Retro-Technics • author: a3310i • last modified: 2022.01.03 •

Local Area Networks (LAN's) allow microcomputers to communicate with each other over distances of no more than a few kilometres. Originally designed for use with office equipment, later with personal computers, they are now even making their way into robotics.
As a rule, the network topology can be defined as star, ring or bus. Large computers and minicomputers from companies such as IBM, Honeywell, Unisys, Hewlett-Packard form star-shaped networks. But such networks, although very popular, cannot always be counted as LANs. According to the local area network principle, each node (computer or intelligent peripheral device) must be able to connect to any other node without intermediaries, using a single cable. These conditions are fulfilled by bus and ring shaped networks. Teletransmission networks, on the other hand, do not fall into this category, not providing connections between nodes through modems and telephone networks. Local area networks can be divided into proprietary, dedicated and general-purpose networks. Proprietary networks are owned by a single company and may be subject to patents and copyrights. Although these networks meet the requirements for LANs, they tie the owner to a single hardware and software provider. Dedicated networks have a very specific application, there may be a network to connect robots and other devices from different manufacturers in a common car assembly line. The most common are general-purpose networks that can cooperate with a large number of devices and different software. Ethernet-type networks, already accepted as valid principles, fall into this category.

[001] LAN Ring architecture.

Information in the network is transmitted in portions, so called frames. Each frame usually contains synchronization bits, sender and receiver addresses, data type, data and a check sum, sometimes extended by a self-correcting code.
As only one device can send information in a network at a given time, the rules for allocation of the cable connecting the nodes (computers) are different. In 1980, the international organisation IEEE set standards for three different methods of cable usage: token passing ring, token passing bus and carrier sensing multiple access CSMA. All the methods are based on each node being able to transmit and receive information at certain intervals. The first two methods work with a frame circulating in the network. If node A wants to connect to node C, the following operations will be performed sequentially: node A intercepts a free frame, fills it with data and addresses it to C; the frame starts circulating in the network again, it passes node B because the recipient's address does not match; node C recognizes that the frame is addressed to it and receives the information indicating this fact in the frame; node D passes the frame, while node A intercepts it again; data verification takes place and checks if the data has been received. If everything is in order, the frame is released and let back into the ring or bus. Although the two methods are topologically different, the idea of transmitting information using a tagged frame remains the same. The arbitration of access to the frame via the tag is deterministic, every so often each node gets a chance to send information. Another method, CSMA, captures the image of a room filled with people, each of whom has something to say, but first has to check that someone else is not speaking. Each node can drop a data frame into the network and wait for an echo. If the echo is identical to what was sent, the sending node knows that the frame was received. If the echo differs, that node and at least one other node know that a collision has occurred. After a certain delay, different for each node, the attempts are repeated.

[002] Lan bus architecture.

The individual nodes of the network can be connected by ordinary wire, coaxial cable or fibre optic cable. These allow information to be transmitted at increasingly higher speeds. Fibre optic is the most expensive but offers the fastest connectivity. The electrical signals of each node are converted into light pulses and transmitted in this form via optical fibre. The receiving device converts them back into electrical impulses understandable for the computer device. Under certain conditions this is necessary because of resistance to electromagnetic interference. The cable itself should be laid out in such a way as to enable the network to be extended with new nodes in the future. Links realised on fibre optic cables have many advantages. Apart from speed, there is also security. It is virtually impossible for unauthorised persons to connect to such a network via a cable. It is important in applications where information is secret - government systems or banking.
Until recently, a computer was a system in itself. Now, the system is the network and the computer has become only one of its elements. Further development of computer networks is therefore inevitable.