Beruflich Dokumente
Kultur Dokumente
P
(Z
L
L-1
P=0
)Z
-P
=X
q
(Z
L
)
L-1
q=0
Z
-q
E
L-1
=0
(Z
L
)Z
-
(2)
Fig. 1. Two-parallel FIR filter implementation
using the FFA.
Fig. 2.Three-parallel FIR filter implementation using the FFA.
WhereX
q
(Z) = Z
-k
k=0
x(Ik +q),X
q
(Z) =
Z
-k
k=0
x(Ik +q),And
p
(Z) =
Z
-k
k=0
y(Ik +p) for p,q,r=0,1,2,,L-1
A. 2 2 FFA (L =2)
According to (2), a two-parallel FIR filter can be expressed
as [1], [3]
0
=E
0
X
0
+Z
-2
E
1
X
1
)
1
=E
0
X
1
+Z
-2
E
1
X
0
(3)
However, (3) can be written as
0
=E
0
X
0
+Z
-2
E
1
X
1
1
=(E
0
+E
1
)(X
0
+X
1
) E
0
X
0
E
1
X
1
(4)
The two-parallel (L =2) FIR filter implementation using the
FFA obtained from(4) is shown in Fig. 1.
B. 3 3 FFA (L =3)
By the similar approach, a three-parallel FIR filter using the
FFA can be expressed as [1], [3]
However, (3) can be written as
Y0=E
0
X
0
Z
-3
E
2
X
2
+x
2
[(E
1
+E
2
)(X
1
+X
2
)
E
1
X
1
]
1=[(E
0
+E
1
)(X
0
+X
1
) E
1
X
1
] (E
0
X
0
Z
-3
E
2
X
2
)
2
=[(E
0
+E
1
+E
2
)(X
0
+X
1
+X
2
)]
[(E
0
+E
1
)(X
0
+X
1
) E
1
X
1
]
[(E
1
+E
2
)(X
1
+X
2
) E
1
X
1
] (5)
The implementation obtained from(5) is shown in
Fig. 2.
III. PROPOSED STRUCTURES FOR SYMMETRIC
CONVOLUTIONS OF ODD LENGTH
To exploit the symmetry of coefficients, the main idea is to
manipulate the polyphase decomposition to earn as many
subfilter blocks as possible, which contain symmetric
coefficients so that half the number of multipliers within a
single subfilter block can be utilized for the multiplications of
whole taps.
Fig. 3A. Implementation of the proposed structure
A. 3 3 Proposed FFA (L =3)
International Journal of Engineering Trends and Technology- Volume4Issue2- 2013
ISSN: 2231-5381 http://www.internationaljournalssrg.org Page 83
The existing two-parallel FFA structure naturally has
benefitsto symmetric convolutions in odd length. When it
comes to a set of odd-length symmetric coefficients, two out
of three subfilters contain symmetric coefficients, i.e., H0 and
H1, shown in Fig. 1. However, the existing three-parallel FFA
structure is not as advantageous. In this section, new three-
parallel FIR filter structures are proposed, which enables more
multipliers sharing in the subfilter section and, therefore, can
save more hardware cost over the existing FFA.
1) Proposed Structure 3A, ((N mod 3) =0): From(5), it
can also be presented as (7). For a set of symmetric
coefficients in odd length N, when (N mod 3) equals zero, (7)
can earn
two more subfilter blocks containing symmetric coefficients
than (5). The implementation of the proposed three-parallel
FIRfilter based on (7) is shown in Fig. 3.An example is
demonstrated here for a clearer perspective.
Example 1: Consider a 27-tap FIR filter with a set of
symmetric coefficients as follows:
{ h(0), h(1), h(2), h(3), h(4), h(5),
h(6), h(7), h(8), h(9), . . . , h(26)}
where h(0)=h(26), h(1)=h(25), h(2)=h(24), h(3)=h(23),
h(4) =h(22), h(5) =h(21), . . . , h(12) =h(14), applying to
the proposed structure 3A, and then, we gain two more
subfilterblocks with symmetric coefficients as
Fig. 4.Subfilter block implementation with symmetric coefficients.
Fig. 5. Comparison of subfilter blocks between the existing FFA and the
proposed structure 3A.
TABLE I
COMPARISON OF PROPOSED AND THE EXISTI NG FFA STRUCTURES
NUMBER OF REQUIRED MULTIPLIERS (M.), REDUCED MULTIPLIERS
(R.M.), NUMBER OF REQUIRED ADDERS IN SUBFILTER SECTION (SUB.),
NUMBER OF REQUIRED ADDERS IN PRE/POSTPROCESSING BLOCKS
(PRE/POST.), NUMBER OF THE INCREASED ADDERS (I.A.)
In addition, the proposed and FFA structures are implemented
in Verilog HDL for an 81-tap linear-phase FIR filter with
word length of 32 bits. The synthesis results by Xilinx Design
Complier and the results are shown in Table II.
Advanced HDL Synthesis Report Table analysis:
L
Len
gth
Struc
ture
Used
Multi
pliers
Reduc
ed
Multi
pliers
Required
Adders
Incre
ased
Adde
rs
Subtr
actor
Pre
/
Pos
t
pro
cess
2
1
tap
FFA 3
- 2
5
3 Prop
osed
3 8
9
tap
FFA 27
6
18
45
27 Prop
osed
21 72
27
tap
FFA 81
8
54
135
81 Prop
osed
73 216
81
tap
FFA 243
24 162
405
243 Prop
osed
219 648
3
1
tap
FFA 3
1 2
5
3 Prop
osed
2 8
9
tap
FFA 27
7 16
45
27 Prop
osed
20 72
27
tap
FFA 81
11 52
135
81 Prop
osed
70 216
81
tap
FFA 243
23 190
405
243 Prop
osed
220 648
A comparison between the proposed and the existing FFA
International Journal of Engineering Trends and Technology- Volume4Issue2- 2013
ISSN: 2231-5381 http://www.internationaljournalssrg.org Page 84
structures with various lengths under different levels of
parallelismis presented in Table I.
TABLE II
SYNTHESIS RESULTS FOR AN 81-TAP LINEAR-PHASE
FIR DIGI TAL FILTER
TABLE III
Area Analysis Graph.
L Length Structure
Number
of Slices
Number
of Slice
Flip
Flops
Number
of 4 input
LUTs
L=2
1 tap
FFA 202 40 353
Proposed 860 45 1627
9 tap
FFA 1445 301 2706
Proposed 4603 333 8422
27 tap
FFA 4270 881 7998
Proposed 12875 950 23707
81 tap
FFA 12713 2631 23875
Proposed 37900 2859 69559
L=3
1 tap
FFA 287 71 528
Proposed 592 83 1152
9 tap
FFA 2133 552 3884
Proposed 4679 595 8720
27 tap
FFA 6277 1639 11441
Proposed 13837 1747 25764
81 tap
FFA 18583 4902 33895
Proposed 41244 5203 76770
Gragh.
SIMULATION RESULTS
Two Parallel Single TAF Filter
9 TAF FILTER
27 TAF FILTER
0 4000080000
FFA
FFA
FFA
FFA
1
t
a
p
9
t
a
p
27
t
a
p
81
t
a
p
1
t
a
p
9
t
a
p
27
t
a
p
81
t
a
p
L
=
2
L
=
3
Number of
bonded IOBs
Number of 4
input LUTs
Number of Slice
Flip Flops
International Journal of Engineering Trends and Technology- Volume4Issue2- 2013
ISSN: 2231-5381 http://www.internationaljournalssrg.org Page 85
81 TAF FILTER
IV.CONCLUSION
In this brief, we have presented new parallel FIR filter
structures, which are beneficial to symmetric convolutions of
odd length. Multipliers are the major portions in hardware
consumption for the parallel FIR filter implementation. The
proposed new structures exploit the nature of symmetric
coefficients of odd length and further reduce the amount of
multipliers required at the expense of additional adders. Since
multipliers outweigh adders in hardware cost, it is profitable
to exchange multipliers with adders. Moreover, the number of
increased adders stays still when the length of FIR filter
becomes large, whereas the number of reduced multipliers
increases along with the length of the FIR filter.
REFERENCES
[1] D. A. Parker and K. K. Parhi, Low-area/power parallel FIR digital filter
implementations, J. VLSI Signal Process. Syst., vol. 17, no. 1, pp. 7592,
Sep. 1997.
[2] J . G. Chung and K. K. Parhi, Frequency-spectrum-based low-area
lowpower parallel FIR filter design, EURASIP J. Appl. Signal Process.,
vol. 2002, no. 9, pp. 444453, J an. 2002.
[3] K. K. Parhi, VLSI Digital Signal Processing systems: Design and
Implementation. New York: Wiley, 1999.
[4] Z.-J. Mou and P. Duhamel, Short-length FIR filters and their use in
fast nonrecursive filtering, IEEE Trans. Signal Process., vol. 39, no. 6,
pp. 13221332, J un. 1991.
[5] J. I. Acha, Computational structures for fast implementation of L-path
and L-block digital filters, IEEE Trans. Circuits Syst., vol. 36, no. 6,
pp. 805812, J un. 1989.
[6] C. Cheng and K. K. Parhi, Hardware efficient fast parallel FIR filter
Structures based on iterated short convolution, IEEE Trans. Circuits Syst.
I, Reg. Papers, vol. 51, no. 8, pp. 14921500, Aug. 2004.
[7] C. Cheng and K. K. Parhi, Furthur complexity reduction of parallel FIR
filters, in Proc. IEEE ISCAS, May 2005, vol. 2, pp. 18351838.
[8] C. Cheng and K. K. Parhi, Low-cost parallel FIR structures with 2-stage
parallelism, IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 54, no. 2,
pp. 280290, Feb. 2007.
[9] I.-S. Lin and S. K. Mitra, Overlapped block digital filtering, IEEE
Trans. Circuits Syst. II, Analog Digit. Signal Process., vol. 43, no. 8,
pp. 586596, Aug. 1996.
[10] Y.-C. Tsao and K. Choi, Area-efficient parallel FIR digital filter
structures for symmetric convolutions based on fast FIR algorithm, IEEE
Trans. Very Large Scale Integration. (VLSI) Syst., vol. 20, no. 2, pp. 366371,
Feb. 2010.