Claims Analysis
Total Claims: 10
Claim 1
Independent Claim
123 words
A digital filter method comprising the following steps: input signal data x(n) is sampled at the input sampling frequency Fe and is shared between and stored in a plurality of Q memories of shift register or FIFO type, respectively associated with the Q branches of the filter; the filter coefficients g m (p) are stored in at least one other memory referred to as a "coefficient" memory; interpolation by L is performed on each branch; said data samples x(n) are multiplied by the said coefficients g m (p) at the rate of the output frequency Fs; and decimation by M, enabling the working frequency of the filter to be lowered to the output frequency Fs=(L/M)Fe, where L and M are integers and L/M<1.
Claim 2
Dependent Claim
123 words
A digital filter method comprising the following steps: input signal data x(n) is sampled at the input sampling frequency Fe and is shared between and stored in a plurality of Q memories of shift register or FIFO type, respectively associated with the Q branches of the filter; the filter coefficients g m (p) are stored in at least one other memory referred to as a "coefficient" memory; interpolation by L is performed on each branch; said data samples x(n) are multiplied by the said coefficients g m (p) at the rate of the output frequency Fs; and decimation by M, enabling the working frequency of the filter to be lowered to the output frequency Fs=(L/M)Fe, where L and M are integers and L/M<1.
Claim 3
Dependent Claim
41 words
A method according to claim 1, wherein said data samples x(n) in each branch remain present for a plurality ≈Fe/Q of multiplications and coefficients are rotated in each branch by permutation of addresses in said coefficient memory at each clock period.
Claim 4
Dependent Claim
59 words
A method according to claim 1, wherein said multiplications are performed in parallel on a plurality p of filter planes, each filter plane having Q branches, said p filter planes being connected in a parallel architecture, thereby enabling the working frequency of the filter to be lowered to the output frequency Fs=(L/pM)Fe with synchronous sequencing between the parallel filters.
Claim 5
Dependent Claim
55 words
A method according to claim 3, wherein said multiplications are formed in cascade by a plurality k of parallel filters, each parallel filter comprising p planes, said parallel filters being connected in series, the p outputs from each preceding parallel filter being connected to the p inputs of each following parallel filter in the cascade.
Claim 6
Dependent Claim
139 words
Digital filter architecture comprising: means for sharing input sampled data x(n); means for temporary storage of said input data x(n), e.g. a plurality Q of memories of shift register or FIFO type; Q branches of the filter; at least one other memory referred to as a "coefficient" memory for storing filter coefficients g m (p); means for multiplying data x(n) by the coefficients g m (p); means for performing interpolation by L; means for decimation by M, with integer L and M and with L/M<1; said temporary storage means for said input data being associated respectively with said Q branches of said filter to provide said data to said multiplication means; said at least one coefficient memory being connected to said multiplication means to supply said coefficients to said multiplication means; and said multiplication means outputting filtered data y(m).
Claim 7
Dependent Claim
139 words
Digital filter architecture comprising: means for sharing input sampled data x(n); means for temporary storage of said input data x(n), e.g. a plurality Q of memories of shift register or FIFO type; Q branches of the filter; at least one other memory referred to as a "coefficient" memory for storing filter coefficients g m (p); means for multiplying data x(n) by the coefficients g m (p); means for performing interpolation by L; means for decimation by M, with integer L and M and with L/M<1; said temporary storage means for said input data being associated respectively with said Q branches of said filter to provide said data to said multiplication means; said at least one coefficient memory being connected to said multiplication means to supply said coefficients to said multiplication means; and said multiplication means outputting filtered data y(m).
Claim 8
Dependent Claim
16 words
Parallel digital filter architecture comprising a plurality of filters according to claim 5 connected in parallel.
Claim 9
Dependent Claim
13 words
Cascaded digital filter architecture comprising a plurality of filters according to claim 5.
Claim 10
Dependent Claim
19 words
A digital filter circuit according to the architecture of claim 5 for implementing a method according to claim 1.