RFID deployment guide

This page is for the moment the code already runs in the lab and it is time to take hardware on site: choosing antennas, setting power, establishing the real read zone. The readers are usually the integration developer together with the warehouse manager or the installation technician. The radio terms on this page, polarization, near field, Q, session, all have short explanations in the glossary.

An RFID system runs reliably not because of one good device but because the whole chain agrees: tags suit the goods, antennas suit the zone, power suits the site, and the software understands the stream coming back. Tuning one link while forgetting the rest is where most troubled RFID projects come from.

The deployment sequence

Keep the order: choosing devices before pinning the use case is shopping by guesswork, and signing off before measuring with real goods is signing off on paper.

Start from the use case

Each kind of read point, dock door, conveyor, retail ceiling, desktop POS, handheld counting, needs a completely different read zone, so everything starts from "what must we detect, and where". Answer and write down these six questions before choosing any equipment:

• What carries the tags: material, packaging, and which way the tags face while passing the read point.
• How many tags per pass, and whether goods move fast or slow.
• What read rate is required, and how much bleed-over from the neighboring area is acceptable.
• Where the read point sits, and whether tagged goods that do not belong to the pass are nearby.
• What event each pass must produce in Easy Inventory, WMS, ERP, or POS.
• Which bands and devices the installation country permits.

Survey the site

1. Draw two zones: where reads must happen, and where reads must never happen.
2. Mark the trouble sources on the sketch: metal shelving, liquid-heavy goods, lights and motors, network gear, neighboring readers.
3. Test with the actual tag on the actual product, in the worst orientation reality will produce.
4. Measure at three power levels: low, medium, and the intended setting.
5. Test with real densities and real motion; never conclude from one tag standing still.
6. Record everything: antennas, cables, power, session and Q, frequency profile, and the results.

Three pieces of physics to keep in mind while reading results: metal reflects and makes tags mounted on it hard to read; liquids absorb; nearby readers interfere with each other. And raising power extends the zone into unwanted places just as quickly as into wanted ones.

Choose and place antennas

• Pick polarization from tag orientation: tags always facing one fixed way get the longest range from a linear antenna; tags stuck every which way need circular polarization so every orientation still reads.
• Pick the antenna style from the distance: near field for tight, close zones like tagging stations and smart shelves; far field for dock doors, conveyors, and multi-meter ranges.
• Keep cables as short as possible and low-loss: every dB lost in the cable is energy that never reaches the antenna.
• Never mount an antenna directly on metal unless the antenna or bracket was designed for it.
• Overlap coverage just enough to kill blind spots without spilling into the neighboring read point's area.

While tuning, start at low power and raise it gradually, watching two numbers side by side: how many wanted tags read, and how many out-of-zone tags bled in.

Multiple readers

The usual production architecture:

Prefer wired Ethernet and a dedicated VLAN where the infrastructure allows, and monitor each reader's heartbeat: a dock-door reader silently offline means shipments through that door vanish from the system with nobody warned.

Readers standing near each other must be coordinated so they do not trample each other's air. The strategies, used alone or combined:

Strategy	Behavior	Trade-off
TDMA	Readers take turns in time slots	Simple, but each reader gets less scan time
FDMA	Each reader uses a different channel group	Good for fixed layouts; limited by legal channels
LBT	Listen for a clear channel before transmitting	Adaptive; depends on region and device
FHSS and channel masks	Hopping or complementary masks	Reduces sustained collisions when configured well
Dense Reader Mode	Narrow transmission mode, more interference-tolerant	May reduce per-reader throughput

Enable Dense Reader Mode when several readers stand within a few meters or adjacent doors see each other; for a lone reader, enable it only when testing proves a benefit.

Avoid tag starvation

One strong tag right next to the antenna can hog the air so weaker tags never get a word in. The symptom shows in the numbers: total reads run into the thousands while unique EPCs come in below the real item count.

The mitigations, tried top to bottom:

• Tune Q to the tag population size instead of leaving the default.
• Use session persistence so already-read tags stay quiet a while, yielding air to unread ones.
• Scan antenna ports sequentially instead of all at once.
• Lower power toward the strong tag and raise it only where genuinely needed.
• Alternate targets A and B between scan rounds.
• Split the population with Select commands when the above are not enough.

Starting templates

These templates are starting points for testing, not guaranteed designs; your site will demand further tuning:

Scenario	Starting template
Dock door	Four antennas, two per side, facing inward; around 30 to 33 dBm; tune until the neighboring door stops reading
Conveyor tunnel	Antennas on four sides of the belt; around 25 dBm; RF shielding between lines
Retail ceiling	Circular antennas on a 3 to 4 m grid; around 20 to 24 dBm; test metal shelving and overlap carefully
Handheld counting	Session S1, moderate power around 25 dBm; hold 0.5 to 1 m from goods and walk about 1 m per second
Desktop POS	Near-field antenna or a very tight zone; the lowest power that still reads everything on the desk

Acceptance tests

Before signing off a read point, run all eight and record each result:

1. Test with the least, the typical, and the most goods this point will ever see.
2. Rotate goods through every realistic orientation, in the real packaging.
3. Count three numbers: tags read correctly, tags missed, and out-of-zone tags bleeding in.
4. Switch on the neighboring readers and test again, proving they do not break each other.
5. Pull the network and reboot the reader mid-run; confirm the system recovers by itself.
6. Send the same item through twice; confirm the system records exactly one stock movement.
7. Power everything off and on; confirm the configuration survives.
8. Pick any EPC and trace it backwards from the read to the final warehouse document or order.

When all eight pass, save the RF configuration together with the firmware and application versions as one set, so the next read points install identically.

Read next

• Production checklist: the overall sign-off once each read point passes.
• RFID glossary for developers: the radio terms used on this page.
• Reading tags and deduplication: the software counterpart of the hardware you just installed.