Study On Multicast Technology In MPLS Networks

With the development and popularization of Internet, network applications go forward to multiculture. At the same time, supporting network bandwidth and multimedia bring forward higher and higher demands. These new Applications will consume network bandwidth sharply. Consequently, it speeds up the network congestion in a certain extent. IP multicast is the best choice to account for this question, and with the diversity of users' demands, multicast is required to meet different QoS (Quality of Service). Internet provides a best-effort service, which is provided with low-speed forwarding, complex routing mechanism, redundant services and low-efficiency usage of network resources. MPLS produces for these questions and new requirements. It has high-speed forwarding, traffic engineering (TE), QoS, Virtual Private Network (VPN) and etc.In this paper, we study on multicast trees algorithm in MPLS network, analyze and compare the presented schemes and propose some new schemes for resolving MPLS multicast questions. At the same time, we have combined with a number of important issues facing the current multicast technologies, study and explore the MPLS Multicast technologies respectively from scalability, flow balancing, fault recovery and reliability. Finally, we put forward some new views about MPLS Multicast technology.Multicast and MPLS are two completementary technologies. Considering multicast services are provided in MPLS network, which can make good use of the excellence of MPLS in order to ensure the QoS of IP Multicast. At the same time, IP multicast is reasonably applied in MPLS, which can save the number of using labels in MPLS environment, economize the memory, reduce the controlling spending, play down network comgestion and improve network security. Thus, the main research contents and innovative points are the follows:(1) Research of MPLS Multicast technology in Aggregated Multicast treeStudying on the MPLS aggregated multicast network from three aspects including pre-establishing the aggregated multicast trees, choosing the aggregated multicast tree for multicast group and matching the multicast group to the multicast tree, we have overcome the shortcomings of the one-sidedness, complexity and waste of resources about the previous aggregated multicast tree schemes. In the pre-established the multicast tree, we have proposed choosing the edge nodes far away the source node as the destination node for choosing the destination nodes set. In addition, we have explained the advantages of choosing the two methods. The establishment of the optimal spanning tree is always an NP-Completer problem, where is no optimal solution and the commonly used the heuristic algorithm to obtain the optimal solution. In this paper, we have presented a heuristic algorithm HLCA which is a heuristic A* search algorithm to improve and it can overcome defects of the existence of A* algorithm which is unable to determine on the estimated value of h(i). For selecting the candidate multicast tree, we introduce the definition of matching and propose aggregation tree selection algorithm. The algorithm can reduce the number of multicast trees in the multicast group-tree matching process, which is not involved in the past multicast technology. In the group-tree matching, we have proposed GTBA algorithm because the previous multicast group-tree matching algorithms had an extra waste of bandwidth. The GTBA algorithm can quickly achieve matching the multicast group or a subset of multicast group members to aggregated multicast tree under the premise of minimal waste bandwidth.(2) Research of MPLS Multicast techlology in traffic engineeringStudying on the multicast supported by the traffic engineering in MPLS network from four aspects including the establishment of paths between the source and destination nodes, aggregating some P2P LSPs into the MP2P LSPs, choosing the multicast routing to build the minimum cost multicast tree by applying the genetic algorithm to the MPLS network and establishing a network model of multicast traffic engineering in order to achieve the purpose of saving the labels space , minimize the maximum link utilization and the MPLS label space. We have proposed the st-AFSP algorithm which is based on the source-destination node pairs to build the feasible paths. It can find all possible paths based on the routing constraints in network, which is a easy-achieving and low complexity algorithm. In order to further save label space, we have raised the S-DA algorithm, which implements all feasible LSPs for further aggregation. It is not mentioned in the previous schemes. For the establishing the minimum cost multicast routing tree in MPLS network, we use the genetic algorithm to improve the convergence. Not involved the label space problem in the past traffic engineering model, we have proposed the MLLSU model which objective is to minimize the MPLS network, maximum link utilization and the using of label space. (3) Research of Multicast fault-tolerant technology in MPLS networkStudying on the MPLS multicast fault recovery techniques from the three aspects of the network fault detection, fault notification and fault recovery respectively, we have achieved a reduction in recovery latency and save network bandwidth and other resources. We have proposed the SBMR recovery program which is the improved AMFM fault-tolerant, and improvements include the following areas. When the network is in the normal state, the backup path can be used by a low priority business. When the network fails, the failing business flow can seize the backup path in order to the waste of bandwidth. Partial failure recovery algorithm can avoid a lot of bandwidth, delay and waste of resources due to the tree recovery AMFM. When the part of the failure returns to normal, the business traffic is re-transferred to the working path to continue to transfer. It can guarantee that multicast routing optimality, improve the network reliability and provide the QoS support.