近年来，伴随互联网技术的快速发展，越来越多新型通信需求随之出现，尤其是日益兴起的视频会议、在线教育、IP电话等多媒体实时业务。此类的应用往往会对网络的通信能力提出更高的要求，同时要求计算机在支持多媒体业务时，使用更好的组播通信方式。多媒体实时业务对延时、带宽、费用、丢包率等QoS参数有不同的需求，多约束QoS组播路由算法已经成为互联网技术研究领域的热点问题之一。 本文在研究多约束QoS组播路由算法现状的基础上，构建了QoS组播路由问题的数学模型，并提出了一种将遗传算法和蚁群算法有效的结合起来的新型算法—遗传蚁群混合优化算法(CGAACA，the Combination of Genetic Algorithm and Ant Colony Algorithm)。算法前期利用遗传算法生成若干组优化解；算法中期，为了确保遗传算法和蚁群算法在适当时机能够融合，本文在这里设置了一个遗传算法进化程度函数，通过遗传算法的进化程度，动态地控制两种算法的最佳融合时机；算法后期，把遗传算法的若干优化解转换为蚁群算法里的信息素初值，利用蚁群算法得到满足一定QoS约束条件的最优解。此外，本文在算法前期和后期加入了邻近搜索的概念，采用了最大差异性交叉策略、保优选择策略和双重信息素更新策略。这些新策略使该算法既克服了遗传算法后期进化缓慢和蚁群算法前期信息素缺乏等缺点，又保留了遗传算法的并行性和蚁群算法正反馈等优点。 本文将遗传蚁群混合优化算法应用于QoS组播路由问题，使用Matlab进行仿真。实验证明，相比较于基本遗传算法和基本蚁群算法，本文算法不仅具有可行性、有效性，而且具有更好的全局收敛性，实现了对网络资源的有效优化，对未来网络的发展提供了理论依据。

With the rapid development of computer network technology, more and more communication demands appear, especially the rise of distance learning, video conference, IPTV, network game and so on. These applications have proposed higher and more urgent requirements on the bearing capacity of existing network, at the same time, the computer is required to use a better way of communication namely the multicast communication when it supports multimedia business. Multimedia real-time business has different requirements on QoS parameters, such as time delay, bandwidth, cost, packet loss rate. Multi-constraints QoS multicast routing algorithm has became one of hot issues in the field of computer network research. On the basis of multi-constraints QoS multicast routing algorithm, the paper built the mathematical model of QoS multicast routing, and proposed a new algorithm named the Combination of Genetic Algorithm and Ant Colony Algorithm (CGAACA), which combined genetic algorithm and ant colony algorithm effectively. On the early stage of algorithm, a few groups of optimization solutions were generated; on the middle stage, in order to ensure the fusion of genetic algorithm and ant colony algorithm at the best time, the genetic algorithm evolution degree function was given to control the best fusion time of the two algorithms; on the late stage, some optimization solutions of genetic algorithm were converted into pheromone initial value of ant colony algorithm, and then used ant colony algorithm to get the optimal solution which meet QoS constraints. In addition, the paper added the concept of neighborhood search on the early and late stage of algorithm, and used the maximum difference crossover strategy, optimal choice strategy and dual pheromone update strategy. Because of these new strategies, the algorithm not only overcame shortcomings of the slow evolution of genetic algorithm’s late stage and the lack pheromone of ant colony algorithm’s early stage, but also remained advantages of parallelism of genetic algorithm and positive feedback of ant colony algorithm. The paper applied the Combination of Genetic Algorithm and Ant Colony Algorithm to QoS multicast routing, and used Matlab to simulate. The results have showed that this algorithm not only has feasibility and effectiveness but also has better global convergence comparing with the genetic algorithm and ant colony algorithm. At the same time, it achieves effective optimization of network resources and offers a certain theoretical support on the future development of network.