computer network designs is by their scope or scale. LAN (Local Area Network) MAN (Metropolitan Area Network ) WAN (Wide Area Network ) LAN
LAN is generally a privately owned network
with in a single office, building or campus covering a distance of a few kms. The main reason for designing a LAN is to share resources such as disks, printers, programs and data MAN
A network spanning a physical area larger
than a LAN but smaller than a WAN, such as a city. A MAN might be owned and operated by a single organization, but it usually will be used by many individuals and organizations. WAN
WAN spans a large geographic area, such as
a state, or country. WANs often connect multiple smaller networks, such as local area networks (LANs) or metro area networks (MANs). Error detection
CRC (cyclic redundancy check)
A mathematical algorithm is used on the data block
to be sent to arrive at the CRC (a small block of bits which are appended to the data block and sent by the sender). At the destination, the receiver separates the data block, recomputes the CRC using the same algorithm and matches the received CRC with the computed CRC. A mismatch indicates an error. LAN can use star, bus or ring topology. However, bus topology is popular in the Ethernet LANs and Token Bus LANS and Ring topology is popular in the Token Ring LANs. A modified version of Token Ring is FDDI
Ethernet and Token Ring are the most popular LANs
LAN broad cast their message to all host on that LAN. This is because, all the host share a single transmission wire. The address in the packet or frame enables the destination host to receive that packet, while all other hosts ignore it. Broadcast network can be 1. Static 2. Dynamic In static method each host is given a fixed time slice to send the information. If a host does not have anything to send, that time slice is wasted. That is why this method is not very popular. In dynamic method, a host can send a frame any time. Thus, if two host send a frame at the same time, the two frames could collide with each other. We need some arbitration to solve this problem. A protocol called MAC (Media Access Control) performs this job and decides which node can access the medium and when. Dynamic method can be further subdivided into: 1. Centralized 2. Decentralized In Centralized method, there is a single entity which decides who would send the data next. This is typically a master-slave method. A host wanting to send data can explicitly request this unit for the permission. In star topology, the hub can play the role of the master. However if the arbitration unit goes down, the entire network goes down. In decentralized method, we do not require any external arbitrator for this. So, more efficient and more popular.
This can be implemented in two major ways
1. Ethernet 2. Token Passing Ethernet
Ethernet is the name of a popular packet
switching LAN technology. Ethernet uses a single coaxial cable as the transport medium. All hosts in the Ethernet LAN connect to this cable. A device called transceiver is used to establish the connection between a computer and Ethernet. This transceiver carry the signals. It contains analog circuits for interfacing with the cable and digital circuits for interfacing with the host. It able to determine when another host is using the cable i.e. whether the cable is busy. Each host NIC (network interface card) controls the operation of its transceiver using the network software inside the host. At any point of time, the Ethernet bus can be in any of the three states: Bus is idle Carries a legitimate signal i.e. it is busy Carries a erratic signal generated by a collision
The transceiver constantly listens to the bus i.e. it
measures the signal level to determine the state of the bus. The transceiver does not connects to the host directly. Instead, it connects to NIC, a small card plugged on the motherboard of the host.
It has a small CPU, memory and a limited instruction
set. It performs all the network related functions on behalf of the host.
Each NIC bears a unique hardware address that
identifies a host uniquely. Ethernet properties Broadcast network Every message travels across the entire cable and all NICs receive it, but only the NIC of the correct destination accepts it.
Best effort delivery
Ethernet hardware or software do not take care of missing frame.
Decentralized access control
There is no single centralized authority that dictates if a host can transmit data. Instead, the Ethernet uses the approach of carrier sense multiple access with collision detection (CSMA/CD). Multiple host can access the Ethernet bus simultaneously through their transceivers and can determine if it is idle by looking for the presence/absence of a carrier wave on the bus. For this reason it is called CSMA
When a host determines that no other host is using the
Ethernet, it transmits its own data on to the Ethernet.
When the transceiver of a host begins transmission, the signal
does not immediately reach all parts of the network. It taken some finite time. Therefore, until the signal reaches another host continues to believe that cable is idle. Thus, two transceiver believe that Ethernet is free for transmission and can transmit data almost exactly at the same time. When this happens, the electrical signals of the two transmissions intermingle and neither remains a meaningful transmission. Such incidents termed as collisions, which produce erratic signals, which the transceiver of both the sending hosts can detect. To resolve collisions, while a host transmits the data, the transceiver of the host continues to listen to the Ethernet to see is a collision has occurred. If it has, the transceiver informs its NIC about it. NIC stops wait for some time to retransmit the data. However, as soon as it detects a collision, it generates a specific jamming signal across the bus, informing all the nodes of the collision. The NIC of the other node trying to send some data also detects the jamming signal and backs off. now, both the nodes wanting to send data wait for a while. The question is : How long should they wait? The Ethernet standard specifies a binary exponential back off policy where a sender waits for a random time after a first collision, twice as long if a retransmission also results in a collision, four times as long if the retransmission also results in a collision and so on. Ethernet Frame Format
Preamble Destination Source Frame Frame data CRC (4
(8 bytes) Address (6 Address (6 Type (64-1500 bytes) bytes) bytes) (2 bytes) bytes) Token ring Token ring network is based on ring topology A token ring network employs a mechanism called token passing. All host on a token ring share the same physical medium, as all the hosts are arranged to form a circular ring. The sending computer transmits a frame, which travels across the ring. Each host on the ring check the destination address and if it is not meant for it, forward it along. Only the actual destination makes a copy of it i.e. accepts it. In last frame comes back to the sender. In this, only the host which have a token will be able to send frame. However, at the destination node, before it forwards a frame, that host checks the CRC to ensure that there are no errors and then not only accepts the frame, but also changes a flag bit in the frame to indicate the receipt of a correct frame. The frame comes back to the sender, the sender can check the flag bit to verify if the frame was received by the destination successfully, or if there were any errors during transmission. Such an acknowledgement scheme is implemented only in the token ring, unlike Ethernet, where there is no such theme. Medium Access Control (MAC)
Token ring does not employ CSMA/CD.
The token ring hardware ensures that the permission for data transmission is granted to each host on the token ring in turn. The hardware facilitates this coordination by using a special three byte frame called token A token contains a bit pattern that is completely different from any other data frame, so that both can be easily distinguished from each other. Token ring frame format
of optical fiber could support high data rates (100 mbps) These days copper wires support such rates. So, the copper version of FDDI is called CDDI. FDDI uses glass fibers for data transmission therefore encodes data bits in the form of pulses of light. FDDI properties
1. Token passing for media access control
2. Self Healing Mechanism This is possible by second ring when a network error occurs, or a host is down. In such a case, the NIC uses the second ring, which is used as backup for such failures for data transmission. This is called loopback. FDDI Frame