Towards an Optimized Multi-Cyclic Queuing and Forwarding in Time Sensitive Networking with Time Injection

TL;DR

This paper proposes an optimized Multi-Cyclic Queuing and Forwarding approach with Time Injection in TSN, using genetic algorithms to improve flow schedulability and convergence speed over traditional methods.

Contribution

It introduces a set of constraints and a hybrid GA-Simulated Annealing algorithm for efficient Multi-CQF configuration, incorporating Time Injection to enhance TSN flow scheduling.

Findings

Significantly increases scheduled TT flows by 15%

GASA converges 20% faster than baseline SA

Outperforms existing methods in efficiency and speed

Abstract

Cyclic Queuing and Forwarding (CQF) is a Time-Sensitive Networking (TSN) shaping mechanism that provides bounded latency and deterministic Quality of Service (QoS). However, CQF's use of a single cycle restricts its ability to support TSN traffic with diverse timing requirements. Multi-Cyclic Queuing and Forwarding (Multi-CQF) is a new and emerging TSN shaping mechanism that uses multiple cycles on the same egress port, allowing it to accommodate TSN flows with varied timing requirements more effectively than CQF. Despite its potential, current Multi-CQF configuration studies are limited, leading to a lack of comprehensive research, poor understanding of the mechanism, and limited adoption of Multi-CQF in practical applications. Previous work has shown the impact of Time Injection (TI), defined as the start time of Time-Triggered (TT) flows at the source node, on CQF queue resource utilization. However, the impact of TI has not yet been explored in the context of Multi-CQF. This paper introduces a set of constraints and leverages Domain Specific Knowledge (DSK) to reduce the search space for Multi-CQF configuration. Building on this foundation, we develop an open-source Genetic Algorithm (GA) and a hybrid GA-Simulated Annealing (GASA) approach to efficiently configure Multi-CQF networks and introduce TI in Multi-CQF to enhance schedulability. Experimental results show that our proposed algorithms significantly increase the number of scheduled TT flows compared to the baseline Simulated Annealing (SA) model, improving scheduling by an average of 15%. Additionally, GASA achieves a 20% faster convergence rate and lower time complexity, outperforming the SA model in speed, and efficiency.