- Point-to-point link and broadcast links
- Point-to-point links
are links pertaining to
switched networks
These links typically spans large geographical distances (across continents)
Due to the length, transmission errors (bit errors) are very likely (1 in ~10000 in copper and 1 in ~10000000 in optical fiber)
Hence:
- the datalink layer for point-to-point link deals primarily with reliablilty (in addition to framing)
- Broadcast links
in contrast span only
very short distances
(local area network)
Transmissions over short distance are very reliable
Hence:
- the data link layer of broadcast network has only rudentary error control functions !!!
- Point-to-point links
are links pertaining to
switched networks
- Different kind of problem:
- Problem of broadcast networks:
- Only one transmission
at a time is allowed
(Otherwise, you have collisions)
- Only one transmission
at a time is allowed
- Problem of broadcast networks:
- Main issue dealt with by data link layer of
broadcast networks:
- How to gain (exclusive) access to the shared broadcast medium
We will discuss medium access protocols in more details
- Contention based MAC protocols
Contention: be the first
- The node that transmits first
will be allowed to
complete its transmission
- When the node finishes with its transmission, all nodes with frames will compete again for the right to transmit...
A.k.a.: Multiple Access protocols
- In these protocols, it is
possible
(and happens frequently) that
multiple nodes
transmit at the same time.
And when multiple nodes transmits at the same time, all transmissions become corrupted.
- The collision must be detected and when collision is detected,
nodes will have to retry their transmission.
But each node must try to retransmit is such a way that they don't collide with other nodes... (we will see how later)
- An inherent property
of these protocols is that
nodes may need multiple tries
before they are successful
in transmitting
For this reason, this class of protocol is known as multiple access protocols
Question:
- Why not use frequency division multiplexing (like radios (i.e., each node transmits on a different radio frequency) and have the nodes share the broadcast medium ???
Answer:
- The purpose is to
communicate
with each other
- Nodes using
different frequencies
cannot receive each other's transmissions
And can thus unable to communicate with each other.
- Analogy:
- You and your friend can
share the
medium by
using
different channels
of a
walkie-talkie
(Both of you can talk at the same time
But, you will not be communicating with each other, because you cannot receive each other's messages
- You and your friend can
share the
medium by
using
different channels
of a
walkie-talkie
- The node that transmits first
will be allowed to
complete its transmission
- Contention FREE MAC protocols
Regulated access to the broadcast medium:
- Each computer is
given a turn to transmit
when it receives
the token message.
-
When a computer receives the token and that computer
has some message
to transmit, it transmits
one message
and
passes the token
to the next computer.
When the computer receives the token and that computer does not have any message to transmit, it simply passes the token to the next computer.
-
Main problem with these schemes:
- How to maintain the list of
computers
- What to do when a
new computer join the
network
The new computer will not receive the token "automatically"
- What to do when an existing
computer leaves the
network
The ring is broken !
- What to do when a
new computer join the
network
- How to maintain the list of
computers
- Each computer is
given a turn to transmit
when it receives
the token message.
- The datalink layer for broadcast links consists of
two functional layers:
- Logical Link control:
- Provide reliability functionality
(i.e.: sliding window and ACK scheme).
In practice:
- Most real-life (Ethernet, Token Ring)
datalink layers for broadcast links
omits this function
.... because the broadcast link is very reliable.
- The real-life (Ethernet, Token Ring) datalink layers rely on the Transport Layer (TCP !) to implement the reliability function.
- Provide reliability functionality
- Medium Access Control:
- This layer implements functionalities to regulate orderly and fair access (what is fair ?) to the shared transmission medium.
- We will study a number of commonly used MACs...