Truck Sensor Networks

1. Overview and operating context

This report covers three truck variants: long-haul tractor-trailers, reefer / cold chain (Thermo King / Carrier Transicold style refrigeration units) and tankers. Heavy-duty trucks expose most engine, drivetrain and brake data on a J1939 CAN bus at 250 kbit/s, which a telematics gateway reads and uplinks over cellular. Trailer-side data (EBS, TPMS, reefer unit, cargo door) is added through ISO 11992, BLE/RF sensors, or J1939 / RS-232 / RS-485 links.

Evidence tags used throughout: S backed by a cited source; K domain knowledge or engineering estimate stated in the research notes; Inference a conclusion drawn here from the notes and not stated in them. The research notes state that the fetched sources did not give J1939 scaling, so decoding must be verified against SAE J1939-71 / J1939DA. OEM reefer logging intervals and alarm codes were not found in sources; reefer numbers are [K].

Typical truck deployment facts: GNSS HDOP 0.7-1.5 on highway; reporting 10-60 s while moving and 1-15 min while parked; rural trucks see 3-10% of drive time without cellular service, in episodes of 1-30 min and some over 1 h [K]; a long-haul truck generates roughly 5-30 records/s and 10-60 MB/day as JSON (2-15 MB binary), about 300-1500 MB/month, as a planning figure [K].

2. Sensor inventory

2.1 Common J1939 signals (applicable to all truck variants)

J1939 raw value conventions: "not available" is 0xFF (1 byte) or 0xFFFF (2 bytes); error indicator is 0xFE / 0xFExx. Temperatures use a -40 degC offset at 1 degC/bit; kPa scaling varies per SPN.

2.2 Long-haul truck-specific sensors

2.3 Reefer / cold chain sub-variant

Reefer units are read via CAN / J1939 or RS-232 / RS-485; remote setpoint, mode and defrost commands exist; BLE probes capture every 30 s [S]. Remaining values are [K].

2.4 Tanker sub-variant

All tanker values are [K].

3. On-vehicle networks and protocols

Typical J1939 broadcast periods [K]: EEC1 10-20 ms; pedal/load 50 ms; CCVS and fuel rate 100 ms; ET1 and most temperatures 1 s; DM1 1 s while a fault is active; TIRE1 on change or every 10 s.

3.1 Bridge-layer facts relevant to trucks

4. Duty cycles and operating profiles

4.1 Long-haul

• 8-11 h driving per day, 700-1000 km; highway 80-90 km/h (EU limiter 85-90, US 100-105 km/h); night driving common [K].
• Overnight rest 4-10 h with idle or APU / sleeper A/C at 1.5-3.5 L/h [K].
• Cruise at 1100-1400 rpm; 28-36 L/100 km loaded (US 6-8 mpg); empty about 15% lower [K].
• Climbs: rpm 1500-1800, fuel 50-80 L/h, coolant +5-10 degC; downhill engine brake or retarder, fuel near 0 [K].
• Idle share of engine hours: long-haul 6-12%; local/vocational 30-40% [K].
• Fuel stops: step up of 200-600 L over 5-15 min at about 10-20% level; DEF use is 5-10% of diesel use; DPF active regen every 500-1500 km lasting 20-40 min [K].
• Weekend freight is lower; trailer swaps change TPMS and trailer IDs [K].

4.2 Reefer distribution

• 10-40 drops per day, door open 5-45 min each, RAT spikes then recovers; the unit keeps running during stops [K].
• Dynamics [K]: pre-cool 30-90 min; pulldown of 20 degC to setpoint in 30-60 min (chilled) or 90-180 min (frozen); steady oscillation of +-0.3-1 degC with start-stop cycling every 5-30 min; air time constant 5-10 min, product 30-120 min.
• Excursion definition [K]: chilled above +5 degC or frozen above -15 degC for more than 30-120 min.

4.3 Tanker

• Loaded outbound, empty on return; level steps down at each delivery; 2-8 deliveries per run [K].
• Pump delivery 20-60 min at 500-2000 L/min [K].

4.4 Regulatory driving state

Tachograph state machine (driving, work, rest) follows EU 4.5 h then 45 min break, 9 h daily, 11 h rest; US 11 h in 14 h with a 30 min break after 8 h [K]. Inference: the duty-cycle state machine should force breaks consistent with these limits.

5. Faults and anomaly signatures

Injection rates are [K] estimates unless noted.

6. Events, alerts and thresholds

7. Generator guidance for the simulator

8. How the Sense simulator implements this

8.1 Scenario codes (tenant subvariant)

The Fleet tenant's variant truck selects the vehicle kind and its subvariant selects the scenario. The report's three variants are the platform's subvariant codes exactly as they are: long-haul, reefer, tanker. A tenant with another subvariant runs no truck; the simulator's own selector is used only when the tenant has no subvariant.

8.2 Protocols emitted and the one-vehicle-per-tenant model

Wialon input bits: 0 ignition, 1 cargo door (reefer) or hatch (trailer door at the dock, tanker dome hatch), 2 trailer connected, 3 alarm (reefer alarm, tanker tilt alarm, long-haul harsh event or active fault code). The iButton field carries the driver card in hex. CAN slow groups are sent every 10 s and the VIN every 600 s in all scenarios.

8.3 Signals actually emitted

Sense makes one reading per J1939 signal and per Wialon param, named as in the tables. NMEA is as for buses: GGA latitude, longitude, fix quality, satellites and altitude; RMC status, latitude, longitude, speed (knots) and course.

DM1 codes sent (amber lamp): SPN 110 FMI 16 (coolant over 102 degC, overheat episode of 10-25 min), SPN 1761 FMI 18 (DEF under 10%), SPN 117 FMI 1 (air pressure under 600 kPa), SPN 3719 FMI 16 (DPF soot load over 90%).

8.4 What Sense can and cannot read yet

The state below follows TRUCK_POINTS.md section 8 (checked against the working copy of the Sense listener sources, read only) and Sense's BUILD.md.

• Read by the working copy of Sense, not by the released build of 12 Oct 2026: TIRE1 (65268) and B1 (65274); GGA HDOP and the RMC time; the Wialon record time; a Wialon unit name table. J1939 groups read besides the engine ones: IC1 (65270, boost, intake manifold and exhaust gas temperature), ETC2 (61445, gear), TRF1 (65272, transmission oil), EBC1 (61441, brake pedal position and ABS active), AT1S (64891, DPF soot load) and the DEF tank temperature in 65110. The unit table covers the names of Table 14 that have a unit (for example trans_oil_t, egt, brake_t, dpf_dp, tk_flow, tk_tilt and tk_c<n>); names it does not know, flags and states stay without a unit.
• Still not read: axle and gross combination weight (65258), the J1939 vehicle number (65260; the OBD-II VIN works), the DPF differential pressure group, the tachograph driver state and card (65132 gives only the speed), the trailer EBS of ISO 11992 and PTO state. BUILD.md gives the reason for the weight, vehicle number and differential pressure as encoding or scaling not verified.
• No standard bus for the cargo equipment: the reefer controller (set point, return and supply air, probes, evaporator, mode, defrost, alarms) and the tank gauges (compartment levels, temperature, pressure, tilt, valves, pump flow, earthing) have no group in Sense; the simulator carries them in Wialon params and in Cargo ambient temperature. Their alarms arrive as ordinary 0/1 readings (rf_alarm, tk_alarm, rf_exc, tk_overfill); Sense does not turn them into events with thresholds.
• Not read at all: string params and the iButton field of Wialon IPS (driver ids go as the number driver_card); the native UDP form of Wialon IPS 2.0; Sense does not use seq or boot_id, so replays and duplicates count as new readings.
• Vehicle profiles: the Fleet compatibility check chooses a profile by variant and subvariant (long-haul, reefer, tanker); an empty or unknown subvariant uses the plain Truck profile.

8.5 Faults and the ground-truth file

Latent traits per truck are as for buses (clock ppm, chronic flag in 1%, frame loss 0.01-0.5%, bus-off 0.05-0.5 a day, weak battery in 5%). The generic faults of the bus work apply unchanged (J1939 not-available after ignition on, spike, encoding error, dropped and duplicated frames, VIN transport abort, stuck-at, drift, GNSS episodes, link outages with store-and-forward, reboot, TPMS leak of 5-20 kPa/day on one wheel in 1.5% of trucks, engine overtemperature, DEF low, low air, fuel theft of 5-12% at the depot); rates are those of Tables 7 and 8, scaled by --truck-faults. Specific to trucks: reefer unit fault (compressor off 15-35 min, 1.5 per 1000 operating hours), cargo probe stuck for 2-10 h (2% a day), door left open 15-40 min extra on a drop, reefer alarm and excursion, tank valve stuck reading open for 1-4 h (1% of deliveries), tilt event and tilt alarm (1 per 7000 s moving), rollover (1 per 2.5 million s moving), harsh brake (8% of traffic stops from over 43 km/h at 0.33 g) and DPF regeneration (normal operation, labelled).

With --truth FILE the simulator writes JSON lines, each with t, utc, vehicle and type. First a run line and one vehicle line per truck (scenario, IMEI, VIN, source, clock error, block hours, set point, compartments). Then the bus labels that apply (link_down, link_up, replayed, gps_*, reboot, frame_dropped, bus_off, tp_abort, j1939_na, spike, encoding_error, stuck_at, drift, tpms_leak, fuel_theft, refuel, engine_overtemp_*) and for trucks stop (end, drop, depot or delivery), fuel_stop, rest_start and rest_end, night_start and night_end, idle_stop, dpf_regen_start and dpf_regen_end, cargo_door_open and cargo_door_close, defrost_start and defrost_end, reefer_alarm_start and reefer_alarm_end, reefer_excursion, reefer_unit_fault_*, probe_stuck_*, door_left_open, delivery and delivery_done, valve_stuck_*, tilt_event, tilt_alarm and rollover_start and rollover_end. One truth file serves buses, trucks and ambulances of the same run. Not simulated: multi-code DM1 transport failures, wrong CAN bit rate, firmware or schema mismatch, clock jump, odometer reset, trailer swap, ISO 11992 trailer EBS frames, tachograph driver-card frames, and geofence and unscheduled-stop alerts (the platform derives those).

8.6 Measured differences between the scenarios

Slow runs of 13 Oct 2026 (nothing faster than clock 2; 24 simulated minutes each, starting 13:00 IST), with the default fault rates in set A (8 long-haul, 12 reefer, 12 tanker) and more trucks in set D for the rare states. Every datagram was identified by Sense's protocol handler as NMEA 0183, OBD-II, J1939 or Wialon IPS (none unknown or unreadable), except that with every fault at 1000 times the rate (set C) the datagrams carrying only the three address-claim frames (9 / 27 / 21 of 4,101 / 5,122 / 5,689) were not identified. In the released listener build TIRE1 and B1 arrived as "parameter group ... (not read by this build)"; the working copy reads both. Because the runs were 24 simulated minutes long, rare states appear only as far as they fell into that window.

Rare states seen in the slow runs: long-haul rest breaks 6, one night rest of 33,263 s with idling, idle stops 11, one fuel stop, DPF regeneration started 7 times; reefer cargo door opened 46 times, defrost started 7 times, alarms 19; tanker deliveries 14 (1.1-13.5 kL, flow 539-1,861 L/min), tilt events 138 of which 47 alarms (44 of those in set C, the fault-boosted run). The reefer excursion label, the end of DPF regeneration, the end of defrost and the end of a rollover were not seen in the slow runs, only in earlier faster smoke runs that checked the state machines and are not reported as numbers.

9. Uncertainty and assumptions