Serial, parallel, bandwidth and bit rate

Serial transmission sends bits one after another down a single wire. Parallel sends several bits simultaneously down several wires.

Parallel sounds obviously faster — and for short distances at low speeds it was. But over distance it fails badly:

• Skew — the wires are never quite identical, so bits sent together arrive at slightly different times. The longer the cable and the higher the clock, the worse it gets.
• Crosstalk — adjacent wires induce interference in each other.
• Parallel needs many more conductors, so cables are bulkier and dearer.

That is why USB, SATA and Ethernet are all serial: one clean, very fast channel beats many mediocre ones.

Synchronous vs asynchronous: synchronous transmission is governed by a shared clock signal. Asynchronous wraps each byte in a start bit and a stop bit (often with a parity bit), so the receiver can re-synchronise on every byte — simpler, but with more overhead.

Your turn — transfer time

Topologies and network models

Physical star: every device has its own cable to a central switch.

• A single cable failure isolates one device only;
• no collisions — the switch sends each frame only to its destination;
• easy to add devices, but it needs a lot of cable and the switch is a single point of failure.

Physical bus: all devices share one backbone cable with terminators at each end. Cheap and cable-light, but every device sees every transmission, collisions rise with traffic, and a break in the backbone downs the whole network.

Physical vs logical: the two can differ. Classic Ethernet was wired as a physical star to a hub but behaved as a logical bus — every device still saw everything. Topology is about how data flows, not just how the cables run.

Wireless (CSMA/CA): a wireless station cannot detect collisions while transmitting, so it avoids them instead: listen, wait a random back-off, and optionally use RTS/CTS to reserve the channel — which also solves the hidden node problem, where two stations can each hear the access point but not each other.

The TCP/IP stack

Four layers. Each layer only talks to the layer above and below, and adds its own header on the way down. That separation is what lets Wi-Fi be swapped for fibre without changing a single web browser.

Packet switching: data is split into packets, each stamped with source, destination and a sequence number. Each packet is routed independently and may take a different route; the receiving TCP layer reassembles them in order and requests any that are missing.

Why this beats circuit switching: no dedicated line sits idle, the network routes round congestion and failure automatically, and capacity is shared efficiently. The cost is variable delay (jitter) and out-of-order arrival — which is exactly why live video often uses UDP: a late packet is worse than a lost one.

The TCP/IP stack

DNS, IP addressing and subnetting

The Domain Name System is the internet's phone book: it resolves a human-readable domain name into an IP address. It is a distributed, hierarchical database — a resolver asks a root server, which points to the top-level domain (.uk), which points to the authoritative name server for the domain. Answers are cached at every step.

An IPv4 address is 32 bits, written as four dotted decimal octets. It has two parts, and the subnet mask is what says where the boundary falls:

The mask for /26 is 255.255.255.192 — because 26 ones followed by 6 zeros makes the last octet 11000000 = 192.

Why subnet at all? It splits one large network into smaller ones, cutting broadcast traffic, isolating faults, and letting you apply different security rules to different departments. Routers only need to know how to reach the network, not every individual host — which is what stops the global routing table from exploding.

Your turn — usable hosts

Your turn — the subnet mask

Your turn — counting subnets

NAT, firewalls and encryption in transit

Network Address Translation (NAT) lets an entire private network share one public IP address. The router rewrites the source address (and port) of every outgoing packet and keeps a translation table so replies find their way back.

• It conserves the badly depleted IPv4 address space;
• it incidentally hides the internal network — an unsolicited inbound packet has nowhere to go;
• port forwarding is needed to deliberately expose an internal server.

Private ranges (never routed on the public internet): 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16.

A firewall controls traffic between networks:

• Packet filtering — accept or drop based on source/destination IP, port and protocol. Fast, but it judges each packet in isolation.
• Stateful inspection — tracks whole connections, so it can tell a legitimate reply to your request from an unsolicited packet pretending to be one.
• Proxy server — makes the request on your behalf, hiding the client, caching content and filtering at the application level.

Encryption in transit: HTTPS uses asymmetric encryption to authenticate the server (via a certificate signed by a trusted CA) and to agree a symmetric session key. The bulk data is then encrypted symmetrically, because symmetric ciphers are far faster. You need both — asymmetric to solve key exchange, symmetric to go fast.

NAT

Packet switching

Which layer of the stack?

Tap an item, then tap the TCP/IP layer it belongs to.

Match the technology to its job

Tap an item on the left, then its partner on the right.

The big ideas to know

That is the whole of AQA 4.9. Press Finish to see your score.