Research On Network Topology Structure For Iptv Live Service

With the popularity of the Internet, more and more multimedia data is delivered across the Internet. The IPTV live services belong to one of services. IPTV live service has strict requirements of Quality of Service (QoS). IP Multicast is an effective technology, but the limited availability of IP Multicast in the Internet motives researchers to implement multicast in the application layer. In the application-layer multicast, applications arrange themselves as a logical tree. Data are transferred along the edges of the logical tree using unicast. The application-layer multicast for IPTV live service can be modeled as the minimum diameter, degree-bounded spanning tree (MDDBST) problem which has been proved to an NP-Complete problem.Application-layer multicast is a Peer-to-Peer (P2P) style. Applications don’t have reliable information of the underlying network topology and select distant peers, which generates much unnecessary traffic. Thus in the application-layer multicast for IPTV live service, application-layer traffic optimization (ALTO) should be also considered. The problem becomes the MDDBST problem considering ALTO.This paper proposes a Topology-aware Ant Colony Optimization based on Forest Growth which is a mesh-first solution. At first we combine the Delaunay triangulation based on the geographic location with the binning scheme based on the round-trip-time (RTT) to establish an overlay mesh. Both geographic location and RTT can be used to evaluate the underlying network topology. Thus the overlay mesh above is topology-aware. Then on the basis of this topology-aware overlay mesh we use an Ant Colony Optimization algorithm based on Forest Growth (ACOF) to construct a minimum diameter, degree-bounded spanning tree. In ACOF ants communicate with each other by the pheromones. The behaviors of each ant form a forest. Initial forest is empty. Each step of the ant is to select an edge in the overlay mesh by perceiving the densities of pheromones and makes the current forest grow further. The final aim is that the forest becomes a tree with all nodes in the overlay mesh. After all ants in the ant colony establish trees, ants update the pheromones according to the locally optimal tree and the globally optimal tree. The above steps are repeated until the algorithm converges. The unique principle abided by is the positive feedback mechanism of Ant Colony Optimization. The results of a serious of simulations show the overlay mesh combining the Delaunay triangulation and the binning scheme (DTB) is more related to the underlying network than the random overlay mesh and the Delaunay triangulation overlay mesh, and ACOF generates less traffic and smaller diameter than the CT(Compact Tree) algorithm on the DTB overlay mesh.