DocumentCode :
353643
Title :
Transient behavior of fixed point LMS adaptation
Author :
Gupta, Riten ; Hero, Alfred O., III
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci., Michigan Univ., Ann Arbor, MI, USA
Volume :
1
fYear :
2000
fDate :
2000
Firstpage :
376
Abstract :
We relate the distinguishing features of the fixed point power-of-two step size LMS algorithm´s learning curve to the precision of its data and coefficient variables. In particular, we show that the increase in the steady state MSE floor due to finite precision effects is determined primarily by data quantization while the decrease in convergence rate due to finite precision is determined by both data and coefficient quantization. We also derive a condition under which the slowdown phenomenon can be eliminated, given the reference variance and lower bounds on the minimum MSE and optimal weight vector magnitude
Keywords :
adaptive signal processing; convergence of numerical methods; fixed point arithmetic; identification; least mean squares methods; quantisation (signal); transient analysis; FIR filter; LMS algorithm learning curve; coefficient quantization; coefficient variables; convergence rate; data quantization; data variables; finite precision effects; fixed point LMS adaptation; lower bounds; minimum MSE magnitude; optimal weight vector magnitude; power-of-two step size; reference variance; slowdown phenomenon elimination; steady state MSE floor; system identification; transient behavior; Adaptive equalizers; Algorithm design and analysis; Convergence; Finite impulse response filter; Fixed-point arithmetic; Least squares approximation; Mean square error methods; Quantization; Steady-state; System identification;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Acoustics, Speech, and Signal Processing, 2000. ICASSP '00. Proceedings. 2000 IEEE International Conference on
Conference_Location :
Istanbul
ISSN :
1520-6149
Print_ISBN :
0-7803-6293-4
Type :
conf
DOI :
10.1109/ICASSP.2000.861981
Filename :
861981
Link To Document :
بازگشت