Packet Scheduling Using Deficit Round Robin for Fair Bandwidth Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing packet scheduling algorithms face challenges in achieving fair bandwidth allocation, reducing congestion, and minimizing packet losses, particularly during busy periods, with many techniques resulting in large queuing delays and complex hardware implementations.
Innovation Solution
A packet scheduling method that updates credit or deficit values for each queue in a round-robin fashion, prioritizing packets with no deficit for transmission and allocating credits based on proportional weights to ensure fair bandwidth distribution and reduce congestion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If timestamp schedulers are used to emulate GPS operation, then fairness in bandwidth allocation is improved, but device complexity increases due to sorting bottleneck and complicated data structures
Solution Approach 1:
The patent uses simple circular buffer structures with fixed-size entries that are continuously overwritten, replacing complex dynamic data structures. Each buffer entry is a simple structure with packet pointer, length, and timestamp fields, designed for efficient hardware implementation without complex memory management.
Solution Approach 2:
The patent replaces the software-based sorting mechanism of timestamp schedulers with a hardware-friendly circular buffer approach that uses simple comparison logic. Instead of implementing full sorting algorithms, the system uses deficit counter comparisons and round-robin selection that can be directly implemented in hardware logic.
2Device complexity
If round-robin schedulers are used to achieve O(1) time complexity, then device complexity is reduced, but delay properties worsen due to transmitting all data of a flow at once
Solution Approach 1:
The patent segments the transmission of each flow's data across multiple rounds rather than transmitting all data in a single round. The deficit counter mechanism allows a flow to transmit only the portion of its data that fits within its allocated quantum, with the remainder deferred to subsequent rounds, thereby spreading transmission over time and reducing burstiness.
Solution Approach 2:
The patent introduces dynamic credit allocation based on flow weights and actual transmission needs. The deficit counter is dynamically adjusted each round based on the flow's weight and the amount of data actually transmitted, allowing the scheduler to adapt to varying traffic conditions while maintaining O(1) complexity through simple counter updates.
3Device complexity
If Deficit Round Robin is used to achieve O(1) time complexity, then device complexity is reduced, but delay and burstiness properties worsen especially for flows with large weights
Solution Approach 1:
The patent modifies the deficit round robin parameters by introducing weight-based credit allocation and a minimum transmission threshold. Flows with large weights receive proportionally larger credits, but the actual transmission is constrained by the minimum threshold and the flow's available data, preventing excessive burstiness while maintaining fairness for high-weight flows.
4Loss of time
If Smoothed Round Robin is used to improve delay properties, then packet delay is reduced, but device complexity increases due to Weight Spread Sequence and Weight Matrix data structures
Solution Approach 1:
The patent replaces the complex Weight Spread Sequence and Weight Matrix data structures with simple circular buffers and deficit counters. The scheduling decision is based on straightforward comparisons of deficit counters and weight-based credit allocations, eliminating the need for complex pre-computed sequences and matrices while achieving similar delay performance.
Data Source
AI summary
A methods, apparatus and computer memory are provided for packet scheduling. A processor polls queues in a round robin fashion and schedules for transmission onto a link a packet in each queue with no deficit before scheduling for transmission onto the link a packet in each queue with a deficit. A credit is allocated to each queue with the deficit based on a proportional weight, until each queue with the deficit has a credit.