| In the recent years, with the popularization and volatile development of the Internet, there appear so many new applications of Internet. Most of these are multimedia applications which need high bandwidth, such as telconferencing, audio/video broadcast, stock market issue, distance education, computer supported cooperate work. All these bring many problems, for example, rapid consumed of bandwidth and the net congestion. In order to solve the net bottleneck, people introduce the technology of IP multicast. As a pattern of communication among host computers, IP multicast searches address through multicast address, sending multicast data to group members. Multicast provides an effective measure of communication and transmission, and it also can fully make use of net resource. Multicast can make the network performance optimized, it also make distributed application possible. In economy, the application (such as VOD), which can not be realized using unicast or broadcast can be applied. At first, this paper introduces the base knowledge of multicast .In order to implement multicast, we need two kinds protocols: multicast routing protocol and group management protocol. Multicast routing protocol includes two types: Dense-Mode(DM) and Sparse-Mode(SM).DM multicast routing protocol depends on flooding to send multicast data to each router's interface. The common multicast routing protocol of Dense-Mode includes PIM-DM, DVMRP, MOSPF.SM constructs some Rendezvous Points (RP) as data exchanging place. RP takes charge of transmission between data resources and receivers. The common multicast routing protocol of Sparse-Mode includes CBT, PIM-SM. Besides, Multicat has a group management protocol IGMP, which is the precondition of multicast.This paper introduces PIM-SM detailedly. PIM-SM routing protocol has several virtues: explicit "Join" mode, effective data transmission, stream flooding control, multicast not limited in one area, the standard and realizing technology mature, at the same time, PIM-SM is the base of border multicast routing protocol, so it relatively fit multicast model. So after careful researching, we implement the basal function by using raw socket. This program includes six modules:Mcast_ip, Mcast_packet, Mcast_pim, Mcast_igmp, MRIB and RIB.When the program receives IP packet, it is disposed by Mcast_ip module. This module triggers all kinds disposing mechanism according protocol code and destination address. In this program , we only care about three cases: multicast data, PIM-SM control message and IGMP control message. In the part of implementation, for each router ,what it do is to creat multicast routing table by sending and receiving packet.　From the whole of implementation, the whole domain PIM-SM routers compose a net topology.In this topology ,each router sends control message to other routers and exchanges routing information.Through these way ,each router knows the net structure and construct RPT and SPT and transform from RPT to SPT.During the work, we found that the RP (Rendezvous Point) as the root of shared-based tree plays an important action. How to choose a RP is the key of the PIM-SM protocol implementation .But the existing selections of RP presented in PIM-SM protocol don't consider the net characteristic and the RP paralysis problem.To several existing chosen function of RP,we present a new mechanism—"two degree RP",which managing the RPs in two grades, considering the topology of the network, present the candidate RP selecting policy. The first degree RP explore the change of network timely, increased the dynamic of RP, and adapted the instance of the members join/exit the group frequently better. Furthermore, if the second degree RP paralysis the first degree RP can take over the work ,it can avoid the data loss problem effectively . To implement the ameliorate, we should modify PIM-SM protocols and add some control messages. Through the test in the real network, We found although "two degree RP" can add burden of net and routers, but after an...
|
PDF Full Text Request