TY - GEN

T1 - Optimal fixed-point implementation of digital filters

AU - Roozbehani, Mardavij

AU - Megretski, Alexandre

AU - Feron, Eric

N1 - Generated from Scopus record by KAUST IRTS on 2021-02-18

PY - 2007/12/1

Y1 - 2007/12/1

N2 - We consider the problem of finding optimal realizations of discrete-time digital filters w.r.t. implementation on a finite-memory machine with a fixed-point processor. A realization for which the performance of the fixed-point implementation under quantization effects is closest (in some sense) to the ideal performance (assuming precise arithmetic) is considered optimal. The transfer function corresponding to the optimal implementation is not necessarily the same as the transfer function of the original filter. Therefore, the optimal implementtaion cannot be obtained via a change of coordinates. This problem is inherently a nonlinear control problem due to the presence of the quantizer. We use a signal + noise model to linearize the system. After linearization, the problem of finding the optimal realization is still a nonconvex optimization problem. We show that under some mild technical assumptions the problem can be convexified losslessty, and transformed into an equivalent set of LMIs for which numerical solutions can be obtained efficiently. ©2007 IEEE.

AB - We consider the problem of finding optimal realizations of discrete-time digital filters w.r.t. implementation on a finite-memory machine with a fixed-point processor. A realization for which the performance of the fixed-point implementation under quantization effects is closest (in some sense) to the ideal performance (assuming precise arithmetic) is considered optimal. The transfer function corresponding to the optimal implementation is not necessarily the same as the transfer function of the original filter. Therefore, the optimal implementtaion cannot be obtained via a change of coordinates. This problem is inherently a nonlinear control problem due to the presence of the quantizer. We use a signal + noise model to linearize the system. After linearization, the problem of finding the optimal realization is still a nonconvex optimization problem. We show that under some mild technical assumptions the problem can be convexified losslessty, and transformed into an equivalent set of LMIs for which numerical solutions can be obtained efficiently. ©2007 IEEE.

UR - https://ieeexplore.ieee.org/document/4283132/

UR - http://www.scopus.com/inward/record.url?scp=46449108840&partnerID=8YFLogxK

U2 - 10.1109/ACC.2007.4283132

DO - 10.1109/ACC.2007.4283132

M3 - Conference contribution

SN - 1424409888

SP - 3565

EP - 3569

BT - Proceedings of the American Control Conference

ER -