分布式云环境中的流媒体内容分发是个多约束的有向斯坦纳树问题。针对典型的多信道和多用户区域的应用场景,提出启发式内容分发算法,同时考虑分发拓扑、各云数据中心计费方式、用户位置和请求速率等因素,通过多种方式构建有向斯坦纳树,以尽量低的代价满足流媒体服务质量要求。实验表明在不同的拓扑下,该启发式算法均能以较低的时间复杂度获得近优解,可方便地应用于商业的流媒体内容分发系统。
Streaming media content distribution in distributed cloud environments is a directed Steiner tree problem with multiple constraints. We proposed the heuristic content distribution algorithm for typical application scenarios with multiple channels and multiple user areas, and took into account the factors including topology structure, billing methods in each cloud data centre, and users location and request rate at the same time. Through building directed Steiner tree in various ways, the algorithm meets the requirements of streaming media services quality at the cost as low as possible. Experiments showed that with different topologies the heuristic algorithm could all obtain near- optimal solutions with lower time complexity, and could be easily applied in commercial streaming media content distribution systems.