This whitepaper is for historical reference only. Some content might be outdated and some links might not be available.
Intelligent data collection with AWS IoT FleetWise
For most automotive companies, the primary motivation for a connected vehicle platform is to provide a mechanism to retrieve data off the vehicle and monetize the data for downstream services. Therefore, a common pattern exists where this data from the vehicle needs to be collected, normalized and aggregated to the cloud. With AWS IoT FleetWise, the undifferentiated heavy lift of building a data management platform of your connected vehicle is removed and another building block to the overall connected vehicle platform on AWS is delivered.
Most vehicle manufacturers have, in some form or another, collected telemetry data from vehicles over the past several years to help diagnose potential issues, identify preventative maintenance assistance and potential recalls. These automotive companies are beginning to shift towards building vehicles with more advanced sensors that generate orders of magnitude of larger data volume —with LIDAR and camera data, this can mean up to 2 terabytes of data every hour.
With these vast amounts of data now being generated by the vehicle, automotive companies need access to this data in the cloud to derive insights that can help improve vehicle quality, safety, and autonomy. As mentioned, transferring this data off the vehicle to the cloud can be complicated and expensive. Additionally, with the multitude of added rich data sensors in vehicles that generate data in different proprietary formats create a complex array of data across vehicles. Collecting this data in an efficient and cost-effective manner requires a custom-built in-vehicle data-collection system, which, for automotive companies, can be a difficult task. As a result, these automotive companies over index on building data management capabilities, rather than focusing on their own differentiators to allow their data scientists deliver insights and create new experiences for users in a highly performant manner.
With AWS IoT FleetWise the undifferentiated heavy lift of building this data collection platform is removed. These challenges of collecting vehicle data are now performed by a fully managed service that customers can use to collect, transform, and transfer vehicle data to the cloud in near real time. With AWS IoT FleetWise, automotive companies can now collect and organize data from vehicles with differing protocols and proprietary data formats. AWS IoT FleetWise helps to transform CAN and OBD telemetry binary frames into human-readable data and then standardizes that data into a vehicle model in the cloud for data analyses. Vehicle manufacturers can then define different data collection campaigns to remotely determine which vehicle data to collect and how frequently to transfer that data to the cloud.
Figure: AWS IoT FleetWise high-level architecture
Figure : AWS IoT FleetWise - high-level architecture
Data modeling
AWS IoT FleetWise provides a vehicle model orchestrator that automotive companies can use to build digital twins of their vehicles in the cloud. Vehicle signals, signal catalogs, vehicle models, and decoder manifests are the core components that help deliver data from the vehicle to the cloud efficiently and effectively.
Signals
Signals are fundamental structures that customer utilize to define to contain vehicle data and its metadata. A signal can be an attribute, a branch, a sensor, or an actuator. For example, an automotive OEM can create a signal to receive in-vehicle temperature values, and to store its metadata, including a sensor name, a data type, and a unit.
Signal catalog
A signal catalog contains a collection of signals. Signals in a signal catalog can be used to model vehicles that use different protocols and data formats. For example, there are two cars made by different automakers: one uses the Control Area Network (CAN bus) protocol; the other one uses the On-board Diagnostics (OBD) protocol. You can define a sensor in the signal catalog to receive in-vehicle temperature values. This sensor can be used to represent the thermocouples in both cars.
Vehicle model
Vehicle models are declarative structures that you can use to standardize the format of your vehicles and to define relationships between signals in the vehicles. Vehicle models enforce consistent information across multiple vehicles of the same type. You add signals to create vehicle models.
Decoder manifest
Decoder manifests contain decoding information for each signal in vehicle models. Sensors and actuators in vehicles transmit low-level messages (binary data). With decoder manifests, AWS IoT FleetWise is able to transform binary data into human-readable values. Every decoder manifest is associated with a vehicle model.
Data collection
Once the vehicle has been modeled, and the signal catalog has been created, the customers are now able to create data collection campaigns using signals created within the model.
A campaign is an orchestration of data collection rules. Campaigns give the Edge Agent for AWS IoT FleetWise software instructions on how to select, collect, and transfer data to the cloud.
All campaigns are created in the cloud. After the campaigns have been marked as approved by team members, then AWS IoT FleetWise automatically deploys them to vehicles. Automotive teams can choose to deploy a campaign to a specific vehicle or a fleet of vehicles. The Edge Agent software will not start collecting data of the vehicle network until a running campaign is deployed to the vehicle.
Data protection considerations
Next generation vehicle communication requires robust encryption mechanisms. There are different internal and external requirements based on threat models that inform your data protection decisions. For encryption in transit, AWS IoT Core supports TLS 1.2 and 1.3. AWS IoT Core also provides security policy options that include several different ciphers. AWS IoT Core allows you to select TLS security policies. You can choose a predefined policy that supports the TLS protocols and ciphers that meet your requirements.
Customers might want to encrypt sensitive data client-side before sending it to the
cloud, or before sending data to the vehicle. AWS Key Management Service
OEMs collect a significant amount of data from the vehicle. This can include consumer data such as driving behavior, insurance carriers, PII (for example, name and email address), VIN or ECU IDs, and navigation services. AWS IoT Core and AWS IoT FleetWise provide the ability to store data centrally in Amazon S3, as described later in this whitepaper.
One challenge OEMs face is identifying sensitive data coming from the vehicle to
determine the types of sensitive data stored in the backend. You can use Amazon Macie
Data analytics
Once the campaigns have been executed in the vehicle, the destination of your data is determined by the campaign setup. For near-real time analytics and visualization dashboards of your data, Amazon Timestream would be the selected destination for telemetry data. When looking to create a performant data lake, centralized data storage and data processing pipelines, AWS IoT FleetWise offers storage in Amazon S3 with Apache Parquet or JSON data formats.
With flexible data storage options using AWS IoT FleetWise, automotive companies can customize their usage of AWS in their connected vehicle platform to collect the data as they see fit; some data needs to be pulled real time for vehicle tracking use cases, other data can be batched and stored for further processing to fulfill predicative maintenance use cases where data can be loaded into machine learning (ML) models to help predict issues within the fleet before they happen.
