Coded data generation or conversion – Analog to or from digital conversion – Differential encoder and/or decoder
Reexamination Certificate
2002-10-10
2004-06-08
Young, Brian (Department: 2819)
Coded data generation or conversion
Analog to or from digital conversion
Differential encoder and/or decoder
C341S144000
Reexamination Certificate
active
06747584
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to signal processing systems. More specifically, the present invention relates to an aggregate beamformer for use in a directional receiving array such as an array of microphones.
BACKGROUND OF THE INVENTION
Beamforming is a method of combining signals that are received by an array of sensor elements by adjusting the phase relationships between signals and adding the signals, to cause enhanced sensitivity in a particular direction. Since the array of sensor elements receive the signals at different times, the signals are steered and focused in the desired direction by applying appropriate delays from each array sensor element so that the signals transmitted from a desired point add constructively. The delay for each signal is selected such that a virtual beam is focused at the desired point. In other words, beamforming electronically forms a virtual beam through steering and focusing the signals. Beamforming may serve to determine the location of the target points when it is known which beams detected that target signal. Beams represent the directional response of a system, and the direction of a beam is an angle relative to the array. The beam direction is generally focused to a point, which may or may not be at infinity.
Beamformers are utilized with arrays of electromagnetic and sonic receiving elements, for combining signals of the receiving elements to produce beams of electromagnetic and sonic energy. The term beam is used both for radiant energy received from a particular direction as well as for a beam of transmitted radiant energy since the receiving and transmitting radiation patterns of an array of receiving or radiating elements are identical. Beamformers for receiving arrays employ linear circuits for summing together the signals of the respective receiving elements and for imparting selective delays, or sometimes only phase shifts, to signals of the respective receiving elements. The selection of specific values of time delay is based on the direction of the desired beam relative to the array.
In some situations, the signals of the elements are sampled repetitively to produce sequences of signal samples from each of the elements. The sequences of samples are then transmitted to the beamformer, which forms one or more beams as is desired.
FIG. 1
illustrates a conventional beamformer. If the (sampled) signal at the m-th sensor element is denoted by x
m
(n&Dgr;t), where n is the sample number and &Dgr;t is the time interval between samples, the conventional beamforming procedure results in the output signal y(n&Dgr;t) according to the equation (1):
y
(
n
Δ
t
)
=
∑
m
W
m
x
m
(
n
Δ
t
-
d
m
)
(
1
)
wherein
W
m
are weight factors that determine the shape of the directional response pattern;
x
m
(n&Dgr;t) is the output from sensor element m at time n&Dgr;t;
&Dgr;t is the time interval between samples; and
d
m
are delays that determine the direction in which the response is maximized.
In digital systems, the analog signals x
m
are sampled and digitized with the delays for each sensor element being an integer number of sample intervals taken as near as possible to the delay necessary for the steered direction (&thgr;). Once the signal are converted to digital data they are typically combined by beamforming using weighted sums of the data in tapped delay lines.
The generalized beamformer takes the weighted sum of current and past samples to form the beamformer output. It is more flexible in regard to obtaining desired beam shapes because it allows beam shape to be determined as a function of signal frequency. The counterpart of equation (1) for the generalized beamformer would be
y
(
n
Δ
t
)
=
∑
m
∑
k
w
m
,
k
x
m
(
n
Δ
t
-
k
-
d
m
)
,
wherein w
m,k
are weight factors that determine the shape of the directional response pattern, m is the array component index, and k is an index for the delayed samples.
The generalized formulation can be considered to be a special case of the conventional beamformer expressed by equation (1) if the delayed input signals as treated as separate (virtual) array components (channels) so that the array input channels are x
p
(n&Dgr;t)=x
m
(n&Dgr;t−k), wherein p is an index for the sensor-delay pairs (m,k). In this case, the index p would simply replace the index m in the equation (1).
Since one physical input channel is used for several delayed input channels, more collisions and voids may occur for the generalized beamformer than would be the case where delayed channels are not considered.
A complete state-of-the-art acoustic signal processing system uses a collection of components such as analog to digital converters (ADC), application specific integrated circuits (ASICs), digital signal processors (DSPs), microcontrollers (&mgr;C), memory buffers, etc. integrated onto a set of printed circuit boards connected by one or more communications busses. In order to process the data received from the array of sensor elements, the front-end processor, and more specifically the beamformer, is used to process the data from multiple sensor elements substantially all at the same time. The beamformer includes a data acquisition system (DAS) for converting the plurality of sets of data received from the detectors into corresponding signals that can be processed by a signal processor.
Various problems exist with respect to current beamformer designs. The number of circuit components is large, and increases with the number of input signal components, causing high cost and complexity in the designs. The front-end components which provide coupling and anti-alias filtering are not easily integrated into an integrated solid state circuit (IC). A large number of arithmetic operations are required for the calculation of each beam and this number increases with the sampling rate and with the number of sensor elements. The precision to which delays can be realized is limited by the sampling rate (per channel) unless an interpolation filter is used. The circuit and computational complexity generally scales with the number of signal components (channels), making beamformer implementation very expensive or impractical for very large numbers of channels.
Some of these problems can be partially overcome, with associated loss of performance or increase in cost. For example, a multiplexer or switching circuit may be used to share one high speed ADC amongst several signal channels, thereby reducing the number of ADC required but there is generally a trade-off between ADC speed and resolution. The beamformer components subsequent to the ADC involve digital signal processing and can be integrated in an IC or implemented in a high speed digital signal processor (DSP) computer specifically designed for such applications but front-end components are not amenable to low cost integration for arrays of numerous sensor elements, particularly at lower audio frequencies. The large size of the non-integrated components necessitates moving them some distance from the sensor element array—requiring a high-bandwidth umbilical cord and driving circuitry in most cases.
The numeric computations can, to some extent, be sped up by using high performance processors that perform them in parallel using redundant computation units or pipeline aspects of the processing. For ultrasound applications, the incoming signal from each sensor element may be shifted to a lower frequency, by a heterodyning circuit, to reduce the subsequent circuit and computational requirements but the heterodyning circuits add cost and complexity, and are subject to variability. The precision to which channel delays can be applied is limited by the sampling interval unless interpolation filters are used to estimate the signal components at times between samples but interpolation is computationally expensive and may be inaccurate.
Arrays of s
(Marks & Clerk)
Lauture Joseph
National Research Council of Canada
Young Brian
LandOfFree
Aggregate beamformer for use in a directional receiving array does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Aggregate beamformer for use in a directional receiving array, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Aggregate beamformer for use in a directional receiving array will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3339384