基于负载均衡和流量优先级的网络拓扑设计

交换式以太网网络的拓扑结构设计是一个带约束的优化问题，需要同时考虑多种约束条件。本文中定义了两个主要的准则：交换机负载均衡和流量最短路径。根据设计目标而衡量每条准则的权重．对拓扑进行评分而进行网络的拓扑结构设计。该方法以终端节点间网络流量需求矩阵和终端设备间流量优先级矩阵为输入，利用遗传算法从所有的拓扑结构中找出最优拓扑，决定交换机生成树拓扑和终端节点的分布位置。通过网络仿真，可以证明此方法的有效性。

Ethernet topology design based on load balance and traffic priority

Topology design of switched Ethernet Networks is a constrained optimization problem which need to consider multiple criteria with different units, simultaneously. This paper defines two major criterions： load balancing on switches and traffic shortest path selection. We can weigh the importance of each criterion based on our goal. We study a design method of network topology using average rank that is calculated by averaging all the ranks of each criterion. With terminal-node＇s traffic demand matrix and traffic priority matrix as input, this method can search global optimized topology by employing genetic algorithm. This method requires small amount of calculation, and the principle of topology evaluation is easy to understand. Through some network simulations, we show effectiveness of our approach.