Checkpoint Message Latency Detection in Vehicle Computing Paths
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional systems for detecting message faults and latency in high-performance computing systems, such as those in autonomous vehicles, are insufficient as they only monitor publish and receive points, failing to account for complex architectures and multiple endpoints, leading to delayed detection of latency and faults.
Innovation Solution
Implementing checkpoints along the communication path between publishers and subscribers to measure latency and identify faults by monitoring message IDs and timestamps at multiple points, allowing for early detection and triggering remedial actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional point-to-point latency measurement is used, then the measurement approach is simple, but the detection precision is insufficient and latency detection is delayed
Solution Approach 1:
The communication path is segmented into multiple checkpoints (publisher, intermediaries, subscriber) where each checkpoint records timestamps for incoming and outgoing messages. This segmentation enables precise latency measurement at each segment rather than only at the endpoints, resolving the contradiction by improving measurement precision through multi-point monitoring without requiring complete system redesign
Solution Approach 2:
Timestamp records are introduced as intermediary data elements that carry timing information through the communication path. These timestamps act as mediators between publishers and subscribers, enabling latency calculation without requiring direct complex coordination or additional communication protocols, thus improving precision while maintaining relative simplicity
2Reliability
If traditional schema-based fault detection is used, then the implementation is straightforward, but the reliability of fault detection is insufficient in complex systems
Solution Approach 1:
The system performs preliminary actions by recording timestamps at each checkpoint before messages are processed. This advance timing recording enables later reliable detection of faults and latency issues without requiring complex real-time analysis during message transmission, improving reliability while keeping the implementation straightforward
Solution Approach 2:
The system establishes feedback mechanisms where checkpoints monitor message timing and provide information back to the monitoring system. This feedback loop enables reliable fault detection by continuously comparing actual timing data against expected parameters, resolving the contradiction through systematic monitoring rather than complex schemas
3Loss of time
If simple latency measurement is used, then the system overhead is low, but the loss of time for fault detection is excessive
Solution Approach 1:
The monitoring system performs self-service by having checkpoints automatically record their own timestamps without requiring external intervention or complex coordination. This self-service approach minimizes the loss of time for fault detection while keeping resource consumption low, as each checkpoint independently performs simple timing operations rather than requiring system-wide synchronization or heavy processing
Data Source
AI summary
This document discloses system, method, and computer program product embodiments for detecting a fault or latency in a computing system, such as that of an autonomous vehicle. In the computing system, messages are received from subsystems of the vehicle that are publishers and transferred along a communication path to subsystems of the vehicle that are subscribers. The method includes receiving, from checkpoints along the communication path, for each of a group of messages that are transmitted along the communication path, a message ID and a timestamp indicating when the checkpoint receives the message, publishes the message, or publishes a result derived from the message. The method also includes using the message IDs and the timestamps to measure an amount of latency that occurred in the communication path before that message reached its corresponding subscriber. When the measured latency does not satisfy a threshold, the vehicle may take a remedial action.