The idea
Error detection tells a receiver a frame is damaged. It says nothing about getting a correct copy — for that, something has to ask again. ARQ, Automatic Repeat reQuest, is that something: a set of rules for when to retransmit, running without a human deciding case by case.
There is a second problem sitting right next to it. Even a perfectly correct stream of frames can overwhelm a receiver that simply cannot keep up — a different failure mode, called by a different name: flow control. ARQ’s retransmission logic is where error control and flow control meet.
ACK, NAK and the timeout
How it works
Three ways a frame's fate gets settled
- Positive acknowledgment (ACK) — sent back when a frame arrives with no detected error.
- Negative acknowledgment (NAK) — sent back when a frame arrives with a detected error, asking for it again.
- Retransmission after a timeout — the sender starts a timer when it transmits. If nothing comes back before it expires, the sender retransmits anyway.
The timeout case matters because ACK and NAK both assume the frame arrived at the receiver in the first place. If the frame itself is lost, the receiver never gets a chance to send anything back. If the frame arrives fine but the ACK is lost on the way back, the sender is left in the same position — no reply, regardless of which side the loss happened on. The timer is what breaks that stalemate.
Where marks get lost
A timeout does not mean the frame was damaged
It is tempting to treat “no ACK arrived” as proof the frame was corrupted. It is not — the lecture is explicit that either the frame or the acknowledgment may have been lost. The sender cannot tell which, and does not need to: retransmitting is the correct move either way.
Flow control and error control
| What it manages | What goes wrong without it | |
|---|---|---|
| Flow control | The rate data is sent at | A fast sender overruns a slower receiver's buffer |
| Error control | The integrity of what arrives | Corrupted data is accepted as though it were correct |
The lecture places ARQ as a flow control mechanism operating on the datagram, run in the transport layer — naming TCP specifically as where this runs in practice.
The slides name stop-and-wait ARQ, go-back-N ARQ and selective repeat ARQ as required topics, but do not develop their mechanics — no window sizes, sequence-number widths or retransmission rules for any of the three appear in the lecture material, and none of that was independently verified for this page. If an exam question asks for the specific windowing behaviour of one of these, this page cannot give it — check the tutorial solutions or the textbook chapter directly rather than relying on a number this page did not check.
PPP
How it works
Point-to-Point Protocol
PPP is derived from DLC (Data Link Control) and was built mainly for WAN links. The lecture notes it is no longer relevant for LAN.
11111111 — PPP is a point-to-point link with exactly one possible recipient, so the field exists for framing compatibility with HDLC rather than to actually select an address.- Flag — 1 byte,
01111110. Marks the start and end of the frame. - Address — 1 byte, always the broadcast value
11111111. A point-to- point link has exactly one possible recipient, so there is nothing for the field to select — it exists for framing compatibility rather than to make an addressing decision. - Control — constant values; PPP does not use this field for flow control.
- Protocol — defines what the frame is carrying, user data or other control information.
- FCS — 2 or 4 bytes, a CRC used for error detection.
Check yourself
Two checks:
- A sender’s timer expires with no ACK or NAK received. Was the frame definitely corrupted?
- Why does the PPP Address field never actually select a destination?
Answers: no — either the frame or the returning ACK could have been lost, and the sender cannot distinguish the two cases, so it just retransmits. And: because PPP connects exactly two points, so there is only ever one possible recipient — the field is fixed at the broadcast value rather than doing real addressing work.
In the exam
- ARQ’s three mechanisms. ACK (clean arrival), NAK (detected error, request retransmission), and timeout-triggered retransmission (covers the case where nothing comes back at all).
- Why timeout exists alongside ACK/NAK. Either the frame or the ACK may be lost — the sender cannot tell which, and retransmits regardless.
- Flow control vs error control. Rate versus integrity — Tutorial 3’s Q5 pairing.
- Where the lecture places ARQ. Flow control for the datagram, run in the transport layer (TCP named specifically).
- PPP field values. Flag
01111110, Address11111111(broadcast, always), Control constant (no flow control), FCS 2 or 4 bytes. - Stop-and-wait / go-back-N / selective repeat. Named in the unit, not developed in the slides. This page states the gap rather than guessing.