Low Complexity LLR-Based Successive-Cancellation List Decoder for Polar Codes

Abstract

Polar codes are the latest breakthrough in coding theory as they can provably achieve the channel capacity Among existing decoding schemes successive-cancellation list (SCL) decoding is recognized as the main approach that can be applied to achieve the best error-correction performance However hardware implementation of the conventional SCL decoder usually consumes a large area and can only achieve low throughput Recent research works were mainly focused on either reducing the hardware requirement or improving the resulting throughput According to the log-likelihood-ratio (LLR) based SCL decoding algorithm and the concept of symbol decision this thesis presents a highly reusable LLR memory structure and simplifies the partial-sum generator to reduce the overall area requirement of polar decoders In addition new frozen-location patterns of rate-0 and repetition codes were employed to decrease the number of sorting stages for approximate maximum likelihood decoding and modified rate-0 and rate-1 codes were used to further reduce the decoding cycles Experimental results reveal that the SCL decoder designed using the proposed algorithm and optimization schemes for a 4-bit (1024 512) polar code with a list size of 2 can achieve at least 1 22 times the hardware efficiency of related works

Date of Award	2019
Original language	English
Supervisor	Ming-Der Shieh (Supervisor)