UHF RFID for Warehouses and Asset Tracking: Basics and ERP Integration

RFID promises something every warehouse manager wants: records that match what is physically on the shelf, without someone scanning each item. The technology works, but results depend on engineering decisions that are easy to underestimate. This article covers the essentials of UHF RFID and the integration work that turns raw reads into reliable business events.

How UHF RFID works, briefly

A reader sends a radio signal through its antennas. Tags within range respond, returning an identifier. UHF (ultra-high frequency) systems, in the passive tag form most warehouses use, can read tags at distances of several metres and can read many tags at once, without line of sight. Frequency bands and permitted power differ by region, so equipment and configuration must match local regulation.

The components

• Tags. Passive tags come in different form factors and chips. The right choice depends on what the tag is attached to: cardboard, plastic, metal, liquids or fabric each behave differently.
• Readers and antennas. Fixed readers cover portals, conveyors and shelves. Handheld readers support stock counts and exceptions. Antenna type, placement and polarisation shape the read zone.
• Edge software. Collects reads, filters noise, de-duplicates and buffers.
• Backend services. Convert reads into business events and store history.
• Integration with ERP or WMS. Publishes the events where users and processes need them.

We work with Alien Technology readers on embedded and warehouse projects, and the principles below apply across vendors.

Designing read points

A read point is a place where you want to know that tagged items passed or are present: a dock door, a gate, a conveyor, a shelf or a room. For each, define the question you are answering. "Did pallet X leave the building?" demands a different design from "What is on aisle 4?".

Good design depends on:

• Environment. Metal racking and liquids reflect and absorb radio energy. A site survey with test tags saves expensive surprises.
• Antenna layout. Overlapping zones help coverage but can cause stray reads from neighbouring areas.
• Power and settings. Tune power to read what you need and nothing else.
• Tag placement and orientation. Put tags consistently, and test the worst case rather than the best.

Always pilot at one site before rolling out.

From reads to events

Raw reads are noisy: the same tag is seen many times, some tags are missed, and some are read from the next room. The edge layer should:

1. De-duplicate repeated reads within a time window.
2. Filter by signal strength, tag type or expected ranges.
3. Infer direction at portals using sequences across antennas.
4. Buffer locally when the network is down and synchronise later.
5. Emit events such as "tag arrived at read point", "left read point" and "seen in zone".

Keep raw reads for a limited time so you can reprocess when rules improve.

Integrating with ERP and inventory systems

The ERP does not want millions of reads; it wants meaningful business transactions. A reliable pattern is:

• Map identifiers. Maintain a registry that links each tag identifier to an item, asset or logistics unit. Encoding standards such as GS1 EPC can help where applicable.
• Convert events into transactions. A goods-receipt event at a dock portal, matched to an expected delivery, becomes a receipt confirmation or an exception.
• Use idempotent interfaces. Messages will be retried, so the same event must not post twice.
• Handle exceptions visibly. Unknown tags, missing items and mismatches go to a dashboard that operators can act on, rather than silently into logs.
• Reconcile. Schedule comparisons between RFID-observed stock and ERP stock and investigate differences, which also reveal process problems.
• Secure the chain. Authenticate devices and services, and protect data in transit.

Asset tracking and access control

The same stack supports asset tracking, where tags on equipment record movement between rooms or sites, and gate control at seaports and airports, where reads trigger authorisation checks. These cases raise stronger requirements for audit trails, availability and fail-safe behaviour.

Pitfalls to avoid

• Skipping the survey and discovering interference after installation.
• Treating read rate as the only metric, rather than business accuracy.
• Leaving tag registration out of the process design.
• Ignoring operators: if exceptions are hard to resolve, the data degrades.

Summary

RFID succeeds when the radio design, edge logic and ERP integration are treated as one system and validated in a pilot against a real business question. Start small, measure accuracy against physical counts and expand with confidence.