随机计算方法在信号处理中的应用综述
【摘要】随机计算（Stochastic Computing，SC）是一种可以节省硬件资源的电路结构，但是不可忽视的误差和过长的运行周期使随机计算在完全基于误差，电路资源消耗和吞吐量等传统标准的、于使用了位宽减小（bit-reduction）的传统二进制的电路的比较中不占优势，从而一定程度上阻碍了其在传统数字信号处理领域的应用。然而这并不意味着随机计算没有改进的空间。随机序列的选择对运算结果精度的影响非常明显，随机电路的设计也会极大地改变整个系统的硬件资源消耗。本文介绍了随机计算方法在信号处理中的应用，并着重分析了提高随机计算精度的方法。
【关键词】随机计算 数字信号处理 低差异序列 随机数生成器 乘积累加器
Overview of Stochastic Computing and its applications in Digital Signal Processing
Zihao Wang 917108180234
【Abstract】Stochastic Computing(SC) is an emerging computing diagram with low hardware resource consumption, but its inherent error and exponentially long runtime make it less favorable compared to conventional binary impleme ntation with bit-reduction based purely on classic criteria including error, resource consumption and throughput. However, it is still possible to enhance Stochastic Computing systemrsquo;s performance. The type of sequence adopted in Stochastic Number Generator(SNG) or the design of Stochastic Circuit can greatly affect the output error. The Stochastic Computing and its application in DSP system is introduced in this paper, and emphasis is attached to the method to enhance accuracy of SC.
【Keywords】Stochastic Computing; Digital Signal Processing; Low-discrepancy sequence; Stochastic Number Generator; Multiply-accumulator
1 引言
