HK1114963A

HK1114963A - Apparatus, methods and computer program products for delay selection in a spread-spectrum receiver

Info

Publication number: HK1114963A
Application number: HK08104698.9A
Authority: HK
Inventors: Douglas A. Cairns; Gregory E. Bottomley; Ali S. Khayrallah
Original assignee: Telefonaktiebolaget Lm Ericsson (Publ)
Priority date: 2004-10-06
Filing date: 2005-10-04
Publication date: 2008-11-14

Description

Apparatus, method and computer program product for delay selection in spread spectrum receivers

Technical Field

The present invention relates to radio frequency communications, and more particularly, to an apparatus, method and computer program product for processing spread spectrum communication signals.

Background

Spread spectrum signal transmission techniques are widely used in communication systems such as Code Division Multiple Access (CDMA) cellular telephone networks. Referring to fig. 1, an information symbol is typically modulated by a spreading sequence prior to transmission from transmitting station 110 such that the symbol is represented by a number of chips in the transmitted signal. At the receiver 120, the received signal is despread using a despreading code, which is typically the conjugate of the spreading code. Receiver 120 includes a radio frequency processor 122 that performs downconversion, filtering, and/or other operations to generate a baseband signal that is provided to a baseband processor 124. The baseband processor 124 despreads the baseband signal to produce symbol estimates that are provided to an additional processor 126, which may perform additional signal processing operations, such as error correction decoding.

In coherent direct sequence CDMA (DS-CDMA) systems, coherent RAKE reception is typically used. This type of receiver despreads the received signal by correlating to a sequence of chips to produce despread values that are weight-combined based on the estimated channel coefficients. The weighting may remove the phase rotation of the channel and scale the despread values to provide "soft" values representing the transmitted signal.

Multipath propagation of the transmitted signal can result in time dispersion that can cause multiple resolvable echoes of the transmitted signal to arrive at the receiver. In a conventional RAKE receiver, the correlators are typically matched to selected echoes of the desired signal. Each correlator produces despread values that are weighted and combined as described above. While RAKE receivers may be effective in certain environments, self-interference and multi-user interference can cause loss of orthogonality between channels defined by spreading sequences, thereby degrading performance.

A "generalized" RAKE (G-RAKE) receiver has been proposed to provide improved performance in such interference environments. Conventional G-RAKE receivers typically use combining weights that vary with channel coefficients and noise covariance, including information about the interfering signal. These weights w can be expressed as follows:

w＝R^-1c， (1)

where R is the noise covariance matrix and c is the channel coefficient vector.

A typical baseband processor for a G-RAKE receiver is shown in fig. 2. The chip samples are provided to a finger placement unit 230 which determines where the correlation unit 210 places fingers (selecting delays for one or more antennas). Correlation unit 210 despreads one or more traffic channels and generates traffic despread values. The selected paths are also provided to a weighting computer 240 that computes combining weights that are used to combine the despread values in combiner 220 to produce soft values.

As shown in fig. 3, a similar function may be provided with a chip equalizer structure. In such a configuration, the chip samples are provided to a tap placement unit 330 that determines where to place the filter taps (i.e., which delays to use for one or more antennas) of the Finite Impulse Response (FIR) filter 310. The selected tap positions are also provided to a weight calculator 340 that calculates filter coefficients (or weights) for the filter 310. Filter 310 filters the sliced samples to produce a signal that is despread by correlator 320 to produce symbol estimates.

A conventional weighting computer for a G-RAKE receiver is shown in fig. 4. The signal samples are provided to a correlation unit 410 that despreads the symbols from the pilot or traffic channel to produce initial despread values. The symbol modulation may be removed from these values with a modulation remover 420 and the resulting values provided to a channel tracker 430 that generates a channel estimate. The despread values and the channel estimates are provided to a noise covariance estimator 450, which generates a noise covariance estimate for the delay set in use. The channel estimates and the noise covariance estimates are provided to a weight calculator 440, which calculates combining weights (filter coefficients) therefrom.

The G-RAKE receiver differs from a conventional RAKE receiver in that it takes into account delays other than those corresponding to the desired signal echoes. These other delays are typically selected to provide information about the interference so that the receiver can suppress the interference.

In a practical RAKE receiver (conventional RAKE or G-RAKE), hardware and/or software constraints generally limit the number of "fingers" that can be used at any given time. In a conventional RAKE receiver, the fingers are typically selected to collect the maximum amount of desired signal energy. However, in a G-RAKE receiver, the finger selection criteria may also collect interference signal information to enable a desired amount of interference suppression.

Various strategies have been proposed for selecting fingers for a RAKE receiver. United states patent 5572552 to Dent et al describes a process whereby fingers are placed according to a calculated signal-to-noise ratio (SNR) metric that varies with channel coefficient, power level, and optional spreading code. United states patent 6363104 to Bottomley describes estimating the SNR as a function of the channel estimate and impairment correlation matrix estimate for each candidate combination for different finger position combinations and selecting the finger combination that maximizes the SNR. United states patent 6683924 to ottoson et al describes a finger selection process based on the time difference and relative signal strength of the signal paths. Other selection techniques are described by Kutz et al in the "Low complexity implementation of downlink CDMA conventional RAKE receiver" (Low complexity implementation of a downlink CDMA generated RAKE receiver ") and in the" On the performance of the downlink CDMA conventional RAKE receiver "(On the performance of a practical downlink CDMA conventional RAKE receiver) published by proc.

Disclosure of Invention

According to some embodiments of the present invention, methods are provided for recovering a signal from a composite signal comprising signals of one or more sources. Channel and correlation characteristics may be determined for the composite signal. Based on the determined channel and correlation characteristics, a respective combining weight of the composite signal information is determined for each of a plurality of candidate delays. A set of delays, e.g., RAKE correlator delays or chip equalizer filter taps, is selected from a plurality of candidate delays based on the determined weights. Information from the symbol signal may be processed for selected delays based on the spreading code to generate symbol estimates.

In some embodiments of the present invention, a time domain channel response and magnitude correlation may be determined, and a corresponding weight may be determined based on the time domain channel response and magnitude correlation. The quantitative correlation may be a noise covariance. The selected set of delays may include a set of delays having the largest associated weights, e.g., correlator delays or chip equalizer filter taps.

In other embodiments of the present invention, a frequency domain scheme may be used to determine the correlator delays and/or the weights of the chip equalizer filter taps. A weighted frequency response comprising noise information may be determined and corresponding weights may be determined from the weighted frequency response, e.g., by converting the effective channel response to the time domain to determine coefficients of a corresponding time domain effective channel model. The delay with the largest coefficient may be selected.

According to further embodiments of the present invention, the delays to be included in a set of delays may be selected in an incremental manner. For example, the first delay is selected for the set of delays using the techniques described above or other selection techniques. A signal and noise content estimate is estimated for the second delay based on the weights associated with the first delay. A second delay is selected for the set of delays based on the generated signal and noise content estimate. A correlation between the first delayed composite signal information and the second delayed composite signal information may be determined, and a second delayed signal and noise content estimate may be determined based on the correlation and a weight associated with the first delay. According to other embodiments of the present invention, signal and noise content estimates may be generated without inverting the noise covariance matrix.

In some embodiments of the invention, a respective signal and noise content estimate for each of a plurality of second delays may be generated from a weighting associated with the first delay. Selecting the second delay for the group of delays based on the generated signal and noise content estimate may include selecting from a plurality of second delays based on the signal and noise content estimate. Such a process may be performed iteratively, e.g., a first one of the second delays may be selected, a new respective weight determined for each delay of a selected group of delays including the selected first one of the second delays, a new signal and noise content estimate generated for a respective delay of the plurality of second delays that has not been selected, and a further one of the second delays selected based on the new signal and noise content estimate. Selecting a new candidate delay may include replacing a previously selected delay in the selected set of delays based on a comparison of the signal to the noise content estimate.

In still further aspects of the invention, delays such as RAKE correlator delays and/or chip equalizer filter taps may be evaluated in an aggregated manner, e.g., as "super fingers". An aggregate signal and noise content estimate may be generated for the delay set. Based on the aggregated signal and noise content estimates, a set of delays to be included in the selected delay group is evaluated.

According to other embodiments of the present invention, channel and correlation characteristics may be determined for a composite signal. Based on the determined channel and correlation characteristics, a respective weight of information in the symbol signal is determined for each of a plurality of candidate delays. A set of delays is selected from the plurality of candidate delays with weights that meet a predetermined criterion. Information in the symbol signal may be processed for a selected delay based on the spreading code to produce a symbol estimate.

In still other embodiments of the present invention, a signal may be recovered from a composite signal. Channel and correlation characteristics of the composite signal may be determined. Based on the determined channel and correlation characteristics, a respective combining weight of information in the symbol signal is determined for each of a plurality of candidate delays. A first delay is selected to be included in a set of delays. A signal and noise content estimate is generated for the second delay based on a weighting associated with the first delay. Based on the generated signal and noise content estimate, a second delay to be included in the set of delays is selected. Information in the symbol signal may be processed for the selected delay based on the spreading code to generate a symbol estimate.

In other embodiments of the present invention, a spread spectrum communication receiver includes a radio frequency processor configured to receive a radio frequency signal containing signals from a plurality of sources and to generate a composite signal containing the signals from the plurality of sources. The receiver further includes a baseband processor configured to: determining channel and correlation characteristics of the composite signal; determining a respective weight of the composite signal information for each of a plurality of candidate delays based on the determined channel and correlation characteristics; selecting a set of delays from a plurality of candidate delays based on the determined weights; and processing information in the symbol signal for the selected delay based on the spreading code to produce a symbol estimate.

According to further embodiments of the present invention, a computer program product comprises computer program code, the computer program code comprising: code configured to determine channel and associated characteristics of the composite signal; code configured to determine, for each of a plurality of candidate delays, a respective combining weight for information in the symbol signal based on the determined channel and correlation characteristics; code configured to select a set of delays from a plurality of candidate delays based on the determined combining weights; and code configured to process information in the symbol signal for the selected delay based on the spreading code to produce a symbol estimate.

Drawings

Fig. 1 is a block diagram illustrating a conventional spread spectrum communication system.

Fig. 2 is a block diagram illustrating a conventional baseband processor having a conventional RAKE structure.

Fig. 3 is a block diagram illustrating a conventional baseband processor having a chip equalizer structure.

Fig. 4 is a block diagram illustrating a weighting computer for a generalized RAKE receiver.

FIG. 5 is a block diagram illustrating a processing device according to some embodiments of the invention.

Fig. 6 is a block diagram illustrating a radio frequency receiver according to other embodiments of the present invention.

Fig. 7 is a block diagram illustrating a signal processing apparatus having a generalized RAKE structure according to other embodiments of the present invention.

Fig. 8 is a block diagram illustrating a signal processing apparatus having a chip equalizer structure according to further embodiments of the present invention.

Fig. 9 is a block diagram illustrating a signal processing apparatus with a probing finger element according to other embodiments of the present invention.

Fig. 10-16 are flow diagrams illustrating different correlator finger and filter tap selection operations according to different embodiments of the present invention.

Fig. 17 is a block diagram showing a system model of a spread spectrum communication system.

Fig. 18 is a flow diagram illustrating correlator finger and filter tap selection operations according to other embodiments of the invention.

Fig. 19 is a block diagram illustrating a non-parametric noise covariance estimator.

Fig. 20 is a block diagram illustrating a parametric noise covariance estimator.

Detailed Description

The present invention now will be described more fully hereinafter with reference to the accompanying drawings, in which embodiments of the invention are shown. However, the invention should not be construed as being limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art. Like reference numerals refer to like parts throughout the several views.

It will be understood that the term "comprises/comprising" when used herein is open-ended, i.e. denotes one or more stated elements, steps and/or functions, and does not exclude one or more unstated elements, steps and/or functions. It will also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items. It will also be understood that when a transfer, communication, or other interaction is described as occurring "between" elements, such transfer, communication, or other interaction may be unidirectional and/or bidirectional.

The present invention is described below with reference to block diagrams and/or operational illustrations of methods, apparatus, and/or computer program products according to embodiments of the invention. It will be understood that each block of the block diagrams and/or operational illustrations, and combinations of blocks in the block diagrams and/or operational illustrations, can be implemented by analog and/or digital hardware, and/or computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, ASIC, and/or other programmable data processing apparatus, such that the instructions, which execute via the processor of the computer and/or other programmable data processing apparatus, create means for implementing the functions/acts specified in the block diagrams and/or operational illustrations. In some alternative implementations, the functions/acts noted in the figures may occur out of the order noted in the block diagrams and/or operational illustrations. For example, two operations shown as being performed in succession may, in fact, be executed substantially concurrently, or the operations may sometimes be executed in the reverse order, depending upon the functionality/acts involved.

According to some embodiments of the invention, an electronic device may include a radio frequency receiver configured to provide the operations described herein. Such receivers may include any of a number of types of devices, including but not limited to: cellular handsets and other wireless terminals, cellular base stations and other types of radio frequency network nodes and wireline receiver devices. Computer program code for carrying out operations of the present invention may be written in the following languages: an object-oriented programming language, a procedural programming language, or low-end code, such as assembly language and/or microcode. The program code may execute entirely on a single processor and/or across multiple processors, as a stand-alone software package or as part of another software package.

According to various embodiments of the present invention, the determination of delays, e.g., signal paths such as RAKE correlator fingers or equalizer filter taps, may be accomplished by determining weights for candidate delays based on channel and correlation characteristics of the composite signal. In some demonstrative embodiments, a maximum weighting criterion may be used to select the correlator delays and/or the equalizer filter taps. In some demonstrative embodiments, a signal-to-noise ratio metric derived from the weighting may be used to identify the required delay or tap. In other embodiments, the channel frequency response may be used to calculate weights, which may be selected with maximum weights or other criteria.

Fig. 5 illustrates a signal processing apparatus 500 according to some embodiments of the inventions. The apparatus 500 includes a symbol estimator 510, e.g., a combination of a correlation unit and a combiner used in a conventional RAKE receiver architecture or a combination of a FIR filter and a correlator used in a chip equalizer. The symbol estimator 510 processes the composite signal according to the spreading code using a delay, e.g., a correlator delay or filter tap, selected from a plurality of candidate delays by a delay selector 520. The delay selector 520 selects the delay according to the combining weight generated by the weight determiner 530. Combining weights 530 are generated based on the channel and correlation characteristics determined by channel and correlation determiner 540.

As described in greater detail below, the apparatus and operations described in terms of apparatus 500 may be implemented in a number of different ways according to different embodiments of the invention. For example, delay selection may be based on a maximum weighting criterion, such as selecting correlator finger delays, equalizer filter taps, and/or signal sources (e.g., different antennas), in accordance with various embodiments of the present invention. In other exemplary embodiments, signal-to-noise ratio metrics derived from such weights may be used to identify the required correlator delays (fingers) or taps. In other embodiments, the channel frequency response may be used to calculate weights, which may be selected with a maximum weight or other criteria.

It is to be understood that the apparatus and methods according to various embodiments of the present invention may be generally implemented using analog and/or digital electronic circuits. For example, function blocks 510 and 540 may be implemented with program code executed on a data processing device, such as a microprocessor or Digital Signal Processor (DSP), or in data processing circuitry included in a special purpose electronic device, such as a communications ASIC. The invention may also be embodied as computer code configured to provide the operations described above when executed on a data processing apparatus.

Fig. 6 shows a receiver 600 comprising a signal processing device as shown in fig. 5. The rf signal, which may include information for a number of transmission sources, is received by antenna 670 and provided to rf processor 650, which performs filtering, frequency down conversion, and other processes to generate a composite baseband signal. The baseband signal is provided to a baseband processor that includes a symbol estimator 610, a delay selector 620, a channel and correlation determiner 640, and a weight determiner 630, which may have the same functions as described above for the corresponding parts of fig. 5. The symbol estimates produced by the symbol estimator 610 are provided to an additional processor 660 that may perform other signal processing functions, such as error correction decoding.

Fig. 7 shows a signal processing device 700 according to a further embodiment of the invention. Apparatus 700 includes a symbol estimator 710 that processes a composite signal to generate symbol estimates. The symbol estimator 700 comprises a correlation unit 712 and a combiner 714. The correlation unit 712 correlates the composite signal with the spreading code by using a correlation delay selected from the candidate delays by the correlation delay selector 720 according to the weight determined by the weight determiner 730. The weight determiner 730 determines weights based on channel and correlation characteristics determined by the channel and correlation determiner, e.g., a combination of channel estimates and quantitative correlation estimates, such as a noise covariance estimate.

Fig. 8 shows a signal processing apparatus 800 with an alternative chip equalizer structure according to further embodiments of the present invention. Apparatus 800 includes a symbol estimator 810 that processes a composite signal to generate symbol estimates. The symbol estimator 810 includes a filter 812 that filters the composite signal and a correlator 814 that correlates the output of the filter 812 with the spreading code. The taps of the filter 812 are selected from a plurality of candidate taps by the filter tap selector 820 based on the combining weights generated by the weight determiner 830. The coefficients of filter 812 correspond to the coefficients (i.e., weights) of the selected taps. Coefficient determiner 830 determines coefficients based on the channel and correlation characteristics of the composite signal determined by channel and correlation determiner 840.

Candidate delays, e.g., correlator delays and/or filter taps, provided to a signal processing apparatus such as those shown in fig. 5-8, may be generated in any of a number of different manners. For example, a "probe finger" approach may be used to generate candidate delays as described in the aforementioned U.S. patent application 09/845950 (published as U.S. patent application publication US 2001/0028677), filed on 30/4/2001.

One such embodiment is shown in fig. 9. As shown in fig. 9, a signal processing apparatus 900 according to some embodiments of the present invention includes a symbol estimator 910 that includes a correlation unit 912 that correlates a composite signal with a traffic channel spreading code and a combiner 914 that combines the correlations generated by the correlator 912 in a weighted manner. Correlator 912 uses the delay selected from the plurality of candidate delays by correlation delay selector 920 based on the weights generated by weight determiner 930. The weights are generated based on the channel and correlation characteristics generated by channel and correlation determiner 940. The candidate delays provided to the correlation delay selector 920 are generated by the probing unit 950. The detection unit 950 may identify candidate delays by, for example, correlating the composite signal with pilot channel codes, as described in the above-mentioned U.S. patent application 09/845950. It is understood that other ways of identifying candidate delay or filter taps may also be used with the present invention.

Fig. 10 is a flow diagram illustrating exemplary signal processing operations according to some embodiments of the present invention. Channel and correlation characteristics of a composite signal including a desired signal and interfering signals from one or more transmission sources are determined (block 1010). A respective weight is determined for each of the candidate delays based on the channel and correlation characteristics of the composite signal (block 1020). A set of candidate delays is selected based on the weights, e.g., by selecting those delays having the largest associated weights or desired signal-to-noise estimates derived from the weights (block 1030). Here, "maximum association weight" may be regarded as a maximum amplitude, a maximum amplitude squared, or the like. Information in the symbol signal from the selected delay is processed according to the spreading code to generate a symbol estimate of the desired signal in the composite signal (block 1040).

According to some aspects of the invention, a maximum weighting criterion may be used to select delays, e.g., G-RAKE correlator delays or chip equalizer filter taps. With the G-RAKE structure, it is assumed that a set of N candidate correlator delays have been identified, and that there is a noise covariance matrix R for the N candidate delays_NForm of correlation feature and channel coefficient vector C_NForm of channel characteristics, combined weights W of candidate delays_NCan be expressed as follows:

combining weights W_NCan be determined using, for example, a direct matrix inversion or another method of solving a system of linear equations (e.g., the gaussian-seidel method). According to some embodiments of the invention, a subset of the N candidate delays may be selected by selecting those delays having the greatest associated weights.

Fig. 11 shows an exemplary signal processing operation according to this approach. Channel coefficients and covariance matrices for the composite signal and the set of candidate delays are determined (block 1110). Weights for the candidate delays are generated based on the determined channel coefficients and the covariance matrix (block 1120). The candidate delay with the greatest weight is selected (block 1130). The composite signal is correlated with the spreading code at the selected delay (block 1140). The resulting correlations are combined according to the recalculated weights to generate symbol estimates (block 1150). The recalculation may use equation (2), where N is substituted with the selected number of delays.

Fig. 12 illustrates exemplary operation in an alternative chip equalizer implementation. Channel coefficients and covariance matrices for the composite signal and the candidate filtered tap sets are determined (block 1210). Weights for the candidate filter taps are generated from the determined channel coefficients and covariance matrix (block 1220). The candidate filter tap with the greatest weight is selected (block 1230). The composite signal is filtered using a Finite Impulse Response (FIR) filter having non-zero coefficients for only selected taps (block 1240). This may be implemented, for example, with a filter having a programmable delay. The filter coefficients are obtained by recalculating the weights. The output of the FIR filter is correlated with the spreading sequence to generate symbol estimates (block 1250).

It will be appreciated that the delay selection may be accomplished in a number of different ways within the scope of the invention. For example, the weights may be sorted in ascending or descending order of magnitude, and the top (or last) M delays selected. Alternatively, iterative search techniques may also be used. For example, the maximum magnitude weighting may be searched, the associated delay selected, and the delay (and weighting) disregarded. It is also understood that "quantitative correlation" outside the noise covariance may be used. For example, for Minimum Mean Square Error (MMSE) combining, a data correlation matrix generated corresponding to pilot despread values may be used.

In accordance with other aspects of the invention, RAKE correlator delays, chip equalizer filter taps, or other delays may be incrementally selected by determining signal and noise content changes associated with adding a particular delay or set of delays. For the purposes of the following discussion, a G-RAKE structure is assumed and a set of M candidate fingers has been selected. For such a set of M fingers, the M x 1 vector c contains the corresponding channel taps and the M x M matrix R contains the corresponding correlation coefficients. The matrix R is a Hermitian matrix (Hermitian) and is reversible except in the degenerate case. An mx 1 vector w contains the combined weights of each of the M fingers. The combiner output z is the weighted sum of the despread values represented by the mx 1 vector r:

z＝w^Hr。 (3)

for a given combining weight vector w, the signal-to-noise ratio (SNR) can be expressed as:

for the G-RAKE implementation, the combining weights that theoretically maximize SNR can be expressed as follows:

w＝R^-1c。 (5)

accordingly, the SNR can be expressed as follows:

γ＝c^HR^-1c。 (6)

if a new finger is to be added to the selected group, the impact of adding the new finger on the SNR may be determined based on the weights and noise covariance of the previously selected fingers. Placing the new finger at the end of the updated channel tap vector c':

where χ represents a new channel tap corresponding to a new finger. The new correlation matrix R' can be represented as follows:

wherein the Mx 1 vector ρ contains a correlation coefficient between the old finger and the new finger, and σ²Is the noise variance of the new finger. Inverse matrix R 'of new correlation matrix R'^-1Can be expressed as follows:

wherein

α^-1＝σ²-ρ^HR^-1ρ (10)。

Using equation (9), the new optimal weighting vector can be expressed as follows:

wherein the weighted delta vector Δ w is represented as follows:

Δw＝αR^-1ρρ^Hw-αR^-1ρχ， (12)

and represents the change in the old best weight due to the addition of a new finger. The weight w for the new finger may be expressed as follows:

w＝-αρ^Hw+αχ。 (13)

assuming that a "noise" finger is being evaluated for the addition, i.e., a finger that is not aligned with the desired signal echo, the channel tap χ may be set to zero, thus:

for such cases, equation (12) can be simplified to:

Δw＝αR^-1ρρ^Hw， (15)

and equation (13) can be simplified as:

w＝-αρ^Hw。 (16)

new SNRThe old SNR γ can be expressed as follows using equations (9) and (10):

γ′＝c′^H R′^-1c′＝γ+Δγ， (17)

wherein the SNR increment Deltay can be expressed as follows

The SNR delta deltay represents the SNR change due to the addition of a new finger.

Analysis of equation (18) shows that the numerator can be maximized by matching the new correlation vector ρ with the old weighting vector w, and that the inverse matrix R of the old correlation is reached at the new correlation vector ρ^-1When matching, the denominator can be minimized. Thus, to maximize the SNR increment Δ γ, the new correlation vector ρ may be in the old correlation inverse R^-1Is calculated on the eigenvector of the largest eigenvalue of (c). Since the old correlation matrix R is invertible, it does not have zerosSpace, therefore, for this optimal condition, the new correlation vector ρ comes to lie in the null space of the old correlation matrix R. This can be understood to mean that the new finger should provide information about the dimension that the existing finger does not effectively cover.

According to some embodiments of the present invention, a simplified method of determining SNR delta without going to the inverse of the matrix may be provided. First order approximation of new weightsCan be expressed as follows:

the weight of a new finger may be set without using equation (13)But rather determine thatSNR of (1)Up to maximum weightingThe value of (c). Using equations (4), (8), and (14):

the numerator of equation (20) is independent of weighting. The denominator can be minimized with the following formula:

this yields the approximate SNR given belowMaximum value of (d):

approximate SNR delta ΔCan be expressed as follows:

this expression does not require matrix inversion and is therefore easier to calculate than equation (18). Theoretically, approximate SNR Should be less than the SNR γ' calculated according to equation (17). Approximate SNR if the number of old fingers is not too smallThe SNR γ' calculated according to equation (17) can be very close. Accordingly, approximate SNR(or increment. delta. Δ)) May be used to evaluate a new finger.

Fig. 13 illustrates exemplary operations for incrementally selecting G-RAKE fingers in accordance with some embodiments of the present invention. Assume that one or more members of the delay set have been selected and the associated weights and covariance matrix have been determined (block 1310). SNR estimates for one or more candidate delays are generated based on the predetermined weights for the previously selected delays and the covariance matrix (block 1320). Based on the SNR estimate, one or more new delays are selected (block 1330), and the new delays are used to process the composite signal (block 1340). The selection may be an addition and/or involve replacing a previously selected delay, e.g., if an "old" delay is added by an incremental process, its SNR estimate may be kept in memory and compared to the newly determined SNR estimate for the new candidate finger to determine whether to replace the old delay with the new delay.

Figure 14 illustrates exemplary operations for incrementally selecting delays in accordance with other embodiments of the present invention. Likewise, assume that a set of M delays has been selected from up to L candidate delays, and that the associated weighting and covariance matrices have been determined (block 1410). SNR estimates for L-M remaining candidate delays are generated based on the weighted sum covariance matrices for the M delays (block 1420). These SNR estimates correspond to a set of M selected delays plus one additional delay. The set of selected delays may be updated by selecting the remaining candidate delays with the largest SNR estimate (block 1430) and using the updated set of delays in processing the composite signal (block 1440).

An alternative incremental approach according to other embodiments of the present invention is shown in fig. 15. A set of M delays is selected from up to L candidate delays and an associated weight and covariance matrix is determined (block 1510). SNR estimates for the remaining candidate delays are generated based on the weights and covariance matrices for the previously selected delays (block 1520). The delay with the best SNR estimate is selected (block 1530). If the desired amount of new delay has been selected (block 1540), the composite signal is processed with the updated selected set of delays (block 1550). If, however, the desired number of delays is not selected, a new weight and a new covariance matrix for the selected delay group including the newly selected delay are calculated (blocks 1560, 1510), new SNR estimates for the remaining candidate delays are generated (block 1520), and another new delay is selected (block 1530). The method illustrated in fig. 15 may be more computationally burdensome than the operations of fig. 14, but it may utilize more information and thus may provide more accurate results, e.g., may perform better in avoiding redundancy options.

According to further embodiments of the present invention, an incremental method of selecting delays or taps may be implemented by processing delay groups in an aggregated manner, i.e., treating the delay groups as "super fingers". Due to complexity considerations, the receiver may not be able to evaluate each finger individually. To reduce complexity, the receiver may identify several finger sets to be evaluated as one set. The group as a whole may be added to the selected finger set without the need to evaluate each finger in the group separately if the group as a whole shows the required degree of SNR improvement.

An exemplary operation of such a method is shown in fig. 16. A set of M delays is selected from up to L candidate delays and an associated weight and covariance matrix is determined (block 1610). From the weights and covariance matrices for the previously selected delays, an aggregate SNR estimate for a set (or sets) of remaining candidate delays is generated (block 1620). The group of R delays with the best SNR estimate is selected (block 1630). If the selected delay update has completed (block 1640), the composite signal is processed with an update set that includes the selected group of R delays (block 1650). If not, the updated weights and new covariance matrix are determined (blocks 1660, 1610), new SNR estimates are generated (block 1620), and a new set of delays are selected (block 1630). The one-time criterion to complete is that all available fingers have been used. It will be appreciated that the group-based approach of fig. 16 may be combined with the separate evaluation approach of fig. 15, e.g., a super finger may be evaluated with a single finger.

It may be noted that the aggregation process may compromise useful information associated with a single finger in a super finger when determining channel taps and related parameters for the super finger. In other words, component fingers may exhibit a useful correlation with the selected fingers, but the correlation may be cancelled in the super finger aggregation. For the super finger, an exact (e.g., equation (18)) or approximate (e.g., equation (23)) SNR estimate may be used. If the SNR estimate is large, it is shown that the group or subgroup is of interest. If the receiver has excess computational power, it can decide to investigate each finger in the group in detail. Alternatively, it may confirm the group more thoroughly in future evaluations. If the improvement in SNR is small, the receiver may discard the group. If a super-finger can provide a significant SNR improvement, it can be used directly in the G-RAKE combiner, i.e., since the same weight can be applied to each constituent finger, it is not necessary to process the constituent fingers in the combiner separately.

The super fingers may be considered an average of a plurality of fingers. It can also be considered as a low-pass combination, which can result in increased information loss as the number of fingers increases. This morphology can be augmented by using a Hadamard-like structure. For example, assume that a group includes 2^jA +1/-1 weighting vector may be added to the fingers and the results summed. Hadamard sequence provides all (2)^j-1) Balanced sequences (equal numbers of +1 and-1). These sequences may be used to attempt to obtain a super finger with a significant SNR improvement. This can be seen as a simplified pre-combination using only +1 and-1.

Other aspects of the invention may be described in terms of chip equalizer structures, although these methods are also applicable to the selection of G-RAKE correlator delays. Selecting a combining finger in a G-RAKE receiver can be viewed as analogous to setting all coefficients of a chip equalizer FIR filter to 0 except for a small subset. The formulation of the chip equalizer can be done in the frequency domain.

Fig. 17 illustrates a representative downlink structure of spread spectrum signals (e.g., IS-2000 and CDMA) for terminals connected to a single base station. The terminology used in fig. 17 is as follows: x is the number of_i(t) represents a spread/scrambled signal of the ith user; e_iRepresenting the energy per symbol of the ith user signal; h is_TXIs the impulse response of the transmit filter; c (t) is the time-varying channel response; h is_RXIs the impulse response of the receive filter; n (t) is a signal having a variance N₀Complex additive white gaussian noise.

The colored noise matched filter for the system in fig. 17 is as follows:

wherein H_TX(ω) is h_TXFourier transform of (H)_c(ω) is c_(t)Fourier transform of (1), X₀(ω) is x₀(t) Fourier transform, and

in the context of a hypothetical system model and a hypothetical chip equalizer structure, the term of equation (24) may be associated with a receive filter, a FIR filter, or a weighting filter and correlator. These terms are readily observed:

a receiving filter:

H_TX ^*(ω)；

FIR filter:

and (26)

A correlator:

X₀ ^*(ω)。

the problem is how to calculate the weighted frequency response over the weighting filter. This weighting filter may be represented by (26).

For a given system, product H_TX(ω)H_c(ω)H_RX(ω) is the Fourier transform H of the net channel coefficient_NET(ω). H is generally known exactly or approximately at the receiver_RX(ω). Thus, H^* _TX(ω)H^* _c(ω) can be calculated as follows:

and only signal energy E is required_ISum noise variance N₀To calculate the frequency response H_off(ω). To obtain signal energy E_ISum noise variance N₀Can perform the procedures described, for example, in U.S. patent 10/943274 entitled "Method and Apparatus for CDMA Receivers" filed 2005, 9/17, to douglas a. cairns and Leonid Krasny.

Fig. 18 illustrates an exemplary embodiment of selecting correlator delays or filter taps with a weighted frequency response according to some embodiments of the invention. Channel coefficients, receive filter responses, signal energy, and noise variance are determined (block 1810). A weighted frequency response (e.g., as in equation (26)) is determined from these factors (block 1820). This weighted frequency response is converted to the time domain to produce a time domain weighted frequency response (1830). Correlator delays or FIR filter taps are selected based on the weights (block 1840). The composite signal is processed using the selected delay or tap (block 1850).

The block diagrams and flowchart illustrations of fig. 5-16 and fig. 18 illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present invention. It should also be noted that, in some alternative implementations, the acts noted in the figures may occur out of the order noted in the figures. For example, two operations shown as being performed in succession may, in fact, be executed substantially concurrently, or the operations may sometimes be executed in the reverse order, depending upon the functionality involved.

The present invention also includes other variations on the apparatus, methods, and computer program products described above. Such correlations may be obtained using parametric or non-parametric methods, for example, in techniques that employ "quantitative correlation" such as noise covariance or data correlation. A non-parametric approach may involve measuring the number correlation, for example, with a probing finger or set of existing fingers. The parametric approach may involve, for example, computing the noise correlation using channel estimates and/or other quantities.

A conventional non-parametric noise covariance estimator that may be used with the present invention, as described in U.S. patent 6363104 to bottomley et al, is shown in fig. 19. The channel estimates are provided to a remove signal unit 1910 that extracts the channel estimates from the despread values (with modulation removed) to form error values. The error processor 1920 receives the error value and includes a noise correlation computer 1922 in which the error is multiplied by the conjugate of the other error to produce a noise correlation value. These noise correlation values are then provided to a smoother 1924 that averages the values over a period of time (e.g., a number of time slots).

Fig. 20 shows a parametric noise covariance estimator that may be used with the present invention as described in U.S. application 10/800167 entitled "Method and Apparatus for parameter estimation in a Generalized RAKE Receiver", filed 3/12/2004. The removal signal unit 2010 generates an error signal which is used to perform noise correlation measurements in a noise correlation computer 2022 of the sub-processor 2020. The measurements are fitted to a noise covariance model, which includes the structural elements provided by the structural element computer 2024 and the scaling parameters determined by the G-RAKE parameter estimator 2026. The scaling parameters and structural elements are combined by the noise covariance computer 2028 to produce a noise covariance estimate.

For non-parametric methods, fingers may be placed at candidate delays to measure data or noise correlations. For a chip equalizer, the samples may be correlated with other samples based on the candidate delay. For parametric methods, the concept of "virtual" probing fingers may be introduced, i.e. parametric methods allow to evaluate candidate delays or taps even if no actual corresponding physical delay or tap information is available.

It is also understood that the invention also encompasses embodiments using multiple receive antennas. For example, finger or tap selection may be done on an antenna-by-antenna basis, or a combined number correlation may be used to select delays from multiple antennas. Embodiments that provide transmit diversity are also encompassed by the present invention. For example, for transmit diversity with feedback, pilots from multiple transmit antennas may be used to calculate a composite channel estimate. For soft handoff, a weighted solution may be computed separately and used to select from the combined set of candidate delays. Alternatively, a joint SNR metric may also be used. For transmit diversity using Alamouti codes, the same finger selection may be used for both transmitted signals. For multiple-input multiple-output (MIMO) applications, delay selection may be performed for each signal individually, or a method that maximizes the minimum SNR may be used.

In the drawings and specification, there have been disclosed typical illustrative embodiments of the invention and, although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation, the scope of the invention being set forth in the following claims.

Claims

1. A method of recovering a signal from a composite signal containing signals from one or more sources, the method comprising the steps of:

determining channel and correlation characteristics of the composite signal;

determining, for each of a plurality of candidate delays, a respective combining weight for information in the composite signal based on the determined channel and correlation characteristics;

selecting a set of delays from the plurality of candidate delays based on the determined combining weights; and

information of the composite signal is processed for the selected delay based on the spreading code to generate symbol estimates.

2. The method of claim 1, wherein selecting a set of delays from the plurality of candidate delays based on the determined combining weights comprises selecting a set of delays having a largest corresponding combining weight.

3. The method of claim 2:

wherein the step of determining the channel and correlation characteristics comprises determining a time domain channel response and a quantity correlation; and

wherein the step of determining respective combining weights comprises determining the respective combining weights in dependence on the time domain channel responses and the quantitative correlations.

4. The method of claim 3, wherein the quantitative correlation comprises a noise covariance.

5. The method of claim 2, wherein:

the step of determining channel and correlation characteristics includes determining a weighted frequency response including noise information; and

the step of determining the respective combining weights comprises determining the respective combining weights in dependence on the weighted frequency response.

6. The method of claim 5, wherein determining a weighted frequency response comprises:

determining a channel frequency response;

determining a white noise variance and an aggregated signal energy for the plurality of sources; and

determining the weighted frequency response from the channel frequency response, the white noise variance, and the aggregated signal energy.

7. The method of claim 5, wherein:

a step of determining the respective combining weights from the weighted frequency responses, including converting the weighted frequency responses to a time-domain weighting model including a plurality of signal taps and corresponding coefficients; and

selecting a set of delays from the plurality of candidate delays based on the determined combining weights comprises selecting a set of signal taps from the plurality of signal taps based on the coefficients.

8. The method of claim 1, wherein selecting a set of delays from the plurality of candidate delays based on the determined combining weights comprises:

selecting a first delay to be included in the delay group;

generating a signal and noise content estimate for a second delay based on a weight associated with the first delay; and

selecting the second delay to be included in the set of delays based on the generated signal and noise content estimates.

9. The method of claim 8, wherein generating the signal to noise content estimate comprises:

determining a correlation between information in the first delayed composite signal and information in the second delayed composite signal; and

generating the signal and noise content estimate for the second delay as a function of the correlation and the weighting associated with the first delay.

10. The method of claim 9, wherein generating a signal and noise content estimate based on the correlation and the weighting associated with the first delay comprises generating the signal and noise content estimate without inverting a noise covariance matrix.

11. The method of claim 8, wherein:

generating a signal and noise content estimate for a second delay in accordance with a weighting associated with the first delay, including generating a respective signal and noise content estimate for each of the plurality of second delays in accordance with the weighting associated with the first delay; and

selecting the second delay to be included in the set of delays based on the generated signal and noise content estimate comprises selecting from the plurality of second delays based on the signal and noise content estimate.

12. The method of claim 11, wherein selecting from the plurality of second delays based on the signal and noise content estimates comprises selecting the second delay whose associated signal and noise content estimate exhibits the greatest signal to noise ratio improvement.

13. The method of claim 11, wherein:

selecting from the plurality of second delays based on the signal and noise content estimates comprises selecting a first one of the second delays;

determining a respective combining weight for information of said composite signal for each of a plurality of candidate delays based on the determined channel and correlation characteristics, including determining a new respective combining weight for each delay of a selected group of delays including a selected first one of said second delays;

generating a respective signal and noise content estimate for each of the plurality of second delays based on the weighting associated with the first delay, including generating a new signal and noise content estimate for each of the plurality of second delays that has not been selected; and

selecting from said plurality of second delays based on said signal and noise content estimates comprises selecting a second of said second delays based on said new signal and noise content estimates.

14. The method of claim 13, wherein selecting a second one of the second delays based on the new signal to noise content estimate comprises replacing a previously selected one of the selected set of delays with the second one of the second delays based on a comparison of the signal to noise content estimate.

15. The method of claim 11, wherein selecting from the plurality of second delays based on the signal to noise content estimate comprises:

generating an aggregated signal and noise content estimate for the second set of delays; and

determining whether to select the second set of delays to include in the group of selected delays based on the aggregated signal and noise content estimates.

16. The method of claim 1, wherein processing information in the composite signal for the selected delay based on a spreading code to generate symbol estimates comprises:

generating, for each of the selected delays, a correlation of the respective composite signal with the spreading code; and

combining the generated correlations to generate the symbol estimates.

17. The method of claim 1, wherein processing information in the composite signal for the selected delay based on a spreading code to generate symbol estimates comprises:

filtering the composite signal with a filter having filter taps corresponding to each of the selected delays; and

correlating the filtered spread spectrum communication signal with the spreading code to generate the symbol estimate.

18. The method of claim 1, wherein determining the channel and correlation characteristics of the composite signal comprises determining the correlation characteristics parametrically and/or non-parametrically.

19. The method of claim 1, wherein determining the channel and correlation characteristics of the composite signal comprises determining the correlation characteristics based on a correlation of the composite signal over the plurality of candidate delays.

20. A method of recovering a signal from a composite signal containing signals from one or more sources, the method comprising the steps of:

determining channel and correlation characteristics of the composite signal;

selecting a set of delays from the plurality of candidate delays having a combining weight that satisfies a predetermined criterion; and

information in the composite signal is processed for the selected set of delays according to a spreading code to generate symbol estimates.

21. The method of claim 20, wherein selecting a set of delays from the plurality of candidate delays having combining weights that satisfy a predetermined criterion comprises selecting a set of delays having a largest corresponding combining weight.

22. The method of claim 21, wherein:

the step of determining channel and correlation characteristics includes identifying time domain channel response and quantity correlation; and

the step of determining the respective combining weights comprises determining the respective combining weights in dependence on the time domain channel responses and the quantitative correlations.

23. The method of claim 22, wherein the quantitative correlation comprises a noise covariance.

24. The method of claim 20, wherein:

the step of determining channel and correlation characteristics includes determining a weighted frequency response containing noise information; and

25. The method of claim 24, wherein determining a weighted frequency response comprises:

determining a channel frequency response;

26. The method of claim 24, wherein:

determining the respective combining weights from the weighted frequency responses, including converting the weighted frequency responses to a time-domain weighting model containing a plurality of signal taps and corresponding coefficients; and

27. A method of recovering a signal from a composite signal containing signals from one or more sources, the method comprising the steps of:

determining channel and correlation characteristics of the composite signal;

selecting a first delay of a set of delays;

generating a signal and noise content estimate for a second delay according to a weighting associated with the first delay; and

selecting the second delay to be included in the set of delays based on the generated signal and noise content estimate; and

processing information in the composite signal for the selected first and second delays according to a spreading code to generate symbol estimates.

28. The method of claim 27, wherein generating the signal to noise content estimate comprises:

determining a correlation between the first delayed composite signal information and the second delayed composite signal information; and

generating the signal and noise content estimate for the second delay based on the correlation and a weighting associated with the first delay.

29. The method of claim 28, wherein generating a signal and noise content estimate based on the correlation and the weighting associated with the first delay comprises generating the signal and noise content estimate without inverting a noise covariance matrix.

30. The method of claim 27, wherein:

generating a signal and noise content estimate for a second delay in accordance with the weight associated with the first delay, including generating a respective signal and noise content estimate for each of the plurality of second delays in accordance with the weight associated with the first delay; and

31. The method of claim 30, wherein selecting from the plurality of second delays based on the signal and noise content estimates comprises selecting the second delay whose associated signal and noise content estimate exhibits the greatest signal to noise ratio improvement.

32. The method of claim 30, wherein:

selecting from said plurality of second delays based on said signal and noise content estimates, including selecting a first one of said second delays;

determining a respective combining weight for each of a plurality of candidate delays with respect to information in the composite signal based on the determined channel and correlation characteristics, including determining a new respective combining weight for each delay of a selected group of delays including a selected first one of the second delays;

selecting from said plurality of second delays based on said signal and noise content estimates comprises selecting a second one of said second delays based on said new signal and noise content estimates.

33. The method of claim 32, wherein selecting a second one of the second delays based on the new signal to noise content estimate comprises replacing a previously selected delay in the selected set of delays with the second one of the second delays based on a comparison of the signal to noise content estimate.

34. The method of claim 30, wherein selecting from the plurality of second delays based on the signal to noise content estimate comprises:

35. A spread spectrum communication receiver, comprising:

a radio frequency processor configured to receive a radio frequency signal comprising signals from one or more sources and to generate a composite baseband signal comprising signals from one or more sources; and

a baseband processor configured to: determining channel and correlation characteristics of the composite signal; determining, for each of a plurality of candidate delays, a respective combining weight for information in the composite signal based on the determined channel and correlation characteristics; selecting a set of delays from the plurality of candidate delays based on the determined combining weights; and processing information of the composite signal for the selected delay based on the spreading code to generate symbol estimates.

36. The receiver of claim 35 wherein the baseband processor is operative to select a set of delays having a largest corresponding combining weight.

37. The receiver of claim 36, wherein the baseband processor is configured to identify a time domain channel response and a quantity correlation and to determine the corresponding combining weights based on the time domain channel response and the quantity correlation.

38. The receiver of claim 37, wherein the quantitative correlation comprises a noise covariance.

39. The receiver of claim 36 wherein the baseband processor is operative to determine a weighted frequency response containing noise information and to determine the corresponding combining weights based on the weighted frequency response.

40. The receiver of claim 35, wherein the baseband processor is operative to generate a signal and noise content estimate for a second delay based on a weight associated with a first delay, and to select the second delay to be included in the set of delays based on the generated signal and noise content estimate.

41. The receiver of claim 40, wherein the baseband processor is operative to generate an aggregate signal and noise content estimate for the second set of delays and to determine whether to select the second set of delays to include in the group of selected delays based on the aggregate signal and noise content estimate.

42. The receiver of claim 35, wherein the baseband processor is operative to generate a correlation of the respective composite signal with the spreading code for each of the selected delays, and to combine the generated correlations to generate the symbol estimates.

43. The receiver of claim 35 wherein the baseband processor filters the composite signal using a filter having filter taps corresponding to each of the selected delays and correlates the filtered spread spectrum communications signal with the spreading code to generate the symbol estimates.

44. A computer program product for recovering a signal from a composite signal comprising signals from one or more sources, the computer program product comprising computer program code stored in a computer readable storage medium, the computer program code comprising:

code configured to determine channel and correlation characteristics of the composite signal;

code configured to determine, for each of a plurality of candidate delays, a respective combining weight for information in the composite signal based on the determined channel and correlation characteristics;

code configured to select a set of delays from the plurality of candidate delays based on the determined combining weights; and

code configured to process information in the composite signal for selected delays based on a spreading code to generate symbol estimates.