Basic Data Communications Series
1.0 Networks - Introduction
Networks: An interconnected collection of autonomous users.
Networks are required in order to connect a set of nodes (users – usually computers) together in order that communication may be established between each participant. The set of nodes may be fixed or variable and may cover small geographical areas, whole continents or the globe.
Hence the scope of the network may be as follows:
A Local Area Network (LAN) which may be limited to a single building or room in a building.
A Wide Area Network (WAN) which may cover a country or continent.
A Metropolitan Area Network (MAN) which lies between that of a ‘LAN’ or ‘WAN’ i.e. spanning a small town or a city.
Also there are three basic types of Networks as follows:
Circuit (line) switched
Message switched
Packet switched.
Local Area Networks
Local Area Networks consist of networks of interconnected computers, printers, workstations and other intelligent systems within a building or small group of buildings. The upper limit in size for a LAN is approximately 10km and a lower limit of about 1m. LANs are relatively fast compared to WANs and involve bit speeds typically from 1 to 10 Mb/s - although it is possible to achieve speeds of 100 Mb/s using fiber optic technology. Also the transmission error rates are much lower.
LANs use shared media networks of varying topologies, for example a bus structure, which consists of a single cable connecting the various nodes together. This configuration is also known as a multi-drop line. Another topology is that of a Star or radial structure in which the various nodes are connected to a central controller. A third structure is that of the ring or loop in which the various nodes are connected via a closed loop. Each member of the ring passes data from its immediate neighbor to the next node on the loop; thus data passes in one direction through each node.
Since each node shares the media, this leads to the notion of a multiple access channel. The rules, which control the access to the channel, are known as multiple access protocols. The multiple access channels allow broadcast messages to be sent in which every node connected to the channel may potentially receive any transmitted message. Whereas a unicast message is sent from a transmitter to a single receiver.
Wide Area Networks
Wide area networks operate over large areas covering countries or continents and as a result are subject to much lower performance in terms of bit rates, error rates and propagation delays. Typical bit rates are less than 1M/s. Also the network will no longer be under the ownership and control of a single organization. Some networks will be privately owned whilst others are publicly owned and thus available for a fee to all who want to use them.
1.3 Circuit Switched
A circuit switched network establishes communication via a private transmission path or circuit between two users or group of users. Once the path is established (switched) it is held for as long as required and other users are blocked from joining in. Thus the ubiquitous telephone network is an example of a circuit switched network. Once the communication is established between two users the ‘line is engaged’ until released. Circuit switched networks are the earliest form of networks.
Circuit switched telephone networks are Wide Area Networks and have traditionally been analogue in nature i.e. real time voice transmitted between users using electromechanical switching and routing i.e. the classical Strowger exchange. Since the 1960’s however digital networks have been increasingly introduced with digital switches using computers introduced in the 1970’s.
1.4 Message Switched
A distinctive feature of computer communications networks is that traffic is bursty and is separated by long intervals of no transmission. So if a line were to connect two computers together it may be used for only 1 per cent of the time and therefore uneconomical.
In message switching no such complete connection is needed and facilities are shared. Instead a computer will transmit a message to a switching center were it will be stored in a buffer until such time that it can be routed to the destination computer. This has the disadvantage that queuing delays will be introduced due to the buffering but the advantage those long distance fast expensive lines are left idle for shorter intervals. Since the system involves storing and forwarding of messages the network is refereed to as a store and foreword network.
1.5 Packet Switched
Message switching has features that are less than desirable as follows.
The bursty messages may be long or quite short and to accommodate the long messages large buffers will be required. However during short bursts the expensive hardware of large buffers is wasted.
Even if a line is available, a switching computer will not begin to relay a message until the complete message is received. Hence if the message is Tm long and is relayed R times then a delay of R.Tm is produced ignoring any waiting (queuing) time.
It may be that a link between centers is needed only for a short time to transmit a short message, however the short message may have to wait a long time while this same link is transmitting a long message. It would be useful if the long message could somehow be interrupted in order to send the short message on its way, however this is not easily implemented in a message switched system. Hence a short message my be delayed for a period which is long compared to the message length while the system is tied up.
In a packet switched network on the other hand, whilst the facilities are shared, the communication messages are divided into blocks of data called packets from source to destination. The message is then transmitted packet by packet. Like the message switched network, packets are stored in buffers at the subsystem nodes and then forwarded. A packet switching network is thus a store and foreword network also.
One disadvantage with a packet switched system is that packets containing elements of a message may be routed via different paths. Therefore the delays may be different and the packets of a message may arrive out of order. Hence extra information must be included with each packet (known as overhead) in order to route the packets and reassemble the messages in the correct order. Packet switching however is in general more economical and gives better performance than message switching. Packet switched networks are relatively new having been developed since the 1960’s with the advent of the digital computer.
1.6 Integrated Services Digital Network
Finally there are Integrated Networks which is a general term for any network that combines all of the above networks in order to give a generic service connecting homes or businesses together with service companies using a digital network.
Thus an ISDN may contain Local Area Networks which are connected to Wide Area Networks via gateways. These in turn may have high speed fiber optic links to other WAN’s or radio links via satellites to other counties or continents. The traffic handled may be a combination of data, voice and video.
1.6 Open Systems Interconnect Reference Model
So far nothing has been said about the data or stations that transmit the data over the network.
Historically the data communications industry evolved slowly from different companies and countries designing and producing different equipment with very little attempt at standardization. Also the various parts of the system from the computers and operating systems down to the network transmission equipment and network system components themselves were all seperatlty developed. This lead to the inevitable interfaces problems and non-compatibility between the various manufacturers equipment. This naturally led to a layered type of systems architecture was the individual system functionality is separated off into a set of functional layers.
One of the first layered architectures to be developed was IBM’s System Network Architecture SNA. This was developed in the late 1960’s and the first products delivered around 1973. The Digital Equipment Corporation developed DNA and Boroughs BNA, all of which are incompatible. This state of affairs lead the International Standards Organization (ISO) to launch an intensive effort to develop a worldwide communications architecture standard that would allow systems to communicate openly. This effort was begun in 1978 and was completed in 1980 in a proposal for the OSI reference model.
This proposal incorporated the concept of seven layered communication architecture borrowing heavily from SNA and other computer manufacturers and was finally approved as a standard in May 1983.
The seven layers of the model are as follows:
Application Layer
Presentation Layer
Session Layer
Transport Layer
Network Layer
Data Link Layer
Physical Layer
The top four layers of this archetecture are normally resident in the host computer or other intelligent system that requires to communicate via the network to another host or system.
The bottom three layers are normally resident in the communication equipment that controls the access and transmission/ reception of the data.
The individual functions are as follows.
1.6.1 Application layer
The application layer is the highest layer and is concerened with the semantics of the data to be exchanged. That is that the two host communicating understand each other and provides a virtual terminal environment to the user completely isolated from the details of the actual data transmission and communication. It may for example be part of the computers operating system.
1.6.2 Presentation layer
The presentation layer isolates the application layer from the differences in syntax of the data actually transmitted. The presentation layer may therefore perfrom any data compression and data encryption required for security and efficiency of transmission. In addition the presentation layer must initiate and terminate connection, manage layer states and handle error correction and control.
1.6.3 Session layer
The session layer controls the setting up of the session between users and to identify themselves when wanting access to the network. It controls the actual interchange between the users, the modes of operation (duplex, half duplex etc), any synchronisation and repeat of messages due to errors etc. In summarry it controls the session.
1.6.4 Transport layer
The transport layer isolates the session layer from the details of the actual network, wether it be a LAN or WAN, packet switched or circuit switched. If the network is a packet switched network it ensures that the the packets are delivered in the correct order, it provides buffereng and flow control and correct connection from the host in one network to the host in another network. It is responsible for delivering Quality of service: transit delay, erroir rate and transfer failure probability.
1.6.5 Network layer
The network layer is concerened with routing to control congestion in the network and interfca emessage processor. It is also reponsible for access control to the network media and thus impliments the various access protocols (eg X25 or CSMA/CD etc) controlling collisions, data acknowledgements and repeats of the data frames transmitted.
1.6.6 Data Link layer
The data link link layer is responsible for setting up the data and communications between each of the Data Communications Equipement (DCE). The data Link Layer therefore sets up the message frames containing data and control bits for the correct operation of the DCE. It is responsible for setting up the correct connection to the media required by the Network Layer protocol and the correct signalling speed for the session.
1.6.7 Physical layer
The physical layer contains the physical hardware for modulating, demodultaing, transmitting, receiving the actual data bits onto the physical medium. The physical layer will also contain the necessary control logic to set up the handshake between the mediums control and status lines for correct operation of the medium.
