Fundamentals of communication and networking.

HookWhy your message survives even when the network is on fire

ModelCommunication basics: serial, parallel, and the bit-rate trap

Data can be sent one bit at a time down a single wire (serial transmission) or several bits at once down parallel wires (parallel transmission). Parallel is faster over very short distances but suffers from skew (bits arriving at slightly different times) and crosstalk over longer runs, which is why nearly all modern links — USB, Ethernet, SATA — are serial. Transmission also has a direction mode: simplex is one-way only (a broadcast signal to a TV), half-duplex is two-way but only one direction at a time (a walkie-talkie), and full-duplex is two-way simultaneously (a phone call).

The vocabulary that trips candidates is speed. Bit rate is the number of bits transmitted per second (bps). Baud rate is the number of signal changes (symbols) per second. They are equal only when each symbol carries exactly one bit; if a symbol encodes several bits, the bit rate is a multiple of the baud rate. Bandwidth is the range of frequencies a channel can carry, and higher bandwidth allows a higher bit rate. Latency is the delay between sending and the data arriving. A protocol is the agreed set of rules that lets two devices communicate. The exam's standard relationship is: bit rate = baud rate x number of bits per symbol.

ModelTopologies and network models

A network topology is the arrangement of connected devices, and the specification focuses on two physical layouts. In a bus topology all devices share a single backbone cable, tapping into it directly; it is cheap and uses little cable, but the shared cable is a single point of failure and only one device can transmit at a time, so collisions and congestion rise with traffic. In a star topology every device has its own cable to a central switch; it costs more in cabling but is faster (the switch can direct traffic and support simultaneous conversations), more secure, and resilient — one broken cable takes out only its own device, not the whole network. A logical distinction also matters: modern Ethernet is physically a star but historically behaved logically as a bus.

Separately, hosts relate to each other in one of two models. In a client-server network, dedicated servers provide resources (files, authentication, web pages) that client machines request; it centralises control, security and backup but needs costly server hardware and expertise. In a peer-to-peer network every machine is an equal that can both request and provide resources, with no central server; it is cheap and simple for small setups and underpins file-sharing and some cryptocurrencies, but it lacks central security and backup and degrades as it grows. Match the model to the need: a school uses client-server for control; a handful of home PCs sharing a printer can be peer-to-peer.

MechanismWireless networking and sharing the air

A wireless network (Wi-Fi) connects devices by radio rather than cable, which brings mobility and easy installation but weaker security and signal that degrades with distance and obstacles. Every device joins a network identified by its SSID (the network name it broadcasts). Because radio is a shared medium and a transmitting device cannot simultaneously listen for other transmissions, wireless cannot detect collisions the way wired Ethernet does; instead it tries to avoid them using CSMA/CA — Carrier Sense Multiple Access with Collision Avoidance. A device that wants to send first listens; if the channel is busy it waits a random back-off period and listens again. Optionally, for hidden-node situations, it uses RTS/CTS (Request To Send / Clear To Send): the sender asks the access point for permission, and only transmits once the access point broadcasts a Clear To Send, which momentarily reserves the channel and stops two devices that cannot hear each other from talking over the access point at once.

Security on wireless is essential precisely because anyone in range can receive the radio signal. Traffic should be encrypted with WPA2 or WPA3 (the older WEP is broken and should never be used), access can be restricted by MAC address filtering, and the network relies on a shared key or authentication server so that intercepting the signal yields only ciphertext.

ModelThe Internet and how it actually works

The Internet is a global network of networks, and its defining technique is packet switching: a message is divided into packets, each carrying the source and destination addresses, a packet number for reassembly, and a checksum for error detection. Routers forward each packet toward its destination, independently choosing a route based on current conditions, so packets from one message may travel different paths and arrive out of order, to be reassembled at the destination. A gateway is needed where two networks use different protocols — it translates between them so they can interconnect. The alternative, circuit switching, reserves a dedicated path for the whole conversation (as the old telephone network did); it wastes capacity when idle and fails if the path breaks, which is why the Internet chose packets.

Humans use names, not numbers, so the Domain Name System (DNS) exists to translate a human-readable domain such as vofti.com into the numeric IP address a router needs. DNS is a distributed hierarchy of servers; your device asks a resolver, which works down from the top-level domain until it finds the authoritative answer, then caches it. A full web address (a URL) combines the protocol, the fully qualified domain name (FQDN) and the path to a resource — so https://www.vofti.com/learn names the protocol (https), the host (www.vofti.com) and the resource (/learn).

MechanismInternet security: firewalls, encryption and signatures

The first line of defence is the firewall, which sits between a private network and the Internet and controls traffic against a rule set. Packet filtering inspects each packet's source, destination and port and allows or blocks it; stateful inspection goes further, tracking whole connections so it can tell a legitimate reply from an unsolicited packet. A proxy server sits between clients and the wider Internet, making requests on their behalf; it hides the internal addresses, can cache frequently requested pages to speed access, and can filter or log traffic. The threats these guard against include viruses (malicious code that attaches to a host program and spreads when it runs), worms (self-replicating malware that spreads across a network without needing a host), and trojans (malware disguised as legitimate software).

Confidentiality relies on encryption. In symmetric encryption the same secret key encrypts and decrypts, which is fast but requires the key to be shared securely first — the hard part. In asymmetric (public-key) encryption there are two mathematically linked keys: a public key anyone may use to encrypt, and a matching private key that only the owner holds to decrypt, which solves the key-distribution problem and is how HTTPS establishes a secure channel. Reverse the keys and you get a digital signature: the sender encrypts a hash of the message with their private key, and anyone can verify it with the sender's public key, proving authorship and that the message was not altered. A digital certificate, issued by a trusted Certificate Authority, binds a public key to an identity so you can trust that a public key really belongs to the site it claims to.

ModelThe TCP/IP stack and application protocols

Networking is organised into layers, each with one job and each hiding the layers below, so that a change in one layer (say, swapping Wi-Fi for cable) does not force changes in the others. The AQA TCP/IP model has four layers. The application layer is where user-level protocols operate, producing the data. The transport layer (TCP) splits data into segments, numbers them, manages reliable delivery and reassembly, and uses port numbers to identify the correct application. The network (internet) layer (IP) adds source and destination IP addresses to make packets and handles routing across networks. The link layer puts packets into frames for the physical network and adds MAC addresses for the next hop. As data descends the stack each layer adds its own header (encapsulation); at the receiver each layer strips its header off on the way up.

The application-layer protocols to know by name and purpose: HTTP and its encrypted form HTTPS for web pages; FTP for transferring files; SMTP for sending email between servers; POP3 and IMAP for retrieving email (POP3 typically downloads and removes mail from the server, while IMAP keeps it on the server and syncs across devices); and SSH for secure encrypted remote access to another machine. A common exam distinction is that SMTP pushes mail out while POP3/IMAP pull it in, and that HTTPS is simply HTTP running over an encrypted (TLS) channel.

DataIP addressing, subnetting and NAT

An IPv4 address is a 32-bit number written as four dotted decimal octets (e.g. 192.168.1.20), split into a network part and a host part. A subnet mask marks which bits are the network part: performing a bitwise AND of an IP address with its subnet mask yields the network address of the subnet the host belongs to. Subnetting lets an organisation divide one address range into several smaller networks for management and security. Because 32 bits gives only about 4.3 billion addresses (2 to the power 32) — far too few — IPv6 was introduced with 128-bit addresses (2 to the power 128, around 3.4 x 10 to the power 38), written as eight groups of hexadecimal.

Addresses split into public (globally unique, routable across the Internet) and private (reusable inside any local network, never routed on the public Internet). The reserved private ranges are 10.0.0.0-10.255.255.255, 172.16.0.0-172.31.255.255 and 192.168.0.0-192.168.255.255. Three mechanisms make this work in practice. DHCP (Dynamic Host Configuration Protocol) automatically leases an IP address and settings to a device when it joins, rather than an administrator configuring each by hand. NAT (Network Address Translation) lets many devices on a private network share one public IP: the router rewrites the source address (and port) of outgoing packets and remembers the mapping so replies return to the right internal host — which also conserves scarce IPv4 addresses. Port forwarding configures the router to send incoming traffic on a particular port to a specific internal machine, so an outside client can reach, say, a games or web server hidden behind NAT.

ModelThe client-server model and thin versus thick clients

Zooming back out, most Internet services run on the client-server model: a client (your browser, an email app) sends a request, and a server (a web server, a mail server) processes it and returns a response. The interaction is defined by a protocol such as HTTP, and one server handles many clients concurrently. This centralises data, security and backup on the server, but makes the server a bottleneck and a single point of failure that must be provisioned for peak demand.

A final, frequently examined distinction is thin versus thick client computing, which is about where the processing and storage happen. A thin client is a lightweight device that relies almost entirely on a powerful central server to do the work and hold the data — cheap to buy, easy to maintain and secure centrally, and useless without a reliable network connection, since it cannot function offline. A thick (fat) client does most of its own processing and storage locally — more expensive and harder to manage across many machines, but able to work offline and to make use of local hardware. The trade-off is the recurring theme of the topic: centralise for control, cost and security, or distribute for resilience and independence. Streaming a game from a server is a thin-client model; a full PC running the game locally is a thick-client one.

VocabularyKey terms the mark scheme pays for

TrapsMisconceptions that cost marks

ExamWhat examiners want