As a key technology of the fifth generation （5G） wireless communications, sparse code multiple access （SCMA） system has quite high frequency utilization, but its message passing algorithm （MPA） decoder still has...As a key technology of the fifth generation （5G） wireless communications, sparse code multiple access （SCMA） system has quite high frequency utilization, but its message passing algorithm （MPA） decoder still has high time complexity. By the aid of the proposed multi-level dynamic threshold, the complexity of the MPA decoding can be greatly reduced with little error performance loss. In order to reduce a great deal of insignificant computational amounts of message update, we compare the multi-level symbols probability products with optimized multi-level thresholds step by step before they are used for message update calculation. The dynamic threshold configuration refers to the three factors： the level of input symbol probabilities, signal noise ratio （SNP.） and the number of iterations. Especially, in the joint iterative MPA-Turbo decoding procedure, since most encoded bits have good convergence, the inp^t thresholds can avoid more unnecessary computational overhead of message update and reduce the decoding time more significantly. The simulation results show that the proposed multi-level dynamic thresholds considerably reduce the decoding delay in both additive white Gaussian noise （AWGN） channel and frequency selective fading channel.展开更多
由于传统的互补金属-氧化物-半导体(Complementary Metal Oxide Semiconductor,CMOS)神经元电路与生物学的契合性较差且电路复杂,提出了一种基于忆阻器的多端口输入的泄露-整合-激发(Leaky-Integrate-Fire,LIF)神经元电路。该电路由运...由于传统的互补金属-氧化物-半导体(Complementary Metal Oxide Semiconductor,CMOS)神经元电路与生物学的契合性较差且电路复杂,提出了一种基于忆阻器的多端口输入的泄露-整合-激发(Leaky-Integrate-Fire,LIF)神经元电路。该电路由运放、逻辑门等器件以及忆阻器构成,主要分为信号叠加模块和神经元信号产生模块。通过施加多个双尖峰脉冲信号并调节输入信号的数量和频率,模拟了生物神经元受到的不同程度刺激。研究发现施加到神经元上信号的数量和频率达到一定的值,神经元电路才会输出电压信号,这与生物体中只有受到一定程度的刺激时才会做出反应的现象是一致的。进一步,调节该电路中神经元信号产生模块的阈值电压大小,研究发现输入相同的信号,只有当电路的阈值电压较低时,神经元电路才能输出电压信号,这与生物中不同部位受到相同的刺激,神经元兴奋程度越高,越容易做出反应的现象一致。由此,该文所提出的LIF神经元电路不仅解决了传统电路输入信号单一、输入信号波形与生物信号波形差异大等问题,而且能模拟生物神经元的兴奋程度,这为人工神经网络的设计提供理论依据。展开更多
基金supported by the Graduate Scientific Research and Innovation Foundation of Chongqing(CYS16023)the Foundation and Frontier Research Project of Chongqing(cstc 2016jcyj A0365)
文摘As a key technology of the fifth generation （5G） wireless communications, sparse code multiple access （SCMA） system has quite high frequency utilization, but its message passing algorithm （MPA） decoder still has high time complexity. By the aid of the proposed multi-level dynamic threshold, the complexity of the MPA decoding can be greatly reduced with little error performance loss. In order to reduce a great deal of insignificant computational amounts of message update, we compare the multi-level symbols probability products with optimized multi-level thresholds step by step before they are used for message update calculation. The dynamic threshold configuration refers to the three factors： the level of input symbol probabilities, signal noise ratio （SNP.） and the number of iterations. Especially, in the joint iterative MPA-Turbo decoding procedure, since most encoded bits have good convergence, the inp^t thresholds can avoid more unnecessary computational overhead of message update and reduce the decoding time more significantly. The simulation results show that the proposed multi-level dynamic thresholds considerably reduce the decoding delay in both additive white Gaussian noise （AWGN） channel and frequency selective fading channel.
文摘由于传统的互补金属-氧化物-半导体(Complementary Metal Oxide Semiconductor,CMOS)神经元电路与生物学的契合性较差且电路复杂,提出了一种基于忆阻器的多端口输入的泄露-整合-激发(Leaky-Integrate-Fire,LIF)神经元电路。该电路由运放、逻辑门等器件以及忆阻器构成,主要分为信号叠加模块和神经元信号产生模块。通过施加多个双尖峰脉冲信号并调节输入信号的数量和频率,模拟了生物神经元受到的不同程度刺激。研究发现施加到神经元上信号的数量和频率达到一定的值,神经元电路才会输出电压信号,这与生物体中只有受到一定程度的刺激时才会做出反应的现象是一致的。进一步,调节该电路中神经元信号产生模块的阈值电压大小,研究发现输入相同的信号,只有当电路的阈值电压较低时,神经元电路才能输出电压信号,这与生物中不同部位受到相同的刺激,神经元兴奋程度越高,越容易做出反应的现象一致。由此,该文所提出的LIF神经元电路不仅解决了传统电路输入信号单一、输入信号波形与生物信号波形差异大等问题,而且能模拟生物神经元的兴奋程度,这为人工神经网络的设计提供理论依据。
