LOW SENSITIVITY DESIGN OF MULTIPLIERLESS ELLIPTIC IIR DIGITAL FILTERS

Similar documents
A New Medium Access Control Protocol for TDD Mode Wideband CDMA Wireless Local Area Networks

Superposition, Thevenin and Norton. Superposition

Sinusoidal Steady State Analysis

Rejection of PSK Interference in DS-SS/PSK System Using Adaptive Transversal Filter with Conditional Response Recalculation

Simplified Algorithm and Hardware Implementation for the (24, 12, 8) Extended Golay Soft Decoder Up to 4 Errors

The Pilot Alignment Pattern Design in OFDM Systems

To: Professor Avitabile Date: February 4, 2003 From: Mechanical Student Subject: Experiment #1 Numerical Methods Using Excel

Passive Filters. References: Barbow (pp ), Hayes & Horowitz (pp 32-60), Rizzoni (Chap. 6)

Uncertainty in measurements of power and energy on power networks

Pre-distortion Linearization for 64-QAM Modulation in Ka-Band Satellite Link

PRACTICAL, COMPUTATION EFFICIENT HIGH-ORDER NEURAL NETWORK FOR ROTATION AND SHIFT INVARIANT PATTERN RECOGNITION. Evgeny Artyomov and Orly Yadid-Pecht

Cooperative Collision Avoidance between Multiple Mobile Robots

29. Network Functions for Circuits Containing Op Amps

Walsh Function Based Synthesis Method of PWM Pattern for Full-Bridge Inverter

RC Filters TEP Related Topics Principle Equipment

DESIGN AND IMPLEMENTATION OF NETWORKED PREDICTIVE CONTROL SYSTEMS. S C Chai, G P Liu and D Rees

The Existence, Uniqueness and Error Bounds of Approximation Splines Interpolation for Solving Second-Order Initial Value Problems

i S1 V IN i C1 i N i C2 i S2

Dynamic Optimization. Assignment 1. Sasanka Nagavalli January 29, 2013 Robotics Institute Carnegie Mellon University

LECTURE 9: QUADRATIC RESIDUES AND THE LAW OF QUADRATIC RECIPROCITY

A High-Sensitivity Oversampling Digital Signal Detection Technique for CMOS Image Sensors Using Non-destructive Intermediate High-Speed Readout Mode

A MODIFIED DIFFERENTIAL EVOLUTION ALGORITHM IN SPARSE LINEAR ANTENNA ARRAY SYNTHESIS

Section 16.3 Double Integrals over General Regions

antenna antenna (4.139)

Switched-Capacitor Filter Optimization with Respect to Switch On-State Resistance and Features of Real Operational Amplifiers

5 October 2015 Stereo Cross-feed Network for Headphones 1 of 12 Copyright 2015 Peter H. Lehmann. All Rights Reserved.

TEMPERATURE PREDICTION IN TIMBER USING ARTIFICIAL NEURAL NETWORKS

CHAPTER 2 LITERATURE STUDY

Automatic Level Difficulty Adjustment in Platform Games Using Genetic Algorithm Based Methodology

Efficient Large Integers Arithmetic by Adopting Squaring and Complement Recoding Techniques

MTBF PREDICTION REPORT

Evaluate the Effective of Annular Aperture on the OTF for Fractal Optical Modulator

Comparative Analysis of Reuse 1 and 3 in Cellular Network Based On SIR Distribution and Rate

A Novel Control Method for Input Output Harmonic Elimination of the PWM Boost Type Rectifier Under Unbalanced Operating Conditions

VI.C CIRCUIT BREAKERS

Phasor Representation of Sinusoidal Signals

Synchronous Machine Parameter Measurement

Target Response Adaptation for Correlation Filter Tracking

ICI Mitigation with CFO Compensation for OFDM in Mobile-to-Mobile Channel

THE GENERATION OF 400 MW RF PULSES AT X-BAND USING RESONANT DELAY LINES *

Chess players fame versus their merit

Chapter 13. Filters Introduction Ideal Filter

NATIONAL RADIO ASTRONOMY OBSERVATORY Green Bank, West Virginia SPECTRAL PROCESSOR MEMO NO. 25. MEMORANDUM February 13, 1985

Determination of the Multi-slot Transmission in Bluetooth Systems with the Estimation of the Channel Error Probability

Module 9. DC Machines. Version 2 EE IIT, Kharagpur

High Speed, Low Power And Area Efficient Carry-Select Adder

aperture David Makovoz, 30/01/2006 Version 1.0 Table of Contents

POLYTECHNIC UNIVERSITY Electrical Engineering Department. EE SOPHOMORE LABORATORY Experiment 1 Laboratory Energy Sources

MATH 118 PROBLEM SET 6

DESIGN OF CONTINUOUS LAG COMPENSATORS

problems palette of David Rock and Mary K. Porter 6. A local musician comes to your school to give a performance

Modeling and simulation of UPFC using PSCAD/EMTDC

Shunt Active Filters (SAF)

An Iterative Parameters Estimation Method for Hyperbolic Frequency Modulated Signals with Colored Noise

STUDY OF MATRIX CONVERTER BASED UNIFIED POWER FLOW CONTROLLER APPLIED PI-D CONTROLLER

ECE315 / ECE515 Lecture 5 Date:

A Novel GNSS Weak Signal Acquisition Using Wavelet Denoising Method

GLOBAL POWER SYSTEM CONTROL USING A UNIFIED POWER FLOW CONTROLLER

First Round Solutions Grades 4, 5, and 6

1 GSW Multipath Channel Models

MAXIMUM FLOWS IN FUZZY NETWORKS WITH FUNNEL-SHAPED NODES

AN ALTERNATE CUT-OFF FREQUENCY FOR THE RESPONSE SPECTRUM METHOD OF SEISMIC ANALYSIS

International Steering Committee:

fiziks Institute for NET/JRF, GATE, IIT JAM, M.Sc. Entrance, JEST, TIFR and GRE in Physics ELECTRONICS AND EXPERIMENTAL METHODS NET/JRF (JUNE-2011)

A Comparison of Two Equivalent Real Formulations for Complex-Valued Linear Systems Part 2: Results

Desensitized Kalman Filtering with Analytical Gain

DIMENSIONAL SYNTHESIS FOR WIDE-BAND BAND- PASS FILTERS WITH QUARTER-WAVELENGTH RES- ONATORS

The Clock-Aided Method for GPS Receiver Positioning in an Urban Environment

Torque Ripple in Switched Reluctance Motor Based on Asymmetrical Bridge Converter Pairote Thongprasri

Power Quality Enhancement Using Energy Storage Devices

An Optimal Method for Using Multiple Gateways in Cellular IP Networks

Digital Transmission

Design of an FPGA based TV-tuner test bench using MFIR structures

Calculation of the received voltage due to the radiation from multiple co-frequency sources

BUREAU INTERNATIONAL DES POIDS ET MESURES. Final report

GAME THEORY AND INTERFERENCE AVOIDANCE IN DECENTRALIZED NETWORKS

Degree Distribution Optimization in Raptor Network Coding

Synchronous Machine Parameter Measurement

Francis Gaspalou Second edition of February 10, 2012 (First edition on January 28, 2012) HOW MANY SQUARES ARE THERE, Mr TARRY?

Battery-aware Static Scheduling for Distributed Real-time Embedded Systems

IEE Electronics Letters, vol 34, no 17, August 1998, pp ESTIMATING STARTING POINT OF CONDUCTION OF CMOS GATES

Modeling and Control of a Cascaded Boost Converter for a Battery Electric Vehicle

ECE 2133 Electronic Circuits. Dept. of Electrical and Computer Engineering International Islamic University Malaysia

Graph Method for Solving Switched Capacitors Circuits

Chaotic Filter Bank for Computer Cryptography

Minimal K-Covering Set Algorithm based on Particle Swarm Optimizer

Space Time Equalization-space time codes System Model for STCM

(1) i. In this paper it is assumed that the process dynamics can be described with reasonable accuracy by a second-order-plusdead-time

Impact of Interference Model on Capacity in CDMA Cellular Networks. Robert Akl, D.Sc. Asad Parvez University of North Texas

FAST ELECTRON IRRADIATION EFFECTS ON MOS TRANSISTOR MICROSCOPIC PARAMETERS EXPERIMENTAL DATA AND THEORETICAL MODELS

AC network state estimation using linear measurement functions

ANNUAL OF NAVIGATION 11/2006

Figure 1. DC-DC Boost Converter

Control of assessment for demersal fish stocks in ICES area: analysis for 36 stocks and investigation of some potential bias sources.

Medium Term Load Forecasting for Jordan Electric Power System Using Particle Swarm Optimization Algorithm Based on Least Square Regression Methods

Practical Considerations of ASK Modulated Passive Tags

The Analysis and Simulation of Robot Kinematics and Dynamics Based on RoboAnalyzer

Lecture 20. Intro to line integrals. Dan Nichols MATH 233, Spring 2018 University of Massachusetts.

Multipath Mitigation in GPS/Galileo Receivers with Different Signal Processing Techniques

Comparison of Reference Compensating Current Estimation Techniques for Shunt Active Filter

Transcription:

LOW SENSITIVITY DESIGN OF MULTIPLIERLESS ELLIPTIC IIR DIGITAL FILTERS Vlentn Ilev Anzov Dertment of Telecommunctons, Techncl Unversty of Sof Techncl Unversty, Dertment of Telecommunctons, 1756 Sof, Bulgr, e-ml: v@tu-sof.bg In ths er n mroved method of desgn of low senstvty multlerless elltc IIR dgtl flters s roosed. The desgn technque s bsed on the senstvty nlyss nd selecton of roer llss sectons. The llss sectons used to relze the llss structures re wth mnmzed senstvtes. Usng the symbolc nlyss, the exressons for hse senstvtes to multler coeffcents for the selected llss sectons hve been derved. The mltude senstvtes of the obtned relztons hve been nvestgted. The comrson between the resulted dgtl multlerless IIR flters hs been mde for dfferent coeffcents wordlength. The results re confrmed exermentlly. Keywords: IIR dgtl flters, llss sectons, low senstvty, multlerless 1. INTRODUCTION In VLSI mlementtons multler s very costly. Therefore t s very ttrctve to reresent the multlctons s seres of shft nd dd oertons. In ths wy n multlerless relztons the overll flter comlexty s sgnfcntly reduced. The desgn of multlerless IIR dgtl flters s very dffcult becuse of the hgh senstvty of ther frequency resonse to the multler coeffcents. Therefore, t s very mortnt to mnmze the frequency resonses senstvty. When the flter senstvty s low t s ossble to shorten coeffcents wordlength wthout voltng the flter secfctons. It s well known tht the flter structure consstng of two llss subflters n rllel s hvng very low coeffcent senstvty (eseclly n the ssbnd). Ths structure s lso very economcl t s relzng two n th - order trnsfer functons usng only n multlers. In [1][][3], t hs been shown tht the rllel-llss-flter-structure my hve smll number of shft nd dd oertons f the trnsfer functon (TF) s derved by the blner trnsformton from n elltc mnml Q-fctors (EMQF) nlog rototye. Becuse of the roertes of EMQF flters [4], (n+1)/ multler coeffcents cn be mlemented wthout quntzton. Then the remnng (n-1)/ less senstve coeffcents re quntzed usng cnonc sgn dgt (CSD) code. Ths clss of flters hs ll coeffcents mlemented wth smll quntzton error. The urose of ths contrbuton s to modfy ths roch nd to ly t to desgn of nrrowbnd lowss low senstvty multlerless IIR flters. The llss sectons used to relzed the llss structures re wth mnmzed senstvtes for trnsfer functon coeffcents vlues corresondng to oles ner z 1 [5][6]. The frequency resonses of the multlerless relzton so obtned nd the hse senstvty to multler coeffcents re comred wth those of multlerless IIR flter mlemented usng llss sectons selected by Mlć nd Lutovc [1][].

. IIR DIGITAL FILTERS DESIGN USING AN EMQF ANALOG PROTOTYPE The desgn method suggested by Mlć nd Lutovc [1][] strts wth hse resonse senstvty nlyss of the llss sectons tht mlements the desred dgtl IIR flter to ts multler coeffcents. The (n+1)/ most senstve coeffcents cn be mlemented s exct vlues wth mnml number of shft nd dd oertons by n rorte desgn of EMQF trnsfer functon. Then the remnng (n-1)/ less senstve coeffcents re quntzed usng cnonc sgn dgt (CSD) code. It s shown n [1][] tht n IIR flter, f derved from n EMQF nlog rototye, cn be desgned to hve n ll second order sectons of the llss networks one common coeffcent whose vlue deends only on the frequency t whch the flter ttenuton s nd my be djusted ccordng to the redetermned number of shft nd dd oertons. In ths wy, the drect control s estblshed over the vlues of (n-1)/ multlcton constnts nd they cn be mlemented s exct vlues, elmntng thus the nfluence of ther hgh senstvtes. In ths contrbuton, multlerless elltc flter desgn s bsed on the TF H formed by the blner trnsformton from n EMQF nlog rototye [1][3][4]. An mortnt roerty of EMQF flter s tht ts oles re lced on crcle wth the center t the orgn of the s lne. The rdus of the crcle s Ω, where the normlzed stobnd edge frequency s Ω nd for s j Ω the flter hs the ttenuton of. The relton between f (ssbnd edge frequency of the dgtl flter) nd f (stobnd edge frequency of the dgtl flter) nd the stobnd edge frequency of the nlog rototye Ω hs been estblshed n [3]. For gven f nd f, Ω s estmted from the exresson Ω. (1) The frequency where dgtl flter hs the ttenuton of corresonds n the nlog flter domn to the frequency Ω, nd cn be determned by the relton tg π f. () The trnsfer functon of n odd order IIR flter H cn be resented s sum of two llss functons H nd H b. The llss subflters H nd H b re usully relzed s cscde connecton of frst nd second order sectons. In order to obtn nrrowbnd lowss low senstvty multlerless IIR flter relztons, we need llss sectons hvng low senstvtes when relzng oles ner z 1. Such sectons hve been roosed nd nvestgted n [5][6]. We shll cll them ST1 (Fg. 1) (the frst order) nd STA (Fg. 1b) (the second order secton) nd they relze the followng trnsfer functons: (1 ) + z 1 d (1 d)(1 c) z + z H ST 1( z), H ST A ( z). (3) 1 (1 ) z 1 (1 d)(1 c) z + (1 d) z

) b) Fg. 1 Low senstvty llss sectons: ) frst order (ST1); b) second order (STA) For comrson the sme desgn rocedure wll be led for the oulr lttce wve sectons selected by Mlć nd Lutovc n [1]. The TFs of these sectons (clled GM-sectons) re: H b + z 1( z) GM bz, 1 (1 1) z + z H GM ( z). (4) 1 (1 1) z 1z The coeffcents of the dgtl llss sectons hve been derved from the corresondng frst nd second order nlog rototyes Eq.(5) by blner trnsformton s + B s A 0 s + B S( s), S( s) s + B0 s. (5) + A s + B If both the relton between the oles oston nd the coeffcents of the TF nd Eq.(1), Eq.() re used, the followng exressons for coeffcents n llss sectons cn be obtned: 1 B0 b 1 b. (6) 1 + B0 A B 1 1 r. (7) A + B + 1 1 1 + B B B 0 1 + B0 c B 1 + B A d A + B +1 where r s the comlex ole rdus. 1 tg πf. (8) tg πf 1 +. (9) πf f 1 r tg c. (10) tg π d, (11) 3. SENSITIVITIES INVESTIGATIONS We hve nvestgted the worst-cse (WS) mltude resonse senstvtes of the obtned elltc IIR flters (STAST1, GM1GM) for two dfferent TFs orders (3

nd 5) usng the ckge PANDA [7]. The results for the WS senstvtes vs. frequency re gven n Fg. (thrd order TF) nd Fg. b (ffth order TF). The IIR flters obtned usng llss sectons ST1 nd STA re mrked s STAST1. Resectvely, the flter relztons wth GM-sectons re mrked s GM1GM. It s clerly seen tht IIR flters STAST1 hve mny tmes lower senstvtes wthn the entre frequency rnge, but the dfference s consderbly stronger n the stobnd nd eseclly t mxml onts. The WS senstvty nlyss n Fg. demonstrtes tht dfference between senstvtes of both relztons (STAST1 nd GM1GM) for ffth-order TF becomes even more sgnfcnt. ) b) Fg. Worst-cse senstvtes of dfferent IIR flter relztons: ) thrd order; b) ffth order 4. EXPERIMENTS We desgned nd nvestgted two sets of flters wth secfctons s gven n Exmles 1,. In both cses frst, the desgn rocedure s led for llss secton ST1 nd STA (wth TFs gven n Eq.(3)). Then, for comrson, the sme rocedure s used for GM-sectons (wth TFs gven n Eq.(4)). Usng the symbolc nlyss, the exressons for hse senstvtes to multler coeffcents for the selected sectons hve been derved. Fgs. 3, 4, 5 nd 6 dsly the hse resonse senstvtes to coeffcents of the corresondng frst or second-order sectons of the both exmles. Exmle1 The flter secfctons re s follows: F 0.006, F 0. 0, A 0. db, A 0dB. The flter coeffcents cn be dvded nto two grous tht wll be consdered sertely. The frst grou ncludes the most senstvty coeffcents (tht s - GM1GM nd c - STAST1) tht cn be controlled by TFs, nd the coeffcents 1 (GM1GM) or d (STAST1) reresentng the ole r closest to the unt crcle. The second grou ncludes the remnng coeffcents b (GM1GM) nd (STAST1). For gven secfctons, the rnge of ermtted vlues of or c s determned from F nd F ccordng to [1]. Then, ll vlues n the whole rnge re quntzed n CSDcode. The vlues of or c wth the smllest quntzton error s chosen. From Eq.(8) nd Eq.(10) the frequency f 3 db s evluted. Keeng the vlues of f 3 db unchnged, the

trnston bndwdth s to be djusted to rovde the sutble vlue for 1 or d ccordng to [1][3]. At the end ll remnng coeffcents re quntzed wth dfferent coeffcents wordlength. Tble I gves the results for ll coeffcents. Exmle The flter secfctons re: F 0.006, F 0. 01, A 0. 1dB, A 0 db. The secfctons re fulflled wth ffth order TF flter. The second order sectons ' '' hve common coeffcent (GM1GM) or c ' c' ' (STAST1). The coeffcents of the obtned multlerless IIR flters re gven n Tble I. We hve nvestgted the frequency resonses of the multlerless relztons so obtned. It s clerly seen from Fgs. 7 nd 8 tht the shes of the ttenuton chrcterstcs of the multlerless flters obtned ccordng to our roch (STAST1) do not dffer from the del even though coeffcents of the second grou re quntzed to 3bts. TABLE I MULTIPLIERLESS REPRESENTATION OF THE FILTER COEFFICIENTS STAST1 GMGM1 d 5 7 1 4 6 c 10 9 1 4 8 + + b 3 STAST1 GMGM1 d ' 8 10 + 1' 7 9 + c ', c' ' 11 ', ' ' 0 1 d '' 6 7 + 1 ' ' 4 6 5 6 + b 4 + Fg. 3 Phse resonse senstvty of 3 th order GMGM1 TF Fg. 4 Phse resonse senstvty of 3 th order STAST1 TF Fg. 5 Phse resonse senstvty of 5 th order GMGM1 TF Fg. 6 Phse resonse senstvty of 5 th order STAST1 TF

Fg. 7 Attenuton chrcterstcs Fg. 8 Attenuton chrcterstcs of 3 th order TFs of 5 th order TFs 5. CONCLUSIONS In ths er modfcton of the desgn method ntlly suggested by Mlć nd Lutovc hs been roosed. The hse senstvtes to multler coeffcents for the selected llss sectons hve been derved. The hlf most senstve coeffcents hve been reresented wth smll number of shfters nd dders. The remnng coeffcents re quntzed wth dfferent coeffcents wordlength. Consequently, the obtned multlerless IIR relztons re wth smll quntzton error nd low comlexty. It s shown tht the senstvtes of the obtned flters cn be consderbly decresed f llss sectons used re wth mnmzed senstvtes. It s frmly demonstrted (through exerments) tht flters desgned ccordng to our roch re styng wthn the gven secfctons even fter very severe quntzton of the multler coeffcents whle the mgntudes of the reference flters (s desgn n [1][]) re totlly destroyed. 6. REFERENCES [1] Mlc L, M. Lutovc, Desgn of multlerless elltc IIR flters wth smll quntzton error, IEEE Trns. Sgnl Processng, vol.47,.469-479, no., Feb.1999 [] Lutovc M, L. Mlc, Desgn of comuttonlly effcent elltc IIR flters wth reducednumber of shft-nd-dd oertons n multlers, IEEE Trns. Sgnl Processng, vol.45,.4-430, Oct.1997 [3] Mlc L, M. Lutovc, Low-comlexty lttce wve dgtl flters, Proceedngs of 13 th Euro Conf. ECCTD, Budest, Hungry,.1141-1146, Aug.1999 [4] Rbrenovc D, M. Lutovc, Elltc flters wth mnml Q-fctors, Electron. Lett, vol 30, no.3,. 06-07, Feb.1994. [5] Stoynov G, H. Clusert, A comrtve study of frst order dgtl ll-ss structures, Frequenz, vol. 48, No 9/10, Set./Oct. 1994. 1-6 [6] Stoynov, G., V. Dmtrov, Senstvty nd nose nlyss of second-order dgtl ll-ss sectons (n Bulg.). Proc. 3-th Sc. Conf. "Communcton, electroncs nd comuter systems'93", Sof, Bulgr, My 1993, vol. 1,. 104-115. [7] Sugno H, A. Nshhr, Frequency-domn smultor of dgtl networks from the structurl descrton, Trns.of the IEICE of Jn, Vol. E73, No.11,.1804-1806, Nov.1990.