Botez Robert, Pasca Andres-Gabriel, Sferle Alin-Tudor, Ivanciu Iustin-Alexandru, Dobrota Virgil
Communications Department, Technical University of Cluj-Napoca, 400114 Cluj-Napoca, Romania.
Sensors (Basel). 2023 Jun 30;23(13):6053. doi: 10.3390/s23136053.
This paper presents a novel approach for network slicing in 5G backhaul networks, targeting services with low or very low latency requirements. We propose a modified A* algorithm that incorporates network quality of service parameters into a composite metric. The algorithm's efficiency outperforms that of Dijkstra's algorithm using a precalculated heuristic function and a real-time monitoring strategy for congestion management. We integrate the algorithm into an SDN module called a path computation element, which computes the optimal path for the network slices. Experimental results show that the proposed algorithm significantly reduces processing time compared to Dijkstra's algorithm, particularly in complex topologies, with an order of magnitude improvement. The algorithm successfully adjusts paths in real-time to meet low latency requirements, preventing packet delay from exceeding the established threshold. The end-to-end measurements using the Speedtest client validate the algorithm's performance in differentiating traffic with and without delay requirements. These results demonstrate the efficacy of our approach in achieving ultra-reliable low-latency communication (URLLC) in 5G backhaul networks.
本文提出了一种用于5G回程网络中网络切片的新颖方法，目标是满足低或极低延迟要求的服务。我们提出了一种改进的A*算法，该算法将网络服务质量参数纳入一个复合指标中。使用预先计算的启发式函数和用于拥塞管理的实时监测策略，该算法的效率优于迪杰斯特拉算法。我们将该算法集成到一个名为路径计算元件的软件定义网络模块中，该模块为网络切片计算最优路径。实验结果表明，与迪杰斯特拉算法相比，所提出的算法显著减少了处理时间，特别是在复杂拓扑结构中，有数量级的提升。该算法成功地实时调整路径以满足低延迟要求，防止数据包延迟超过既定阈值。使用Speedtest客户端进行的端到端测量验证了该算法在区分有无延迟要求的流量方面的性能。这些结果证明了我们的方法在5G回程网络中实现超可靠低延迟通信（URLLC）的有效性。
