Research Of Overlay Multicast Based On Nodes QOS In Heterogeneous Network

Internet shifts mobile terminal and the mobile nodes and the traditional PC nodes are very different. That is, heterogeneous network is about trends and the future. Nevertheless, the traditional Application Layer Multicast (ALM) models are no longer applicable to heterogeneous networks. So a multicast model applicable to heterogeneous networks is proposed.The current methods of constructions of multicast tree only take delay into consideration, using a distance priority method. As a consequence, some weak nodes may be at upper layer in the multicast tree, serving a lot of nodes. Nevertheless, these nodes can’t be able to serve so much nodes. These nodes become bottleneck and the lower nodes are all affected. So a multicast model applicable to heterogeneous networks is proposed.Firstly, a method of constructing cluster tree based on node degree is used to improve multicast tree topology of ZIGZAG to get an algorithm called ALM algorithm based on ZIGZAG considering Degree (AZCD). A tree is constructed in each tree, and the children of each node can’t exceed its degree while each node has as many children as possible. As a result, each node can be able to serve all its children. In the meantime, the height of the cluster tree is kept small. By simulating, because of above mentioned method, latency is reduced and load is balanced.Secondly, the (Quality of Service) QoS of a node is not only the degree of the node. So all QoS attributes should be taken into consideration to obtain the QoS assessment. This QoS assessment is used to improve ZIGZAG to get an algorithm called ALM algorithm based on ZIGZAG considering QoS (AZCQ). This is an improvement of the previous method. Compared with Short Delay Degree-constrained Multicast (SDDM), the simulations show the delay of AZCQ is smaller than SDDM. Moreover, the link pressure of AZCQ is smaller than SDDM which means AZCQ has better balanced load. Moreover, node join, node departure, cluster split and cluster merging are improved by the above method. And a short term strategy is describe to optimize tree structure.