Pulse or digital communications – Equalizers – Automatic
Reexamination Certificate
1998-06-29
2001-07-31
Chin, Stephen (Department: 2634)
Pulse or digital communications
Equalizers
Automatic
C375S350000, C708S323000
Reexamination Certificate
active
06269117
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates generally to improvements in digital signal processing and specifically to a system and method for enhancing the efficiency of downsampling operations.
2. Description of the Background Art
The increased use of digital audio and video in the consumer market has challenged designers to produce digital signal processing technology with superior performance while using economical components. Often digital techniques which are well-known in the art require expensive processing hardware. In order to make use of inexpensive hardware, new techniques must be invented.
One technique that is used in digital signal processing is downsampling. The need for downsampling arises when the source of digitized signals provides digital samples at a higher sampling rate than the receiver of the digital signals can accept. In the case where the source sampling rate MR is an integral multiple M of the receiving sampling rate R, it would appear that simply deleting (M−1) samples out of M samples would yield a digital signal of the correct sample receiving rate R. In reality, this significantly reduces the accuracy of the resultant signal.
In order to downsample, defined as converting a higher sampling rate digital signal to a lower sampling rate digital signal, a downsampling filter is required. This downsampling filter makes use of the signal content of a number of neighboring samples from the digital signal at the MR rate to give a best representation of the signal at the R rate. The downsampling filter is often implemented as a Finite Impulse Response (FIR) filter. Let the digital signal at the rate of MR be represented by the function x(n) of the discrete variable n, and the corresponding digital signal at the rate of R be represented by the function y(m) of the discrete variable m. In this case the discrete variable m occurs once for every M'th occurrence of discrete variable n. If x(n) is known, a FIR filter representation of the calculation of y(m) from x(n) may be expressed by the following equation:
y
(
m
)
=
∑
k
=
0
N
-
1
h
(
k
)
x
(
Mm
-
k
)
where k is the summation index, h(k) are constants called the filter coefficients, and N is called the length of the filter.
In the case of digital audio, it is often necessary to downsample from a digital audio bitstream arriving at a rate twice that which can be accepted by a low-cost digital-to-analog converter (DAC). However, in the case of digital audio, very high quality is required and this causes the number N of filter coefficients h(k) to be large. The result is that both the processing power required from a digital signal processing (DSP) microprocessor and the size of the memory required to store the set of N coefficients cannot be realized by the most economical devices.
SUMMARY OF THE INVENTION
The present invention includes a system and method for enhancing the downsampling operations for digital data. Downsampling is an operation in which sample values, which are the digital representation of an analog signal, are converted to have a lower sample rate. In the preferred embodiment, this invention reduces the processing power and storage required for downsampling by using a finite impulse response (FIR) filter technique. This filter consists of forming the sum of the products of adjacent sample values of the digital data with a corresponding series of constants called filter coefficients. A filter is designed with the number of filter coefficients preferably equal to 4(a)+1, where (a) is an integer called the filter length parameter. In this embodiment, the non-central even-numbered coefficients are, to the degree of accuracy required, equal to zero. Also, the central even-numbered coefficient is equal to ½. Because multiplication by ½ in a binary digital system may be executed by a simple right shift of a binary word, the total number of coefficient-sample multiplications may be reduced to 2(a). This process represents a savings of slightly over 50% in the number of multiplications required in the execution of the FIR filtering process. As the multiplications represent the most computation-intensive use of a microprocessor, a 50% reduction in the number of multiplications represents an overall savings in computer use of approximately 50%.
In accordance with the present invention, the filter design wherein the even-numbered coefficients equal zero also has the additional benefit of reducing the number of storage locations necessary for the filter coefficients. Because non-volatile storage is at a premium in most cost-effective microprocessors, and since many applications such as digital audio require a filter with a large number of filter coefficients for the necessary accuracy, this is a significant benefit of the present invention.
A further aspect of the present invention is the segregation of the input data stream into even and odd numbered sample buffers. With input buffers thus organized, it is possible to use software pointers and obviate the use of modulo-power-of-2 addressing. Because addressing using this mode is not compatible with assembly language, the present invention allows the designer to write the application code in assembly language to achieve better system performance.
REFERENCES:
patent: 5053984 (1991-10-01), Cavallotti et al.
patent: 5222144 (1993-06-01), Whikehart
patent: 5513223 (1996-04-01), Shoji
patent: 5515402 (1996-05-01), Chester
patent: 5732337 (1998-03-01), Wargnier et al.
patent: 5872480 (1999-02-01), Huang
patent: 5894495 (1999-04-01), Kase
patent: 5905757 (1999-04-01), Kundmann et al.
Applications Engineering Staff of Analog Devices, DSP Division, “Digital Signal Processing Applications Using the ADSP-2100 Family”, 1990, pp. 86-97.
Chin Stephen
Ha Dac V.
Koerner Gregory J.
Simon & Koerner LLP
Sony Corporation
LandOfFree
System and method for enhancing downsampling operations does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for enhancing downsampling operations, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for enhancing downsampling operations will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2468332