Industrial IoT and Industry 4.0

Why it matters

Factories are connecting machines that used to be islands: PLC and sensor data now flow to edge computers and cloud platforms for condition monitoring, energy management, quality analytics and production dashboards. Mechatronics engineers are expected to know what Industry 4.0 actually involves — the architectures, protocols and metrics — and to size the data, bandwidth and power of an IIoT solution sensibly rather than repeat buzzwords.

Key ideas

Four industrial revolutions. 1.0: water and steam mechanisation. 2.0: electricity and mass production. 3.0: electronics, PLCs and IT automation. 4.0: cyber-physical systems — physical processes tightly coupled with computation and networking, so machines, products and systems exchange data and adapt.

Cyber-physical system (CPS). Sensors measure the physical process, computation (embedded, edge or cloud) decides, actuators act, and a network links them — the same sense–decide–act loop as a PLC, extended across the plant and enterprise.

Industrial IoT (IIoT) architecture.

1. Devices: smart sensors, PLCs, drives, robots, wireless sensor nodes.
2. Edge/gateway: collects data from fieldbuses (Modbus, PROFINET, EtherCAT), pre-processes it (filtering, feature extraction, buffering), translates protocols and enforces security.
3. Platform (on-premises or cloud): storage, time-series databases, analytics and machine learning, device management.
4. Applications: dashboards, OEE monitoring, predictive maintenance, energy management, MES/ERP integration. The automation pyramid is flattening into a network where data can go from a sensor to an analytics service directly — but real-time control and safety still stay in PLCs and controllers.

Key protocols and technologies.

• OPC UA: platform-independent, secure (certificates, encryption) client–server and publish–subscribe communication with a rich information model (objects, types, units) — the standard for machine-to-MES/cloud data.
• MQTT: lightweight publish/subscribe over TCP through a broker; clients publish to topics and subscribe to topics; QoS 0 (at most once), 1 (at least once), 2 (exactly once); 2-byte minimum fixed header — ideal for constrained devices and unreliable links.
• Wireless: Wi-Fi, Bluetooth LE, LoRaWAN and NB-IoT (low-power wide-area, small data, years on a battery), private 5G (low latency, many devices).
• TSN (Time-Sensitive Networking): IEEE 802.1 extensions bringing deterministic real-time traffic to standard Ethernet.

Edge vs cloud. The edge gives low latency, works when the internet link is down, and cuts bandwidth by sending features instead of raw data; the cloud gives elastic storage and compute and fleet-wide analytics. Most systems use both.

Applications.

• Predictive maintenance: monitor condition (vibration, temperature, current, oil) and act before failure rather than on a fixed schedule (preventive) or after breakdown (reactive). Vibration RMS, kurtosis and spectral bands are typical features.
• Digital twin: a virtual model of an asset or process kept synchronised with real data, used for virtual commissioning, what-if simulation and optimisation.
• OEE monitoring: automatic collection of downtime, speed and reject counts from PLCs.
• Additive manufacturing, AGVs/AMRs, collaborative robots, AR-assisted maintenance are often grouped under Industry 4.0.
• Reference architecture RAMI 4.0 and the asset administration shell describe how assets present their data in a standard way.

Security. Connecting OT to IT and the cloud enlarges the attack surface. Apply IEC 62443: zones and conduits, a DMZ, outbound-only connections from the edge, encrypted and authenticated protocols (OPC UA security, MQTT over TLS), certificate management, patching and least privilege.

Formulas

OEE = A × P × Q

• Overall equipment effectiveness (dimensionless). World-class is often quoted as about 0.85.

A = Run time / Planned production time P = (Ideal cycle time × Total count) / Run time Q = Good count / Total count

• A = availability, P = performance, Q = quality; times in the same unit (min or s).

R = N × n_bytes × 8 × f_s

• Raw data rate (bit/s): N sensors, n_bytes per sample, f_s samples per second per sensor.

Reduction factor = R_raw / R_edge

I_avg = D × I_active + (1 − D) × I_sleep, D = t_active / T

• Average current (A) of a duty-cycled wireless node; D = duty cycle.

Battery life = C / I_avg

• Life (h) for capacity C (A·h or mA·h, matching units); ignores self-discharge and temperature effects.

Worked examples

Example 1 (standard). A machine is scheduled for a 480-min shift with a 30-min planned break. It is down for 45 min, has an ideal cycle time of 0.5 min per part, makes 720 parts and 684 are good. Find A, P, Q and OEE.

1. Planned production time = 480 − 30 = 450 min; run time = 450 − 45 = 405 min.
2. A = 405 / 450 = 0.900.
3. P = (0.5 × 720) / 405 = 360 / 405 = 0.889.
4. Q = 684 / 720 = 0.950.
5. OEE = 0.900 × 0.889 × 0.950 = 0.760.

Answer: OEE = 76 % (A = 90 %, P = 88.9 %, Q = 95 %).

Example 2 (GATE level). A pump station has 50 vibration sensors each sampled at 10 kHz with 16-bit samples. (a) Find the raw data rate. (b) An edge gateway instead sends 16 features of 4 bytes per sensor once per second; find the new rate and reduction factor. (c) A wireless temperature node draws 20 mA for 50 ms every 10 s and 10 µA asleep, from a 2400 mA·h battery; estimate its life.

1. R_raw = 50 × 2 × 8 × 10 000 = 8 × 10⁶ bit/s = 8 Mbit/s (1 MB/s).
2. R_edge = 50 × (16 × 4) × 8 × 1 = 25 600 bit/s = 25.6 kbit/s.
3. Reduction = 8 × 10⁶ / 25 600 = 312.5.
4. D = 0.05 / 10 = 0.005.
5. I_avg = 0.005 × 20 + 0.995 × 0.010 = 0.100 + 0.00995 = 0.110 mA.
6. Life = 2400 / 0.110 = 21 830 h ≈ 2.5 years.

Answers: 8 Mbit/s raw vs 25.6 kbit/s from the edge (≈ 312× less); battery life ≈ 2.5 years.

Common mistakes

• Mixing bits and bytes in data-rate calculations (factor of 8).
• Computing performance with the actual rather than the ideal cycle time, or quality from rejects over planned output instead of total count.
• Treating OEE components as additive instead of multiplicative.
• Sending raw high-rate data to the cloud when features computed at the edge would do.
• Putting closed-loop control or safety functions in the cloud.
• Ignoring sleep current in battery calculations — at low duty cycles it can dominate.
• Assuming MQTT itself is secure; security comes from TLS, authentication and broker configuration.

For GATE ME

Industry 4.0 appears as conceptual MCQs (CPS, IIoT layers, digital twin, edge vs cloud, MQTT/OPC UA roles, predictive vs preventive maintenance) and simple numericals on OEE and its components, data rates and storage, and battery life of duty-cycled devices. Practise OEE with all its time definitions and careful unit handling.

Quick check

1. Name the three factors of OEE.
2. A = 0.95, P = 0.90, Q = 0.98. What is OEE?
3. Which MQTT QoS level guarantees exactly-once delivery?
4. What is a digital twin?
5. 100 sensors each send 2 kbit/s. What is the total rate?

Answers: 1. Availability, performance, quality 2. ≈ 0.838 (83.8 %) 3. QoS 2 4. A virtual model of a physical asset kept in sync with its real data 5. 200 kbit/s