Industrial IoT Traceability: From Raw Material to Finished Product

/ IoT / By

Why Industrial Traceability Has Become a Strategic Priority

In an environment of ever-stricter regulations — the Food Safety Modernization Act, the EU Medical Device Regulation (MDR), IATF 16949 in automotive — the ability to trace a product’s complete journey from raw material extraction to final customer delivery is no longer optional but a legal requirement for many sectors. Beyond regulatory compliance, industrial traceability has become a competitive lever: it reduces product recall costs, improves quality, optimizes logistics flows, and reassures prime contractors.

However, traditional methods — manual data entry, linear barcodes, Excel spreadsheets — are reaching their limits when faced with modern supply chain complexity. A recent Gartner study estimates that 80 % of industrial companies that invested in IoT for traceability reduced their quality-related losses by at least 25 % within two years. It is in this context that the Internet of Things provides a concrete technical answer, automating data collection at every stage of a product’s lifecycle.

This article details industrial IoT traceability architectures, available capture technologies (RFID, UWB, BLE, LoRaWAN), end-to-end protocols (GS1 EPCIS, OPC UA, MQTT), and presents real-world use cases in food processing, automotive, and pharmaceuticals.

The Limits of Traditional Traceability Approaches

Manual Entry and QR Codes: The Risk of Human Error

Manual data entry — whether scanning a barcode or writing a lot number on paper — introduces latency and a non-negligible error rate. A 2024 McKinsey study estimated that production lines relying on manual entry had traceability error rates of 3–7 %, rising to 12 % in high-speed environments (over 100 parts per minute).

QR codes and Data Matrix (2D barcodes) improve information density but retain several fundamental limitations:

  • Need for direct line-of-sight between reader and label
  • Hostile environments: dust, moisture, oils, high temperatures quickly degrade printed labels
  • No update capability: once printed, the code is frozen
  • No location without scanning: no real-time visibility on work-in-progress position

ERP Alone Is Not Enough

ERP systems (SAP, Microsoft Dynamics, Odoo) excel at managing transactions and inventory at the macro level, but struggle to capture real-time field events. The gap between leaving one workstation and arriving at the next — what industry calls the WIP tracking gap (Work In Progress) — is a major source of losses: misplaced parts, undetected bottlenecks, incomplete batches.

IoT bridges precisely this gap by adding a layer of sensors, electronic identifiers, and connectivity between field equipment and information systems.

IoT Traceability Architecture

A complete IoT traceability architecture rests on four functional layers, each addressing specific constraints of the industrial environment.

Layer 1 — Marking and Identification (Edge)

Marking is the starting point: every object (raw material, sub-assembly, finished product, pallet) must carry a unique machine-readable identifier. Available technologies:

  • RAIN RFID (UHF, 860–960 MHz): passive tags, range up to 10–12 meters, simultaneous reading of hundreds of tags. ISO 18000-6C / EPC Gen2v2 standard. Ideal for pallets, cartons, and reusable containers.
  • HF RFID / NFC (13.56 MHz): short range (a few centimeters), secure read, smartphone-compatible. Used for individual tracking of high-value products (medical devices, electronic components).
  • UWB (Ultra-Wideband): active tags with centimeter-level location (10–30 cm). Perfect for tracking tools, AGVs, or parts in a machining workshop.
  • BLE (Bluetooth Low Energy) 5.x: low-cost active tags, 2–5 year battery life, RSSI triangulation localization (1–3 meter accuracy). Good cost/accuracy compromise for batch tracking.
  • Laser-engraved QR codes and Data Matrix: for extreme environments (foundries, heat treatment), Direct Part Mark (DPM) laser engraving is sometimes the only viable option. High-temperature RFID tags (up to 200 °C) also exist.

Layer 2 — Reading and Capture (Gateway)

Fixed RFID readers positioned at strategic passage points (zone entrances, dock doors, conveyors, quality control stations) automatically capture marked object passages. Technical choices include:

  • UHF RFID portals: 360° read, up to 600 tags/second. Positioned at zone transitions.
  • Tunnel readers: for high-speed conveyors, 100 % carton read even with overlapping.
  • Floor RFID bridges: for cart and AGV passage, automatic pallet reading.
  • Mobile readers: Bluetooth scanners paired with tablets or Android terminals for handling operations.
  • LoRaWAN gateways: for long-range upstream (outdoor container tracking, cold storage lot traceability).

The choice of transport protocol between readers and the upper layer depends on criticality and data volume:

  • MQTT with QoS 2: reliable, asynchronous, ideal for continuous RFID read streams
  • OPC UA: ISA-95 standard, interoperable with existing PLCs and SCADA systems
  • HTTP/2 (gRPC): for cloud integrations with REST APIs
  • AMQP: advanced message routing, persistent queues for disconnection periods

Layer 3 — Event Engine and Correlation (Middleware)

The traceability middleware (sometimes called Tracking & Tracing Engine) receives raw reader events (tag ID, timestamp, location) and transforms them into business events. This is where all the system intelligence resides:

  • Aggregation: 50 parts passing a portal = one “Batch X entered zone Z” event
  • Filtering: suppression of redundant reads (a tag may be read 15 times in 2 seconds passing through a portal)
  • Correlation: linking a production event (welding, testing, assembly) to the object’s unique ID
  • Sequencing: reconstructing the exact operation order from timestamps
  • Anomaly detection: alerting if an object skips a process step or if inter-operation time exceeds a defined threshold

Layer 4 — Storage and Retrieval (Cloud/On-Premise)

Traceability data is massive: a medium-sized factory can generate several million events per day. Suitable storage solutions:

  • Time-series database (TimescaleDB, InfluxDB): for sensor time series and location history
  • Graph database (Neo4j, ArangoDB): ideal for upstream tracing (“which lots supplied this finished product?”) and downstream tracking (“where are all products made from this material lot?”)
  • Cold data lake (S3/MinIO, Parquet): for long-term retention (10–15 years as required in medical or aerospace) at low cost
  • Industrial blockchain (Hyperledger Fabric, Quorum): for cases where irrevocability and multi-actor sharing are critical (farm-to-fork food traceability, provenance certification)

The retrieval API exposes this data to business applications (ERP, MES, WMS, CRM, customer portal) typically via REST or GraphQL web services, with predefined queries for the most common use cases: track by ID, trace by lot, genealogy by product.

IoT Traceability Protocols and Standards

A traceability system is only valuable if data is interoperable across the supply chain. Several industrial standards play a key role:

GS1 EPCIS 2.0 — The Universal Standard

The EPCIS (Electronic Product Code Information Services) standard — now in version 2.0 ratified by GS1 in 2022 — defines the exchange format for traceability events. Each EPCIS event contains four dimensions:

  • What: which object (EPC / Global Trade Item Number — GTIN)
  • When: at what timestamp
  • Where: at what location (GLN — Global Location Number)
  • Why: which process step (business step: commissioning, shipping, receiving, aggregation, disaggregation)

EPCIS 2.0 brings native JSON-LD support (alongside legacy XML), management of non-GS1 identifiers (proprietary IDs), and standardized traceability queries via the EPCIS Query Interface. It is the exchange standard adopted by retail, pharmaceuticals (DSCSA in the US), and food (EU Food Information to Consumers regulation).

OPC UA Companion Specification for ISA-95

The OPC Foundation published a traceability-dedicated Companion Specification in 2023, aligned with ISA-95. It defines a standardized information model for:

  • Resources (equipment, operators, materials)
  • Operations (production, maintenance, quality)
  • Batches and sub-batches
  • Material consumption and allocations

OPC UA’s advantage is its ability to integrate directly with production APIs (CNC, robots, PLCs) without requiring an additional gateway — the OPC UA server often runs directly on the PLC or industrial PC.

MQTT Sparkplug B — Real-Time IIoT

The Eclipse Foundation’s Sparkplug B standard adds industrial semantics to native MQTT: data type definitions, connection state awareness (birth certificate / death certificate), and equipment hierarchy (host > device > group). It is the preferred protocol for lightweight edge computing traceability architectures, where the local gateway processes events and only sends aggregates to the cloud.

Three Real-World Use Cases

1. Food Processing — From Farm to Fork

In food processing, traceability has become a commercial acceptance criterion. A mid-sized poultry processor (50,000 tons/year) deployed a UHF RFID solution coupled with IoT:

  • Marking: RFID tags inserted in reusable plastic transport crates
  • Reading: RFID portals at receiving docks, cutting area entrance, and each packaging station
  • Environmental data: DS18B20 temperature sensors on 1-Wire bus in cold rooms, connected to a LoRaWAN gateway
  • Middleware: correlation between RFID events and temperature curves, real-time alert if a batch experienced cold chain breach

Results: 40 % reduction in manual investigations during quality checks, batch traceability time reduced from 4 hours to 12 seconds, and most importantly — during a simulated recall exercise — the ability to identify the 3 affected pallets out of 12,000 in under 5 minutes (down from 2 days with the previous system).

2. Automotive — Safety Part Genealogy

A Tier-1 automotive supplier manufacturing brake calipers implemented unit-level traceability using Data Matrix laser marking + RFID:

  • DPM marking: a Data Matrix laser-engraved on each steel caliper at the start of the line
  • RFID pallet tag: the carrier accompanying the caliper on the overhead conveyor carries an HF RFID tag containing the part ID
  • Automated readings: each CNC machining station and each test station communicates results via OPC UA to the MES, which associates them with the part ID
  • Brake testing: torque and pressure curves are timestamped and attached to the unique ID

Result: when a defect is detected at end of line, the system immediately identifies the machine, CNC program, operator, and raw material batch involved. The supplier reduced its scrap rate from 1.8 % to 0.3 % over 18 months, primarily through early detection of process drifts before they affected an entire batch.

3. Pharmaceuticals — DSCSA Compliance and Anti-Counterfeiting

In the US, the Drug Supply Chain Security Act (DSCSA) has mandated unit-level traceability since November 2023. A pharmaceutical laboratory deployed:

  • Serialization: each box receives a unique GS1 DataMatrix with GTIN + serial number + lot + expiry date
  • Hierarchical aggregation: box → carton → pallet, with UHF RFID reading at each level
  • Hyperledger Fabric blockchain: EPCIS events are signed and timestamped in a shared blockchain between manufacturer, wholesaler, and pharmacy
  • Verification API: a pharmacist can scan any box’s DataMatrix and instantly obtain the complete history from production

This multi-million euro investment per packaging line is now a prerequisite for accessing the US market. The system also detects counterfeiting attempts: any box whose serial number has already been scanned elsewhere triggers a real-time alert.

Measurable Benefits of IoT for Traceability

Beyond compliance, industrial players who have made the leap report tangible benefits. Here are the key indicators observed across 12 deployments accompanied by our team between 2022 and 2025:

Indicator Before IoT After IoT Improvement
Batch traceability time 2 to 8 hours 5 to 30 seconds ×100 to ×500
Loss/misplaced rate (WIP) 3 to 7 % < 0.5 % −85 %
Quality investigation duration 1 to 3 days 15 to 60 minutes −90 %
Safety stock needed 15 to 25 % 5 to 10 % −50 %
Targeted product recall (vs. total) Not possible Specific lot/pallet −80 % recall cost

These figures reflect a simple reality: accurate, real-time traceability transforms tracking data into a decision-making asset. Quality, logistics, and production teams work from the same event repository, without redundant data entry or information latency.

How to Deploy an IoT Traceability Solution in 5 Steps

Based on our deployment experience with over twenty industrial companies — from 50-person SMEs to 800-employee production sites — here is the methodology that maximizes ROI while minimizing risk:

Step 1: Audit and Flow Mapping (2–4 weeks)

Identify current bottlenecks (error rates, search time, non-quality costs), map physical flows and associated decisions at each stage. Define which objects to trace (batch, unit, pallet, container) and required granularity.

Step 2: Capture Technology Selection (1–2 weeks)

Select the right tag combination (UHF RFID, HF/NFC, UWB, BLE) for each environment: temperature, metal presence, throughput speed, budget per read point.

Step 3: Middleware Architecture and Integration (4–8 weeks)

Deploy the event engine, connect to MES/ERP via appropriate adapters (OPC UA, REST API, SOAP, EDI). Define correlation and filtering rules with business teams.

Step 4: Field Deployment and Validation (2–4 weeks)

Install portals, bridges, and readers; train operators; validate the complete chain on a pilot line before serial deployment.

Step 5: Production Go-Live and Ongoing Support (continuous)

Measure KPIs, adjust alert thresholds, extend the system to new lines and upstream suppliers as needed.

Conclusion: IoT Traceability — A Fast-ROI Investment

Industrial traceability through IoT is no longer an innovation project: it is an operational and regulatory necessity. Technologies are mature (UHF RFID, standardized EPCIS protocols, industrialized middleware), tag costs have dropped 40 % in five years, and ROI — from loss reduction, inventory optimization, and avoided mass recalls — is measured in months, not years.

At IOTINNOV, we support industrial companies in the diagnosis, design, and deployment of their connected traceability systems — from sensor selection to ERP integration. Every project begins with a free on-site audit of your flows to identify loss pockets and size the solution adapted to your real constraints.

👉 Contact our team for a personalized IoT traceability diagnosis.