Modified Leaky-Bucket Scheduler for Packet Delay and Jitter
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Packet switched networks face challenges in handling real-time traffic classes that require low or no buffering, as traditional leaky-bucket algorithms can lead to increased network delay and jitter, particularly in GPRS traffic classes like VoIP, where buffering exacerbates delays and jitter.
Innovation Solution
Implementing a modified leaky-bucket method that distinguishes between buffered packets and real-time packets by setting a larger bucket size for real-time packets, allowing them to transmit immediately while buffering and rescheduling buffered packets based on the bucket leak rate to prevent overflow, thereby creating a hysteresis effect that prioritizes real-time packets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional leaky-bucket algorithm is used for packet switching, then bandwidth optimization and network robustness are improved, but network delay and jitter increase for real-time traffic classes
Solution Approach 1:
The patent segments the single bucket into two separate buckets: a first bucket for buffered packets and a second bucket for real-time packets. This segmentation allows different packet types to be handled independently, preventing real-time packets from experiencing the delays caused by buffering in the traditional single-bucket approach.
Solution Approach 2:
The patent applies different quality characteristics to different buckets: the first bucket uses standard leaky-bucket buffering for best-effort traffic, while the second bucket operates with minimal or no buffering for real-time traffic. This local differentiation of handling qualities resolves the contradiction by optimizing each traffic type according to its specific requirements.
2Reliability
If buffering is applied to all traffic classes, then congestion handling is improved, but packet delay and jitter increase for real-time traffic
Solution Approach 1:
The patent divides the traffic into two separate queues/buckets, allowing selective buffering application. The first bucket applies buffering for congestion handling, while the second bucket minimizes buffering to maintain low packet delay for real-time traffic, thus resolving the contradiction between congestion handling and packet delay.
Solution Approach 2:
Different buffering strategies are applied locally to different traffic types: aggressive buffering for best-effort traffic to handle congestion, and minimal or no buffering for real-time traffic to maintain low latency. This localized quality differentiation eliminates the need to compromise real-time performance for overall congestion handling.
3Quantity of substance
If bucket size is increased to accommodate more packets, then buffering capacity is improved, but network delay increases for real-time packets
Solution Approach 1:
The patent segments the buffering capacity into two separate buckets with different size characteristics. The first bucket can have larger capacity for best-effort traffic, while the second bucket maintains smaller capacity or no buffering for real-time traffic, preventing the universal delay effect of large buffers on real-time packets.
Solution Approach 2:
Different buffering capacity characteristics are applied locally: larger buffering capacity for best-effort traffic to maximize throughput, and minimal or zero buffering capacity for real-time traffic to minimize delay. This local differentiation resolves the contradiction between buffering capacity and network delay.
Data Source
AI summary
The invention provides a leaky-bucket method for delivering packets from a first node (A) to a second node (B) in a packet switched network, comprising the steps of: receiving at the first node (A) of at least one transmission comprising buffered packets, and at least one transmission comprising real-time packets; transmitting the received packets to the second node (B) and increase a bucket counter for the bucket, which bucket has a leak rate that is equal to or larger than the byte transmitting ratio of the real-time packets. The method is particularly characterize in that: the bucket size for the buffered packets is set to a first size; the bucket size for real-time packets is set to a second size that is larger than the first bucket size; real-time packets are transmitted to the second node (B) upon arrival at the first node (A), provided that a bucket counter indicates that the second bucket size is not overflowed; buffered packets are transmitted to the second node (B) when the bucket counter indicates that it has leaked enough to receive a buffered packet without overflowing said first bucket size. The invention is also providing an improved node arranged to perform the improved method for a leaky-bucket.