摘要：

由于卫星网络星座规模大、拓扑时变、链路不稳定等特点, 软件定义网络(SDN)应用在卫星网络时需要部署多个控制器来协同工作, 这就会引起一些问题, 例如控制器部署位置和数量以及平衡控制器数量、控制区域与网络性能之间的关系等。针对上述问题, 提出一种基于改进麻雀搜索算法(SSA)的软件定义卫星网络(SDSN)多控制器动态部署方案。该方案充分考虑卫星网络拓扑的周期性变化, 以降低网络时延、平衡控制器负载、提升网络可靠性作为优化目标建立控制器动态部署模型, 利用k-means聚类算法对SSA进行初始化改进, 并动态调整SSA的步长参数, 通过改进的SSA优化迭代出最优的控制器部署方案。仿真结果表明, 与对比实验中的基于粒子群优化(PSO)算法的SDSN多控制器部署方案CDS-LR相比, 该部署方案不仅使网络时延降低了约10%, 而且还使控制器负载均衡度和网络可靠性分别提升了约25%和40%, 充分保证了动态卫星网络的整体性能。

关键词: 控制器部署, 软件定义网络, 卫星网络, 负载均衡, 麻雀搜索算法

Abstract:

Given the characteristics of satellite network constellations, which include a large scale, time-varying topology, and unstable link, Software-Defined Networks (SDNs) must deploy multi-controllers to function effectively in satellite networks. This multi-controller approach raises several issues, such as determining the optimal locations for controller deployment and the required number of controllers as well as how to balance the number of controllers with control zones and overall network performance. To address these challenges, this study proposes a dynamic deployment scheme for multi-controllers in a Software-Defined Satellite Network (SDSN) based on an improved Sparrow Search Algorithm (SSA). This scheme considers the periodic changes in satellite network topology and establishes a dynamic deployment model focused on optimizing three key objectives: reducing network delay, balancing controller load, and improving network reliability. The k-means clustering algorithm is used to initialize the SSA, and the SSA's step parameters are dynamically adjusted. Through the improved SSA, the optimal controller deployment scheme is iterated. Experimental results show that the proposed deployment scheme significantly outperforms the existing SDSN multi-controller deployment strategy based on the Particle Swarm Optimization (PSO) algorithm, known as CDS-LR. Specifically, the proposed deployment scheme not only reduces network delay by approximately 10% but also improves controller load balancing and network reliability by nearly 25% and 40%, respectively, thereby ensuring superior overall performance for satellite networks.

Key words: controller deployment, Software-Defined Network (SDN), satellite network, load balancing, Sparrow Search Algorithm (SSA)