multicast.tex

柯老师网站上找到的

\chapter{Multicast Routing}
\label{chap:multicast}
This section describes the usage and the internals of multicast
routing implementation in ns.
We first describe the user interface to enable multicast routing,
specify the multicast routing protocol to be used and the
various methods and configuration parameters specific to the
protocols currently supported in ns.
Then we describe in detail the internals and the architecture of the
multicast routing implementation in ns.

\section{Multicast API}
\label{sec:mcast-api}
Multicast routing is enabled in the simulation by setting 
the \code{EnableMcast_} Simulator variable to 1
before any node, link or agent objects are created.
This is so that the subsequently created 
node, link and agent objects are appropriately
be configured during creation to support multicast routing.
For example the link objects are created with interface labels that
are required by some multicast routing protocols, 
node objects are created with the 
appropriate multicast classifier objects
and agent objects are made to point to the 
appropriate classifier at that node
[need to point to descriptions of interfaces, multicast-enabled node
and multicast classifiers].

A multicast routing strategy is the mechanism by which
the multicast distribution tree
is computed in the simulation.
The multicast routing strategy or protocol
to be used is specified through the mrtproto command.
A handle is returned to an object that has 
methods and configuration parameters specific to a
particular multicast routing strategy or protocol.
A null string is returned otherwise.
There are currently 4 multicast routing strategies in ns: Centralized
Multicast, static Dense Mode, dynamic Dense Mode (i.e., adapts to
network changes), Protocol Independent Multicast - Dense Mode.
Currently only Centralized Multicast returns an object that has
methods and configuration parameters.

Need to describe how an agent joins and leaves multicast groups.
Also need to talk about multicast addresses.

An example configuration would be:
\begin{program}
	set ns [new Simulator]
	Simulator set EnableMcast_ 1 \; enable multicast routing;
        set node0 [$ns node]
\end{program}

\subsection{Protocol Specific configuration}

\paragraph{Centralized Multicast}
A Rendezvous Point (RP) rooted shared tree is built
for a multicast group.
The actual sending
of prune, join messages etc.
to set up state at the nodes is not simulated.
A centralized computation agent is used
to compute the fowarding trees and set up 
multicast forwarding state, (*,G) at the relevant nodes
as new receivers join a group.
Data packets from the senders to a group
are unicast to the RP.
Note that data packets
from the senders are unicast to the RP
even if there are no receivers for the group.
Description of available methods:
	shared tree
		setting nodes to be candidate RPs
                getting RP for group
	switching to source-specific trees
Description of behavior under network dynamics
	mention causality violations
Should give example configuration

\paragraph{Static Dense Mode}
The Static Dense Mode protocol 
is based on DVMRP with the exception
that it does not adapt to network dynamics.
It uses parent-child lists as in DVMRP
to reduce the number of links over which the
data packets are broadcast.
Prune messages for a particular group
are sent upstream by nodes in case they do not lead
to any group members.
These prune messages instantiate prune state
in the appropriate upstream nodes to prevent multicast
packets from being forwarded down links
that do not lead to any group members.
The prune state at the nodes times out
after a prune timeout value that is 0.5s by default.
The prune timeout is a class variable in Agent/DM.
Should give example configuration.

\paragraph{Dynamic dense mode}
DVMRP-like dense mode protocol that
adapts to network changes is simulated.
'Poison-reverse' information
(i.e. the information that a particular neighbouring node
uses me to reach a particular network)
is read from the routing tables of neighbouring nodes
in order to adapt to network dynamics
(DVMRP runs its own unicast routing protocol
that exchanges this information).
Prune Timeout value etc.
Should give example configuration.

\paragraph{PIM dense mode}

\section{Internals of multicast routing}
\label{sec:mcast-internals}

We first describe the main classes that are
used to implement multicast routing and
then describe how each multicast routing strategy
or protocol is implemented.

\subsection{The classes}
The main classes in the implementation are
the McastProtoArbiter class and the McastProtocol class 
that is the base class for the
various multicast routing strategy and protocol objects.
In addition some methods and configuration parameters
have been defined in the Simulator, Node, Agent 
and Classifier objects for multicast routing.

\paragraph{McastProtoArbiter class}

\paragraph{McastProtocol class}

\paragraph{Simulator class}

\paragraph{Node class}

\paragraph{Agent class}

\paragraph{Classifier class}

\subsection{Protocol Internals}
\label{sec:mcastproto-internals}

We describe the implementation of the
multicast routing protocol agents in this section.

\paragraph{Centralized Multicast}

\paragraph{Static Dense Mode}

\paragraph{Dynamic Dense Mode}

\paragraph{PIM Dense Mode}

\endinput