High performance priority queue scheduling mechanism application driven by AI in SDN data plane design

Abstract

With the emergence of new network services, they have put forward extremely high quality of service requirements for transmission links. A high-performance priority queue scheduling mechanism driven by artificial intelligence is proposed to address the low efficiency and easy blocking of traditional priority queue scheduling mechanisms in software defined network data planes. This mechanism avoids queue head blocking by improving the deficit round robin algorithm, ensures fair allocation of output bandwidth for differentiated fine-grained priority queues, and guarantees maximum forwarding delay. The results show that this method significantly improves performance compared to traditional algorithms, as it can process and forward data packets faster, thereby reducing latency and increasing throughput. Specifically, the average round-trip time of the deficit round robin algorithm is 8.21 ms, compared to 3.89 ms for this method, which reduces 4.72 ms. The highest average throughput of the deficit round robin algorithm is 60.01 Mbps, compared to 105.99 Mbps for this method, which improves by 76.62%. After improving the algorithm, the software defined a network data plane system with an average frame delay of 2.81 ms and a packet loss rate of 4.50%. The amount of data transmitted by the improved system was basically the same as when it was not attacked, with an availability rate of 98.6%. The improved algorithm can better ensure the smoothness of transmission services. This provides a new direction for future research on new network architecture technologies and has certain economic value.