Method and apparatus for signal reconstruction
AssigneeU S West Advanced Technologies, Inc.
InventorsYao Wang, V. Ramamoorthy
Life of this patent
Estimating the term…
Abstract
A method for reconstructing an incomplete signal in a sub-band transmission system is disclosed. Signal degradation from signal loss in a sub-band transmission system may be avoided by formulating an error function, and finding an optimal signal which minimizes the error between the sub-band signals interpolated from received sub-band signals and the sub-band signals obtained by convolving an optimal reconstructed signal with the original analysis filters. The minimized system equation is solved to provide an optimal reconstructed output signal.
Claims15
A method of constructing an output signal approximating an original signal from a set of sub-band signals, said method using a system having a system controller, a signal reconstruction system, and synthesis filters, wherein the set of sub-band signals is produced by inputting the original signal into a set of analysis filters, each of the filters having a known impulse response, and decimating the output from the set of analysis filters to produce the set of sub-band signals, wherein said original signal has a good segment and a bad segment, said good segment being a portion of the original signal where all sub-band signals in the set of sub-band signals are received by said system, said bad segment being a portion of the original signal where the set of sub-band signals is incomplete, the method comprising the steps of:
- (a) constructing a good region of an output signal corresponding to the good segment by inputting the sub-band signals of said good segment into said synthesis filters which are matched to the analysis filters;
- and
- (b) constructing a bad region of said output signal corresponding to the bad segment by:
- (i) receiving at least one sub-band signal from the set of sub-bands signals in said bad segment;
- and
- (ii) identifying an output signal ##EQU12## where i is the index identifying said received sub-band signals, B.sub.r are the indices of the received sub-band signals. H.sub.i is a matrix defining the impulse response of the analysis filters, w.sub.i are weighting coefficients, t is the operation of matrix transposition followed by complex conjugation, and f.sub.i.sup.B are the interpolated received sub-band signals, whereby f.sub.opt is said output signal;
- and
- (c) combining the good region of said output signal with the bad region of said output signal to form a complete output signal.
A receiver for constructing an output signal which approximates an original signal from a set of sub-band signals, said set of sub-band signals produced by inputting the original signal into a set of analysis filters and decimated, each analysis filter having a known impulse response, the set of sub-band signals being transmitted over a packet switched network to the receiver, the original signal having a good segment and a bad segment, the good segment being a portion of the original signal of which the receiver receives all sub-band signals of the set, the bad segment being a portion of the original signal of which the receiver receives at least one of the sub-band signals from the set of sub-band signals, the receiver including:
- (a) means for determining the portions of the original signal which are good segments and bad segments;
- (b) means for constructing a good region of an output signal corresponding to a good segment by inputting said good segment into a set of synthesis filters matched to the analysis filters;
- (c) means for constructing a bad region of said output signal corresponding to a bad segment by identifying an output signal ##EQU15## where i is the index identifying said received sub-band signals, B.sub.r are the indices of the received sub-band signals, H.sub.i is a matrix defining the impulse response of the analysis filters, w.sub.i are weighting coefficients, t is the operation of matrix transposition followed by complex conjugation, and f.sub.i.sup.B are the interpolated received sub-band signals, whereby f.sub.opt is said output signal;
- and
- (d) combining said good region of said output signal with said bad region of said output signal to form a complete output signal.
A method of constructing an output signal approximating an original signal from a set of sub-band signals, said method using a system having a system controller, a signal reconstruction system, and synthesis filters, wherein the set of sub-band signals is produced by inputting the original signal into a set of analysis filters, each of the filters having a known impulse response, and decimating the output from the set of analysis filters to produce the set of sub-band signals, wherein said original signal has a good segment and a bad segment, said good segment being a portion of the original signal where all sub-band signals in the set of sub-band signals are received by said system, said bad segment being a portion of the original signal where the set of sub-band signals is incomplete, the method comprising the steps of:
- (a) constructing a good region of an output signal corresponding to the good segment by inputting the sub-band signals of said good segment into said synthesis filters which are matched to the analysis filters;
- and
- (b) constructing a bad region of said output signal corresponding to the bad segment by:
- (i) receiving at least one sub-band signal from the set of sub-band signals in said bad segment;
- and
- (ii) identifying an output signal f.sub.opt (x,y) which minimizes the following error:
- ##EQU16## where i is the index identifying a sub-signal, B.sub.r is a set of indices for said received sub-band signals, L is a set of points in said bad segment, h.sub.i (x,y) is an impulse response to an i-th analysis filter, f.sub.i.sup.B (x,y) is an interpolated sub-band signal from said received sub-band signals, modified at boundary points of set L in accordance with said good region, and w.sub.i are weighting coefficients.
Dependencies are inferred from the claim text.