Restart Mechanisms for the Successive-Cancellation List-Flip Decoding of Polar Codes

Area of Science:

• Coding Theory
• Digital Communications
• Signal Processing

Background:

• Polar codes with cyclic redundancy check (CRC) are standard in 5G, using successive-cancellation list (SCL) decoding.
• High list sizes in SCL increase hardware complexity; flip decoding offers a low-complexity alternative.
• List-flip algorithms like SCLF and DSCLF achieve near SCL-32 performance with reduced list sizes but have variable execution times.

Purpose of the Study:

• To propose a restart mechanism for list-flip decoding algorithms to reduce variable execution times.
• To investigate methods for selecting restart locations to balance performance and memory overhead.
• To optimize the decoding process for practical applications requiring consistent performance.

Main Methods:

• Introduced a restart mechanism for list-flip decoding algorithms.
• Proposed limiting restart locations to a specific set to mitigate memory overhead in dynamic SCLF (DSCLF).
• Developed and compared various construction methods for these restart location sets.

Main Results:

• The proposed restart mechanism reduced average execution time by 41.7% for a polar code (N=1024, rate=1/4) under DSCLF decoding (L=2, Tmax=300).
• This was achieved with four strategically chosen restart locations, incurring only a 1.5% memory overhead.
• Error-correction performance remained unaffected by the implemented restart strategy.

Conclusions:

• The proposed restart mechanism effectively reduces execution time for list-flip polar decoders.
• Limiting restart locations is crucial for managing memory overhead in DSCLF.
• This approach offers a practical solution for applications demanding efficient and consistent decoding performance.