Networks — Edexcel GCSE Computer Science

A network links two or more computing devices so they can exchange data and share resources. Computers are connected because sharing is cheaper and more flexible than giving every user a separate copy of everything. A school office can send a document to a networked printer instead of buying one printer per desk, and staff can open the same shared folder rather than swapping USB sticks. Connections also let users communicate through email, messaging and video calls, and let a central server hold files, accounts and backups that many machines reach at once. Central management means one administrator can update software, set permissions and monitor use across all devices, which is faster and more consistent than visiting each machine. Networks also allow users to log on at any connected computer and still reach their own files.

Networks are classified by the area they cover and who controls them. A local area network, or LAN, covers a small area such as one building or campus, for example a school where computers, printers and a file server are linked by the school's own cabling and wireless access points. A wide area network, or WAN, covers a large geographical area and typically links separate LANs using infrastructure that the organisation does not own, such as leased telecoms lines. A bank with branches in several cities runs a WAN so each branch LAN can reach the central accounts system. The key distinctions are scale, ownership of the communication medium and the hardware needed: a LAN usually uses owned cabling, switches and wireless access points, while a WAN depends on external links and routers.

The internet is a network of networks: many independently owned networks are joined so that any device can exchange data with any other. Each device needs an address so data reaches the right destination. An IP address identifies a device or interface on a network; a public address is globally unique, while a private address is reused inside a local network and translated at the boundary. Data is split into packets, each carrying a destination IP address. Routers are the devices that read that destination address and forward each packet along the best available path, hop by hop, until it arrives. Because routers choose routes dynamically, traffic can take different paths at different times, and a packet may pass through several routers. A router also connects separate networks, for example a home network to an internet service provider.

Wired and wireless connections differ in how they carry data, and this affects performance. Wired links, such as Ethernet, use a physical cable, which usually gives higher speed, greater bandwidth and lower latency, but the device is limited by the cable length and must be plugged in. Wireless links, such as Wi-Fi, carry data through the air, so devices can move within range, but speed and bandwidth generally fall as distance increases or as walls and interference weaken the signal. Latency is the delay before data starts to arrive; wired links often have lower latency, while wireless links can suffer more delay and variation. Bandwidth is the amount of data that can be transferred per unit time, so a higher-bandwidth link can carry more data at once. Range is how far a signal can travel while remaining usable.

Network speed is a rate: how many bits pass a point each second, written in bits per second (bit/s). The prefixes scale by 1000: 1 kilobit per second (kbit/s) = 1000 bit/s, 1 megabit per second (Mbit/s) = 1000 kbit/s = 1,000,000 bit/s, and 1 gigabit per second (Gbit/s) = 1000 Mbit/s = 1,000,000,000 bit/s. File size is a quantity of bits, so time = file size ÷ transmission rate. Convert units before dividing: 2 MB = 2 × 8 × 1,000,000 = 16,000,000 bits; at 8 Mbit/s, time = 16,000,000 ÷ 8,000,000 = 2 s. Rearranged expressions give file size = rate × time and rate = file size ÷ time. Keep bits and bytes distinct: 1 byte = 8 bits.

A protocol is an agreed set of rules that lets devices communicate. Without shared protocols, different hardware and software could not interpret each other's signals, so protocols are needed for reliable, standardised exchange. Ethernet defines wired local-network communication; Wi-Fi defines wireless local-network communication; TCP/IP governs how data is packetised, addressed, routed and reassembled across networks, with TCP checking delivery and IP handling addressing. HTTP transfers web pages, HTTPS does the same with encryption for security, and FTP transfers files. For email, SMTP sends messages from client to server and between servers, POP3 downloads messages to one device, and IMAP keeps messages on the server so they can be synchronised across devices.

The TCP/IP model splits network communication into four layers, each with a distinct job. The application layer creates data for a service such as sending an email or requesting a web page. The transport layer breaks that data into segments, adds port numbers and, with TCP, sequence numbers and checksums so lost or corrupted segments can be resent. The internet layer adds source and destination IP addresses so routers can forward packets across networks. The link layer frames the data with MAC addresses for delivery on the local network segment. At the receiver the process reverses: link, internet, transport, then application. Each layer adds or removes its own header, which is encapsulation and de-encapsulation. A useful method is to trace one message down the sender's stack and up the receiver's stack, naming the header added at each layer.

A network topology is the arrangement of nodes and connections. In a bus topology all nodes share one backbone cable; data travels along the backbone and terminators absorb signals at the ends. It needs little cable but a break in the backbone can stop the whole network and performance falls as traffic increases. In a star topology every node connects to a central switch or hub; a failed node or cable affects only that node, but the central device is a single point of failure. Star is common in schools and offices because it is easy to add nodes and troubleshoot. In a mesh topology nodes connect to multiple other nodes; a full mesh gives every node a direct link to every other node, so it is highly resilient but needs many cables and ports. Partial mesh connects some nodes by more than one path. A useful method is to compare cost, fault tolerance and ease of adding nodes.