Distributed Leaky Bucket Scheduler for Multi-Modem Vehicle Data Transport
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current communication systems for vehicles, such as driverless cars, face challenges in meeting bandwidth, latency, reliability, and cost requirements for reliable and real-time data transport, particularly in scenarios like remote teleoperation and car-to-car mapping, where data needs to be transmitted efficiently over multiple paths to ensure safe and effective vehicle operation.
Innovation Solution
A distributed leaky bucket based multipath communication scheduler is implemented for in-vehicle computing devices, which determines delivery delays for multiple modems based on leaky bucket levels, burst sizes, and end-to-end delay estimates, and assigns packets across multiple modems to ensure efficient data transmission, prioritizing paths based on availability and quality of service.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If single-path communication is used, then device complexity is reduced, but reliability and real-time performance deteriorate
Solution Approach 1:
The communication system is segmented into multiple independent transmission paths (modems), each capable of carrying data packets. The scheduler divides the data stream and assigns packets to different modems based on their current status and delivery delay estimates, thereby achieving reliable communication through path diversity while keeping each individual path relatively simple
Solution Approach 2:
The system transitions from single-dimensional (single-path) communication to multi-dimensional (multi-path) communication by introducing the dimension of parallel transmission paths. This allows data to be transmitted simultaneously through multiple modems, improving reliability and real-time performance while distributing the complexity across multiple independent channels
2Speed
If multiple modems are used for multipath transmission, then real-time performance and reliability improve, but device complexity increases
Solution Approach 1:
Each modem maintains its own leaky bucket buffer and provides self-reported status information (queue depth, delivery rate, delay estimates) to the scheduler. The scheduler uses these self-service metrics to make autonomous routing decisions without requiring complex centralized control or manual configuration, thereby managing multi-modem complexity through decentralized information gathering
Solution Approach 2:
The scheduler dynamically changes routing parameters (packet assignment to modems) based on real-time status parameters (delivery delay, queue depth, transmission rate) reported by each modem. This parameter-driven adaptive routing enables the system to optimize transmission speed across multiple modems while keeping the scheduling logic relatively simple and responsive to changing conditions
3Loss of time
If packets are assigned based on delivery delay estimates, then latency is reduced, but measurement precision requirements increase
Solution Approach 1:
The system implements feedback mechanisms where each modem continuously reports its status (queue depth, delivery rate, delay estimates) back to the scheduler. The scheduler uses this feedback to update delivery delay estimates and adjust packet routing decisions in real-time, thereby reducing overall delivery latency while managing measurement precision requirements through continuous adaptation rather than requiring ultra-precise initial measurements
Data Source
AI summary
Systems, methods, and devices of the various embodiments provide a multipath communication scheduler for an in-vehicle computing device, such as a vehicle's autonomous driving system, vehicle's telematics unit, vehicle's control system, etc. In various embodiments, a distributed leaky bucket based scheduler for an in-vehicle computing device may assign packets for transport to a plurality of modems based at least in part on the determined delivery delays. In various embodiments, delivery delays may be determined based on leaky bucket levels, burst sizes, delivery rates, and end to end delay estimates for each of the plurality of modems. In various embodiments, the scheduler may be one of a plurality of schedulers each associated with a separate stream of packets assigned to the plurality of modems and the leaky bucket levels may be determined on a per stream basis.