Polar Coding with Introducing Half-Frozen Bits
Polar Coding with Introducing Half-Frozen Bits
复制标题
DOI：
10.1109/glocomw.2018.8644449
复制
发表时间：
2018-12
期刊：
影响因子：
--
通讯作者：
Kazumasa Nagano;T. Saba
中科院分区：
文献类型：
--
作者：
Kazumasa Nagano;T. Saba
Polar编码是Arıkan提出的纠错编码。在极化编码中，由于信道极化，每个比特位置的解码结果的错误率不同。在无限码字长度的极性编码情况下，解码比特序列的每个位置的译码错误率被极化为0.5或0。因此，当能够采用足够长度的码字时，通过将冻结比特放置到发送端译码错误率为0.5的比特位置，信息比特将被无错误地解码。另一方面，在具有有限码字长度的极性编码的情况下，每个比特位置处的解码错误率取0.5至0之间的值。因此，在放置信息比特的比特位置处也会发生解码错误。在本文中，我们将半冻结比特引入极性编码中。半冻结位是信息位，其作用类似于冻结位。半冻结比特被放置在冻结比特位置内难以发生错误的位置以及信息比特位置内容易发生错误的位置处。作为半冻结比特，部分信息比特经过另一次纠错编码后被使用。这改善了导致高解码错误率的信息比特位置处的解码错误率。因此，半冻结比特提高了极性编码的整体误码率（BER）性能。通过计算机仿真，我们表明引入半冻结比特的极性编码比原始极性编码具有更高的纠错性能。
Polar coding is the error correction coding proposed by Arıkan. In polar coding, the error rate of the decoding result differs for each bit position owing to the channel polarization. In the case of polar coding with an infinite codeword length, the decoding error rate at each position of the decoded bit sequence is polarized to 0.5 or 0. Therefore, when a codeword of a sufficient length can be adopted, information bits will be decoded without errors by placing the frozen bits to bit positions where the decoding error rates are 0.5 at the transmitter. On the other hand, in the case of polar coding with a finite codeword length, the decoding error rate at each bit position takes a value between 0.5 and 0. Therefore, the decoding errors also occur at the bit positions where the information bits are placed. In this paper, we introduce half-frozen bits into polar coding. Half-frozen bits are the information bits which play roles like frozen bits. Half-frozen bits are placed at positions where errors hardly occur within the frozen bit positions as well as where errors prone to occur within the information bit positions. As half-frozen bits, a part of information bits is used after being applied another error correction coding. This improves the decoding error rates at information bit positions which result in high decoding error rates. Therefore, half-frozen bits improve the overall bit error rate (BER) performance of polar coding. Through computer simulation, we show that the polar coding with introducing half-frozen bits provides a higher error correction performance than the original polar coding.