Multiplierless decimation filters based on amplitude sharpening and compensation

doctoral thesis

• Title (croatian): Decimacijski filtri bez množila temeljeni na izoštravanju i kompenzaciji amplitude
• Author: Aljoša Dudarin; MBZ: -355257
• Mentor: Mladen Vučić (mentor)
• Mentor: Goran Molnar (komentor)
• Committee member: Mladen Vučić (član povjerenstva)
• Committee member: Goran Molnar (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Electronic Systems and Information Processing); Zagreb
• Defense date and country: 2019, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing; Data Processing
• Universal decimal classification (UDC): 621.3 - Electrical engineering

Abstract

The simplest multiplierless decimation filter is the cascaded-integrator-comb (CIC) filter. However, CIC filters introduce a passband droop, which is intolerable in many applications. The droop can be reduced by connecting a linear-phase finite-impulse-response filter called compensator in cascade with CIC filter. Since CIC filters are multiplierless, CIC compensators with multiplierless structures are preferable. In the thesis, two methods for the design of multiplierless CIC compensators have been proposed. Both methods are based on minimization of the maximum passband deviation. However, the first method provides an efficient compensation by using coefficients expressed as sums of powers of two (SPT), whereas the second method brings simple compensator's structures by representing each coefficient as signed power of two. In both approaches, the optimum coefficients are found by using global optimization. In processing of wideband signals, CIC filter is often incapable of meeting the requirement for high folding-band attenuations. To improve CIC filter folding-band response, various structures have been developed. An efficient structure arises from polynomial sharpening of the folding-band response. This structure implements a so-called sharpened CIC (SCIC) filter. To obtain very high folding-band attenuations of SCIC filters, the minimax sharpening of the folding bands is proposed. In addition, to obtain multiplierless SCIC structures, polynomials with SPT coefficients are used. However, the SCIC response also introduces a high passband droop. The droop can be reduced by connecting a compensator in cascade with the SCIC filter. For the multiplierless SCIC filters, multiplierless compensators are preferable. In the thesis, two approaches for design of multiplierless SCIC compensators are proposed. The first approach brings a closed-form method based on maximally flat approximation. Such an approximation is suitable for narrowband SCIC filters. The second approach results in a global method based on the minimization of the maximum passband deviation. This method is preferable for wideband SCIC filters.

Abstract (croatian)

Najjednostavniji decimacijski filtar bez množila sastoji se od kaskade integratorskih i češljastih sekcija (Cascaded Integrator-Comb, CIC). Međutim, CIC filtri uvode veliki pad amplitudne karakteristike u području propuštanja, koji nije prihvatljiv u mnogim primjenama. Ovaj pad može se smanjiti spajanjem filtra s konačnim impulsnim odzivom i linearnom fazom, zvanog CIC kompenzator, u kaskadu sa CIC filtrom. S obzirom da CIC filtri ne sadrže množila, preferiraju se i CIC kompenzatori bez množila. U radu su predložene dvije metode za projektiranje CIC kompenzatora koji ne sadrže množila. Obje metode su temeljene na minimizaciji maksimalnog odstupanja u području propuštanja. Prva metoda osigurava učinkovitu kompenzaciju korištenjem koeficijenata izraženih sumama potencija broja dva (Sum of Power of Two, SPT), dok druga metoda donosi jednostavne strukture kompenzatora izražavajući svaki koeficijent kao predznačenu potenciju broja dva. U oba pristupa, optimalni koeficijenti se dobivaju primjenom globalnih optimizacijskih postupaka. U obradi širokopojasnih signala, CIC filtar često ne može osigurati dovoljno velika gušenja u područjima preklapanja spektra. Za poboljšanje amplitude CIC filtra u ovim područjima razvijene su razne strukture. Jedna od učinkovitih struktura proizlazi iz polinomnog izoštravanja amplitude. Ova struktura implementira takozvani izoštreni CIC (Sharpened CIC, SCIC) filtar. Kako bi se postiglo veliko gušenje preklopljenih signala kod SCIC filtera, predloženo je izoštravanje područja preklapanja spektra temeljeno na minimax aproksimaciji. Osim toga, da bi se dobile SCIC strukture bez množila, korišteni su polinomi s SPT koeficijentima. Međutim, odziv SCIC filtra ima veliki propad u području propuštanja. Propad se može smanjiti spajanjem kompenzatora u kaskadi s SCIC filtrom. Za SCIC filtre bez množila, također su poželjni kompenzatori bez množila. U radu se predlažu dva pristupa za projektiranje ovakvih kompenzatora. Prvi pristup rezultira eksplicitnom metodom temeljenom na maksimalno glatkoj aproksimaciji. Takva aproksimacija je prikladna za uskopojasne SCIC filtre. Drugi pristup rezultira globalnom metodom koja se temelji na minimiziranju maksimalne devijacije u području propuštanja. Ova metoda je prikladna za širokopojasne SCIC filtre.

• Keywords: cascaded-integrator-comb (CIC) filters; compensation; decimation; finite-impulse-response (FIR) filters; maximally flat approximation; minimax approximation; multiplierless; sharpening; signed power of two; sum of power of two (SPT)
• Keywords (croatian): kaskada integratorskih i češljastih sekcija; kompenzacija; decimacija; filtar sa konačnim impulsnim odzivom; maksimalno glatka aproksimacija; minimax aproksimacija; strukture bez množila; izoštravanje; predznačne potencije broja dva; suma potencija broja dva
• Language: english
• URN:NBN: urn:nbn:hr:168:596477
• Study programme: Title: Electrical Engineering and Computing; Study programme type: university; Study level: postgraduate; Academic / professional title: Doktor znanosti elektrotehnike i računarstva (Doktor znanosti elektrotehnike i računarstva)
• Type of resource: Text
• Extent: 99 str. ; 30 cm
• File origin: Born digital
• Access conditions: Open access
• Created on: 2020-01-21 10:32:35