The Distance Vector Multicast Routing Protocol (DVMRP) is the extension of the Routing Information Protocol (RIP) which is used in unicast routing. It uses the sourcebased tree approach to multicasting. It is important to mentionin that each router in this protocol that receives a multicast packet to be forwarded implicitly creates a source-based multicast tree in three steps:

1. The router uses an algorithm called reverse path forwarding (RPF) to simulate creating part of the optimal source-based tree between the source and itself.
2. The router uses an algorithm called reverse path broadcasting (RPB) to create a broadcast (spanning) tree whose root is the router itself and whose leaves are all networks in the internet.
3. The router uses an algorithm called reverse path multicasting (RPM) to create a multicast tree by cutting some branches of the tree that end in networks with no member in the group.

Reverse Path Forwarding (RPF)

The first algorithm, reverse path forwarding (RPF), forces the router to forward a multicast packet from one specific interface: the one which has come through the shortest path from the source to the router.

How can a router know which interface is in this path if the router does not have a shortest-path tree rooted at the source? The router uses the first property of the shortest-path tree, which says that the shortest path from A to B is also the shortest path from B to A. The router does not know the shortest path from the source to itself, but it can find which is the next router in the shortest path from itself to the source (reverse path).

The router simply consults its unicast forwarding table, pretending that it wants to send a packet to the source; the forwarding table gives the next router and the interface the message that the packet should be sent out in this reverse direction.

The router uses this information to accept a multicast packet only if it arrives from this interface. In multicasting, a packet may arrive at the same router that has forwarded it. If the router does not drop all arrived packets except the one, multiple copies of the packet will be circulating in the internet.

Reverse Path Broadcasting (RPB)

The RPF algorithm helps a router to forward only one copy received from a source and drop the rest.

However, when we think about broadcasting in the second step, we need to remember that destinations are all the networks (LANs) in the internet.

To be efficient, we need to prevent each network from receiving more than one copy of the packet. If a network is connected to more than one router, it may receive a copy of the packet from each router.

RPF cannot help here, because a network does not have the intelligence to apply the RPF algorithm; we need to allow only one of the routers attached to a network to pass the packet to the network.

One way to do so is to designate only one router as the parent of a network related to a specific source. When a router that is not the parent of the attached network receives a multicast packet, it simply drops the packet.

There are several ways that the parent of the network related to a network can be selected; one way is to select the router that has the shortest path to the source. If there is a tie in this case, the router with the smaller IP address can be selected.

Reverse Path Multicasting (RPM)

RPB does not multicast the packet, it broadcasts it. This is not efficient. To increase efficiency, the multicast packet must reach only those networks that have active members for that particular group. This is called reverse path multicasting (RPM).

To change the broadcast shortest-path tree to a multicast shortest-path tree, each router needs to prune (make inactive) the interfaces that do not reach a network with active members corresponding to a particular source-group combination.

This step can be done bottom-up, from the leaves to the root. At the leaf level, the routers connected to the network collect the membership information using the IGMP protocol. The parent router of the network can then disseminate this information upward using the reverse shortest-path tree from the router to the source.

When a router receives all of these membership-related messages, it knows which interfaces need to be pruned.Since these packets are disseminated periodically, if a new member is added to some networks, all routers are informed and can change the status of their interfaces accordingly. Joining and leaving is continuously applied.