OSI is a conceptual model that abstracts the nuances of specifics and technology of the telecommunication system and at the same time defines and standardizes its communication functions. The model divides a communication scheme into hierarchical abstraction layers to achieve the functionalities necessary for data

Twisted copper pair cable

FIGURE 3.6 Twisted copper pair cable.

ISO/OSI protocol stack - layered architecture

FIGURE 3.7 ISO/OSI protocol stack - layered architecture.

communication. It aims to facilitate interoperability of different communication schemes with standard communication protocols. The model has seven layers as shown in Figure 3.7.

Physical layer: The physical layer, hierarchically the lowermost layer of the model, realizes actual physical transmission and reception of raw information in the form of signals between a network device and the transmission medium. The layer functionally implements the system or devices that convert the digital bits representing data into electrical, radio or optical signals depending on the choice of the medium of communication. Various specifications and functionalities of the layer specify characteristics such as data rates, signal modulation strategy, power and voltage levels of the signal, timing of voltage changes, transmission range, channel access procedure and physical devices like connectors.

Data link layer. The data link layer provides functionalities for creating, maintaining and destroying links between neighboring nodes for node to node data transfer. The layer realizes features for framing, medium access, error control and flow control in addition to link management. Framing is the process of dividing the complete packet of data transferred from the network layer to smaller fragments so as to fit the buffer space in the physical layer. Medium access control is the process by which decisions as to how and when a node should access a medium for transmission or reception of data is decided. Error control includes error detection and possible error correction or retransmission. Flow control involves synchronized operation, in terms of data transfer rate and frequency of data message flow, among neighboring nodes. IEEE 802 divides the data link layer into two sublayers:

  • 1. Medium access control (MAC) layer - responsible for medium access control and assigns MAC or physical address.
  • 2. Logical link control (LLC) layer - responsible for framing, error control, flow control and link management.

Network layer. The network layer provides functionalities for sending data available in this layer in the form of packets from one node to another node in the same or different network. The process of sending packets from a source node to a destination node by providing the data and the addresses is called routing. A packet can have varying size depending on the amount of data pushed down from the transport layer. Another important functionality of the network layer is logical or IP address allocation, management and deallocation.

Transport layer: It provides functionalities for transferring data sequences of varying length from a source node to a destination ensuring quality of service. This is achieved by having protocols for end-to-end delivery of packets, flow control and error control. The two main schemes employed in transport layer are transmission control protocol (TCP), which is connection oriented and hence acknowledgement based, and user datagram protocol (UDP), which is connectionless and uses no acknowledgement. The transport layer does segmentation and reassembly of messages in which long messages from the application layer are divided into smaller messages; whereas smaller packets from the network layer are combined to form larger messages. The data is represented in the form of segments or datagram.

Session layer: The session layer provides functionalities for establishing, managing and terminating the connections between applications on the local as well as remote devices. It facilitates procedures for starting check pointing, suspending, restarting and terminating a session in full-duplex, half-duplex, or simplex operation modes.

Presentation layer: The presentation layer provides functionalities to set up a link between application layer services of the source and destination devices; it also establishes a mapping for the application layer services to use varying syntax and semantics. This layer represents data as presentation protocol data units which are packed into session protocol data units and sent down the OSI protocol stack. The presentation layer acts as data translator between the application and the network layers’ data formats.

Application layer: The application layer, the topmost layer of OSI stack, is the direct user interface with the communication system. This layer provides functionalities that enable interaction of user side software applications and systems and software that implement a communicating component. This is achieved by identifying communication peers, evaluating resource availability, and providing communication synchronization.

TCP/IP Model

The TCP/IP stack has only four hierarchical layers in achieving all the communication functionalities necessary for networking multiple devices for fruitful exchange of data. The functionalities of the physical and data link layers of OSI are provided in TCP/IP in the host to network interface layer. The functionalities of network and transport layers of OSI are provided in TCP/IP in the internet and transport layers respectively. Functionalities of the top three layers of OSI - namely, the application, presentation and session - are achieved using the application layer in TCP/IP model (Figure 3.8).

Network Topologies

The main purpose of any network technology is to realize multipoint to point, point to multipoint or multipoint to multipoint connectivity and message passing in addition to the straightforward point to point communication. Ethernet in this effort allows the following topologies for interconnection of devices for realization of a network (Figure 3.9).

Bus topology: A single central cable acts as the shared medium in the bus network and interconnects all members of the network. Any member node of the network can transmit a data packet to any other node by sending the data to the bus. Though all connected nodes can receive the data, only the intended receiver will process it.

Ring topology: In the ring network topology, every member node is connected to two other nodes such that all of them together form a ring, as the last node will be connected to the first one. The messages can travel in clockwise or anticlockwise direction.

Star topology: Every member node is connected to a common node forming a star with one central point and several end points, in star network topology. The communication happens between the central node and the end nodes; there is no provision for direct communication between the end nodes.

Mesh topology: All member nodes in mesh network (commonly known as the full mesh) topology are connected to each other in highly redundant fashion. When some member nodes are connected to multiple nodes, but not to all other nodes of the network, it is called as partial mesh topology. This topology provides multiple paths for data flow between one node and another.

Network topologies in Ethernet

FIGURE 3.9 Network topologies in Ethernet.

Tree topology: A tree or hierarchical topology is a specialized network topology in which every parent node has two or more child nodes connected to it and a single root node forms one end point of the network. It is a combination of bus and star topologies. This is a hybrid topology and there exist large possibilities for creating various hybrid topologies by combining the above-mentioned topologies or improvising on one.

Network Devices and Terminologies

Any communication network consists of a large number of nodes witli multiple functions to fulfill the requirements for which the network is constituted. The following are some of such specialized functional nodes in ethernet network (Figure 3.10).

Repeater: A repeater is a device that is used to boost the power level of the signal by regenerating it when the signal power level becomes too weak after long distance transmission. It is a two port device that operates in the physical layer and helps to increase the range of the network.

Hub: A hub is basically a multiport version of a repeater. The difference between hub and repeater is the multiple input/output lines available in a hub. Any signal that is received in any line of a hub is regenerated and sent to devices connected to all other lines of the same hub. Both hub and repeater are not intelligent devices and do not read the data frames, instead they just repeat sending these to all lines except the incoming one.

Bridge: A bridge can be considered as two port intelligent repeater as it can read the MAC address and pass the incoming message to the outgoing line.

As it works looking at the MAC address it is operating in the data link layer.

A bridge can interconnect two similar networks.

Switch: A switch is a multiport bridge with a buffer that can perform error checking before forwarding the data according to the MAC address. It can interconnect various similar networks and traditionally works in data link layer.

There are switches that incorporate some routing strategies and hence work in the network layer.

Routers: A router is a device that operates in the network layer which routes data packets based on the IP address. Routers interconnect multiple similar networks together and dynamically update routing table based on which decisions are made as which packet is to be forwarded to which node or network.

Gateways: Gateway is a network device used to interconnect two or more networks operating on different protocols. These are basically protocol converters and can operate in any layer of the protocol stack and act as a bridge between multiple networks by interpreting the address on the packet and forwarding to the correct destination node or network.

MAC Address

MAC address is a 48 bit unique identifier assigned to all network interface controllers of all network devices. MAC address is usually assigned by device manufacturers and hence is referred to as burned-in address or physical address. The address includes an organizationally unique identifier represented by the first three octets and a network interface controller address represented by the last three octets (Figure 3.11).

Addresses can be administered universally or locally. In the universal case, the manufacturer assigns the unique address. The first three octets are called the organizationally unique identifier and are used to identify the manufacturer. The remaining three octets can be assigned by the organization in any manner they choose, but the entire number must maintain its uniqueness. A locally administered address is usually assigned by the network administrator or user by overriding the original physical address. If the second least-significant bit of first octet is zero, the address is

universally administered and if this bit is one, the address is locally administered. If the least-significant bit of the first octet is set to zero, then the frame is supposed to reach only one receiving network interface card (NIC) realizing unicast addressing and if this bit is set to one, then it is multicast addressing.

IP Address

Internet Protocol (IP): Internet protocol address or the logical address of any device on a network following internet protocol shows host or network identification and location. Unlike MAC address, IP address is not unique globally. IP address specifically identifies the network in which a node is present and also the location of the specific node in the network. IP address can be assigned to a device in a network either by the network administrator which can be static or dynamic or by the user which is static. It is used in the network for routing the packets so as to guide it from source node to destination node. The router or gateway has a routing table which has IP addresses to help routing. Two versions of IP addressing are in use today.

IP version 4: IP version 4 uses 32 bit addressing which allows 232 devices to be part of the network with unique address. In this, some address blocks are reserved for private networks and multicast addresses. The IP address consists of two parts: the network identifier and the host identifier. There are two defined styles for IPv4 addressing - namely, classful and classless inter domain routing. In classful addressing, network classes are created for address definition and one, two or three octets are allocated for network or subnet addressing and the remaining for host addressing. As this scheme limits the possible maximum network addresses to 221, a new addressing scheme was developed that allocates address to networks and end users or hosts on any address bit boundary. The 32 bits of IPv4 is usually represented in four octets separated by dots and each octet value is written in decimal form (Figure 3.12).

IPv6 address representation

FIGURE 3.13 IPv6 address representation.

IP version 6: This style of addressing uses 128 bits thereby allowing 2128 devices with unique addresses to be part of the network. Thus, its address space is large compared to that of IPv4. IPv6 is represented as eight groups with values separated by full colons while each group has 16 bits written in hexadecimal form in lower case. IPv4 addresses are mapped to IPv6 by making the least significant 32 bits of IPv6 the same as IPv4 address and the remaining most significant bits in IPv6 format. IPv6 uses the last 64 bits always as interface address (Figure 3.13).

< Prev   CONTENTS   Source   Next >