A deeper guide to the data-link layer services that make raw physical-layer bit streams usable: framing, addressing, error detection, error correction, and flow control.
The physical layer moves bits, but bits alone do not tell the receiver where a message begins or whether it was corrupted. The data-link layer adds structure and local reliability across one link.
The chapter names the data-link duties as framing, addressing, error control, and flow control. A network-layer packet is encapsulated inside a frame, and the receiver removes the frame wrapper during deframing.
A frame is like an envelope. The letter is the packet. The envelope carries local delivery information and marks boundaries so postal workers know what belongs together.
Data-link delivery is local: one hop or one link at a time. End-to-end delivery is handled by higher layers, especially network and transport.
| Concept | Meaning | Exam clue |
|---|---|---|
| Framing | Creates frame boundaries around payload | Keywords: frame, flag, header, trailer |
| Addressing | Adds layer-2 source/destination information | Keywords: physical address, MAC address |
| Error control | Detects or corrects corrupted data | Keywords: parity, CRC, checksum, FEC |
| Flow control | Coordinates sender speed with receiver capacity | Keywords: buffer, overflow, slow receiver |
The slide set shows two framing styles. Both use a flag pattern to mark frame boundaries, and both need stuffing when the flag-like pattern appears inside the data.
Character-oriented protocols treat the frame as a sequence of bytes. An 8-bit flag marks the beginning and end. If the same flag byte appears in the payload, the sender inserts an extra ESC byte before it. The receiver removes the ESC during unstuffing.
Bit-oriented protocols treat the frame as a stream of bits. If the data contains five consecutive 1s, the
sender inserts one extra 0. This prevents the receiver from accidentally seeing the flag pattern
01111110 inside the payload.
First identify whether the protocol is character-oriented or bit-oriented. Then apply the correct insertion rule. Byte stuffing inserts ESC before flag bytes. Bit stuffing inserts 0 after five consecutive 1s.
| Concept | Meaning | Exam clue |
|---|---|---|
| Character-oriented | Frame length is a multiple of bytes | Stuff ESC when flag byte appears |
| Bit-oriented | Frame is interpreted bit by bit | Stuff 0 after five consecutive 1s |
| Flag | Marks start and end of frame | Often shown as 01111110 |
The chapter distinguishes error shape before teaching detection. This matters because some methods catch single-bit errors easily but differ in burst-error performance.
A single-bit error means exactly one bit in the data unit changes from 0 to 1 or from 1 to 0. It is simple to detect with parity.
A burst error means two or more bits in the data unit have changed. The changed bits do not need to be consecutive. The burst length is counted from the first corrupted bit to the last corrupted bit.
If a burst lasts for a certain time, the number of affected bits depends on bandwidth. Higher bandwidth means more bits pass during the same time, so the same 2 ms burst can affect far more bits on a 1 Gbps link than on a 1 Kbps link.
Detection methods add redundant bits so the receiver can test whether the received data still follows an expected pattern.
A parity bit is added so the total number of 1s is even for even parity, or odd for odd parity. It detects all single-bit errors but misses errors when an even number of bits flips in the same protected unit.
Data is arranged in rows and columns. Parity is added for each row and for the overall block. This catches more patterns than one parity bit and can help locate some errors.
CRC treats the bit pattern like a polynomial. Sender and receiver share a generator polynomial. The sender appends a remainder; the receiver divides by the same generator to check whether the result is valid.
The message is divided into fixed-size words, commonly 16-bit or 8-bit in examples. The sender adds all words using one’s-complement arithmetic, complements the sum, and sends it as the checksum. The receiver adds all words including the checksum to test for error.
| Concept | Meaning | Exam clue |
|---|---|---|
| Simple parity | Adds one bit to force even or odd number of 1s | Detects all single-bit errors |
| 2D parity | Adds row/column-style redundancy | Stronger than simple parity |
| CRC | Uses generator polynomial and binary division | Excellent for burst-error detection |
| Checksum | Uses one’s-complement addition | Common exam step-by-step method |
Correction goes beyond detecting a problem. It tries to recover the original data or replace damaged data with a clean copy.
Retransmission is simple: detect the error and request another copy. It is efficient when errors are rare and delay is acceptable, but inefficient for wireless or real-time applications with frequent loss.
FEC sends enough redundant information for the receiver to correct a limited number of errors without waiting for retransmission. It trades extra bits for lower recovery delay.
Interleaving spreads neighboring chunks apart before transmission. A burst then damages small pieces of several chunks instead of destroying one continuous chunk. After deinterleaving, FEC can often repair the smaller distributed errors.
Use this as the last-pass memory page before a quiz or exam.
These questions target the facts, comparisons, and calculations that are easiest to test.
Answer: It separates one transmitted unit from another.
Explanation: The receiver must know where each frame begins and ends before it can interpret the payload.
Answer: One 0 after five consecutive 1s.
Explanation: This prevents payload bits from looking like the flag pattern.
Answer: Because if an even number of bits changes, parity may still look correct.
Explanation: Parity only checks whether the total number of 1s has the expected even/odd property.
Answer: The generator polynomial.
Explanation: Both sides use the same divisor to create and verify the CRC remainder.
Answer: It can correct limited errors without waiting for a retransmission.
Explanation: Retransmission delay can be too expensive when errors are frequent or timing matters.
Answer: From the first corrupted bit to the last corrupted bit.
Explanation: The bits inside the span do not all need to be corrupted.