Weighted Differential Scheduler for NoC Jitter Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current scheduling algorithms for Network-on-Chip (NoC) systems, such as deficit round robin, often result in high jitter due to uneven packet forwarding, which is undesirable for ensuring Quality of Service (QoS) requirements like low latency in applications like IP telephony.
Innovation Solution
A low-jitter packet scheduling algorithm that uses weight ratios and error thresholds to manage packet forwarding from multiple sources, ensuring fair allocation and reducing jitter by incrementing and decrementing error values based on calculated weight ratios and thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If deficit round robin scheduling is used, then implementation complexity is reduced, but jitter increases due to uneven packet forwarding intervals
Solution Approach 1:
The patent implements feedback mechanisms through error value accumulation and threshold comparison. The error value is incremented by weight ratios when packets are skipped and decremented when packets are forwarded, creating a feedback loop that automatically adjusts forwarding timing to maintain consistent intervals and reduce jitter while keeping the algorithm simple to implement
Solution Approach 2:
The patent changes the scheduling parameter from simple round-robin selection to weight-based error value adjustment. By introducing weight ratios and dynamic error thresholds, the system adapts packet forwarding timing based on source priorities while maintaining low computational complexity through simple arithmetic operations
2Manufacturing precision
If smoothed round robin is used, then jitter is reduced, but hardware implementation difficulty increases due to matrix processing requirements
Solution Approach 1:
The patent extracts the essential function of smoothed round robin (reducing jitter through weight-based scheduling) while removing the complex matrix processing requirement. It keeps only the weight ratio calculation and error value accumulation, eliminating the need for storing and processing large weight matrices in hardware
Solution Approach 2:
The patent replaces the complex, persistent weight matrix structure with simple, lightweight error value counters that are easily incremented and decremented. These temporary error values serve the same scheduling purpose without requiring complex storage and processing infrastructure
3Reliability
If timestamp-based scheduling is used, then QoS guarantees are improved, but algorithmic complexity increases from constant to linear
Solution Approach 1:
The patent segments the QoS guarantee mechanism into simple weight ratio calculations and error value thresholds for each packet source, rather than requiring global timestamp comparisons across all packets. This segmentation maintains QoS guarantees through localized, constant-time decisions at each scheduling point
Data Source
AI summary
A method for managing packets, including: identifying a first plurality of packets from a first packet source having a first weight; identifying a second plurality of packets from a second packet source having a second weight; obtaining a first weight ratio based on the first weight and the second weight; obtaining an error threshold and a first error value corresponding to the second packet source, where the error threshold exceeds the first error value; forwarding a first packet from the first packet source in response to the error threshold exceeding the first error value; incrementing the first error value by the first weight ratio; forwarding a first packet from the second packet source, after incrementing the first error value and in response to the first error value exceeding the error threshold; and decrementing the first error value.