Data Packet Scheduling with Reference-Time Queues for Deterministic Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing scheduling methods fail to meet the deterministic latency requirements for data packet transmission, leading to unpredictable end-to-end delays and jitter due to burst accumulation in IP networks.
Innovation Solution
A data packet scheduling method that determines a reference moment based on time information to select a target queue, ensuring the end-to-end delay is controlled by setting an upper limit on the theoretical time across network devices, using a scheduling rule to manage multiple queues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing scheduling methods (weighted fair queue, EDF) are used, then scheduling simplicity is maintained, but deterministic latency requirement cannot be met
Solution Approach 1:
The patent segments the network device's queue system into multiple queues, each associated with different time intervals. By dividing the scheduling process into discrete time segments and assigning packets to specific queues based on their deadline requirements, the system achieves deterministic latency control without requiring complex real-time scheduling algorithms. Each queue can be independently managed with simpler scheduling rules.
Solution Approach 2:
The patent applies preliminary action by calculating and setting the theoretical time upper limit for each queue before scheduling occurs. The scheduling mechanism pre-establishes time bounds for packet processing in each queue, allowing deterministic latency to be guaranteed in advance rather than requiring complex runtime adjustments. This pre-calculation of time parameters simplifies the actual scheduling execution.
2Loss of time
If theoretical time upper limit is set for each queue, then end-to-end delay is controllable, but scheduling rule complexity increases
Solution Approach 1:
The patent implements periodic action by organizing queues into time intervals and applying periodic scheduling cycles. Each queue corresponds to a specific time period, and the scheduling mechanism cycles through queues in a regular pattern. This periodic structure makes the time upper limits predictable and controllable while simplifying queue management through rhythmic, repeating operations rather than complex adaptive control.
Solution Approach 2:
The patent uses parameter changes by dynamically adjusting the theoretical time upper limit parameters for different queues based on traffic patterns and service requirements. By modifying these time parameters rather than changing the fundamental scheduling structure, the system can control end-to-end delay while maintaining relatively simple queue management logic. The parameters are adjusted to match actual network conditions.
3Reliability
If multiple queues are used with different time intervals, then deterministic delay is achieved, but device resource consumption increases
Solution Approach 1:
The patent applies universality by designing the queue system so that multiple queues share common management structures and resources. Rather than implementing completely independent queue systems, the patent uses a unified queue management framework that serves multiple time intervals and packet types. This multi-functional approach allows deterministic delay control across multiple queues while avoiding the resource overhead of separate management systems for each queue.
Data Source
AI summary
This application provides data packet scheduling methods and apparatuses. One method includes: A first network device receives, at a first moment, a data packet from a second network device in a network, the first network device determines a first reference moment based on the first moment and time information carried in the data packet, the first network device determines, based on the first reference moment, a target queue from a plurality of queues included in a first queue system and adds the data packet to the target queue, and the first network device processes the target queue according to a scheduling rule of the plurality of queues.