Title :
Non-linear signal processing with improved STAM algorithm
Author :
Zhixiang Li ; Yinlong Wang ; Xiufeng Gao ; Xiwu Wang ; Qianjin Li ; Xi Li
Author_Institution :
Dept. of Inf. Eng., Ordnance Eng. Coll., Shijiazhuang, China
Abstract :
Symmetric Table Lookup Addition Method (STAM) is an efficient approach to implement non-linear function approximation with hardware, in which only lookup tables and addition are involved, and size of table could be reduced remarkably due to symmetry of the second term of Taylor series. Nevertheless, the dominant memory occupier is the first term, which still uses relatively big table. An improved method based on differential decomposition was brought forth aiming at relieving the problem, and hyperbolic tangent sigmoid function was implemented on FPGA using the improved algorithm. Experiment results showed that 17% to 30% of memory was saved after improvement.
Keywords :
approximation theory; field programmable gate arrays; nonlinear functions; signal processing; table lookup; FPGA; STAM algorithm; Taylor series; differential decomposition; field programmable gate array; hyperbolic tangent sigmoid function; lookup table; nonlinear function approximation; nonlinear signal processing; symmetric table lookup addition method; Algorithm design and analysis; Approximation algorithms; Approximation methods; Field programmable gate arrays; Hardware; Signal processing algorithms; Taylor series; FPGA; Hyperbolic tangent; STAM algorithm; non-linear;
Conference_Titel :
Information Technology and Electronic Commerce (ICITEC), 2014 2nd International Conference on
Print_ISBN :
978-1-4799-5298-4
DOI :
10.1109/ICITEC.2014.7105603
