A2.3.3 Explain how packet switching is used to send data across a network.

Big idea
Packet switching is the organising principle that breaks a large data stream into small, individually addressed packets, lets each packet find its own way through a mesh of routers and switches, and re-assembles the original stream at the destination. By treating every packet as an autonomous unit with its own header, the network can statistically share links among many conversations, maximise utilisation, and survive link or node failures without session teardown.

---

Precise definitions

Term	Definition (concise)
Packet	A finite block of user data (the payload) encapsulated with a header that carries control fields such as source and destination addresses, sequence/fragment numbers, time-to-live (TTL), and protocol identifiers.
Packet switching	A store-and-forward forwarding method in which each packet is routed hop-by-hop, independently of others, based solely on the information in its header and the router’s forwarding table.

---

How the process works

1 Segmentation (Encapsulation)

1. An application hands a byte stream (e.g., 2 MB of a video frame) to the transport layer.
2. The transport layer (TCP or UDP) chops the stream into segments (≤ 65 535 B for UDP; ≤ MSS for TCP) and appends transport-layer headers that hold port numbers, sequence numbers, and checksums.
3. The network layer (IPv4/IPv6) adds its header—source/destination IP, TTL/Hop Limit, protocol number.
4. The data-link layer (Ethernet, Wi-Fi) adds a frame header/trailer for intra-hop delivery.

Result: a hierarchy of headers, each wrapping the payload like nested envelopes.

2 Independent forwarding

• Store-and-forward: At each hop, the switch/router receives the entire frame, checks its integrity (CRC), and either drops it or forwards it according to its lookup table.
• Hop-wise routing decisions: Routers do not keep track of outstanding segments of the same flow; each packet’s destination IP is matched to the longest-prefix entry in the forwarding information base (FIB).
• Out-of-order possibility: Because different packets may traverse different paths, they can arrive out of order or even be duplicated.

3 Control information for reliable delivery

• Sequence numbers/Fragment offsets: Allow the destination host to place packets back in the original order and detect missing fragments.
• Acknowledgements (TCP): The receiver advertises the next expected sequence number; lost packets are retransmitted.
• TTL/Hop-Limit: Prevents routing loops by limiting lifetimes.
• Checksums: At multiple layers detect corruption and trigger retransmission or discarding.

4 Reassembly at the destination

The transport layer buffers incoming segments, uses sequence numbers to reorder them, requests retransmission if gaps persist, and eventually delivers an error-free, in-order byte stream to the application.

---

Role of switching devices

Device	OSI layer	Forwarding basis	Function in packet-switched network
Layer-2 switch	Data-link (Ethernet MAC)	48-bit MAC addresses	• Sorts frames within a local segment (same VLAN)• Uses a CAM table learned from source MACs• Offers micro-segmentation and full-duplex, but is transparent to IP routing decisions
Router / Layer-3 switch	Network	32-bit IPv4 or 128-bit IPv6 addresses + longest-prefix match	• Maintains routing tables (static, OSPF, BGP, etc.)• Makes per-packet path choices across subnets• Performs fragmentation (IPv4) or drops oversize packets (IPv6 “don’t fragment”)• Optionally runs NAT, ACLs, QoS, tunnelling

Explain level insight: L2 switches accelerate local forwarding by hardware learning and cut-through techniques, but cannot move traffic between subnets. Routers understand IP addressing, enforce policy, and find alternate paths when a link fails—enabling the dynamic, resilient behaviour that defines packet switching.

When you explain, always articulate mechanism and reason—not just a surface description.