Decimation method and decimation filter

• US 6,173,302 B1
• Filed: 07/27/1998
• Issued: 01/09/2001
• Est. Priority Date: 12/04/1996
• Status: Expired due to Term

• Current Assignee: RPX Corporation
• Original Assignee: Nokia Telecommunications Oy (Nokia Corporation)
• Inventors: Piirainen, Olli
• Primary Examiner(s): NGO, CHUONG D
• Application Number:
  US09/117,318
• Time in Patent Office: 897 Days
• Field of Search: 208/300, 208/313, 208/319, 208/290
• US Class Current: 708/313
• CPC Class Codes: H03H 17/0621 with input-sampling frequency and output-delivery frequency which differ, e.g. extrapolation; Anti-aliasing
• CPC Class Codes: H03H 17/0671 Cascaded integrator-comb [CIC] filters

Abstract

The invention relates to a decimation method and a decimation filter, which provide the transfer function of a CIC decimation filter. Decimation of the number of order N is performed by coefficient M, whereby FIR filtering is carried out in the main branch (100) of the decimation filter by M-tap FIR filters (20) N times in sequence, decimation is carried out by decimation means (21) by coefficient M and comb-filtering is carried out by means (22) N−1 times. The effect of sample cycles M preceding the sample cycle M to be processed is taken into account by using side branches (101, 102, 104) and additional branches (103, 105, 106, 107).

Claims

1. A decimation method the transfer function of which substantially corresponds to the transfer function of a CIC decimation filter, and in which a signal including digital samples is decimated by coefficient M, characterized in that at least main processing is performed on the signal to be decimated, the main processing comprising the steps of performing M-tap FIR filtering N times in sequence, decimating by coefficient M and comb-filtering the signal N−

1 times.

2. A method as claimed in claim 1, characterized in that if M-tap FIR filtering is performed in the main processing in sequence more than once {N>

1}, a first additional processing is performed on the signal between each of two successive FIR filterings R and R+1{R+1≦

N}, the first additional processing comprising the steps of

decimating by coefficient M;

delaying the decimated signal by one sample cycle M;

performing FIR filtering on the decimated and delayed signal N−

R times; and

adding the FIR filtered signal to the first comb-filtered signal.

3. A method as claimed in claim 2, characterized in that if FIR filtering is performed on the signal decimated and delayed in the first additional processing more than once {N−

R>

1}, a second additional processing is performed at least once, for which the signal between each of two successive FIR filterings is selected from the first additional processing or from a previous second additional processing, the second additional processing comprising the steps of

delaying the signal by one sample cycle M;

performing FIR filtering on the delayed signal as many times as in the first additional processing or in a previous second additional processing after the selection; and

adding the signal to the Pth {P≦

N−

1} comb-filtered signal according to the number P of delays by sample cycles M.

4. A method as claimed in claim 3, characterized in that in M-tap FIR filtering one is substantially used as the coefficient of taps.

5. A method as claimed in claim 3, characterized in that a power of two is used as the decimation coefficient M.

6. A method as claimed in claim 5, characterized in that FIR filtering of a signal is performed in the first and second additional processing in such a manner that the signal is multiplied by a predefined power of two and/or the signal is multiplied by predefined powers of two and added by multiplying the formed signals.

7. A method as claimed in claim 6, characterized in that multiplying is performed by a logic shift of a signal or by logic shifts of a signal and by adding the logically shifted signals together.

8. A decimation filter the transfer function of which substantially corresponds to the transfer function of a CIC decimation filter, and which is arranged to decimate a signal including digital samples by coefficient M, characterized in that the decimation filter at least comprises a main branch (100), which comprises a decimation means (21), which decimate by coefficient M, N M-tap FIR filters (20) in sequence before the decimation means (21) of the main branch (100), and N−

1 combinations (22) of a comb filter and an adder in sequence after the decimation means of the main branch (100).

9. A decimation filter as claimed in claim 8, characterized in that if the number of M-tap FIR filters (20) is more than one, the decimation filter is arranged to take into account N−

1 sample cycles M preceding the incoming sample cycle M, and thus the decimation filter comprises besides the main branch (100) N−

1 side branches (101, 102, 104), the divergence point (120, 124, 126) of which is operationally connected between each of two FIR filters (20) before the decimation means (21) of the main branch (100) and their other end is operationally connected to an adder (25) in connection with the first comb filter, the side branches comprising

decimation means (21) for decimating a signal coming to the side branch by coefficient M;

delay means (23) for delaying the decimated signal by one sample cycle M;

a number of successive FIR filters (20) or means performing an equivalent operation corresponding to the number of FIR filters between the divergence point (120, 124, 126) and the decimation means (21) in the main branch (100).

10. A decimation filter as claimed in claim 9, characterized in that besides the main branch (100) and side branches (101, 102, 104) the decimation filter comprises additional branches (103, 105, 106, 107), the divergence point (121, 122, 123, 125) of which is operationally connected between each of two successive FIR filters (20) in the side branch (101, 102, 104) or in the additional branch (103, 105, 106, 107) or between the means performing an equivalent operation, and their other end is operationally connected to the Pth {Pε

[1, . . . , N−

1]} adder (25) in connection with the comb filter after the decimation means (21) of the main branch according to the number P of divergence points (120-125) involved in forming an additional branch, the additional branches (103, 105, 106, 107) comprising delay means (23) for delaying the sample cycle entered the additional branch (103, 105, 106, 107) by one sample cycle M;

a number of successive FIR filters (20) or means performing an equivalent operation corresponding to the number of FIR filters between the divergence point (121-125) and the adder (25) in connection with the comb filter in the side branch (101, 102, 104) or in the additional branch (103, 105, 106, 107) where an additional branch branches.

12. A decimation filter as claimed in claim 10, characterized in that in M-tap FIR filtering the coefficients of taps are substantially ones.

13. A decimation filter as claimed in claim 10, characterized in that the decimation coefficient is a power of two.

11. A decimation filter the transfer function of which substantially corresponds to the transfer function of a CIC filter and which is arranged to decimate a signal including digital samples by coefficient M, characterized in that the decimation means at least comprises a main branch (100), which comprises decimation means (21), N M-tap FIR filters (20) in sequence before the decimation means (21) of the main branch (100), which decimate by coefficient M;

and

if the number N of M-tap FIR filters (20) in the main branch (100) is more than one, the decimation filter is arranged to take into account N−

1 sample cycles M preceding the incoming sample cycle M, and thus the decimation filter comprises besides the main branch (100)

selection means (61);

memory (61);

shift means (64);

adder/subtractor means (65) and

control means (66);

N−

1 side branches (101-108), the divergence point (120, 124, 126) of which is operationally connected between each of two FIR filters (20) before the decimation means (21) of the main branch (100) and their other end is operationally connected to the selection means (61), the side branches (101-108) comprising

decimation means (21) for decimating the signal entering the side ranch (101-108) by coefficient M;

delay means (23) for delaying the decimated signal by one sample cycle M;

additional branches (109), which comprise delay means (23) for delaying the signal by one sample cycle M, and the divergence point (130) of which is operationally connected after each delay means (23) in the side branch (101-108) or in the additional branch (109) and their other end is operationally connected to the selection means (61); and

the selection means (61) of the decimation filter are arranged to select one signal at a time from the main branch (100), side branch (101-108) or the additional branch (109) for an arithmetic shift operation to the shift means (64);

the shift means (64) are arranged to perform a necessary arithmetic shift on each signal of the main branch (100), side branch (101-108) or additional branch (109);

the adder/subtractor means (65) of the decimation filter are arranged to calculate difference of the signals according to comb-filtering and adding the signal of each side branch (101-108) or the additional branch to the result provided;

the memory (62) of the decimation filter is arranged to store the results provided for the signals of each main branch (100), side branch (101-108) and additional branch (109) and the sums of the results;

the control means (66) of the decimation filter are arranged to control the function of the decimation filter so that the decimation filter substantially provides the transfer function of the CIC filter.

14. A decimation filter as claimed in claim 11, characterized in that in M-tap FIR filtering the coefficients of taps are substantially ones.

15. A decimation filter as claimed in claim 11, characterized in that the decimation coefficient is a power of two.