Logistics & Mobility// definition

Geofence: a zone, 3 triggers, and the hysteresis that stops it lying

In short

A geofence is 3 things, not 1: a zone — a radius or a polygon — a trigger rule of entry, exit or dwell, and a hysteresis policy setting how far and how long a vehicle must cross before the event fires. Drop the third part and you have built a duplicate-arrival generator, because position scatter at a boundary crosses it repeatedly while the vehicle stands still.

Key takeaways

  • A geofence is a zone, a trigger rule and a hysteresis policy. Two of the 3 is a duplicate-event machine.
  • Give exit a wider boundary than entry. A single shared boundary is what makes a parked vehicle flap.
  • Size the zone against fix quality, not the site plan. A 75-metre zone needs fixes good to 15 metres.
  • A zone smaller than the distance covered between 2 pings can be crossed without ever being observed.

A geofence is a zone with a trigger rule and a hysteresis policy attached. The zone says where. The trigger says which crossing you care about — entry, exit, or having stayed. The hysteresis policy says how much crossing counts, and it is the part that gets skipped. A zone and a trigger with no hysteresis fires 9 alerts for 1 arrival.

The reason sits in the physics of the fix, not the code. A parked vehicle does not report a constant position: consecutive fixes scatter by tens of metres, more in a yard surrounded by steel. Park on a boundary and the position crosses it repeatedly while the engine is off.

Radius or polygon, and when each one is wrong

  • A radius is a point and a distance: cheap to draw, defensible for small round sites — a kerbside drop, a weighbridge, a fuel stop.
  • A radius fails on long sites. A 400-metre distribution centre needs 200 metres from its centre, capturing the access road and the neighbouring unit.
  • A polygon is the site's real footprint: right for depots, ports and shared estates, and it needs redrawing when the yard is re-laid.
  • A stale polygon beats no polygon only until the site changes; then it is confidently wrong.

The 3 triggers, and the question each answers

  • Entry. The vehicle has arrived and the detention clock starts. It is wired straight into what a carrier status message actually says, so a noisy fence becomes a customer-visible lie.
  • Exit. The vehicle has left, the dwell clock closes and the next leg starts. It fires late more often than entry: a vehicle can sit 20 metres outside a fence for an hour.
  • Dwell. Still inside after a stated period — 40 unplanned minutes at a service area, a trailer parked overnight. The only trigger needing a duration rather than a crossing.

A dwell rule meaning 'stopped working' rather than 'still inside' needs vehicle state as well as position, and that state has provenance problems of its own — see what an ignition event actually proves.

Hysteresis: the third part, and the one that gets skipped

The accuracy gate surprises people. A fix accurate to 300 metres says nothing about a 150-metre zone either way, so admitting it is worse than no fix at all. It needs the feed to carry the field, which is why it belongs in the 8 fields a position report must carry.

Sizing a zone by what kind of site it is

Site typeShape and sizeConfirmationWhy
Depot or yard gatePolygon, at least 120 m across2 fixes plus 60 sVehicles queue on the approach; a radius misses them or swallows the highway
Customer kerbside, dense urbanRadius, 150 m3 fixes plus 120 sSignals bounce off buildings and the driver stops around the corner
Unit on a shared estatePolygon per unit2 fixes plus 90 sUnits 40 m apart, so a radius fires for the wrong tenant
Motorway service areaPolygon of the footprint5 min dwell firstPassing traffic on the adjacent carriageway clips the boundary
Weighbridge or small yardRadius, 75 m2 fixes plus 30 sTight zones need good fixes; on a phone-based feed this size is unusable
Site types with the shape, size and confirmation each needs

That last row is the general rule in disguise: zone size follows fix quality, not the site plan. A 75-metre zone needs positions good to 15 metres — reasonable from a hardwired unit with a roof-line antenna, optimistic from a handset in a pocket, the trade-off in installed unit or the driver's phone.

Where a correctly configured fence still fails

  1. The zone is smaller than the gap between pings. At 90 km/h a vehicle covers 750 metres in 30 seconds, so a 100-metre zone on that cadence is never observed.
  2. The fence covers a road it should not, so any vehicle driving past produces an arrival.
  3. Fences overlap. On a shared estate a vehicle sits inside 3 zones at once, and the trigger cannot say which site it arrived at.
  4. The dwell rule crosses midnight in the wrong time zone, so a 4-hour parking alert fires on the wrong side of the shift change.

A zone with no hysteresis does not detect arrivals. It detects the moment a fix fell on the far side of a line — a different event that looks identical in a database.

When events go wrong, check the breadcrumb trail the vehicle actually drove first: it shows whether the vehicle crossed or the position wandered. The zone editor, state machine and event log that keep this maintainable are the small, specific product work we describe under MVP and product builds. Related definitions sit in telematics and the live vehicle data plane, inside our logistics and mobility work.

Frequently asked questions

Short answers to the follow-ups this page tends to raise.

What is a geofence in fleet management?

It is a defined zone around a place, paired with a rule about which crossings matter and a policy for how much crossing counts. Fleets use them to detect arrivals and departures, run dwell clocks and flag vehicles stopping where they should not. The rules, not the shape, make it usable.

What is the difference between entry, exit and dwell triggers?

Entry fires when a vehicle crosses in, exit when it crosses out, and dwell when it has stayed inside for a stated duration. Entry and exit are crossings sharing one boundary logic; dwell is a timer needing no crossing. Most operations want entry and exit for time on site, dwell for unplanned stops.

Should I use a radius or a polygon geofence?

Use a radius for small, roughly circular locations and a polygon for anything with a real footprint. A radius wide enough for a long site also contains the access road and the neighbours, producing arrivals for vehicles that never turned in. A polygon costs maintenance.

Why does the same vehicle trigger a geofence several times?

Because it is parked near the boundary and its position is scattering across it. Consecutive fixes on a stationary vehicle differ by tens of metres, so a shared boundary with no confirmation rule flips repeatedly. Separate the boundaries, require 2 or 3 fixes on the new side, and hold for 60 seconds.

  • geofence
  • telematics
  • arrival detection
  • fleet operations
// shipped work

The work behind this page

Builds from our portfolio that this page draws on.

Working on something in this space?

Tell us where you are in a sentence or two. We'll tell you honestly whether we're the right team, and what a sensible first slice of the work looks like.

Start the conversation