Error Control Schemes

The Problem

When you send data wirelessly, bits can get corrupted.

• Radio interference
• Signal fading
• Noise

How do we deal with errors?

Two Fundamental Approaches

ARQ: Automatic Repeat Request

Simple idea: If something’s wrong, ask for it again.

How it works:

1. Sender transmits data
2. Receiver checks if it’s correct (using a checksum)
3. If correct → send ACK (acknowledgment)
4. If wrong → send NACK (negative acknowledgment)
5. Sender retransmits on NACK

Pros:

• Simple to implement
• Only retransmit when needed
• Guaranteed correct data

Cons:

• Requires back-and-forth communication
• High latency (waiting for ACK/NACK)
• Bad for real-time (voice, video)
• Many retransmissions if channel is noisy

FEC: Forward Error Correction

Simple idea: Send extra bits so receiver can fix errors itself.

How it works:

1. Sender adds redundant bits
2. Receiver uses those extra bits to detect AND correct errors
3. No retransmission needed

But wait… how can you correct an error?

Doesn’t that require knowing what was sent?

The trick: Not all bit patterns are valid.

Example: Repetition Code

Instead of sending 0 or 1, send each bit three times:

Only two patterns are valid: 000 and 111.

What if one bit flips?

Valid codewords are spread far apart. Errors land you in the space between them. Decode by finding the nearest valid codeword.

Pros:

• No waiting for retransmission
• Great for real-time (voice, video)
• Works with no return channel (satellite broadcast)

Cons:

• Always sends extra bits (overhead)
• Can only correct limited errors

ARQ vs FEC: The Trade-off

Can we get the best of both worlds?

HARQ: Hybrid ARQ

Combine FEC + ARQ for the best of both.

How it works:

1. Send data with some FEC (not full strength)
2. Receiver tries to decode
3. If FEC fixes all errors → done
4. If not → request retransmission

Key insight: The retransmission can be combined with the original to improve decoding.

You’re not throwing away the first attempt!

HARQ Types

There are three types of HARQ, differing in what gets retransmitted:

Type I: Chase Combining

Retransmit the exact same data. Combine both copies.

How it helps:

• Two noisy copies → average them
• Noise is random, so averaging reduces it
• Better chance of correct decoding

Downside:

• Not very efficient
• Same information repeated

Type II: Incremental Redundancy

Retransmit different parity bits. Each transmission adds new information.

How it works:

1. First transmission: Data + some parity bits
2. If failed: Send different parity bits (not the same!)
3. Receiver combines both → more redundancy → better correction

Why it’s better:

• Each retransmission provides new information
• More efficient than repeating the same thing
• Builds up redundancy only when needed

Type III: Self-Decodable

Each transmission can be decoded on its own if conditions are good enough.

How it works:

• Every transmission includes enough redundancy to be standalone
• If one transmission is clean enough → decode immediately
• If not → combine with others

Trade-off:

• More overhead per transmission
• But simpler receiver logic

What Does LTE Use?

LTE uses Type II (Incremental Redundancy) with a circular buffer.

Why?

• Most efficient use of bandwidth
• Only adds redundancy when needed
• Each retransmission provides new parity bits

How LTE does it:

1. Encode data into a large set of coded bits
2. First transmission: send a portion of coded bits
3. Retransmission: send a different portion
4. Receiver keeps combining until it can decode

Summary

HARQ Types:

HARQ Type II (Incremental Redundancy) is the winner for modern wireless because it adds redundancy only when needed and each retransmission provides new information.