BENC 4453 :

COMPUTER ARCHITECTURE
Wired and Wireless Ethernet
The OSI Reference Model for
Networking
● Open System Interconnection (OSI) reference model, which
became an International Organization for Standardization (ISO)
standard in 1984.
● The OSI model is not a specification in the same sense that the
IEEE 802.3 Ethernet document is a specification.
● It’s a way of creating a “big picture” view of the many smaller
ideas falling within the larger idea of networking.
● The journey of data across a network connection begins at the
top, moves downward through the model’s layers to the physical
link at the bottom, across the physical link to another computer,
and then up through the layers to the top.
2
All = Application Layer
People= Presentation Layer
Seem = Session Layer
To = Transport Layer
Need = Network Layer
Data = Data Link Layer
Processing = Physical Layer

3
The OSI Reference Model
● Our focus is on the bottom four layers (called the transport set), which
encompass Ethernet and two crucial protocols: Transmission Control Protocol
(TCP) and Internet Protocol (IP).
● Not all networking systems map neatly onto its layers, and some networking
systems (particularly the Internet suite of protocols) have their own layered
reference models that predate the OSI model and span some of its several
layers.

4
The Application Layer
● The journey across a network begins when the user, launch a network-aware
program.
● The application layer provides the “human face” of network client programs like
email, chat, Usenet, web browsers, FTP, Telnet, and so on.
● Once the application layer has worked out the commands and data to be sent
over the network, and the address of the destination host, these are passed
down the stack to the next layer.

5
The Presentation Layer
● It has nothing to do with displaying data.
● It’s really about data conversion, and about how data will be “presented” to the
host on the other end of the connection.
● The presentation layer is where encoding differences like that are ironed out.
● Two other tasks often handled in the presentation layer are encryption and data
compression, both of which are optional but these days quite common.

6
The Session Layer
● The session layer opens the actual communication session with the other host.
● The session layer determines if the other host can in fact be reached.
● It also determines whether the connection between the two hosts is full duplex
or half duplex.

7
The Transport Layer
● The transport layer’s primary tasks are to take the data handed down
from the session layer by one or more processes, optionally divide it into
segments that are small enough to handle conveniently (a process called
segmentation), and queue segments from these processes for transmission over
the network (a process called multiplexing).
● At the receive side the transport layer reassembles segments and routes data
to the appropriate receiving process.
● In the modern Internet, the transport layer is implemented by TCP and UDP.

8
The Network Layer
● The network layer is primarily concerned with routing, determining what path
the data will take while it travels to the other host.
● A network path often includes one or more intermediate “stops” at computers
along the way.
● These intermediate nodes don’t generally unpack or attempt to interpret the
data; they simply look at the destination address on each packet and send the
packets on their way.
● In the context of the Internet, the network layer is where the IP does most of its
work.

9
The Data Link Layer
● The data link layer manages the flow of data over these direct connections,
reorganising data coming from higher layers yet again, this time into frames that
are of a size and format that the hardware implementing the direct connection
can handle.
● There are many different technologies used in connecting computers on a local
network.
● The data link layer may also provide local flow control, which ensures that
frames are not sent so quickly that the destination host’s buffers fill up, and
reliable delivery, whereby successfully received frames are acknowledged by
the receiver and unacknowledged frames are retained by the transmitter and
retransmitted as necessary.

10
The Physical Layer
● The physical layer is where the network connection literally “gets physical”: the
frames handed down from the data link layer are received as strings of bits,
which are converted to signals in the physical medium.
● Most of the physical layer’s operation occurs inside electronic circuitry in a
computer’s Network Interface Controller (NIC), and varies widely between
standards.

11
The Basic Ethernet Idea
● All computers on the network are peers; that is, none have special hardware or
software that is not present in all of the others.
● Any computer on the network can send or receive Ethernet packets to any other
computer on the network.
● Ethernet originated the idea of a MAC address, and every device attached to the
cable has a unique 48-bit numeric address, generally expressed as six groups of two
hexadecimal digits.
● The MAC address is in fact more of an ID code than a true address.
● MAC address says nothing about where its device is located on the network and is
used only to tell nodes apart.

12
Collision Detection and Avoidance
● Packet collision - when the network is quiet, nothing prevents two or more
nodes from beginning a transmission at the same time.
● Collisions in shared-medium Ethernet are detected in an interesting way: when
two pulses from two nodes enter the cable at the same moment, the pulses
“add” electrically, and the signal voltage on the cable is higher than during
normal network traffic.
● The NICs monitor the signal voltage while transmitting, and a higher-than-
normal voltage indicates a collision.
● When any transmitting node detects a collision, it ceases to send the current
packet and begins sending out a jam signal, which is a bit pattern that disrupts
the error-detection bits at the end of the frame.

13
Collision Detection and Avoidance
● Other nodes on the segment will see the packet as damaged and drop
it.
● When the network becomes quiet, those nodes that had collided wait a
random period of time called a backoff period (typically only a few
microseconds) before attempting to transmit again.
● An algorithm called truncated binary exponential backoff is used to vary the
distribution of the backoff period based on collision frequency.
● An initial collision triggers a random backoff period of either 0 or 1 slot (where
a slot is the time normally taken to transmit 512 bits) before attempting
retransmission.

14
Collision Detection and Avoidance
● The overall effect is to slow down network activity during congested periods,
“spacing out” retransmitted packets so that the network doesn’t simply grind
to a halt in a storm of packet collisions.
● This protocol is called CSMA/CD, which stands for Carrier Sense Multiple
Access with Collision Detection.
● “Carrier sense” is a bit of a misnomer here.
● Base band systems like Ethernet have no carrier, which is technically a radio
frequency wave on which signals are imposed via modulation.
● In this case, it only means that nodes on the network have a way to
determine when other stations are transmitting.

15
Collision Detection and Avoidance
● A network segment containing nodes that may transmit packets that collide is
called a collision domain.
● On early Ethernet systems, the entire network was a single collision domain,
which meant that throughput degraded as more nodes were added to the
network and collisions became more frequent.

16
Ethernet Encoding Systems
● The physical level of the OSI model, Ethernet NICs encode the data to be
transmitted by imposing a series of voltages on the network medium.
● The many variations of Ethernet each use a different encoding scheme;
○ 10Mbit standards (10BASE5, 10BASE2 and 10BASE-T)
○ 100Mbit (100BASE-TX)
○ 1Gbit (1000BASE-T) standards.

17
Ethernet Encoding Systems
● 10BASE5 and 10BASE2 (along with 10BASE-T; see the “10BASE-T and Twisted
Pair Cabling” section later in this chapter) encode bits via Manchester encoding.

18
Ethernet Encoding Systems
● Each data bit is encoded in one clock cycle, with a transition at the centre of
the cycle encoding the bit: a transition from negative to positive is considered a
1-bit, and a transition from positive to negative is considered a 0-bit.
● The arrows which transitions encode data and which directions the transitions
take.

19
Ethernet Encoding Systems
● 100BASE-TX Fast Ethernet use 4B/5B scheme, it encodes each four data bits
into five bits for transmission.
● The 5-bit encoded group is called a symbol.
● The encoding is performed using a simple static dictionary, which each
unique 4-bit group translates to a unique 5-bit symbol, shown in Table 7-1.
● This ensures that the transmitted bitstream is self-clocking, even in the
presence of long strings of 0- or 1-bits.

20
Ethernet Encoding Systems

21
PAM-5 Encoding
● Higher data rates, like those of Gigabit Ethernet, require denser encoding.
● PAM-5 encodes 2 bits per pulse by varying the signal voltage over five levels
rather than only two.
● Two of the five levels are positive voltages, two are negative voltages and the
fifth is 0 voltage.
● PAM-5 varies the amplitude of the pulses that make up the encoded data
stream, hence the name pulse amplitude modulation.

22
PAM-5 Encoding

23
PAM-5 Encoding
● The grey bars are pulses, and the wide black line is the amplitude for the
stream of pulses.
● Each pulse is considered a symbol, because one pulse encodes two bits.
● The 0V level does not encode any particular value, and its primary purpose is
to allow the receiver to extract a clock signal from the data, and to facilitate
error correction using a technology called forward error correction (FEC).
● A broad overview of FEC, though, additional bits are added to a data stream,
which allows the receiver to identify and correct a limited number of errors
without reversing the line (hence “forward”) to request retransmission of data.

24
Routers and the Internet
● LANs versus WANs is that LANs are networks of computers, and WANs are
networks of networks.
● With LANs everywhere, the next step is to allow one LAN to network with
other LANs.
● This is what the Internet was designed to do.
● The Internet protocol suite is very much involved in even the smallest local-
area network—a network of one device—that connects to the Internet.

25
Names vs. Addresses
● A node on LAN is identified by its MAC address.
● MAC addresses are (ideally) unique, so theoretically a node should be able to
contact another node on the other side of the world by placing the MAC
address of the faraway node in a packet.
● A MAC address contains no information about where its node actually is.
● Think of people at a meeting around a conference table.
● Everyone can see everyone else around the table, and when anyone talks,
everyone can hear.
● Tat’s how LANs work.

26
Names vs. Addresses
● The Internet uses a system very much like this.
● The collection of rules and techniques that enable packet-based
communication over the Internet is called the Internet Protocol (IP).
● Within the Internet protocol is an addressing scheme based on a type of
numeric address called the IP address.
● The IP is intimately connected with a higher-level protocol called Transmission
Control Protocol (TCP).
● The IP is focused on addressing and routing packets; the TCP is focused on
establishing and maintaining connections between computers so that packets
may be transferred.

27
Names vs. Addresses
● TCP is the Internet’s delivery mechanism: it makes sure that packets actually
get where they’re going, and that the order of a stream of packets is preserved
as it travels from computer to computer.
● IP and TCP work together and are rarely used separately. Tis is why most of
the time you’ll see them referred to as TCP/IP.

28
Routers
● Between the LAN and the outside world is
a router.
● In Figure 7-13, on the right is a LAN with
four hosts connected to a router.
● Internet routing is complex .
● Internal routing table is used to look up
network addresses to discover how to
reach them.

29
Routers
● A given router cannot necessarily access any arbitrary network with one
single connection.
● It may take several sequential connections (called hops) to reach a given
destination.
● At the end of each hop there’s another router, and that router forwards the
packet to the next router along the route.
● Eventually the packet arrives at the destination network, and that network’s
router forwards the packet to the individual host to which the packet was
addressed.

30
Routers
● The routing table in a large router may have hundreds of thousands of
entries.
● The routing table on a home router typically has only one, which contains the
address of the home’s ISP’s router.
● Any packet originating on a home router has only one possible path—
through its ISP—to the rest of the Internet.
● The home router forwards all packets to the ISP’s much larger and more
powerful router, which then selects the next hop on the route.

31
Routers
● Routers can block connections that
use specific port numbers as a
security measure, to prevent remote
connections to unauthorised servers.
● For example, a common way to
combat email spam is for hosting
services to configure their routers to
block port 25, which is assigned to the
Simple Mail Transfer Protocol (SMTP).

32
Wi-Fi Equipment in Use
● The simplest way to think of a wireless network is to replace a conventional
wired Ethernet hub with a Wi-Fi appliance called a wireless access point (AP.)
● The network shown in Figure 7-12 then becomes something very much like
Figure 7-17.
● A Wi-Fi AP is an Ethernet hub, using the Wi-Fi data link and physical layers
rather than the data link and physical layer for twisted pair network
technologies like 10BASE-T, 100BASE-TX or 1000BASE-T.

33
Wi-Fi Equipment in Use

34
End of Semester Survey (EoS)

35
End of Semester Survey (EoS)

36
Thanks!
Any questions?

37