Computer networks that usually cover a limited range, for example within the boundary of a building or a campus. A computer network is two or more computers that communicate with each other. The primary usage of local-area networks (LANs) is the sharing of hardware, software, or information across a network with a limited geography (Fig. 1) and usually on telecommunications lines and equipment owned and/or operated by the local-area network's organization. Resource sharing provided by local-area networks improves efficiency and reduces overhead. See also: Computer; Electronic mail; Multimedia technology
A local-area network can be thought of as similar to a telephone network which uses phone wires installed in a star topology. A local-area network must also use some transmission medium installed with a selected topology.
Four basic types of media used in local-area networks are coaxial cables, twisted-pair wires, fiber-optic cables, and wireless. Each medium has its advantages and disadvantages relative to cost, speed, and expandability. Originally, most local-area networks used coaxial cables. A coaxial cable consists of one or two conducting wires encapsulated by several layers of insulation and shielding (Fig. 2a). Coaxial cables provide high speed and low error rates. Twisted-pair wires (Fig. 2b) are cheaper than coaxial cables, can sustain the speeds common to most personal computers, and are easy to install. Fiber-optic cable (Fig. 2c) is the medium of choice for high-speed local-area networks, operating at speeds of 100 megabits per second or higher. A fiber-optic cable uses light pulses to represent data. Because light signals are not distorted by electric or magnetic fields, fiber-optic cables provide excellent error characteristics. The disadvantages of fiber-optic cables include their high cost and the difficulty of adding or removing stations. Messages in wireless local-area networks are transferred through the air as radio waves rather than through a conductive cable or wire. Hence, wireless local-area networks have the advantage of expandability. See also: Coaxial cable; Communications cable; Fiber-optic circuit; Optical communications; Optical fibers; Wi-Fi
The topology of a local-area network is the physical layout of the network. For wired local-area networks, there are four basic topologies: bus, ring, star, and mesh (Fig. 3). For the bus topology (Fig. 3a), the medium consists of a single wire or cable to which nodes are attached. A message transmitted over a bus propagates in both directions along the bus, passing each tap until it is finally absorbed at the ends. In a ring topology (Fig. 3b), the medium forms a closed loop, and all stations are connected to the loop or ring. Data propagate in one direction. The star topology (Fig. 3c) has a central node to which the computers and devices are separately connected. All the network traffic passes through the central node. In a mesh topology (Fig. 3d), messages are transmitted through several switches that connect nodes. The asynchronous-transfer-mode (ATM) technology favors the use of mesh topology. Modifying and combining some of the characteristics of these basic network topologies may result in hybrid topologies, which often provide greater network efficiency.
Circuit versus packet switching
There are a number of ways in which nodes can communicate over a network. The simplest is to establish a dedicated link between the transmitting and receiving stations. This technique is known as circuit switching. In the topologies outlined above, this would involve dedicating the medium to communication between the nodes. However, data communications over a local-area network tend to be bursty; that is, they are characterized by periods of intensive data transfer, followed by lulls. During this period of inactivity, precious bandwidth capability is being wasted.
A better way of communicating is to use a technique known as packet switching, in which a dedicated path is not reserved between the source and the destination. Data are wrapped up in a packet and launched into the network. In this way, a node only has exclusive access to the medium while it is sending a packet. During its inactive period, other nodes can transmit. Thus, the available bandwidth can be better utilized. The problem of sharing access to the network is reduced to establishing some rules that will allow each node on the network to launch its packet in a fair manner. This is achieved by access protocols, discussed below. See also: Packet switching
A typical packet is divided into preamble, address, control, data, and error-check fields. The field of the preamble contains some bit sequence that never occurs in normal data. It serves to inform all other nodes on the network that transmission of a packet has begun. The address field contains addresses of both the sender node and receiver node. This allows the nodes on the network to recognize where the packet comes from and who should receive it. The control field identifies the purpose of the packet as normal data transmission or a special management purpose. Although a packet usually contains normal data, it can also be sent for a network management purpose (for example, to discover the status of nodes and medium). The data field contains the actual data to be transferred. Although the error rate in a local-area network is typically very low, a packet usually has an error-check field that allows the receiver node to detect or recover transmission errors. The sizes of all these fields are usually fixed except for the data field, which may have a variable length, depending on the amount of data to be transferred.
An access protocol is a set of rules observed by all the nodes in a local-area network so that one node can get the attention of another and its data packet can be transferred. Two example protocols are carrier sense multiple access with collision detection (CSMA/CD) and token passing.
With the CSMA/CD protocol, a node that wants to transmit its data must first listen to the medium to hear if any other node is using the medium. If not, the node may transmit immediately. However, while the transmission is taking place, the transmitting node must continue listening to ascertain if anyone else has begun transmitting. This is possible because, with the propagation delay over the medium, two or more nodes could simultaneously decide that the medium is idle and start transmission virtually at the same time. Then, data transmitted by the nodes collide and get garbled. Hence, if the transmitting node detects that someone else is also transmitting, the node aborts its own transmission, waits for a random amount of time, and then restarts the process until its data transmission succeeds.
With the token-passing protocol, the right to transmit is granted by a token, a predefined bit pattern that is recognized by each node. The token is passed for one node to another in a pre-determined order. For example, in a ring network the token passes from one node to its adjacent node. When a node obtains the token, it has two options: it can transmit a message, if any, or it can pass the token to the next node. If the node transmits a message, then at the end of its transmission it releases the token to the next node. When a token network starts up, an initial token must be generated, and when the token is damaged because of node and transmission errors, a token must be regenerated. This function is usually carried out by a designated monitor node. If the token does not arrive at the monitor node within a certain time, it generates a new token. This technique works because the token traveling time around the network is usually predictable.
The success of the local-area network depends on how application and system software interacts with the hardware in setting up the communication capability. Application software is designed to solve users' problems. It is assisted in this goal by supporting system software. Local-area network system software is essentially an extension of the operating system. It insulates applications from hardware details and carries out hardware-oriented local-area network tasks, such as interfacing to the medium, directing printing jobs, and sending and receiving service requests. That is, the applications can make requests for services, and system software contains the logic to carry out these requests for a specific type of hardware. See also: Operating system; Software
The primary usage of local-area networks is resource sharing. The resources can be information, such as data files, multimedia files, email, or software. They can also be hardware devices such as printers, scanners, or storage devices. Resource sharing is realized by services provided by the local-area network. Usually some nodes in the network function as file servers. File services provide the central repository of data or application programs, and offer consistent and easy-to-use file access to all the users and applications. Management services provide uniform interfaces so that a small central staff can monitor and control end-user activities, network changes, security, and system performance across an entire corporate network. Security services aim at establishing consistent security across a variety of network resources. Usually these services are integrated with other network services and are enforced transparently with every network activity. Time services synchronize enterprise-wide services and applications, making network events temporally consistent. See also: Computer peripheral devices; Computer security
While the packet-switching method is cost-effective and acceptable for many common applications such as email and file transfer, it does not provide a prior guarantee regarding the quality of service that an overlying application may receive. Hence, it is insufficient for applications such as multimedia and teleconferencing. See also: Teleconferencing
Connection-oriented communication technology improves the packet-switching method. A connection is an abstract communication service provided by the network to a particular application. It can be considered as a two-way contract: The communicating applications specify the characteristics of traffic which they may generate (say, the maximum rates), and the network agrees to provide the requested quality of service to the applications. The network may disallow a connection if it cannot provide the requested quality-of-service guarantees. Thus, a connection may be viewed as a virtual link that is dedicated for the use of an application and that has a certain traffic-carrying capacity. Connection-oriented communication technology exploits the advantages of packet switching while overcoming its weaknesses.
Development of networks
A representative of the first generation of local-area networks is the Ethernet, a CSMA/CD bus. It uses coaxial cable with a maximum data rate of 10 megabits per second. The best-known second-generation local-area network is the Fiber Distributed Data Interface (FDDI), a token-ring network using optical fibers with a data rate of 100 megabits per second.
Another local-area network technology, which uses the asynchronous transmission mode, is ATM LAN. An ATM LAN is a mesh-based network using the optical-fiber medium with data rates of the order of 150 megabits to 10 gigabits per second. With the high bandwidth provided by ATM LANs, many applications such as integrated real-time data and multimedia services became possible.
The basic premise of virtual local-area networks is to separate logical work groups from physical constraints. This enables devices that are connected to different components throughout the network to be logically grouped into communities of interest. Thus, virtual local-area networks are built by administrative and functional needs rather than attachment. Virtual local-area network technology provides scalability, flexibility, and policy-based management. See also: Data communications