Real-Time Telemetry Streaming & Predictive Maintenance Across 1,400 Connected Freight Assets
Processing high-throughput IoT sensor streams to forecast mechanical degradation and eliminate roadside breakdowns before they occur.
Critical road breakdown frequency plummeted across the active 1,400 vehicle fleet
Predictive lead time for coolant, brake, and transmission anomalies
Eliminated emergency roadside dispatches and contract SLA violation penalties
The Bottleneck
The Operational Challenge
Fleet managers relied on static odometer intervals for service, missing rapid brake and cooling system degradations caused by severe heat and overloading.
High-velocity data problem: 1,400 vehicles equipped with CAN-bus OBD-II telemetry generated 400+ events per second that overwhelmed existing database servers.
Disconnected field operations: when an anomaly was detected, dispatchers still had to phone depot mechanics manually, leading to missed repair windows.
The Engineering Fix
The Sodiac Architecture
Architected Sodiac Sirius as a distributed streaming ingestion pipeline capable of absorbing 15,000 events/sec via Kafka and TimescaleDB.
Trained physics-informed LSTM autoencoders on historical telemetry data to establish vehicle-specific baseline vibration, thermal, and oil pressure profiles.
Built an automated work-order dispatch pipeline: when a vehicle exhibits a degradation score >0.82, a maintenance ticket is staged in the TMS.
Delivered a lightweight mobile app for depot mechanics providing guided diagnostic steps and replacement component part numbers.
System Architecture
4-Step Technical Execution Pipeline
How data, deterministic business logic, and responsible AI guardrails execute in production.
CAN-bus Edge Telemetry Streaming
Onboard vehicle telematics units broadcast speed, oil pressure, coolant temperature, and vibration vectors via MQTT over 4G/5G.
Distributed Kafka Stream & Time-Series DB
Normalizes and filters sensor noise in real time, persisting high-frequency metric windows into partitioned TimescaleDB clusters.
Physics-Informed Anomaly Detection
Compares real-time telemetry against historical wear models to compute health index scores and forecast mean-time-to-failure.
Automated Workshop Dispatch & Spare Parts
Routes vehicles needing preventive service to the nearest depot, pre-allocating inventory spare parts before the vehicle arrives.
“Sodiac transformed our fleet from a reactive firefighting operation into a predictive, precision machine. Detecting a transmission or coolant failure 72 hours before a truck breaks down on the highway has completely changed our business economics.”
Quantified Impact
Verified Production Outcomes
Audited 90-day operational impact after migrating to the Sodiac platform.
Fleet roadside breakdowns decreased by 38% within the first 9 months of full operational deployment.
Average vehicle operational lifespan extended by an estimated 14% due to timely preventive fluid and bearing servicing.
Demurrage and contractual late-delivery penalties fell by 64% across inter-state container transit routes.
Depot turnaround time dropped from 18 hours to 4.5 hours with automated pre-allocated replacement parts.
Technology stack & proprietary products
Products Deployed
Underlying Technologies
Ready to achieve similar results for your team?
Schedule a personalized technical consultation with one of our AI engineering leads.