Multicast

The Problem: One-to-Many Distribution with Unicast
In unicast, each receiver requires its own separate connection.
If a sender (for example, a live video encoder) needs to deliver the same stream to 100 receivers, it must transmit 100 identical copies of the packets.

This consumes:

• 100× the bandwidth on the sender’s interface, and
• 100× the bandwidth on any shared links upstream.

While unicast is ideal for one-to-one communication (web browsing, file transfers), it scales poorly for real-time media distribution to many receivers.

  The Multicast Solution: One Stream, Many Subscribers
With IP multicast, the sender transmits a single stream to a special multicast group address (for example, 239.0.0.1). Receivers that want the stream subscribe to that group using IGMP (for IPv4). The network — switches and routers — replicates packets only where paths diverge, delivering the stream efficiently to all subscribers.

So, the flow looks like this:
Sender → (one stream) → Router → Switch → Multiple Receivers
or
Sender → (one stream) → Network → Replication at branch points → Multiple receivers

Each link carries only one copy of the stream, until replication is required, no matter how many receivers exist downstream.

  Step-by-Step

1. Sender starts streaming
1. The sender transmits packets to a multicast destination IP address (for example, 239.0.0.1) and UDP port (for example, 5004).
   Sends one copy of the stream
2. It doesn’t know or care how many receivers exist.
   Does not receive join/leave information
2. Receivers join the group
1. Each receiver that wants the stream sends an IGMP Membership Report (“Join” used as shorthand) indicating:
   “Please deliver traffic from 239.0.0.1 to me.”
2. The router/switch adds that port/interface to its multicast forwarding table.
• IGMP messages go upstream toward the first-hop router
• Switches do not participate in IGMP as hosts — they observe it
3. Network replicates only as needed as it builds forwarding state
1. Layer-2 switches use IGMP snooping to learn which ports want the multicast group.
2. Layer-3 routers use multicast routing (e.g., PIM) to build a distribution tree toward the sender.
   Result
• Each switch/router adds the relevant port or interface to its multicast forwarding table.
• Routers replicate the multicast stream once per outgoing interface where receivers exist.
• Switches replicate the stream only to ports with interested receivers (based on IGMP snooping).
4. Receivers leave
1. When done, a receiver sends an IGMP Leave message (IGMPv2/v3), or
2. Simply stops responding to IGMP queries (timeout-based removal)
3. The switch/router prunes that port from its table — no more copies wasted there.

In flat 2110 networks, switches do most of the visible work — but routers still matter at VLAN/L3 boundaries.

Efficiency Comparison

Examples
Multicast enables one-to-many streaming by:

1. Using group IP addresses (224.0.0.0–239.255.255.255),
2. Allowing receivers to subscribe/unsubscribe dynamically via IGMP,
3. Making the network, not the sender, responsible for efficient replication.

The process of joining: Some exambles

Example 1 – PC or Server (Linux/Windows CLI)
You can join a multicast group and receive packets using command-line tools:

Linux

1. # Join a multicast group and dump packets
2. sudo ip maddr add 239.0.0.1 dev eth0
3. sudo tcpdump -n host 239.0.0.1

Windows PowerShell
# Windows joins automatically when an app opens a multicast socket.
# Example: using VLC: vlc udp://@239.0.0.1:5004
Opening that address in VLC causes Windows to issue an IGMPv3 Membership Report (Join).

Example 2 – Set-Top Box or IPTV Receiver
A set-top box is usually pre-configured to tune into multicast channels:
# IPTV channel list (M3U format)
#EXTINF:-1, Channel 1
udp://@239.0.0.1:5004
#EXTINF:-1, Channel 2
udp://@239.0.0.2:5004

When the user selects “Channel 1,” the box sends:
IGMP Join → 239.0.0.1
When the user changes the channel, it sends:
IGMP Leave → 239.0.0.1
IGMP Join → 239.0.0.2

Example 3 – Python Program
You can join a multicast group directly from a script:
import socket, struct

mcast_grp = '239.0.0.1'
mcast_port = 5004

sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM, socket.IPPROTO_UDP)
sock.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
sock.bind(('', mcast_port))

mreq = struct.pack("4sl", socket.inet_aton(mcast_grp), socket.INADDR_ANY)
sock.setsockopt(socket.IPPROTO_IP, socket.IP_ADD_MEMBERSHIP, mreq)

print(f"Joined multicast group {mcast_grp}:{mcast_port}")
while True:
 data, addr = sock.recvfrom(10240)  print(f"Received {len(data)} bytes from {addr}")

Running this script automatically sends the IGMP Join, and the network begins forwarding multicast traffic to that port.

Summary