AWS IoT FleetWise
A fully managed service that efficiently collects, transforms, and transfers driving data and vehicle signals from connected cars to the cloud for analysis
Overview
AWS IoT FleetWise is a fully managed service that enables automakers and fleet operators to efficiently collect vehicle signal data from connected cars and transfer it to the cloud. It converts raw data from in-vehicle network protocols such as CAN bus and OBD-II into standardized formats for cloud transmission. Data modeling based on Vehicle Signal Specification (VSS) enables unified data structure analysis regardless of vehicle make or model. Intelligent edge-side data filtering transfers only necessary data, dramatically reducing communication costs and enabling data collection from fleets of millions of vehicles.
Vehicle Data Modeling and Signal Catalog
FleetWise's core capability is vehicle signal data modeling. The Signal Catalog defines all signals available from vehicles (engine RPM, vehicle speed, battery voltage, tire pressure, etc.), and Vehicle Models map these to per-model configurations. Signals are managed hierarchically in a tree structure conforming to the VSS (Vehicle Signal Specification) standard, uniquely identified by paths like Vehicle.Powertrain.Engine.Speed. Defining data type, unit, minimum, and maximum values per signal enables automatic anomaly detection and data quality assurance. Signals common across multiple models (vehicle speed, GPS coordinates, etc.) need only be defined once in the Signal Catalog and reused, requiring only model-specific signal additions when onboarding new vehicle types. This standardized data model facilitates cross-model analysis and benchmark comparisons.
Intelligent Edge Data Collection
The FleetWise Edge Agent runs on in-vehicle computers (ECUs or telematics units) and reads signals from the CAN bus. The Campaign feature defines data collection conditions, enabling event-driven collection of only necessary data - such as "10 seconds of data before and after hard braking" or "a snapshot when battery voltage drops below threshold." Combining time-based periodic collection with condition-based event collection optimizes communication bandwidth and storage costs. Collected data is buffered in vehicle storage and transferred to the cloud when cellular connectivity is available. Connected car books (Amazon) cover the fundamentals of in-vehicle networks. The Edge Agent supports both ARM and x86 architectures and also runs on the AUTOSAR Adaptive Platform.
Large-Scale Fleet Operations and Data Utilization Patterns
For fleets of millions of vehicles, Campaign deployment strategy is key to operations. Rather than deploying to all vehicles simultaneously, grouping by region or model year and rolling out in stages avoids network load concentration. Collected data can be stored in either S3 or Timestream - S3 in Parquet format enables low-cost SQL analysis with Athena, while Timestream provides time-series-optimized storage ideal for real-time dashboard construction. Representative data utilization patterns include predictive maintenance (detecting component degradation trends for pre-failure replacement), driving behavior analysis (calculating safety scores from hard acceleration/deceleration frequency), and OTA update effectiveness verification (comparing fuel efficiency before and after software updates). A pipeline connecting FleetWise data to SageMaker for building anomaly detection models is common in practice.