Standard LAN Protocols

1
Standard LAN Protocols
2
The IEEE 802 Standards

• The IEEE 802 Standards (also known as ISO 8802)
  lay down a set of guide lines as to how a range
  of common networks should work.
• The IEEE 802.1 Standard describes the
  architecture, general management, addressing and
  internetworking of IEEE 802 networks.
• The IEEE 802.2 Standard describes the interface
  between the Network Layer and the Data Link
  Layer.
• This standardised interface is called Logical
  Link Control (LLC) and allows the software from
  the Network Layer upwards to be ported from one
  IEEE 802 LAN to another.

3
IEEE 802.2 Logical Link Control

• Logical Link Control (LLC) is essentially a
  sub-layer in the Data Link Layer of IEEE 802
  networks.
• It forms the upper half of the Data Link Layer
  while Medium Access Control (MAC) forms the
  bottom half.
• The Network Layer passes its packets to the Data
  Link Layer using the LLC access primitives.
• The LLC adds its own header containing sequence
  numbers and piggy-backed acknowledgement numbers.
• The resulting structure is then passed to the
  MAC.

4
IEEE 802.2 Logical Link Control

• The LLC offers three types of service tothe
  Network Layer
• Unreliable datagram service - this type of
  connectionless service does not bother with
  acknowledgements. It is mainly used for sending
  status information.
• Acknowledge datagram service - still
  connectionless but this time packets are
  acknowledge and retransmitted if they are
  corrupted or go missing.
• Reliable connection-oriented service - packets
  are passed up to the Network Layer in the order
  they were transmitted. Packets are acknowledge
  and are retransmitted if they are corrupted or go
  missing.

LLC is based on an older protocol
calledHigh-level Data Link Control (HDLC)
5
Medium Access Control

• IEEE 802.3, 802.4 and 802.5 describes the Medium
  Access Control (MAC) for CSMA/CD Bus, Token Bus
  and Token Ring LANs respectively.
• The roles MAC sub-layer are to determine when a
  frame can be transmitted, transmit the frame and
  to extract incoming frames from the bit stream
  (presented to it by the Physical Layer).
• The MAC sub-layer is particularly important in
  broadcast networks. This is because the MAC
  layer deals with the problem of contention (i.e.
  the situation when two hosts want to transmit at
  the same time).

6
Multiple Access (MA)

• To help us understand the sort of problems MAC
  may have to deal with, imagine a bus topology
  that uses a single cable to connect multiple
  hosts.
• If any host is allowed to transmit data over this
  cable then there is a multiple access (MA)
  channel between the hosts.
• Such a network is actually viable. Each host can
  send a frame at any time and, as long as the
  network is not being used by another host, the
  destination host can receive it.

7
Carrier Sense, Multiple Access (CSMA)

• With just Multiple Access, wecan make a
  cheap and simple network. However, if a host
  transmits a frame while another host is
  transmitting then both frames will be garbled.
  This is known as a collision.
• We can improve the performance of our simple
  network greatly if we introduce carrier sensing
  (CS). With carrier sensing, each host listens to
  the data being transmitted over the cable.
• A host will only transmit its own frames when it
  cannot hear any data being transmitted by other
  hosts.
• When a frame finishes, an interframe gap of about
  9.6?sec is allowed to pass before another host
  starts transmitting its frame.

8
Collision Detection (CD)

• Now and again two hosts will attempt to
  transmit a frame at exactly the same time. Even
  carrier sensing will not help in this case
  because both hosts will already be transmitting
  before they hear the other hosts data.
• This happens more often than you would think.
  There is a small time lag as data propagates
  along the cable. This means there is a realistic
  window of opportunity for both hosts to start
  transmitting without detecting the other.
• Both frames will, of course, be garbled. The
  best thing to do is for both hosts to abandon the
  transmission of their frames and to try again
  later.
• This is done by getting the hosts to listen to
  the data on the cable and comparing it to the
  data they are transmitting. If they are
  different then a collision must have occurred!

9
Collision Detection (CD)

• To ensure that the other hosts knows a
  collision has occurred as soon as possible, the
  first host to detect a collision will transmit a
  48-bit burst of random data called a jam
  sequence.
• This is just to make doubly sure that any hosts
  currently transmitting know to abandon their
  transmissions.
• After a suitable interval (the time it takes for
  the jam sequence to propagate along the whole
  network plus the interframe gap) the hosts can
  attempt to retransmit their frames.
• But hang on! If the collision was due to two
  hosts transmitting at exactly the same time, what
  is to stop them from transmitting at the same
  time again and again?

10
Binary Exponential Back-off Algorithm

• Rather than having two hosts (or occasionally
  more) attempting to retransmit their frames
  immediately after the interval, they wait for a
  random period of time.
• It is unlikely that both hosts will wait for
  exactly the same time period and so a stalemate
  situation can be avoided.
• After the first collision, a host will typically
  wait 0 or 1 time units (usually 1 unit
  51.2?sec).
• If a collision occurs again, it will wait 0,1,2
  or 3 time units.
• And so on. After n collisions a host will
  randomly choose to wait between 0 and 2n-1 time
  units.
• After 10 successive collisions, a host will
  typically give up and report to its user than it
  cannot transmit the data.

11
IEEE 802.3 CSMA/CD Bus LAN

• The 802.3 standard describes the operation
  of the MAC sub-layer in a bus LAN that uses
  carrier sense, multiple access with collision
  detection (CSMA/CD).
• Beside carrier sensing, collision detection and
  the binary exponential back-off algorithm, the
  standard also describes the format of the frames
  and the type of encoding used for transmitting
  frames.
• The minimum length of frames can be varied from
  network to network. This is important because,
  depending on the size of the network, the frames
  must be of a suitable minimum length.
• The standard also makes some suggestions about
  the type of cabling that should be used for
  CSMA/CD bus LANs.
• The CSMA/CD Bus LAN is also called Ethernet.

12
IEEE 802.3 Cable Types

• There are four types of cable use for CSMA/CD
  bus LANs. The most common are 10Base2 (a.k.a.
  thin Ethernet) and 10Base-T (a.k.a. category 5
  UTP).
• (a) 10Base5
• (b) 10Base2
• (c) 10Base-T

13
IEEE 802.3 Frame Format

• Regardless of the type of cable used in
  the CSMA/CD Bus LAN, the format of the frame
  generated by the MAC sub-layer is the same.
• Frames are transmitted using Manchester Encoding.
• The preamble contains the pattern 10101010 (a
  square wave) lasting for 5.6?sec. When a network
  card hears that pattern, it gets ready to listen
  to the address information.
• The Start Of Frame (SOF) byte has the pattern
  10101011 (continuing the square wave until the
  last bit). This change indicates that the
  destination address follows.

14
IEEE 802.3 MAC Addresses

• Every network card in the world has a
  unique 46-bit serial number called a MAC address.
  The IEEE allocates these numbers to network card
  manufacturers who encode them into the firmware
  of their cards.
• The destination and source address fields of the
  MAC frame have 48 bits set aside (the standard
  also allows for 16-bit addresses but these are
  rarely used).
• The most significant bit is set to 0 to indicate
  an ordinary address and 1 to indicate a group
  address (this is for multicasting, which means
  that frames are sent to several hosts). If all
  48 bits are set to 1 then frames are broadcast to
  all the hosts.
• If the two most significant bits are both zero
  then the 46 least significant bits contain the
  MAC addresses of the source and destination hosts.

15
IEEE 802.3 The Other Frame Fields

• The data length field contains the number of
  bytes of data (up to a maximum of 1500 bytes).
• The data field contains the LLC data structure,
  which in turn contains the Network Layer packet.
• If the data field is less that an appropriate
  minimum length (usually 46 bytes but this can be
  changed if necessary) then the pad field is
  filled will extra bytes to ensure the frame is
  long enough.
• The checksum field (a 32-bit cyclic redundancy
  code) is tagged onto the end of the frame so that
  the receiving host can check it for errors.

16
IEEE 802.3 Minimum Frame Length

• When a host transmits a frame, there is a
  small chance that a collision will occur. The
  first host to detect a collision transmits a
  48-bit jam sequence.
• To ensure that any hosts involved with the
  collision realise that the jam sequence is
  associate with their frame, they must still be
  transmitting when the jam sequence arrives. This
  means that the frame must be of a minimum length.
• The worse case scenario is if the two hosts are
  at far ends of the cable. If host As frame is
  just reaching host B when it begins transmitting,
  host B will detect the collision first and send a
  jam signal back to host A.

17
IEEE 802.3 Minimum Frame Length

• The longest time between starting to
  transmit a frame and receiving the first bit of a
  jam sequence is twice the propagation delay from
  one end of the cable to the other.
• This means that a frame must have enough bits to
  last twice the propagation delay.
• The 802.3 CSMA/CD Bus LAN transmits data at the
  standard rate of r 10Mbps.
• The speed of signal propagation is about v
  2?108m/s.

18
IEEE 802.3 Minimum Frame Length

• In order to calculate the minimum frame
  length, we must first work out the propagation
  delay from one end of the cable to the other.
• Say the cable is d 400m long.
• The propagation delay time tp d?v. In our
  example tp 400 ? (2?108) 2 ?10-6 or 2?sec.
• The round-trip propagation delay is, of course,
  twice this. Thus the round trip delay is 2?tp
  4?sec.
• With a data rate of r 10Mbps, each bit has
  duration tb 1/r 1 / 10,000,000 0.1?sec.
• The number of bits we can fit into a round-trip
  propagation delay is 2?tp ? tb 4 ? 0.1 40
  bits.
• The minimum frame length is thus 40 bits (5
  bytes). A margin of error is usually added to
  this (often to make it a power of 2) so we might
  use 64 bits (8 bytes).

19
IEEE 802.3 Minimum Frame Length

• The standard frame length is at least 512
  bits (64 bytes) long, which is much longer than
  our minimum requirement of 64 bits (8 bytes).
• We only have to start worrying when the LAN
  reaches lengths of more than 2.5km.
• 802.3 CSMA/CD bus LANs longer than 500m are
  usually composed of multiple segments joined by
  in-line passive repeaters, which output on one
  cable the signals received on another cable.
• When we work out the minimum frame length for
  these longer LANs, we also have to take the
  delays caused by the passive repeaters (about
  2.5?sec each) into account as well.

20
IEEE 802.3 Non-Deterministic

• The 802.3 CSMA/CD bus LAN is said to be a
  non-deterministic network. This means that no
  host is guaranteed to be able to send its frame
  within a reasonable time (just a good probability
  of doing so).
• When the network is busy, the number of
  collisions rises dramatically and it may become
  very difficult for any hosts to transmit their
  frames.
• A real-time computing application (such as an
  assembly line) will demand that data is
  transmitted within a specified time period.
• Since the 802.3 bus LAN cannot guarantee this,
  its use for real-time applications may not only
  be undesirable but potentially dangerous in some
  situations.

21
802.4 - Standard for Token Buses

• The IEEE 802.4 standard describes the operation
  of token buses.
• Unlike CSMA/CD, token buses have a guaranteed
  response time.
• Designed for use on factory floors, token buses
  are designed to be robust and reliable.
• The token bus uses a common media for sending
  frames (just like that used in CSMA/CD).

22
Tokens

• Unlike CSMA/CD, each host must have permission
  before it can send a frame.
• Permission is given in the form of a token (a
  special packet that is sent to the host).
• There is only ever one token on the network,
  which is passed from host to host.

23
The 802.4 frame

• The 802.4 frame is slightly different from that
  used by the 802.3 specification.
• Even the modulation is different - there are an
  extra three signals other than those that
  represent 0, 1 and idle.
• These signals are used to indicate the start and
  ends of frames.

24
Frame Fields

• Preamble lengths are different and there is no
  data length field.
• The 802.4 frame includes a frame control field
  that specifies the type of frame being sent.
  There are frames for sending data, passing tokens
  and management frames.
• The source address, destination address and
  checksum (CRC) fields are the same as in the
  802.3 standard.

25
802.5 - Standard for Token Rings

• The IEEE 802.5 standard describes the operation
  of token rings.
• Like token buses, these have a guaranteed
  response time depending on network size.

26
Structure of Token Ring

• Unlike bus networks and token buses, token rings
  do not use a single media.
• Instead, each ring interface (NIU) is connected
  to the next by a separate link.
• The links are arranged in a circle.
• Each ring interface has a 1-bit buffer (which
  introduces a 1-bit delay).
• The interface can either store-and-forward the
  bit or send a different bit.

27
The Token

• In a token ring, the token is a special bit
  pattern that circulates around the ring.
• There can be only one token so only one host can
  transmit when it has the token.
• The bits propagate around the ring until a host
  decides it wants to send and removes the token
  from the network.
• Removing the token is done by inverting just one
  bit in the token to turn it into a frame header.

28
Sending Data

• Once a host has the token, it can transmit a
  frame. In fact it can transmit several frames so
  long as it can do so within the token-holding
  time (10msec).
• The frame propagates around the token ring where
  it is seen by the destination ring interface,
  which gives a copy of it to the destination host.
• The frame propagates around the whole token ring
  and is removed by the sending host.

29
Acknowledgements

• Acknowledgements are sent back by modifying bits
  in the frame status field.
• If left unmodified, the sender knows that the
  destination host never received the frame (in
  which case the sender will retransmit).
• When the last frame has been sent, the
  transmitting host reconstructs the token and then
  the next host can take it.

30
The 802.5 Token/Frame

• The token is made of 3 fields SD,AC,ED.
• The SD field indicates the start of a frame.
• The ED field indicates the end of a frame.
• The AC field contains a number of flags.
• The frame also starts SD,AC,

31
The 802.5 Frame

• The frame uses the FC field to distinguish a data
  frame from other management frames (e.g. for
  electing a new monitor station if the existing
  one fails).
• The destination and source address fields are the
  same as in 802.3 and 802.4 frames.
• There is no limit to the amount of data sent as
  long as it is within the token-holding time.
• The checksum is a 32-bit CRC.

32
The Frame Status Field

• The frame status field is used for
  acknowledgements. There are two flags in this
  field, called A and C, which are set to A0 and
  C0 when the frame is transmitted.
• The receiving station can change these bits.
  When read by the sender, they mean
• A0, C0 destination not found
• A1, C0 frame not acknowledged (NAK)
• A1, C1 frame acknowledged (ACK)

33
The Access Control Field

• The access control field of the frame contains
  flags for indicating priorities, differentiating
  between tokens and data frames, and a monitor
  bit.
• When T is 1, it indicates that the frame is a
  token. When it is 0, it is a data frame.

P
P
P
T
M
R
R
R
34
The Monitor Bit of the Access Control Field

• The monitor bit M is used to mark frames as
  having passed the monitor station.
• One station on the network is nominated as the
  monitor when the network is switched on.
• Any frame passing through the monitor is marked.
  If a frame passes through the monitor a second
  time, it is deleted.

35
The Priority Bits

• The 802.5 standard allows 8 levels of priority (0
  being the lowest and 7 being the highest) as
  indicated in the PPP bits.
• A host with an equal or higher priority than the
  token may capture the token.
• A host may also attempt to reserve a token by
  setting the RRR bits of a passing frame to its
  priority. This will then be copied to the new
  tokens PPP bits.

36
Benefits of Token Rings

• Unlike CSMA/CD, the efficiency of token rings is
  actually at its best when the network is being
  used most heavily.
• A token ring can be more difficult to maintain
  and expand than a bus network or a token bus
  network but it is no less reliable than a token
  bus and, in heavy usage environment, much better
  than a bus network.

37
802.2 - Logical Link Control

• It is useful to hide the differences between
  networks by using a common protocol for
  controlling them.
• This is what the Logical Link Control does (its
  like an interface used by hosts for talking to
  the network regardless of type).
• The IEEE 802.2 standard describes Logical Link
  Control for all 802.x networks.

38
Logical Link Control (LLC)

• Basically, the network layer hands packets to the
  LLC (upper part of data link layer).
• The LLC adds some control data to the packet and
  passes it to the data link layer proper.
• The data link layer encapsulates the packets
  inside data frames that are then transmitted on
  the network.

39
LLC service options

• The LLC can operate in several different modes
• unreliable datagram service is useful for
  transmitting video and voice signals. There are
  no acknowledgements and no sequence numbers.
• acknowledged datagram service includes
  acknowledgements and sequence numbers.
• connection-oriented service provides an
  apparently reliable error-free connection.

40
Some Jargon

• A Bridge is a connection that links two similar
  LANs. Packets not meant for the current LAN are
  passed to the bridge and sent on to the other
  LAN.
• A Router connects several similar networks and
  uses the destination address to decide which
  network the packet should be sent to.
• A Gateway connects dissimilar networks and
  performs protocol conversions on packets.

41
Revisiting the OSI Model

• The ISO (International Standards Organization)
  have devised a reference model for computer
  networks.
• Such a model is meant to help network designers
  design networks and protocols.
• The model is called the OSI (Open Systems
  Interconnection) reference model.
• It has 7 layers with distinct functions.

42
The OSI Reference Model

• We imagine that each layer is the servant of the
  layer above it.
• Data and instructions are passed down through the
  layers until the data is physically transmitted
  by the physical layer. The data is then passed
  up through the layers of the destination.

43
Peer-to-peer Communication

• We imagine that each layer in the transmitting
  host is talking directly to the equivalent layer
  (or peer) in the receiving host.
• It is normal to say things like I was talking to
  Fred on the phone. In fact, it was the
  telephone I was talking to, not Fred!
• Fred is my peer (another person) but my voice was
  processed by the telephone and then sent via the
  telephone network.

44
The Layers

• The physical layer transmits raw bits over a
  communication channel.
• The data link layer uses link flow control and
  error correction to make the link appear free of
  errors.
• The network layer controls the local sub-net and
  decides on which direction packets should be
  routed.

45
The Layers

• The transport layer turns the packet based
  communication into a stream of data. It deals
  with assembly of packets and the disassembly of
  message into packets. It also contains
  end-to-end flow control.
• The session layer establishes and maintains a
  logical connection between sender and receiver.

46
The Layers

• The presentation layer can perform various
  functions. Typically it might convert ASCII
  characters into UNICODE characters.
• The application layer represents the programs
  that use network communication such as
  FTP,TELNET, Web Browsers etc...

47
The TCP/IP Reference Model

• Not every network is based on the OSI reference
  model. Many systems are based on the TCP/IP
  model.
• TCP stands for Transmission Control Protocol and
  IP stands for Internet Protocol.

48
How TCP/IP differs from OSI

• TCP/IP has 5 layers rather than 7 (actually it
  traditionally has 4 layers, the data link layer
  and physical layer being combined in the
  Host-to-network layer).
• There are nosession or presentationlayers
  (their rolesbeing performed by the application
  layer if necessary).

49
The TCP/IP Layers

• The TCP/IP layers that exist are essentially the
  same as those in the OSI reference model.
• The host-to-network layer is responsible for
  sending bits across the network and for link
  error control and link flow control (i.e. data
  link layer and physical layer combined).
• The Internet layer switches packets around the
  network and places packets on (or removes them
  from) the network using a packet format called IP
  (Internet Protocol).

50
Finally.The TCP/IP Layers

• The transport layer accepts data (and
  instructions) from the application layer and
  breaks the data up into packets. It also
  reassembles received packets into data. It is
  also responsible for end-to-end flow control.
• The application layer contains all the higher
  level protocols such as TELNET, FTP, SMTP (email)
  etc... that are used by applications.