Method and system for soft-weighting a reiterative adaptive signal processor
First Claim
1. A method of training an adaptive signal processor for filtering interference and noise from an input signal received by the adaptive signal processor, which comprises a plurality of building blocks, each building block comprising an adaptive weight value, the method comprising:
- i) receiving, by the adaptive signal processor, an input signal comprising a plurality of measurements, z and x, in an interference plus noise data environment;
ii) applying, by each of the plurality of building blocks of the adaptive signal processor, a soft weight complex scalar value, wsw, to each of the plurality of measurements, the soft weight value multiplied with the adaptive weight value, wg, thereby modifying the adaptive weight value;
iii) outputting, by the adaptive signal processor, a main channel column output, y, comprising the weighted plurality of measurements in accordance with the equation y=z−
(wsw*wg)*x;
iv) squaring, by the adaptive signal processor, the main channel column output of the adaptive signal processor for each of the plurality of measurements;
v) determining, by the adaptive signal processor, a representative average value of the squared values; and
vi) performing each of steps i) through v), by the adaptive signal processor, the performing step comprising;
verifying, when the representative average value is within a pre-determined signal to power plus noise ratio [SINR]; and
upon a positive verification that the representative average value is within a pre-determined signal to power plus noise ratio [SINR], counting the number of iterations of steps i) through v), whereupon, when the count is less than 2N training samples, N being the number of degrees of freedom of the adaptive signal processor, the performing step is concluded and the adaptive signal processor is now trained to filter between interference and noise of an input signal.
3 Assignments
0 Petitions
Accused Products
Abstract
This invention is an improvement upon the basic adaptive signal processors, the Multi-Stage Wiener Filter and the cascaded canceller. The invention combines the concepts of soft weighting the adaptive weights of either type of processor disclosed herein with reiteratively processing the outputs by returning them to the input of the chosen adaptive signal processor. The combination of these functions improves the Signal to Interference plus Noise Ratio (SINR), the Probability of Detection (Pd), and/or the Bit Error Rate (BER). The invention improves statistical convergence of these types of metrics such that fewer training data samples are needed to achieve a particular satisfactory value of these metrics than would occur using traditional computational adaptive signal processors.
-
Citations
82 Claims
-
1. A method of training an adaptive signal processor for filtering interference and noise from an input signal received by the adaptive signal processor, which comprises a plurality of building blocks, each building block comprising an adaptive weight value, the method comprising:
-
i) receiving, by the adaptive signal processor, an input signal comprising a plurality of measurements, z and x, in an interference plus noise data environment; ii) applying, by each of the plurality of building blocks of the adaptive signal processor, a soft weight complex scalar value, wsw, to each of the plurality of measurements, the soft weight value multiplied with the adaptive weight value, wg, thereby modifying the adaptive weight value; iii) outputting, by the adaptive signal processor, a main channel column output, y, comprising the weighted plurality of measurements in accordance with the equation y=z−
(wsw*wg)*x;iv) squaring, by the adaptive signal processor, the main channel column output of the adaptive signal processor for each of the plurality of measurements; v) determining, by the adaptive signal processor, a representative average value of the squared values; and vi) performing each of steps i) through v), by the adaptive signal processor, the performing step comprising; verifying, when the representative average value is within a pre-determined signal to power plus noise ratio [SINR]; and upon a positive verification that the representative average value is within a pre-determined signal to power plus noise ratio [SINR], counting the number of iterations of steps i) through v), whereupon, when the count is less than 2N training samples, N being the number of degrees of freedom of the adaptive signal processor, the performing step is concluded and the adaptive signal processor is now trained to filter between interference and noise of an input signal. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25)
-
-
26. A system for reiteratively applying a soft-weight value, comprising:
-
a plurality of sensor inputs for measuring signals within an interference plus noise environment with a variable effective rank; a filter comprising an adaptive signal processor; a controller to perform the following steps; input the measured signals, x and z, into the filter; determining y, in accordance with the following equation;
y=z−
(wsw*wg)*x, wherein x and z are input signals, wg is conjugated adaptive weight value, wsw is representative of the soft weight complex scalar valueapplying a soft weight value wsw to the filter, the soft weight value modifying one or more existing adaptive weight values wg associated with the filter; outputting, y of the adaptive signal processor in accordance with the equation y=z−
(wsw*wg)*x;a memory means, wherein the controller retrieves from the memory means a stored number of iterations, such that the operations of the controller are performed a number of times as specified by the number of iterations; and an output means. - View Dependent Claims (27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 41, 42, 43, 44, 45)
-
-
40. The system according to 38 wherein the Multi-Stage Weiner Filter comprises a plurality of stages such that the number of stages is selected to be equal to an effective rank of an interference plus noise environment.
-
46. A method for adaptive signal processing, by an adaptive signal processor, the method comprising:
-
measuring, by one or more sensors of the adaptive signal processor, a signal comprising a plurality of measurements, z and x, in an interference plus noise environment; inputting the measured signal into an adaptive signal processor wherein a controller retrieves from a memory means a stored number of iterations, such that the operations of the adaptive signal processor are performed a number of times as specified by the number of iterations; filtering, by the adaptive signal processor, the measured signal to eliminate interference; applying, by the adaptive signal processor, a soft weight to the algorithm, the soft weight value modifying one or more existing adaptive weight values associated with the adaptive signal processor in accordance with the following equation;
y=z−
(wsw*wg)*x,wherein wg is adaptive weight value, wsw is representative of the soft weight complex scalar value; and outputting, by the adaptive signal processor, the filtered and weighted signal, y. - View Dependent Claims (47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62)
-
-
63. An adaptive signal processor comprising:
-
a filter; a controller for retrieving from a memory means a stored number of iterations, such that operations of the controller, the operations of the adaptive signal processor, and the operations of a soft weight generator are performed a number of times as specified by the number of iterations; a soft weight generator for modifying one or more adaptive weight values associated with the adaptive signal processor; and an output y, determined in accordance with the following equation;
y=z−
(wsw*wg)*x,wherein x and z are input signals, including the measured signal, into the filter, wg is conjugated adaptive weight value, wsw is representative of the soft weight complex scalar value. - View Dependent Claims (64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82)
-
Specification