AFDX SubVL Scheduling with Priority Weights
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In AFDX networks, the round-robin scheduling scheme lacks flexibility in trading off latency and jitter, as it does not distinguish between subVLs based on priority, leading to inefficient resource utilization and increased latency.
Innovation Solution
Implementing prioritization-based scheduling policies such as Static Priority (SPn), Earliest Deadline First (EDF), and Least Laxity First (LLF) to prioritize data frames based on subVL prioritization, allowing for reduced latency and improved bandwidth usage by scheduling higher-priority frames first.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If round-robin scheduling is used for subVLs, then simplicity of scheduling is maintained, but latency and jitter performance deteriorate due to lack of priority differentiation
Solution Approach 1:
The patent changes the scheduling parameter from equal-weight round-robin to priority-based scheduling where each subVL is assigned a priority weight. The scheduler selects the next subVL to transmit based on these weights rather than strict rotation, allowing high-priority traffic to be transmitted more frequently while maintaining a deterministic scheduling framework.
Solution Approach 2:
The patent introduces dynamic elements to the scheduling process by allowing the scheduler to adaptively select subVLs based on current queue states and priority weights rather than following a fixed static pattern. This enables the system to respond to varying traffic conditions while maintaining bounded latency guarantees.
2Stability of the object's composition
If round-robin scheduling is used for subVLs, then fairness among subVLs is maintained, but bandwidth utilization efficiency deteriorates due to inability to prioritize urgent traffic
Solution Approach 1:
The patent transforms the scheduling approach by introducing priority weights as a new parameter that modulates the transmission probability of each subVL. High-priority subVLs with urgent or time-sensitive traffic receive higher weights, increasing their transmission opportunities and improving overall bandwidth utilization for critical applications.
Solution Approach 2:
The patent applies different scheduling qualities to different subVLs based on their specific requirements. Each subVL can be configured with appropriate priority weights matching its traffic characteristics, allowing time-sensitive subVLs to receive preferential treatment while less critical subVLs maintain standard scheduling, thus optimizing overall network efficiency.
3Device complexity
If equal priority is assigned to all subVLs, then simplicity of queue management is maintained, but latency differentiation capability deteriorates
Solution Approach 1:
The patent introduces a priority weight parameter for each subVL that enables latency differentiation without fundamentally changing the queue management structure. The scheduler uses these weights to determine transmission order, providing fine-grained control over latency while maintaining relatively simple queue data structures and management procedures.
Data Source
AI summary
A device for avionics full-duplex switched Ethernet (AFDX) communication can include a transmit port for transmitting AFDX data and a processor. The processor can be configured to obtain AFDX data frames for transmission over a plurality of sub-virtual links (subVLs) of a virtual link (VL), and maintain, for each of the plurality of subVLs, a corresponding data queue by storing AFDX data frames associated with that subVL in the corresponding data queue. The processor can transmit the AFDX data frames from the plurality of data queues on the plurality subVLs via the transmit port according to a scheduling policy that is based on subVL prioritization. The scheduling policy that is based on subVL prioritization can include static priority (SPn) scheduling, earliest deadline first (EDF) scheduling, or least laxity first (LLF) scheduling.