54 Processes and Systems
54.1 Learning Objectives
By the end of this chapter, you will be able to:
- Define processes and systems in the context of IoT device architectures
- Distinguish between open-loop and closed-loop control systems and explain their trade-offs
- Explain feedback mechanisms and their role in achieving system stability
- Describe PID (Proportional-Integral-Derivative) control and when each term is needed
- Apply block diagram decomposition to model IoT systems hierarchically
- Evaluate which control strategy suits a given IoT application scenario
Minimum Viable Understanding
- Processes transform inputs into outputs – every IoT device executes a process (the algorithm) within a system (the hardware and software working together). Understanding this distinction is the first step to designing any IoT solution.
- Feedback loops are what make IoT devices “smart” – without feedback, a device blindly executes instructions; with feedback, it measures its own output, compares it to a target, and self-corrects. This is the foundation of every thermostat, cruise control, and precision agriculture system.
- PID control is the industry workhorse – combining proportional (react to current error), integral (eliminate accumulated error), and derivative (anticipate future error) actions, PID handles the vast majority of real-world control problems from HVAC to robotics.
For Beginners: What Are Processes and Systems?
Every electronic device you use – from a smart thermostat to a self-driving car – works by following a simple pattern: take in information, do something with it, and produce a result.
Process = The “recipe” or set of steps a device follows. Think of it like a cooking recipe: take ingredients (inputs), follow steps (process), get a meal (output).
System = All the physical parts working together. The kitchen itself – oven, pots, utensils, ingredients, and the cook – is the system.
In IoT terms:
| Concept | Everyday Example | IoT Example |
|---|---|---|
| Input | You feel cold | Temperature sensor reads 16 degrees C |
| Process | You decide to turn up the heat | Microcontroller compares 16 degrees C to target 22 degrees C |
| Output | You adjust the thermostat dial | Controller sends signal to turn on heater |
| Feedback | You check if you feel warmer | Sensor measures temperature again |
Two types of control:
- Open-loop (no feedback): A toaster runs for a set time – it does not check if the bread is actually toasted
- Closed-loop (with feedback): A thermostat continuously checks the temperature and adjusts the heater
Why this matters: Understanding processes and systems lets you design IoT devices that actually work reliably in the real world, not just in theory.
Sensor Squad: The Control System Team!
The Sensor Squad was visiting a smart greenhouse, and everything was going wrong!
“The tomatoes are wilting!” cried Sammy the Temperature Sensor. “It’s 40 degrees in here – way too hot! The cooling fans aren’t working properly.”
Max the Microcontroller examined the problem. “I see the issue. The fans are running on a simple timer – they turn on for 10 minutes every hour, no matter what the temperature is. That’s an OPEN-LOOP system. It doesn’t check what’s actually happening!”
“That’s like watering plants with your eyes closed!” said Lila the LED, blinking in alarm.
Bella the Battery had an idea. “What if Sammy keeps watching the temperature, and Max adjusts the fans based on what Sammy sees? When it’s too hot, speed up the fans. When it’s just right, slow them down!”
“That’s called a CLOSED-LOOP system!” Max explained excitedly. “And I know an even better trick called PID control!”
The Squad got to work:
- Sammy (the sensor) measured the temperature: “It’s 40 degrees C – we need 25 degrees C!”
- Max (the controller) calculated: “That’s 15 degrees too hot! Let me use my three super-powers…”
- P-power (Proportional): “15 degrees off? Fans at HIGH speed!”
- I-power (Integral): “We’ve been too hot for 20 minutes? Push fans even harder!”
- D-power (Derivative): “Temperature is dropping fast now – ease off the fans so we don’t overshoot!”
- Lila (the indicator) showed green when the temperature was perfect.
- Bella (the battery) made sure everything had enough power.
Within 10 minutes, the greenhouse was a perfect 25 degrees C, and the tomatoes were happy!
Key Words for Kids:
| Word | What It Means |
|---|---|
| Process | A recipe that a machine follows – like steps for making a sandwich |
| System | All the parts working together – the sandwich maker, bread, fillings, and plate |
| Open-Loop | Doing something without checking the result – like setting a timer and walking away |
| Closed-Loop | Checking the result and adjusting – like tasting soup while cooking |
| PID | Three super-powers: react to NOW (P), remember the PAST (I), predict the FUTURE (D) |
54.2 Chapter Overview
This chapter covers the foundational concepts of processes and systems in IoT architectures – from basic input-process-output models through feedback mechanisms to advanced PID control. The material is organized into four focused sections for progressive learning.
54.2.1 How Processes and Systems Relate in IoT
54.4 Control System Architecture at a Glance
The following diagram shows the generic closed-loop control architecture that underlies all IoT feedback systems covered in this chapter:
54.4.1 Open-Loop vs Closed-Loop Comparison
54.4.2 PID Controller Term Contributions
54.5 Learning Path Recommendation
Beginners: Start with Overview then Feedback Mechanisms
Intermediate: All sections in sequence
Advanced: Jump to PID Control for in-depth control theory
54.6 Worked Example: Smart Greenhouse Temperature Control
This worked example ties together all four parts of the chapter – processes, systems, feedback, and PID control – in a single realistic IoT scenario.
Scenario: You are designing a smart greenhouse that must maintain an internal temperature of 25 degrees C (+/- 1 degree) for optimal tomato growth. The greenhouse has electric heaters, ventilation fans, and temperature sensors connected to an ESP32 microcontroller.
Step 1: Identify the System (Part 1 – Overview)
| Component | Role | Details |
|---|---|---|
| Temperature sensor (DHT22) | Input | Measures greenhouse air temperature |
| ESP32 microcontroller | Controller | Runs control algorithm |
| Electric heater (2 kW) | Actuator (heating) | Warms the greenhouse |
| Ventilation fan | Actuator (cooling) | Expels hot air |
| Setpoint | Reference | 25 degrees C target |
Step 2: Choose Control Strategy (Part 3 – Control Types)
| Approach | Pros | Cons | Verdict |
|---|---|---|---|
| Open-loop (timer) | Simple, cheap | Cannot adapt to weather changes | Not suitable |
| On/Off closed-loop | Simple feedback | Temperature oscillates +/- 3 degrees C | Too imprecise |
| PI closed-loop | Eliminates steady-state error | Slight overshoot possible | Good choice |
| Full PID | Minimizes overshoot | Derivative amplifies sensor noise | Best if sensor is filtered |
Decision: Use PI control (Kp + Ki), since DHT22 sensor noise makes the derivative term unreliable without additional filtering.
Step 3: Design the Feedback Loop (Part 2 – Feedback)
- Sensor reads current temperature: PV = 18 degrees C
- Error calculator: e = SP - PV = 25 - 18 = +7 degrees C (too cold)
- PI controller computes: u(t) = Kp * e(t) + Ki * integral of e(t)
- Controller activates heater at proportional power level
- As temperature rises, error decreases, heater power reduces
- Integral term ensures zero steady-state error (no permanent offset)
Step 4: Tune the Controller (Part 4 – PID)
Using conservative tuning for a thermal system (slow dynamics):
- Kp = 10: For each 1 degree C error, apply 10% heater power
- Ki = 0.5: Accumulate 0.5% additional power per second of error
- Sample period = 5 seconds: Thermal systems change slowly
Expected Response:
- At t=0: Heater at 70% power (Kp * 7 degrees C error)
- At t=60s: Temperature reaches 22 degrees C, heater at 30% + integral contribution
- At t=120s: Temperature reaches 24.5 degrees C, heater at 5% + integral holding
- At t=180s: Temperature stable at 25.0 degrees C (+/- 0.3 degrees C)
Key Design Decision: The integral term eliminates the 1-2 degrees C steady-state error that P-only control would leave. Without Ki, the greenhouse would stabilize at approximately 23 degrees C instead of 25 degrees C.
Try the P-Only Controller Yourself:
Common Pitfalls in Process and Control System Design
1. Using open-loop control where feedback is essential. A timed irrigation system wastes 40-60% more water than a soil-moisture-based closed-loop system. If your output depends on unpredictable environmental conditions, you need feedback.
2. Setting PID gains too aggressively. High proportional gain (Kp) causes oscillation. High integral gain (Ki) causes windup and overshoot. High derivative gain (Kd) amplifies sensor noise. Start conservative and increase gradually.
3. Ignoring sensor noise in the derivative term. The D term computes the rate of change of the error signal. Noisy sensors produce wildly fluctuating derivatives that make the actuator chatter. Always apply a low-pass filter to the sensor signal before computing the derivative, or simply use PI control.
4. Forgetting integral windup protection. When an actuator saturates (e.g., heater at 100%), the integral term continues accumulating error. When the setpoint is finally reached, the bloated integral causes massive overshoot. Implement anti-windup by clamping the integral when the actuator is saturated.
5. Sampling too slowly for the process dynamics. The Nyquist criterion applies to control loops too. If your process can change significantly in 1 second, sampling every 10 seconds will miss critical dynamics and cause instability. Rule of thumb: sample at least 10 times faster than the fastest disturbance you need to reject.
54.7 Knowledge Check
Test your understanding of the key concepts covered across all four parts of this chapter.
54.8 Enhanced Summary
54.8.1 Key Concepts Covered
| Concept | Definition | IoT Relevance |
|---|---|---|
| Process | A sequence of steps transforming inputs to outputs | The algorithm running on the microcontroller |
| System | Interconnected components working as a whole | Sensors + MCU + actuators + communication |
| Open-Loop | Control without output measurement | Simple, cheap, but cannot adapt |
| Closed-Loop | Control with sensor feedback | Self-correcting, essential for precision |
| Feedback Loop | Output measurement fed back to input | Foundation of all smart IoT devices |
| PID Control | Three-term controller (P + I + D) | Industry standard for precision control |
Decision Framework: Selecting Control Strategy for IoT Applications
Step 1: Determine if Feedback Control is Needed
| Question | If YES → Feedback Required | If NO → Open-Loop Acceptable |
|---|---|---|
| Does output depend on unpredictable disturbances? | Wind affects drone → Feedback required | Coffee timer (predictable heating) |
| Must output precisely match a target value? | 78°F ± 0.5°F aquarium → Feedback required | Notification LED (approximate OK) |
| Are consequences of control errors severe? | Medical pump dosing → Feedback critical | Decorative fountain (cosmetic) |
| Does environment change over time? | Battery drains (output varies) → Feedback compensates | Wired device (stable power) |
Step 2: Choose Control Type
| Control Type | When to Use | Pros | Cons | Cost | IoT Example |
|---|---|---|---|---|---|
| Open-Loop | Predictable process, no critical accuracy needs | Simple ($5), reliable | Cannot adapt to disturbances | $5-15 | Timed irrigation, LED blink |
| On-Off (Bang-Bang) | Acceptable to oscillate ± 10% around target | Simple ($15), low tuning effort | Temperature swings ± 2-5°C | $15-30 | Basic thermostat, level control |
| P-Only | Error tolerance 5-10%, fast response | Simple ($20), 1 parameter | Always has steady-state error | $20-35 | Fan speed, LED dimming |
| PI | Zero steady-state error required, moderate speed | Eliminates offset ($30), robust | Slower than PID, tuning effort | $30-60 | HVAC, aquarium heater, pressure |
| PID | Precision ± 1%, minimal overshoot, fast settling | Optimal performance ($50+) | Complex tuning, noise sensitive | $50-100 | Drone, 3D printer, CNC, robotics |
Step 3: Evaluate Process Characteristics
| Process Time Constant | Example Application | Recommended Control | Loop Frequency |
|---|---|---|---|
| Very fast (<100 ms) | Motor position, power converter | PID (edge-only) | 10-100 Hz |
| Fast (100 ms - 10 sec) | Liquid flow, small heater | PI or PID | 1-10 Hz |
| Medium (10 sec - 5 min) | Room temperature, pressure | PI (most common) | 0.1-1 Hz (6-60 sec) |
| Slow (5+ min) | Greenhouse, large tank | PI or On-Off | 0.01-0.1 Hz (10-100 sec) |
Step 4: Assess Sensor Quality
| Sensor Noise Level | Recommended Control | Rationale |
|---|---|---|
| Low (<1% of range) | PID OK | Clean signal supports derivative term |
| Moderate (1-5%) | PI preferred | D-term amplifies noise; filter or skip D |
| High (>5%) | On-Off or P-only | Feedback unstable; keep it simple |
Example 1: Smart Refrigerator
- Process: Cooling with 10-minute time constant
- Target: 4°C ± 1°C (food safety)
- Disturbances: Door openings, ambient temperature
- Sensor: DS18B20 (± 0.5°C accuracy = low noise)
- Decision: PI control (Kp = 8, Ki = 0.2, 30-second loop)
- Result: Maintains 4.0°C ± 0.3°C, eliminates steady-state error from P-only
Example 2: Smart Fan (Room Comfort)
- Process: Fan speed control
- Target: 60% speed (comfortable airflow)
- Disturbances: None significant
- Precision: ± 10% acceptable (subjective comfort)
- Decision: P-only control (Kp = 2.0) or even open-loop (60% fixed)
- Result: Simple, cheap, sufficient for comfort applications
Example 3: Quadcopter Altitude Hold
- Process: Drone altitude with 200 ms time constant
- Target: 10.0 m ± 0.2 m (collision avoidance)
- Disturbances: Wind gusts, battery voltage drop
- Sensor: Barometer + ultrasonic (clean, 100 Hz update rate)
- Decision: Full PID (Kp = 12, Ki = 0.5, Kd = 3.0, 10 ms loop)
- Result: Holds 10.0 m ± 0.1 m in 5 mph wind, 300 ms settling time
Common Mistake Decision Tree:
├─ "I need precise control" → Does NOT automatically mean PID
│ ├─ Check sensor noise first
│ ├─ Check process time constant
│ └─ Try PI before PID (simpler, more robust)
│
├─ "PID is the best control" → NOT always true
│ ├─ 80% of applications work fine with PI
│ ├─ D-term only needed when overshoot is critical + sensor clean
│ └─ On-Off works for many comfort/safety applications
│
└─ "Cloud control is fine for all processes" → DANGEROUS assumption
├─ Fast processes (<1 sec) require edge control
├─ Cloud adds 100-500 ms latency
└─ Use cloud for setpoint optimization, not real-time loopsROI Comparison (smart greenhouse heating):
| Control Type | Hardware Cost | Tuning Effort | Temperature Stability | Energy Use | Total Annual Cost |
|---|---|---|---|---|---|
| Open-Loop (timer) | $10 | 0 hours | ± 5°C swings | 3,500 kWh ($420) | $430 (but kills plants) |
| On-Off | $25 | 1 hour | ± 2°C oscillation | 2,800 kWh ($336) | $361 |
| PI | $45 | 3 hours | ± 0.5°C stable | 2,200 kWh ($264) | $309 |
| PID | $80 | 8 hours | ± 0.3°C stable | 2,100 kWh ($252) | $332 (marginal improvement over PI) |
Verdict for greenhouse: PI control ($45 + 3 hours tuning) is the sweet spot – saves $57/year vs On-Off, achieves ± 0.5°C stability (sufficient for plants), and avoids the complexity of full PID.
54.8.2 Control Strategy Selection Guide
54.8.3 What You Should Remember
- Every IoT device is a system executing a process that transforms sensor inputs into actuator outputs
- Feedback is the key differentiator between simple and smart devices – it enables self-correction
- Closed-loop control always requires a sensor in the feedback path
- P-only control always has steady-state error – you need the integral term to eliminate it
- PI control handles 80%+ of real IoT applications – full PID is only needed for fast, noise-free systems
- Start with conservative tuning and increase gains gradually – aggressive tuning causes oscillation and instability
54.9 Prerequisites
Before diving into this chapter, you should be familiar with:
- Sensor Fundamentals and Types: Understanding how sensors measure physical phenomena
- Actuators: Knowledge of actuators that convert control signals to physical actions
- IoT Reference Models: Familiarity with the layered IoT architecture
54.11 What’s Next
| If you want to… | Read this |
|---|---|
| Start with processes and systems overview | Processes and Systems Overview |
| Study feedback mechanisms | Feedback Mechanisms |
| Learn about control types | Control Types and Cascade |
| Explore PID control implementation | Processes and Systems: PID Control |
| Go to the core PID fundamentals series | Processes and Systems Fundamentals |