Why is interleaving needed if FEC already corrects errors?

FEC codes like LDPC and convolutional codes are designed to correct randomly distributed errors. A deep fade in a wireless channel causes a burst of consecutive errors that overwhelms any single codeword's correction capacity. Interleaving spreads consecutive bits across different codewords so that after de-interleaving at the receiver, each codeword sees only a few scattered errors instead of a concentrated burst, well within FEC correction capability.

What is the difference between block and convolutional interleaving?

Block interleaving writes data row-by-row into a matrix and reads it column-by-column. The interleaving depth equals the number of rows. Convolutional interleaving uses shift registers of progressively increasing delay, producing a continuous stream with lower latency than block interleaving for the same burst protection depth. DVB-T uses convolutional interleaving; 5G NR uses a combination of bit and sub-block interleaving.

How much latency does interleaving add?

Interleaving depth D determines both the maximum correctable burst length and the latency. For a block interleaver with D rows and N columns: latency = D * N * T_bit at both transmitter and receiver. A 48x48 block interleaver at 1 Mbps adds 48*48/1e6 = 2.3 ms each way, 4.6 ms total. This trade-off between burst protection and latency is critical for real-time applications like voice and URLLC.

Digital Communications

Bit Interleaving

Bit Interleaving rearranges the order of transmitted bits before modulation so that consecutive bits in the original data stream are spread apart in time (or frequency). When a burst error occurs in the channel (e.g., from a deep fade), the errored bits are de-interleaved at the receiver and distributed across multiple FEC codewords. Each codeword then sees only a few scattered errors rather than a concentrated burst, within the correction capability of the code. Interleaving converts bursty error channels into approximately random error channels, for which FEC codes are optimally designed.

Category: Digital Communications

Types: Block, Convolutional

Understanding Bit Interleaving

In a fading channel, signal strength can drop below the noise floor for milliseconds, corrupting hundreds or thousands of consecutive bits. A single FEC codeword might have all of its bits in that burst, exceeding its correction capacity. Interleaving ensures that each codeword's bits are spread across the entire transmission frame, so no single fade destroys any one codeword.

Block interleaving writes data row-by-row into a D×N matrix and reads column-by-column. After transmission and de-interleaving, a burst of up to D consecutive errors becomes single errors in D different codewords. Convolutional interleaving uses shift registers with incrementally increasing delays, reducing latency for the same burst protection.

Interleaver Parameters

Block interleaver (D rows, N columns):
Max correctable burst: D bits
Latency: D × N × T_bit (each direction)

Example: D=48, N=48 at 1 Mbps:
Burst protection: 48 bits
Latency: 48×48/10⁶ = 2.3 ms (one way)

Interleaver Type Comparison

Type	Burst Protection	Latency	Memory	Used In
Block	D bits	D×N×T_bit	D×N bits	Wi-Fi, LTE
Convolutional	D bits	D×(D−1)/2×T_bit	Lower	DVB-T, ISDB
Random (S-random)	Variable	Block-length	N bits + map	Turbo codes

Common Questions

Frequently Asked Questions

Why interleave if FEC corrects errors?

FEC corrects random errors, not bursts. Interleaving converts bursts into scattered errors across codewords, keeping each within FEC correction capacity.

Block vs convolutional interleaving?

Block uses a D×N matrix (row-write, column-read). Convolutional uses progressive delay shift registers with lower latency for the same burst depth. DVB-T uses convolutional; 5G NR uses combined bit/sub-block.

Latency impact?

Latency = D×N×T_bit per direction. A 48×48 block at 1 Mbps adds 2.3 ms each way. Critical trade-off for voice and URLLC.

Related Terms

← BER Tester Bit Loading →