%0 Journal Article
%T 基于集成学习的极化码BP译码算法
BP Decoding Algorithm Based on Ensemble Learning
%A 徐坤
%J Software Engineering and Applications
%P 354-365
%@ 2325-2278
%D 2023
%I Hans Publishing
%R 10.12677/SEA.2023.122036
%X 集成学习被广泛用于解决复杂任务,方法是通过学习得到不同的弱学习器,利用其各自的泛化能力,获得强学习器来解决问题。BPL译码算法和WBP译码算法是BP译码算法的两种主要优化算法,两种优化算法均具有不同的优势。基于此,提出了一种基于集成学习的BP译码算法,该算法借助集成学习的优势,来获取更优异的译码性能。通过训练多组具有不同的权重参数的WBP译码器,并将其组合成类似BPL译码算法的一种新的译码器,该译码器的权重参数由离线训练得来。仿真结果表明,该译码算法可以达到与BPL译码算法相近的译码性能,同时具有更低的计算复杂度及更少的迭代次数。
Ensemble learning is widely used to solve complex tasks by learning different individual learners, using their respective generalization capabilities to obtain strong learners to solve problems. BPL decoding algorithm and WBP decoding algorithm are the two main optimization algorithms of BP decoding algorithm, and both optimization algorithms have different advantages. Based on this, a BP decoding algorithm based on ensemble learning is proposed, which obtains better decoding performance by virtue of the advantages of ensemble learning. By training multiple sets of WBP decoders with different weight parameters, and combining them into a new decoder similar to the BPL decoding algorithm, the weight parameters of the decoder are obtained from offline training. Simulation results show that the decoding algorithm can achieve similar decoding performance as the BPL decoding algorithm, and has lower computational complexity and fewer iterations.
%K 极化码,BP译码,加权BP译码,集成学习,弱学习器
Polar Code
%K BP Decoding
%K Weighted BP Decoding
%K Integrated Learning
%K Individual Learner
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=64932