Federated Vehicle Data Analytics Architecture

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current data infrastructure is inadequate to process and analyze the high-dimensional, high-frequency geolocation and sensor data from vehicles, such as autonomous and semi-autonomous vehicles, due to limitations in processing capacity and data variability, leading to reduced confidence in analysis results.

Innovation Solution

A federated machine-learning architecture is implemented, distributing computing resources hierarchically across vehicle, neighborhood, regional, and supra-regional levels, using lossy compression and machine-learning techniques to process and infer insights from vehicle sensor data, combining supervised, unsupervised, and reinforcement learning models to provide actionable data for stakeholders.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Power

If traditional data processing systems are used to handle vehicle sensor data, then system simplicity is maintained, but processing capacity and analysis accuracy deteriorate due to inability to handle high-dimensional, high-frequency data

Engineering Contradiction:
Improveprocessing capacityVSAvoidsystem complexity
Core Design Contradiction:
PowerVSDevice complexity

Solution Approach 1:

The patent divides the data processing system into multiple hierarchical levels: vehicle-level processing units, regional data centers, and cloud-based analytics platforms. Each level handles specific processing tasks, with vehicle units performing real-time processing and higher levels performing aggregate analysis, thereby distributing processing capacity while maintaining manageable system complexity through modular architecture

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a hierarchical spatial dimension to the processing architecture, organizing computing resources across multiple geographic and functional levels. This dimensional expansion allows the system to scale processing capacity by adding more hierarchical layers rather than increasing complexity within a single level, enabling parallel processing across distributed nodes

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Productivity

If more computing power is added to process vehicle data, then processing capacity improves, but infrastructure complexity and costs worsen

Engineering Contradiction:
Improvedata processing throughputVSAvoidinfrastructure complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent combines multiple processing functions into integrated hierarchical nodes that perform sensing, processing, storage, and communication tasks. By merging these functions into unified edge computing devices and regional data centers, the system achieves high processing throughput while reducing infrastructure complexity compared to having separate systems for each function

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent implements self-organizing capabilities where processing nodes automatically discover and coordinate with each other, perform load balancing, and adapt to changing data streams. This self-service behavior reduces the need for complex centralized management infrastructure, allowing high productivity to be achieved with simpler operational overhead

Inventive Principle:
Principle #25Self-service

3Adaptability or versatility

If data from multiple vehicles with different calibrations and standards is processed, then data coverage and insights improve, but analysis reliability deteriorates due to data variability

Engineering Contradiction:
Improvedata compatibilityVSAvoidanalysis confidence
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The patent dynamically adjusts processing parameters, data sampling rates, and aggregation methods based on the specific characteristics of each vehicle's sensor suite and calibration standards. By changing parameters adaptively rather than using fixed processing rules, the system maintains high data compatibility across diverse vehicles while preserving analysis reliability through context-aware processing

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent introduces standardized data abstraction layers and normalization intermediaries that translate diverse vehicle sensor data into a common analytical framework. These intermediary processing stages harmonize data from different calibration standards without losing essential vehicle-specific characteristics, thereby enabling both high adaptability and reliable cross-vehicle analysis

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3820753B1Vehicle-data analytics
Publication Date: 2023.08.02 MOOVE AI
  • EP3820753B1 patent drawingFigure 1
  • EP3820753B1 patent drawingFigure 2
  • EP3820753B1 patent drawingFigure 3

AI summary

A non-transitory, machine-readable medium storing instructions that, when executed by one or more processors, effectuate operations of a process, the process that includes processing and transferring sensor data including geolocations, control-system data and vehicle proximity data from vehicles across different computing levels for machine-learning operations to compute adjustments for vehicle control systems.