Scribd
Upload
2K views

Lattice Structure For FIR Filter

This document discusses lattice filters and their implementation. It begins with a review of previous lectures on discrete-time systems and filter implementations. The objectives are then stated as deriving lattice structures for FIR filters and developing a method to convert lattice filters to direct-form filters. The document proceeds to outline the derivation of lattice structures for FIR filters in multiple stages. It shows that the outputs of each stage can be expressed as forward and backward predictions related to the direct-form filter coefficients. Finally, it discusses converting between lattice and direct-form representations, including working through an example conversion from a three-stage lattice filter to direct-form coefficients.

Uploaded by

jaishpanjwani
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
2K views

Lattice Structure For FIR Filter

This document discusses lattice filters and their implementation. It begins with a review of previous lectures on discrete-time systems and filter implementations. The objectives are then stated as deriving lattice structures for FIR filters and developing a method to convert lattice filters to direct-form filters. The document proceeds to outline the derivation of lattice structures for FIR filters in multiple stages. It shows that the outputs of each stage can be expressed as forward and backward predictions related to the direct-form filter coefficients. Finally, it discusses converting between lattice and direct-form representations, including working through an example conversion from a three-stage lattice filter to direct-form coefficients.

Uploaded by

jaishpanjwani
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

ELEG305: Digital Signal Processing

Lecture 19: Lattice Filters Kenneth E. Barner


Department of Electrical and Computer Engineering University of Delaware

Fall 2008

K. E. Barner (Univ. of Delaware)

ELEG305: Digital Signal Processing

Fall 2008

1 / 19

Outline

Review of Previous Lecture

Lecture Objectives

Implementation of Discrete-Time Systems Lattice Structures for FIR systems Lattice to DirectForm Translation

K. E. Barner (Univ. of Delaware)

ELEG305: Digital Signal Processing

Fall 2008

2 / 19

Review of Previous Lecture

Review of Previous Lecture


Efcient FFT computation for real sequences For x1 (n) and x2 (n) real, set x(n) = x1 (n) + jx2 (n) then X1 (k ) = 1 [X (k ) + X (N k )]; 2 X2 (k ) = 1 [X (k ) X (N k )] 2j

Similar tricks hold for length 2N real sequences Linear ltering computation of the DFT the Goertzel algorithm
k yk (n) = WN yk (n 1) + x(n)

gives X (k ) = yk (n)|n=N Implementations of FIR discretetime systems Direct, Cascade, and Lattice lter structures
K. E. Barner (Univ. of Delaware) ELEG305: Digital Signal Processing Lecture Objectives Fall 2008 3 / 19

Lecture Objectives

Objective Derive lattice structures for FIR lters; Develop a method for converting lattice lter coefcients to directform (FIR) coefcients Reading Chapters 9 (9.2); Next lecture, complete lattice lters, structures for IIR systems (Chapter 9.29.3); start lter design (Chapter 10.110.2)

K. E. Barner (Univ. of Delaware)

ELEG305: Digital Signal Processing

Fall 2008

4 / 19

Implementation of Discrete-Time Systems

Lattice Structures for FIR systems

Objective: Developed a latticestructure realization of FIR lters Approach: Suppose we have a sequence of FIR lters Hm (z) = Am (z) where by denition
m

m = 0, 1, . . . , M 1

Am (z) = 1 +
k =1

m (k )z k

m1

Suppose m = 1. Then the output of H1 (z), for input x(n), is y (n) = x(n) + 1 (1)x(n 1) f0 (n) = x(n) g0 (n) = x(n) f1 (n) = f0 (n) + K1 g0 (n 1) g1 (n) = K1 f0 (n) + g0 (n 1) Question: For what K1 value is () realized, i.e., f1 (n) = y (n)?
K. E. Barner (Univ. of Delaware) ELEG305: Digital Signal Processing Lattice Structures for FIR systems Fall 2008 5 / 19 Implementation of Discrete-Time Systems

()

To build a cascade of 1st order stages, generalize the notation. Let

Single Stage Lattice Filter

To realize (), let K1 = 1 (1). Then f1 (n) = f0 (n) + K1 g0 (n 1) = x(n) + 1 (1)x(n 1) = y (n) Similarly, [order m = 1 lter output] g1 (n) = 1 (1)x(n) + x(n 1) g1 (n) = K1 f0 (n) + g0 (n 1)
Fall 2008 6 / 19

Note: The stage 1 governing equations are f1 (n) = f0 (n) + K1 g0 (n 1)


K. E. Barner (Univ. of Delaware)

where we set the reection coefcient as K1 = 1 (1) to realize ()


ELEG305: Digital Signal Processing

Implementation of Discrete-Time Systems

Lattice Structures for FIR systems

Objective: Add a second stage and equate to the m = 2 lter Note that [m = 2 lter] H2 (z) = A2 (z) = 1 + 2 (1)z 1 + 2 (2)z 2 y (n) = x(n) + 2 (1)x(n 1) + 2 (2)x(n 2) The two stage lattice lter is given by ()

The governing equations are f1 (n) = f0 (n) + K1 g0 (n 1) g1 (n) = K1 f0 (n) + g0 (n 1) f2 (n) = f1 (n) + K2 g1 (n 1) g2 (n) = K2 f1 (n) + g1 (n 1)

Procedure: Set K1 and K2 such that () is realized Note: f0 (n) = g0 (n) = x(n) [pipeline input] and f2 (n) = y (n) [output]
K. E. Barner (Univ. of Delaware) ELEG305: Digital Signal Processing Lattice Structures for FIR systems Fall 2008 7 / 19 Implementation of Discrete-Time Systems

[lattice equations] By substitution

f1 (n) = f0 (n) + K1 g0 (n 1)

f2 (n) = f1 (n) + K2 g1 (n 1)

g1 (n) = K1 f0 (n) + g0 (n 1) g2 (n) = K2 f1 (n) + g1 (n 1)

f2 (n) = f1 (n) + K2 g1 (n 1) = [f0 (n) + K1 g0 (n 1)] + K2 [K1 f0 (n 1) + g0 (n 2)] = [x(n) + K1 x(n 1)] + K2 [K1 x(n 1) + x(n 2)] = x(n) + K1 (1 + K2 )x(n 1) + K2 x(n 2) Recalling () [direct FIR form] y (n) = x(n) + 2 (1)x(n 1) + 2 (2)x(n 2) and equating with () [lattice form] yields 2 (2) = K2 or, equivalently, K2 = 2 (2)
K. E. Barner (Univ. of Delaware)

()

()

and

2 (1) = K1 (1 K2 ) 2 (1) 1 + 2 (2)


Fall 2008 8 / 19

and

K1 =

ELEG305: Digital Signal Processing

Implementation of Discrete-Time Systems

Lattice Structures for FIR systems

Process: This is repeated for M 1 stages, with general recursion f0 (n) = g0 (n) = x(n) fm (n) = fm1 (n) + Km gm1 (n 1), gm (n) = Km fm1 (n) + gm1 (n 1), At the nal stage y (n) = fm1 (n)
m1

m = 1, 2, . . . , M 1 m = 1, 2, . . . , M 1

=
k =0

m1 (k )x(n k )
ELEG305: Digital Signal Processing Lattice Structures for FIR systems Fall 2008 9 / 19

K. E. Barner (Univ. of Delaware)

Implementation of Discrete-Time Systems

Observation: At each stage there are two outputs, fm (n) and gm (n) Consider g2 (n). First & second stage equations: f1 (n) = f0 (n) + K1 g0 (n 1) g1 (n) = K1 f0 (n) + g0 (n 1) g2 (n) = K2 f1 (n) + g1 (n 1) = K2 [f0 (n) + K1 g0 (n 1)] + [K1 f0 (n 1) + g0 (n 2)] = K2 x(n) + K1 (1 + K2 )x(n 1) + x(n 2) using the prior result K2 = 2 (2) and K1 (1 K2 ) = 2 (1), g2 (n) = 2 (2)x(n) + 2 (1)x(n 1) + x(n 2) Result: The coefcients for gm (n) are reverse order those for fm (n)
K. E. Barner (Univ. of Delaware) ELEG305: Digital Signal Processing Fall 2008 10 / 19

f2 (n) = f1 (n) + K2 g1 (n 1) g2 (n) = K2 f1 (n) + g1 (n 1)

Apply substitution similarly to the previous case

Implementation of Discrete-Time Systems

Lattice Structures for FIR systems

Result: The two outputs at arbitrary stage m are expressed as


m m

fm (n) =
k =0

m (k )x(n k )

and

gm (n) =
k =0

m (k )x(n k )

where m (k ) = m (m k ), Taking the ztransform Fm (z) = Am (z)X (z) Similarly Gm (z) = Bm (z)X (z)
K. E. Barner (Univ. of Delaware)

k = 0, 1, . . . , m

Note: fm (n) is the forward prediction; gm (n) the backward prediction Fm (z) Fm (z) = X (z) F0 (z) Gm (z) Gm (z) = X (z) G0 (z)
Fall 2008 11 / 19

or

Am (z) =

or

Bm (z) =

ELEG305: Digital Signal Processing Lattice Structures for FIR systems

Implementation of Discrete-Time Systems

Recall Am (z) =
k =0

m (k )z k

Question: How does Bm (z) relate to Am (z)?


m

Bm (z) =
k =0 m

m (k )z k

[substitute m (k ) = m (m k )] [let l = m k ]

=
k =0 m

m (m k )z k m (l)z lm
l=0 m

= z m
l=0

m (l)z l

= z m Am (z 1 ) Result: B(z) has reciprocal zeros of A(z); B(z) is the reciprocal, or reverse, polynomial of A(z)
K. E. Barner (Univ. of Delaware) ELEG305: Digital Signal Processing Fall 2008 12 / 19

Implementation of Discrete-Time Systems

Lattice Structures for FIR systems

Lattice Filter Representation Summary


Case 1: FIR lter representations
m m

fm (n) =
k =0

m (k )x(n k )

and

gm (n) =
k =0

m (k )x(n k )

Case 2: zdomain representations Fm (z) = Am (z)X (z) Gm (z) = Bm (z)X (z) Also, Bm (z) = z m Am (z 1 ), f0 (n) = g0 (n) = x(n) fm (n) = fm1 (n) + Km gm1 (n 1), gm (n) = Km fm1 (n) + gm1 (n 1),
K. E. Barner (Univ. of Delaware) Implementation of Discrete-Time Systems

or or

Fm (z) X (z) Gm (z) Bm (z) = X (z) Am (z) =

m = 1, 2, . . . , M 1

Case 3: Recursion lattice representations m = 1, 2, . . . , M 1 m = 1, 2, . . . , M 1


Fall 2008 13 / 19

ELEG305: Digital Signal Processing Lattice Structures for FIR systems

Taking ztransform of the recursion lattice representations F0 (z) = G0 (z) = X (z) Fm (z) = Fm1 (z) + Km z 1 Gm1 (z), Gm (z) = Km Fm1 (z) + z 1 Gm1 (z), Dividing by X (z) and using Am (z) =
Fm (z) X (z)

m = 1, 2, . . . , M 1 m = 1, 2, . . . , M 1
Gm (z) X (z)

and Bm (z) =

Case 4: zdomain recursion lattice representations A0 (z) = B0 (z) = 1 Am (z) = Am1 (z) + Km z 1 Bm1 (z), Bm (z) = Km Am1 (z) + z 1 Bm1 (z), or in matrix notation Am (z) Bm (z)
K. E. Barner (Univ. of Delaware)

m = 1, 2, . . . , M 1 m = 1, 2, . . . , M 1

1 Km Km 1

Am1 (z) z 1 Bm1 (z)


Fall 2008 14 / 19

ELEG305: Digital Signal Processing

Implementation of Discrete-Time Systems

Lattice to DirectForm Translation

Design Questions: How do we convert lattice coefcients (K terms) to direct-form lter coefcients ( terms)? How do we convert directform coefcients to lattice coefcients? Case 1: Lattice to DirectForm Translation Approach: Utilize relations A0 (z) = B0 (z) = 1 Am (z) = Am1 (z) + Km z 1 Bm1 (z), Bm (z) = z m Am (z 1 ), m = 1, 2, . . . , M 1 m = 1, 2, . . . , M 1

Solve for terms recursively, starting with m = 1 Example Determine the direct form realization of a threestage (M = 4) lattice lter with coefcients K1 = 1/2, K2 = 1/2, K3 = 1/4.
K. E. Barner (Univ. of Delaware) ELEG305: Digital Signal Processing Lattice to DirectForm Translation Fall 2008 15 / 19 Implementation of Discrete-Time Systems

Start recursion set m = 1 A1 (z) = A0 (z) + K1 z 1 B0 (z) 1 () = 1 + z 1 2 Set () equal to 1 polynomial A1 (z) = 1 + 1 (1)z 1 1 1 (1) = 2 Note, Bm (z) is the reverse polynomial of Am (z) 1 B1 (z) = z 1 A1 (z 1 ) = z 1 1 + z 2 1 1 = +z 2
K. E. Barner (Univ. of Delaware) ELEG305: Digital Signal Processing Fall 2008 16 / 19

Implementation of Discrete-Time Systems

Lattice to DirectForm Translation

For m = 2, use recursion Am (z) = Am1 (z) + Km z 1 Bm1 (z) A2 (z) = A1 (z) + K2 z 1 B1 (z) 1 1 1 z = 1 + z 1 + 2 2 3 1 1 2 = 1+ z + z 4 2 equating terms yields 2 (1) = 3 4 and 2 (2) = 1 2

1 + z 1 2

Reversing the polynomial of Am (z) to get Bm (z) B2 (z) = z 2 A2 (z 1 ) = 1 3 1 + z + z 2 2 4


Fall 2008 17 / 19

K. E. Barner (Univ. of Delaware)

ELEG305: Digital Signal Processing Lattice to DirectForm Translation

Implementation of Discrete-Time Systems

For m = 3, A3 (z) = A2 (z) + K3 z 1 B2 (z) 3 1 1 1 = 1 + z 1 + z 2 + z 4 2 4 7 11 2 1 3 = 1 + z 1 + z + z 8 16 4 Thus 3 (1) = 7 , 8 3 (2) = 11 , 16 3 (3) = 1 4 1 3 1 + z + z 2 2 4

Result: The system in direct-form: (where 3 (0) = 1)


3

y (n) =
k =0

3 (k )x(n k )

7 11 1 = x(n) + x(n 1) + x(n 2) + x(n 3) 8 16 4


K. E. Barner (Univ. of Delaware) ELEG305: Digital Signal Processing Fall 2008 18 / 19

Lecture Summary

Lecture Summary
Lattice Filters general recursion f0 (n) = g0 (n) = x(n) fm (n) = fm1 (n) + Km gm1 (n 1) gm (n) = Km fm1 (n) + gm1 (n 1), Lattice to DirectForm Translation utilize recursion on A0 (z) = B0 (z) = 1 Am (z) = Am1 (z) + Km z 1 Bm1 (z), Bm (z) = z m Am (z 1 ), m = 1, 2, . . . , M 1 m = 1, 2, . . . , M 1

Next lecture Complete lattice lters, structures for IIR systems (Chapter 9.29.3); start lter design (Chapter 10.110.2)
K. E. Barner (Univ. of Delaware) ELEG305: Digital Signal Processing Fall 2008 19 / 19

You might also like