Data Processing Network Execution for Low-Latency Reproducibility

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In driver assistance and automated driving systems, achieving reproducibility and low latency is challenging due to fluctuating runtime behaviors in multi-core systems, which complicates safety measures like SW lockstep and recalculating driving situations accurately.

Innovation Solution

A method that combines data-driven and time-driven execution of data processing components, using a parent structure to aggregate input and output data and control data flow, allowing for predictable and efficient processing with reduced system states.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If a strictly data-driven approach is used to trigger execution of data processing components by data packet arrival, then low latency is achieved, but the number of possible states increases significantly

Engineering Contradiction:

ImprovelatencyVSAvoidnumber of possible states

Core Design Contradiction:

Loss of timeVSDevice complexity

Solution Approach 1:

The system segments the data processing network into multiple data processing modules, each with its own execution context. This segmentation allows the system to divide the large state space into smaller, manageable modules while maintaining low latency through event-driven execution triggers.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system dynamically switches between data-driven execution (for low latency paths) and time-driven execution (for state management). The execution mode adapts based on the specific data processing task, allowing the system to optimize between latency and state complexity in different contexts.

Inventive Principle:

Principle #15Dynamics

2Reliability

If time-driven execution with worst-case execution-time concepts is applied to data processing components, then predictability and reproducibility are improved, but latency increases

Engineering Contradiction:

ImprovepredictabilityVSAvoidlatency

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system implements periodic time-driven execution cycles that provide predictable timing boundaries. Within these cycles, data-driven events can trigger immediate processing when needed, while the periodic structure ensures overall predictability and reproducibility for safety-critical functions.

Inventive Principle:

Principle #19Periodic action

Solution Approach 2:

The system performs preliminary time allocation and resource reservation based on worst-case execution times before actual data processing occurs. This preliminary action ensures predictability is maintained while allowing flexible, low-latency execution when data is actually available.

Inventive Principle:

Principle #10Preliminary action

3Productivity

If data processing components are executed in parallel on multi-core systems to improve performance, then productivity increases, but runtime behavior becomes fluctuating and less predictable

Engineering Contradiction:

Improveprocessing performanceVSAvoidruntime predictability

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system introduces an intermediary execution management layer that coordinates parallel data processing components across multiple cores. This intermediary layer synchronizes execution contexts and manages data flow between parallel components, maintaining predictability while enabling high-performance parallel processing.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Reliability

If the same software is executed simultaneously on two processors for SW lockstep safety measures, then safety and reliability are improved, but hardware resources are doubled

Engineering Contradiction:

ImprovesafetyVSAvoidhardware resources

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The system creates virtual copies of execution contexts and data processing logic through software-based redundancy rather than requiring physical hardware duplication. The same software can be executed on shared hardware resources with proper isolation and synchronization, reducing hardware resource requirements while maintaining SW lockstep safety.

Inventive Principle:

Principle #26Copying

Data Source

PatentUS20230333891A1Method, data processing module, and data processing network for processing data

Publication Date: 2023.10.19 ROBERT BOSCH GMBH

AI summary

A method for processing data with a data processing network including a plurality of data processing modules which each include at least one data processing component, each data processing component being designed for a defined data processing task for processing the data, each data processing module receiving, as input data, data from a data source and/or output data from further data processing modules and generating output data which is network output data of the data processing network and/or input data of further data processing modules. The method for at least one data processing module including: a) receiving a set of input data for performing the data processing tasks; b) receiving a stimulus for activating the at least one data processing component; c) performing the data processing task with the corresponding input data to generate output data; and d) providing the output data.