failures (sensor/actuator malfunction)
 Link failures (broken cable, wireless interference)
 Power failures
 Software or firmware crashes
Fault tolerance ensures continuous operation despite node/link failures via detection,
rerouting, and recovery. Mechanisms include redundant paths, dynamic SDN-based
rerouting, self-healing reconfiguration, and ML-driven anomaly prediction. Adaptive
protocols like AFRR maintain >85% delivery at high fault rates, reducing latency by 30%
through proactive management. Clustering and replication distribute backups across nodes.
Fault tolerance mechanisms
1. Redundancy
o Multiple sensors measuring the same parameter.
o Duplicate communication paths and backup controllers.
2. Network reconfiguration
o Automatic rerouting of data when a link or node fails.
o Common in mesh and ring topologies.
3. Failover mechanisms
 o Immediate switching to backup devices or networks.
4. Health monitoring and diagnostics
o Continuous monitoring of network and device status.
o Faults detected early and isolated before escalation.
5. Graceful degradation
o System continues to operate with reduced performance instead of total failure.
Example
In a power substation, if one communication link fails, the network reroutes traffic through
alternate paths, ensuring uninterrupted monitoring and control.

Relationship Between the Three Requirements

Requirement Role in IIoT
Reliability Ensures correct and consistent data delivery
Real-time capability Ensures data arrives within strict time limits
Fault tolerance Ensures continuous operation despite failures
All three are interdependent:
 Reliable data that arrives too late is useless.
 Real-time data without fault tolerance risks system collapse.
 Fault tolerance without reliability leads to incorrect decisions