It is not efficient for the approach of constraining outdegree to enhance delay and loss rate synthetically for an application-level multicast tree. This paper proposes an application-level multicast tree which achieves minimum cost in terms of delay and loss rate. A cost function which consists of delay and loss rate is given to compute transmission cost of the multicast tree, and the problem model about how to construct the minimum multiplex-cost multicast tree is also given.Unlike existing problem models, this model abandons the approach of constraining outdegree.Furthermore, a variation of the maximal delay path greed algorithm is proposed to address the problem model. This variation algorithm optimizes total cost of a multicast tree. Finally, our extensive experimental results help to give the concrete suggestion about the parameters which the cost function involves, and demonstrate that the proposed application-level multicast tree construction algorithm has higher efficiency in total cost of the multicast tree than another related algorithm has.%针对度约束方式难以减少应用层组播树的延迟和丢包率的问题,提出了一种延迟和丢包率综合代价最小的应用层组播树构树算法.为避免度约束的局限性,给出一个包含延迟和丢包率的复合代价函数,以此来计算传输代价,进而构建了一种最小复合代价组播树的问题模型.为了求解该问题模型,提出了一种基于最大延迟路径贪婪算法的变异算法,同时在构树时对总传输代价进行优化.通过实验,给出复合代价函数的具体参数建议.对比相关算法,文中的构树算法在总传输代价方面有更好的性能.
展开▼
