Supply Chain IoT: RFID, GPS, Condition Monitoring

Supply chains lose goods to theft, spoilage, and misrouting long before a failure shows up in a quarterly report. The fix isn't more paperwork — it's instrumenting pallets, containers, and vehicles so location and condition are known continuously rather than at a handful of checkpoints. Supply chain IoT combines three sensing/tracking layers — RFID for identity, GPS/GNSS for location, and condition-monitoring sensors for environmental state — feeding a common telemetry pipeline to gateways and the cloud. Understanding the trade-offs between these layers is the difference between a system that scales to millions of SKUs and one that drowns in cost or dead batteries.

The Three Sensing Layers

RFID — identity at the item/case/pallet level. Radio-frequency identification tags carry a unique ID (typically an EPC — Electronic Product Code) read by a fixed or handheld reader without line of sight. Two flavors dominate logistics:

Passive UHF is the workhorse: tags cost a few cents to ~$0.15 each, need no battery, and a dock-door portal reader can read hundreds of tags in a pallet load in under a second. The trade-off is range and reliability — RF absorption by liquids and detuning near metal reduce read rates, so portal placement and tag orientation matter as much as the chip itself.

GPS/GNSS — location in transit. Once goods leave a facility, RFID stops working (no reader infrastructure on a highway). GPS/GNSS trackers, usually bundled with a cellular or satellite modem, report position at intervals — typically every 1–15 minutes for road freight, sparser for ocean containers to conserve battery. Key design parameters:

• Fix time and power: a cold GPS fix can take 30–60 s and draw tens of mA; assisted-GPS (using cell tower almanac data) cuts this to a few seconds.
• Reporting interval vs. battery life: a tracker reporting every 5 min over cellular (LTE-M/NB-IoT) can run for 2–5 years on a single primary cell battery; every 30 s might drop that to weeks.
• Geofencing: most trackers evaluate position locally and only transmit on geofence entry/exit or dwell events, not every fix — this is the single biggest lever for battery life.

Condition monitoring — is the cargo still good? Location without state is not enough for pharma, food, or electronics. Condition sensors ride alongside GPS in the same tracker or as separate loggers:

How the Pieces Connect: A Reference Data Flow

[Pallet: passive UHF tags] --(dock portal reader)--> [WMS: case-level receipt]
        |
        v
[Truck: GPS+cellular tracker, temp/shock sensors]
        |  (BLE/LoRa to gateway inside trailer, or direct cellular)
        v
[Gateway / tracker modem] --(LTE-M/NB-IoT or satellite)--> [Cloud ingestion]
        |
        v
[Time-series DB] --> [Alarm rules: excursion, geofence exit] --> [Dashboard / API to TMS]

RFID answers "what is this and did it leave/arrive," GPS answers "where is it now," and condition sensors answer "is it still fit for use." All three converge on the same telemetry pipeline — typically a lightweight payload (CBOR or Protobuf, not verbose JSON, to save airtime) with a timestamp, position, and a handful of sensor readings, published over MQTT or a similar protocol once the tracker has connectivity.

Worked Example: Cold-Chain Pallet, Regional Trucking Route

Assume a pharma pallet ships 400 km, ~6 hours transit, temperature spec 2–8 °C.

Tag/tracker selection: one active tracker per pallet (not per case) — cost-driven, since only pallet-level location/condition is needed; individual cases keep passive UHF tags for warehouse receiving/putaway.

Reporting budget:

• GPS + temperature reading every 5 min → 72 reports over 6 h.
• Each report over NB-IoT: ~200 bytes payload + protocol overhead ≈ 300 bytes actual airtime.
• Total data: 72 × 300 B ≈ 21.6 kB for the trip — trivial for NB-IoT, well inside typical cellular data plans costing cents per device per month.

Battery check: NB-IoT transmit draws roughly 200–250 mA for ~1–2 s per report (deep sleep otherwise, ~10 µA). Energy per report ≈ 0.22 A × 1.5 s ≈ 0.33 A·s = 330 mA·s ≈ 0.092 mAh. Over 72 reports: ≈ 6.6 mAh — negligible against a typical 1200 mAh tracker cell, confirming a multi-year life is dominated by leakage/sleep current and periodic GPS fixes, not the radio itself. This checks out: a well-designed low-rate tracker's dominant power draw is GPS acquisition, not the NB-IoT uplink.

Excursion detection: if temperature exceeds 8 °C for more than 15 continuous minutes (3 consecutive 5-min samples), the tracker raises a local alarm flag and sends it immediately out-of-cycle rather than waiting for the next scheduled report — this is the "alarm" message type distinct from routine "telemetry," and it must not be rate-limited the way routine reports are.

Practical Design Implications

• Match the tag/tracker type to the granularity you actually need. Passive UHF for case/pallet ID inside facilities; active/cellular trackers only on the asset tier where in-transit visibility justifies the cost (often per-pallet or per-container, not per-item).
• Separate telemetry from alarms in the protocol design. Routine position/condition reports can be batched or delayed; threshold violations (temperature excursion, tamper, geofence breach) need immediate, low-latency transmission and should bypass normal reporting intervals.
• Budget power around the GPS fix and radio wake cycle, not just idle current — cold fixes and cellular registration dominate energy use far more than sleep current in most trackers.
• RF environment matters as much as tag spec. Metal containers, dense stacking, and liquid-heavy pallets all degrade passive RFID read rates — validate portal placement and tag position empirically, not just from datasheet range figures.
• Plan for intermittent connectivity. Ocean containers and remote yards lose cellular coverage; trackers need onboard buffering (store telemetry locally, backfill on reconnect) and satellite fallback for high-value or regulated cargo.
• Reconcile RFID and GPS data streams at the application layer. A pallet's RFID "departed dock" event and its tracker's GPS breadcrumb should agree in time and location — discrepancies are often the first signal of a data quality or process problem, not just a sensor fault.

Key Takeaways

• Supply chain IoT layers three complementary technologies: passive/active RFID for identity at fixed checkpoints, GPS/GNSS for continuous in-transit location, and MEMS/analog sensors for condition (temperature, humidity, shock, tilt, light).
• Passive UHF RFID is near-zero cost per tag but limited to short range and reader infrastructure; active RFID and cellular GPS trackers add cost and battery management but deliver continuous, infrastructure-free visibility.
• Battery life in GPS trackers is dominated by GPS fix acquisition and radio wake/registration overhead, not steady-state transmit current — geofencing and event-driven reporting are the primary levers to extend life.
• Condition-monitoring thresholds (e.g., cold-chain excursions) must generate immediate, unthrottled alarm messages, distinct from routine batched telemetry, to be operationally useful.
• Reliable systems reconcile RFID checkpoint events with GPS breadcrumbs and condition data in the cloud pipeline, using compact payload formats (CBOR/Protobuf) to keep airtime and cost low at scale.