IoT Integration on the Factory Floor: Sensors, Protocols, and ERP Connectivity

A manufacturing plant running 100 machines 16 hours per day generates approximately 200 GB of raw sensor data daily. Most of that data is noise -- normal operating readings confirming that equipment is working as expected. The value lives in the 0.1% of readings that indicate something is changing: a bearing beginning to degrade, a process drifting toward the control limit, a tool approaching end of life.

The challenge of factory floor IoT is not collecting data. Modern sensors are inexpensive, reliable, and easy to install. The challenge is transforming raw sensor data into business actions -- maintenance work orders, quality holds, production schedule adjustments, and purchasing triggers -- within timeframes that matter. An alert that a motor is overheating is only useful if it arrives before the motor fails, and if the alert triggers a response in a system where someone is watching.

This article is part of our Industry 4.0 Implementation series. For a comprehensive sensor and network architecture guide, see Smart Factory Architecture.

Key Takeaways

• Sensor selection based on failure modes (not machine types) eliminates the most common IoT deployment mistake -- monitoring what is easy to measure rather than what matters
• MQTT has become the de facto standard for manufacturing IoT due to its lightweight publish-subscribe model that handles thousands of sensors without polling overhead
• Edge computing processes 80-90% of data locally, reducing cloud costs and enabling sub-millisecond response times for safety and quality applications
• The integration layer between IoT platform and ERP determines whether sensor data drives business action or just populates dashboards nobody checks

---

Sensor Strategy: Monitor What Matters

Failure Mode to Sensor Mapping

The right approach is to start with failure modes, not equipment lists:

Sensor Density Guidelines

---

Communication Protocols

Protocol Comparison for Manufacturing IoT

MQTT Architecture for Manufacturing

MQTT has emerged as the standard protocol for manufacturing IoT because of three properties:

1. Publish-subscribe model: Sensors publish data to topics. Consumers subscribe to topics they need. No polling, no wasted bandwidth.
2. Quality of Service levels: QoS 0 (at most once) for high-frequency sensor data where occasional loss is acceptable. QoS 1 (at least once) for alerts. QoS 2 (exactly once) for critical commands.
3. Lightweight footprint: A single MQTT broker can handle 100,000+ concurrent connections on modest hardware.

Topic hierarchy example:

• factory/line-1/cnc-001/vibration/x-axis
• factory/line-1/cnc-001/temperature/spindle
• factory/line-1/cnc-001/status/running
• factory/line-1/cnc-001/alert/threshold-exceeded

---

Edge Computing Architecture

Three-Tier Processing Model

What to Process Where

---

ERP Integration Patterns

The integration between IoT platform and ERP is where factory floor data becomes business intelligence:

Pattern 1: Event-Driven Integration

IoT platform publishes events. ERP subscribes and creates business objects.

Pattern 2: Batch Data Sync

Aggregated data transferred to ERP on a schedule for reporting and planning:

Pattern 3: Bidirectional Integration

ERP sends instructions to the factory floor:

Odoo's REST API enables all three integration patterns. For manufacturing-specific IoT integration with Odoo, contact ECOSIRE.

---

Network Infrastructure Requirements

Industrial Network Design

Cybersecurity for OT Networks

---

ROI of Factory Floor IoT

Payback period: 6-14 months.

---

Getting Started

1. Identify your top 5 failure modes: What unplanned events cost you the most? Match sensors to those specific failure modes.

2. Start with one production cell: Deploy 20-50 sensors on a single cell. Prove value before scaling.

3. Choose MQTT: Unless you have specific OPC-UA requirements, MQTT provides the simplest path to scalable IoT deployment.

4. Integrate with ERP immediately: Do not build dashboards first. Connect IoT events to ERP actions (maintenance requests, quality holds) from day one.

5. Build your integration with Odoo: ECOSIRE builds IoT-to-ERP integration pipelines that transform sensor data into maintenance work orders, quality holds, and production intelligence.

See also: Industry 4.0 Implementation Guide | Smart Factory Architecture | Predictive Maintenance Implementation

---

How many sensors does a typical factory need?

Sensor count depends on manufacturing type and monitoring objectives. A rule of thumb is 5-15 sensors per critical machine. A 100-machine factory typically deploys 500-1,500 sensors in the initial phase, expanding to 2,000-5,000 as use cases prove out. Start with the machines that have the highest unplanned downtime cost and expand based on demonstrated value.

Can existing legacy machines be retrofitted with IoT sensors?

Yes. Retrofit sensors (clamp-on vibration, non-contact temperature, split-core current) can be installed on virtually any machine without modification. For machines with PLC controls, OPC-UA or Modbus gateways can extract existing control data without touching the PLC program. Retrofit IoT is one of the fastest-ROI investments because it extends the useful life of existing equipment while providing modern monitoring capabilities.

What is the difference between MQTT and OPC-UA for manufacturing?

MQTT is a lightweight messaging protocol ideal for sensor telemetry -- high volume, low overhead, simple to implement. OPC-UA is a more comprehensive protocol designed specifically for industrial automation with built-in data modeling, security, and discovery. In practice, many factories use both: MQTT for sensor-to-edge communication and OPC-UA for PLC/SCADA-to-platform communication. They complement rather than compete.