Race Timing with RFID: A Measurement System That Happens to Look Like a Sport

A race director does not buy a tracking system. They buy a result that will survive an appeal. That distinction shapes every engineering decision: where the mat goes, how many antennas point at it, and what happens when four hundred people cross it in ninety seconds.

Timing is a measurement system, not a checkpoint

The output is a number with an uncertainty, and the uncertainty has to be small.

• Resolution and accuracy are different things; a 1/100 second resolution with 0.4 second accuracy is a lie with decimals.
• The timestamp must be traceable to a clock that is synchronised across every mat on the course.
• Drift between the start mat and the finish mat is the error nobody checks and everybody inherits.
• The system must record the read event, not just the fastest read, so disputes can be examined.
• Publish the method, because a result without a stated method invites argument.

Tag on shoe, bib or ankle band

Each placement trades reliability against logistics.

• The shoe tag is the classic choice: close to the ground, short read range, low interference between runners.
• The bib tag reads at chest height, which is convenient for spectators and worse for a crowded mat.
• The ankle band is common in triathlon and survives water, but needs a strap that does not chafe.
• Disposable tags must be verified after the athlete attaches them, not only at packet pickup.
• Reusable tags need a cleaning and a read test between events, because a dead tag is discovered too late.

Antenna layout at the start

The start line is where the crowd physics defeats the radio plan.

• Point antennas across the direction of travel, not along it, to shorten the dwell and reduce collisions.
• A single mat across a wide start is not enough; the field is strongest at the centre and weakest at the edges.
• Add a second row of antennas offset by half a wavelength to fill the nulls.
• Test with the real density: a hundred volunteers walking is not ten thousand runners sprinting.
• The start gun, the mat and the clock must be triggered by the same event, or the splits will not reconcile.

The pile-up problem at the mat

Congestion is the dominant error source, and it is not random.

• When bodies overlap, the tags behind a torso are attenuated and the read probability drops.
• The failure is not uniform: it is worst in the middle of the pack, where the density peaks.
• Antenna diversity helps; a single-polarisation reader at a crowded mat will lose people.
• Reduce dwell time by narrowing the mat and by placing antennas on both sides of the lane.
• Expect the elite to read perfectly and the middle pack to be the risk, which inverts the intuition.

Redundancy and the best-of rule

More reads do not automatically mean a better time.

• Take the earliest read for a start mat and the latest for a finish mat, because that is the physically conservative choice.
• For intermediate mats, take the median or the first crossing, depending on the mat geometry.
• Reject reads that imply an impossible speed between two mats; they are almost always cross-reads.
• Keep the raw read list per athlete, because an appeal will ask for it.
• Never delete a read to make a result look tidy; annotate it instead.

Chip time, gun time and the rules

Which number is official is a decision, not a default.

• Gun time is the ranking basis in most mass-start races, and chip time is the personal result.
• For wave starts, the wave gun is the reference, not the race gun.
• Elite and prize categories usually require gun time for placing and chip time for records.
• Publish both, clearly labelled, or the results page will generate its own controversy.
• The timing protocol should state the rule before the race, not after the protest.

Participants who are not standard runners

The exceptions are where timing systems embarrass their operators.

• Wheelchair athletes cross a mat faster and lower, and the antenna height assumptions break.
• Handcycles and strollers change the tag height and the shielding pattern.
• Pacers, guides and support runners may carry a tag without being competitors.
• Relay teams need per-leg splits, which means each leg has its own tag and its own mat logic.
• DNF and DNS handling must be a designed state, not a blank row.

Data integrity and result publication

The result is published in minutes and questioned for weeks.

• Ingest reads with the reader identity and the antenna port, so a bad mat can be isolated later.
• Detect duplicates by tag and by mat, not by athlete name.
• Run an automated plausibility check before publication, and route the flagged rows to a human.
• Keep an immutable event log separate from the results database.
• Rehearse the correction workflow, because a correction at minute ninety is a different process from one at minute nine.

Weather, water and asphalt

Outdoor events are an RF environment that changes hourly.

• Rain detunes wet clothing and attenuates the signal; the read rate falls before anyone notices.
• Metal barriers, gantries and vehicles near the mat reshape the field in ways the plan did not include.
• Asphalt and concrete reflect differently, and the reflection changes the null pattern.
• Temperature affects the clock hardware and the battery of a reusable tag.
• The triathlon swim exit needs a mat that survives standing water and a wet athlete.

Live splits and the tracking layer

Live tracking is a different product with different guarantees.

• A live split is a best-effort feed, and it must be labelled as provisional.
• The tracking layer aggregates reads into positions; the timing layer keeps the authoritative event.
• Mobile network coverage at remote mats is the usual failure, so plan store-and-forward.
• Give the commentary team a view that is fast and a results team a view that is correct.
• Never let the live feed become the record of the race.

Rehearsal and race-day command

The system is proven in the rehearsal, not in the race.

• Run a full-scale rehearsal with the real mats, the real clocks and a crowd of volunteers.
• Assign one person to the clock, one to the reads and one to the results, with a shared timeline.
• Prepare a manual fallback for a dead mat: a backup camera and a hand-timed backup.
• Have a spare reader, spare antennas and spare tags on site, not in a van across town.
• Debrief within forty-eight hours, while the anomalies are still fresh.