Hi all,
Why do we use Polar for control channels and LDPC for data channels?
One reason I could find is Polar gives better performance for low SNR cases, it doesn’t give a noise floor.
Can someone help me with more eleborated answer?
Channel coding is based on LDPC codes, a code design which was originally proposed in the 1960s but forgotten for many years. They were “rediscovered” in the 1990s and found to be an attractive choice from an implementation perspective. From an error-correcting capability point of view, turbo codes, as used in LTE, can achieve similar performance, but LDPC codes can offer lower complexity, especially at higher code rates, and were therefore chosen for NR.
One reason I could find is that Polar is a sequential decoding procedure.
So, the polar decoder can take longer than LDPC for high code blocks.
Does anyone have any profiling numbers for Polar or LDPC?
in 5g polar codes length n is between 32 and 1024 in uplink however 2 is between 128 and 512 for downlink
1 Like
In the 5G Physical Downlink Control Channel (PDCCH) transmit chain, we see that Polar Encoding is used rather than LDPC encoding.
What is the main reason to have a different encoding method for the control channel?
One of the main reason is: PDCCH majorly carries the Downlink Control Information (DCI). 5G DCI is ~35 bits long.
This is starkingly different from the 1000’s of bit the Physical Data Shared Channel (PDSCH) has to transmit.
So, for PDCCH, we need an encoding method which is optimized for the small block sizes.
LinkedIn: 
1 Like