Rake receiver and de-spreading method thereof

1. A rake receiver, adapted for de-spreading a received signal from a multi-path channel, comprising:

a channel estimation module, for estimating a plurality of channel gains and a plurality of corresponding path delay times in the multi-path channel;

L signal branches, for delaying the received signal for a plurality of signal delay times respectively to obtain a plurality of delayed received signals, de-spreading the delayed received signals respectively to output a plurality of signal components, wherein the signal delay times are respectively proportional to the path delay times, and L is a positive integer;

I noise branches, for delaying the received signal for a plurality of preset times respectively to obtain a plurality of delayed noise signals, de-spreading the delayed noise signals to output a plurality of noise components respectively, wherein I is a positive integer; and

a combiner, for combining the signal components to generate a main signal, combining the noise components to generate an interference-plus-noise estimation value, and outputting a de-spreading signal according to a difference between the main signal and the interference-plus-noise estimation value;

wherein the combiner adjusts weights of the signal components in the main signal according to the channel gains, and adjusts weights of the noise components in the interference-plus-noise estimation value according to I noise weights.

2. The rake receiver as claimed in claim 1 , wherein an i th signal branch among the L signal branches comprises:

a delay unit, for delaying the received signal for an i th signal delay time to obtain an i th delayed received signal; and

a de-spreader, for convolving the i th delayed received signal with a local spreading waveform, and sampling the convolved result to an i th signal component.

3. The rake receiver as claimed in claim 1 , wherein an i th noise branch among the I noise branches comprises:

a delay unit, for delaying the received signal for an i th preset time to obtain an i th delayed noise signal; and

a de-spreader, for convolving the i th delayed noise signal with a local spreading waveform, and sampling the convolved result to an i th noise component.

4. The rake receiver as claimed in claim 1 , wherein the combiner comprises:

a signal combining unit, for combining the signal components according to L signal weights to generate and output the main signal, wherein the signal weights are obtained by the channel gains of the multi-path channel;

a noise combining unit, for combining the signal components according to I signal weights to generate and output the interference-plus-noise estimation value; and

an adder, subtracting the interference-plus-noise estimation value from the main signal, so as to output the de-spreading signal.

5. The rake receiver as claimed in claim 4 , wherein the signal combining unit comprises:

L multipliers, wherein an i th multiplier receives an i th signal component, and multiplies the i th signal component by an i th signal weight, so as to output an i th product; and

a summation unit, for receiving and summing up the products to output the main signal.

6. The rake receiver as claimed in claim 4 , wherein the noise combining unit comprises:

I multipliers, wherein an i th multiplier receives an i th noise component, and multiplies the i th noise component by an i th noise weight, so as to output an i th product; and

a summation unit, for receiving and summing up the products to output the interference-plus-noise estimation value.

7. The rake receiver as claimed in claim 6 , wherein the signal weights are represented as A 0 -A L-1 , and a signal weight vector formed by the conjugate value of the signal weights A 0 -A L-1 is represented as α=[A 0 *, . . . , A L-1 *] T ; the noise weights are represented as B 0 -B I-1 , and a noise weight vector formed by the conjugate value of the noise weights B 0 -B I-1 is represented as β=[B 0 *, . . . , B I-1 ] T , and the rake receiver further comprises:

a noise weight calculating unit, for calculating the noise weights B 0 -B I-1 through a noise matrix R u and an equation β=R n − R ns α, and outputting the noise weights B 0 -B I-1 to the noise combining unit,

wherein the noise matrix R u is a covariance matrix of a noise vector u, with a dimension of (I+L)×(I+L), the noise vector u comprises noise compositions in the signals components and the noise components, with a length of I+L, the matrix R n is a I×I sub-matrix at a bottom right corner of the noise matrix R u , the matrix R ns is a I×L sub-matrix at a bottom left corner of the noise matrix R u , and the superscript (−1) represents an inverse matrix operation.

8. The rake receiver as claimed in claim 7 , further comprising:

a noise branch selecting unit, for selecting first I candidates having larger metrics from a plurality of candidates, so as to serve as the preset times,

wherein the metric is |β i | 2 E[|z n (m i )| 2 ], the i th candidate corresponds to the i th preset time m i , the i th noise branch generates the i th noise component z n (m i ) according to the i th preset time m i , β i is a conjugate value of the noise weight corresponding to the i th noise component z n (m i ), and E[•] represents an expected value operation.

9. A de-spreading method used for a rake receiver, comprising:

estimating a plurality of channel gains and a plurality of corresponding path delay times in a multi-path channel;

receiving a received signal from the multi-path channel;

delaying the received signal for a plurality of signal delay times respectively to obtain a plurality of delayed received signals, de-spreading the delayed received signals respectively to obtain a plurality of signal components, wherein the signal delay times are respectively proportional to the path delay times;

delaying the received signal for a plurality of preset times respectively to obtain a plurality of delayed noise signals, de-spreading the delayed noise signals respectively to obtain a plurality of noise components;

adjusting weights of the signal components in a main signal respectively according to the channel gains, so as to combine the signal components to obtain the main signal;

adjusting weights of the noise components in an interference-plus-noise estimation value according to a plurality of noise weights, so as to combine the noise components to obtain the interference-plus-noise estimation value; and

generating a de-spreading signal according to a difference between the main signal and the interference-plus-noise estimation value.

10. The de-spreading method as claimed in claim 9 , wherein the step of obtaining the signal components comprises:

delaying the received signal for an i th signal delay time to obtain an i th delayed received signal; and

convolving the i th delayed received signal with a local spreading waveform, and sampling the convolved result to an i th signal component.

11. The de-spreading method as claimed in claim 9 , wherein the step of obtaining the noise components comprises:

delaying the received signal for an i th preset time to obtain an i th delayed noise signal; and

convolving the i th delayed noise signal with a local spreading waveform, and sampling the convolved result to an i th noise component.

12. The de-spreading method as claimed in claim 9 , wherein the step of adjusting weights of the signal components in a main signal respectively according to the channel gains so as to combine the signal components to obtain the main signal comprises:

generating a plurality of signal weights according to the channel gains;

multiplying the i th signal component by an i th signal weight, so as to generate an i th product; and

summing up the products to obtain the main signal.

13. The de-spreading method as claimed in claim 12 , wherein the step of adjusting weights of the noise components in an interference-plus-noise estimation value according to a plurality of noise weights so as to combine the noise components to obtain the interference-plus-noise estimation value comprises:

multiplying an i th noise component by an i th noise weight to generate an i th product; and

summing up the products to obtain the interference-plus-noise estimation value.

14. The de-spreading method as claimed in claim 13 , wherein a number of the signal components is L, a number of the noise components is I, the signal weights are represented as A 0 -A L-1 , and a signal weight vector formed by a conjugate value of the signal weights A 0 -A L-1 is represented as α=[A 0 *, . . . , A L-1 ] T , the noise weights are represented as B 0 -B I-1 , and a noise weight vector formed by a conjugate value of the noise weights B 0 -B I-1 is represented as β=[B 0 *, . . . , B I-1 *] T , the de-spreading method further comprises:

calculating a I×I sub-matrix at a bottom right corner of a noise matrix R u to obtain a matrix R n , wherein the noise matrix R u is a covariance matrix of a noise vector u, with a dimension of (I+L)×(I+L), and the noise vector u comprises noise compositions in the signals components and the noise components, with a length of I+L;

calculating a I×L sub-matrix at a bottom left corner of the noise matrix R u to obtain a matrix R ns ;

calculating an inverse matrix of the matrix R n to obtain a matrix R n −1 , wherein the superscript (−1) represents an inverse matrix operation; and

calculating the noise weights B 0 -B I-1 according to an equation β=R n −1 R ns α.

15. The de-spreading method as claimed in claim 14 , further comprising:

calculating a metric of an i th candidate among a plurality of candidates; and

selecting first I candidates having larger metrics to serve as the preset times,

wherein the metric is |β i | 2 E[|z n (m i )| 2 ], the i th candidate corresponds to the i th preset time m i , the i th preset time m i corresponds to the i th noise component z n (m i ), β i is a conjugate value of the noise weight corresponding to the i th noise component, and E[•] represents an expected value operation.