KR20080036223A - Wireless communications device including a joint space-time optimum filter(jstof) using cholesky and eigenvalue decompositions - Google Patents

Abstract

A wireless communications device may include a wireless transmitter and a wireless receiver. The wireless receiver may include a filter for reducing co-channel interference and may include a multi-channel, space-time filter circuit that filters n signal parts that have been split from a communications signal by jointly estimating space-time filter weights and multi-channel impulse responses (CIRs) based upon Cholesky and eigenvalue decompositions. The filter may further include a multi-channel, matched filter circuit that receives multi-channel signals from the multi-channel, space-time filter circuit and has a filter response that is provided by a channel impulse response estimation from the space-time filter circuit.

Description

WIRELESS COMMUNICATIONS DEVICE INCLUDING A JOINT SPACE-TIME OPTIMUM FILTER (JSTOF) USING CHOLESKY AND EIGENVALUE DECOMPOSITIONS}

FIELD OF THE INVENTION The present invention relates to wireless communication systems, such as cellular communication systems, and more particularly to filtering received wireless signals to reduce unwanted interference.

Interference canceling matched filter (ICMF) to meet the requirements for Downlink Advanced Receiver Performance (DARP) standardized by 3rd Generation Mobile Communication Systems and 3rd Generation Partnership Project (3GPP) And joint demodulation (JDM) have been studied. Some of these proposals are described in the following papers and documents.

Liang et al. A Two-Stage Hybrid Approach for CCI / ISI Reduction with Space-Time Processing, IEEE Communication Letter Vol. 1, No. 6, Nov. 1997.

2. Multichannel Receives Performance Comparison In the Presence of ISI and CCI, 1997 13th Intl. Conf. on Digital Signal Processing, July 1997.

Spagnolini's Adaptive Rank-One Receiver for GSM / DCS Systems, IEEE Trans, on Vehicular Technology, Vol. 51, No. 5, Sept. 2002.

4.Feasibility Study on Single Antenna Interference Cancellation (SAIC) for GSM Networks, 3GPP TR 45.903 Version 6.0.1, Release 6, European Telecommunications Standards Institute, 2004.

5.Radio Transmission and Reception (Release 6), 3GPP TS 45.005 Version 6.8.0; European Telecommunications Standards Institute, 2005.

6. Maximum Likelihood Parameter and Rank Estimation in Reduced-Rank Multivariate Linear Regressions, IEEE Trans. On Signal Processing, Vol. 44, No. 12, Dec. 1996.

7. Blind Subspace Identification of a BPSK Communication Channel, Proc. 30 ^th Asilomar O Conf. On Signals, Systems and Computers, 1996.

8. Matrix Computations, Golub et al., 3 ^rd Edition, 1996.

9. Numerical Linear Algebra, 1997, Trefethen et al.

10. Numerical Recipes in C, 2 ^nd Edition, 1992.

Current Global System for Mobile communications (GSM) cellular systems must cope with co-channel interference (CCI) on the mobile station (MS) side as well as with DARP requirements. Some single channel structures and pre-filters have been used to aid in canceling the interference and provide some channel impulse response (CIR) estimates. In addition, some systems jointly designed a single channel space-time filter and a CIR estimate for a single channel using signal-to-interference maximization. Another system designed a single channel spatial filter using constrained minimization of the mean squared error. Another system used a single channel spatial filter designed by rank-one approximation of ML channel estimation. A target application for such a system was a base station in which a physical antenna array comprising a plurality of antennas was available.

Generally speaking, the present invention relates to a wireless communication device that may include a wireless transmitter and a wireless receiver. More specifically, the wireless receiver may include a filter to reduce co-channel interference and from the communication signal by jointly estimating the space-time filter weight and the multi-channel impulse response (CIR) based on Cholesky decomposition. It includes a multi-channel space-time filter circuit for filtering the separated signal portion. The multichannel matched filter circuit receives the multichannel signal from the multichannel space-time filter circuit and has a filter response provided by the channel impulse response estimate from the space-time filter circuit. The standard filter can be operated when the interference level is lower than a predetermined threshold, the standard filter being formed as a matching filter, a cross-correlation circuit, and a switching mechanism that converts the signal portion into the matching filter and the cross-correlation circuit. Can be.

In one aspect, the multi-channel space-time filter circuit includes a plurality of multipliers and delay circuits each receiving n signal portions. The multiplier and delay circuit are operated based on the space-time filter weights. Each multiplier and delay circuit includes two multiplier circuits and one delay circuit. Each multiplier and delay circuit operates with one symbol delay. The joint optimal filter weight and channel estimator are operatively connected to the multichannel space-time filter circuit, receive training sequence (TS) symbols and timing uncertainty data, and generate space-time filter weights for the multichannel space-time filter circuit. A summer circuit sums the data from the multiplier and delay circuit for each channel. The equalizer circuit works with the multichannel matched filter circuit.

Various objects, features and advantages will become apparent from consideration of the accompanying drawings in light of the following detailed description.

1 is a block diagram of a DARP capable receiver based on a joint space-time optimized filter according to an exemplary embodiment.

FIG. 2 is a more detailed block diagram of the joint space-time optimum filter and the multichannel matched filter shown in FIG. 1 in accordance with an exemplary embodiment.

2A is a block diagram of a method according to an exemplary embodiment.

FIG. 3 is a graph illustrating DARP capable receiver performance based on a joint space-time optimized filter for various DARP cases.

4 is a graph illustrating the comparison of a joint space-time optimal filter receiver performance according to an exemplary embodiment with additive white Gaussian noise (AWGN) with or without an automatic switching scheme.

FIG. 5 is a graph illustrating a comparison of a joint space-time optimum filter receiver performance according to an exemplary embodiment with DTS-5 with or without an automatic switching scheme.

6 is a graph comparing the performance of multiple Viterbi equalizers with the performance of a single Viterbi equalizer using an 8-bit SD limiter in a simulation.

7 is a graph illustrating a modified example and the performance of a joint space-time optimized filter receiver according to an exemplary embodiment.

8 is a schematic block diagram of an example model wireless communication device that may be used in accordance with an example embodiment.

9 is a table comparing three approaches for performing Cholesky decomposition, QR decomposition, and singular value decomposition (SVD) calculation in accordance with the present invention.

Various non-limiting embodiments will be described more fully below with reference to the accompanying drawings, in which preferred embodiments are shown. However, these embodiments may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these examples are provided so that this disclosure will be thorough and complete, and will convey the scope of the invention to those skilled in the art. Like numbers refer to like elements throughout, and in alternative embodiments prime (') notation is used to indicate similar elements.

According to one embodiment, in addition to co-channel interference (CCI) on the mobile station (MS) side in current GSM communication systems, it also addresses the requirements of downlink advanced receiver performance (DARP) by 3rd Generation Partnership Project (3GPP). Done.

The embodiment shown in FIG. 1 provides a multi-channel pre-filter that is optimally operable to adaptively cancel interference and provide channel impulse response (CIR) estimation. The prefilter is a non-limiting example of two main components: (1) Joint Space-Time Optimum (JSTOF) based on multiple-input-multiple-output (MIMO). filter); And (2) a multi-input-single-output (MISO) based multi-channel matched filter. In a typical mobile station using a single available antenna, the virtual antenna array may be configured internally by a combination of oversampling and real and imaginary separation receiving the samples.

In one non-limiting embodiment, the signal from the virtual antenna array is fed to the JSTOF, where the optimal weight for the MIMO based interference cancellation filter is estimated. At the same time, the multichannel CIR for the desired signal is jointly estimated. The output of the JSTOF allows the interference to be filtered and fed into the MISO-based multichannel matched filter. The filter response of the matched filter is provided by the CIR estimate from JSTOF.

The output of the multichannel matched filter is passed to a Viterbi equalizer that eliminates intersymbol interference (ISI) and provides soft decision for further processing. The single channel response required by the equalizer can be formed by a combination of CIRs convolved from JSTOF. The prefilter can also automatically switch to a conventional or standard filter in a conventional receiver if any AWGN prevails, and back to a JSTOF based receiver if any interference prevails. This automatic switching capability reduces losses when AWGN dominates.

Examples of prefilters or interference cancellation filters for JSTOF based and DARP capable receivers are shown in FIG. 1 to 10, where the oversampling ratio is 2 and the number of virtual antennas is also indicated as X ₁ (k) to X ₄ (k). As one is 4 ( M = 4). Throughout this specification, the prefilter 10 may be referred to as an interference cancellation filter or a JSTOF filter and acts as a prefilter in a DARP compatible receiver. The receiver including this filter 10 may be described as a JSTOF receiver as shown by the dashed line in FIG. 1.

1 illustrates examples of various circuit blocks used in filter 10. The input signal is received by a derotation circuit 12. The reverse output signal is separated and passed to filter 14 of a conventional receiver including a 2: 1 switch 16, which is matched to a filter 18 that receives a shortened training sequence (TS) symbol. And an output to the cross-correlation circuit 20. The 2: 1 switch 16 is operable to switch between filter 14 and JSTOF based and DARP capable prefilter 10.

The other part of the output signal from the reverse rotation circuit 12 is separated into even and odd samples as part of the virtual antenna 24, and then separated into real and imaginary signals, a JSTOF circuit, also called a multichannel space-time filter circuit. Each X ₁ (k) to X ₄ (k) input signal to 30 is formed. The output signal from the JSTOF circuit is passed to the multichannel matched filter circuit 32, which output signal is passed to the rescaling circuit 34 and then to the multiplier circuit 36 as data d ₁ . do. Multiplier circuit 36 also receives a channel c ₁ response. When conventional filter 14 is connected, multiplier 36 receives data d ₂ and channel c ₂ responses from matched filter circuit 18 and cross-correlation circuit 20. The signal is passed to the Viterbi equalizer 38 as a soft decision output.

Figure 2 illustrates the JSTOF and multichannel matched filter in more detail, where the number of time delayed samples used in the JSTOF circuit is 2 ( N = 2). Various inputs X ₁ (k) to X ₄ (k) are received with JSTOF, which is shown in more detail. JSTOF circuit 30 includes a channelized multiplier, denoted mixer 40, 42, delay unit 44, and summer 46, which is a multichannel matched filter 48 for each of the four illustrated channels. ), And the signal from the matched filter is passed to summer 50. The joint optimal filter weight and channel estimator circuit 52 receives the TS symbol and timing uncertainty signal to generate the weight W _OPT used for the mixers 40 and 42.

Therefore, it is possible to integrate the prefilter function in a conventional GSM receiver by adding a prefilter branch in parallel with the conventional matched filter as shown in FIG. Conventional software / hardware Viterbi equalizer 38 can be used without modification. As one non-limiting example, the DARP case was tested by simulating an integrated DARP capable receiver, where the receiver provided a 1.6 dB to 6.9 dB margin over one of the performance specified for frame error rate (FER) for one of the AMR speech channels. Showed that.

2A is a flow diagram illustrating a high level method associated with the system described, showing various steps as a non-limiting example. These various steps begin with a 100 series of reference signs. The incoming communication signal is reversed (block 100) and passed to the virtual antenna. The communication signal is separated into even and odd samples (block 102), and then each even and odd samples are separated into real and imaginary signals (block 104). The communication signal from the virtual antenna is passed to the JSTOF circuit, where the communication signal is both multiplied and delayed (block 106) and then summed (block 108) as part of the first MIMO joint space-time optimal filter (JSTOF). After summation, the summed signal is passed to a multichannel MISO matched filter circuit (block 110), then summed (block 112), and passed as a single output signal to a Viterbi equalizer (block 114), where a soft decision is made. (Block 116).

In operation, the reverse circuit 12 is operable using a GMSK modulated signal and a frequency offset that is part of its signaling protocol. Before any reverse rotation the signal constellation is dynamic, after the reverse rotation the signal arrangement is static, i.e. any signal is typically concentrated at 0 ° and 180 ° so that the symbol is rotated to these two points. Thus, the GMSK signal can be treated as a conventional binary phase shift keying (BPSK) signal. Reverse rotation at the front end is used for even and odd sampling, and even and odd sampling is useful due to the oversampling ratio. For example, in a conventional receiver this is typically a ratio of 1, ie one sample per symbol.

The virtual antenna 24 can increase the sampling ratio by two samples per symbol coming from the baseband filter in a continuous fashion, forming two separate channels, even and odd. Prior to this process, odd / even samples were successively interleaved. These signals are then further separated into real and imaginary signal parts to form four independent channels of the virtual antenna. As will be appreciated by those skilled in the art, it should be noted that in some embodiments different numbers (eg, one or more) of virtual antennas / channels may be used.

As best seen in Figure 2, these signals are then passed to multipliers 40, 42 and unit delay circuit 44 (e.g., one symbol delay) so that the signals are multiplied and delayed to be processed, The multiplication operation follows, as can be clearly seen by the two multipliers 40 and 42 and one delay circuit 44. Following this operation, summing is done in summer 46 as shown. This part of the system is operable as a multichannel two dimensional filter. One dimension occurs due to the delay over time and the other dimension is introduced from the virtual antenna, ie the spatial dimension is introduced as described, so that the two dimensions form a space-time filter.

It is clear that each incoming signal is used in relation to the other channel and the multiplier receives the weights from the joint optimal filter weight and the channel estimator 52. The joint optimal filter weights and the weights coming from the channel estimator 52 are passed to a multiplier.

The weight is also an 8 × 4 matrix, ie 32 weights in one non-limiting example. For joint optimal filter weights and training sequence symbol inputs from channel estimator 52, there are typically about 26 known symbols in some non-limiting examples and it is known which training sequence the packet contains. As a non-limiting example, +/- 3 or seven location searches may be used to find timing. The impulse response h _opt of the multichannel matched filter can be used so that the system matches the channel response and forms a stronger signal after the matched filter.

As shown in FIG. 1, rescaling is not necessarily required but may occur as hardware or software convenience. This rescaling circuit 34 enables better operation for a 4 bit or 5 bit input as a non-limiting example for the Viterbi equalizer 38. The dynamic range of the signal can be readjusted so that the signal can be sent to a 4 bit or 5 bit circuit.

As previously mentioned, multiplier 36 may take signals d ₂ and c ₂ for data and channel response from conventional filter receiver 14 or signals for data and channel response from JSTOF receiver 10. d ₁ and c ₁ can be taken, and switching between them is possible. The JSTOF receiver will introduce some loss if there is no interference (i.e. there is only pure white noise). In this case, a conventional receiver 14 can be used and it will work properly. Therefore, the circuit can switch back to the conventional filter without introducing loss by the JSTOF receiver and its circuit. This conversion is based on the estimation of SINR _OUT minus SINR _INP . If the amount is less than the threshold, the system determines that there is little interference and does not require interference cancellation of the JSTOF receiver. Thus, the filter of the conventional receiver 14 is used by switching the 2: 1 switch 16.

The circuitry is operable in beam forming systems and other systems. This type of system can also improve the signal-to-noise ratio and improve the bit error rate (BER). This may also affect the higher level protocols used with these circuits, as well as phone calls and other communications issues.

In one embodiment the multi-channel structure of the JSTOF-based filter 10 is used, and the MIMO-based JSTOF circuit 30 provides space-time filter weights and channel estimation, different from prior art solutions. This circuit provides the ability to efficiently remove interference and yield high performance for both synchronous and asynchronous interference. In some simulations, it has been shown that some prior art solutions do not provide the required performance for the DARP case.

The MISO-based multichannel matched filter circuit 32 is characterized by reducing the complexity of the equalizer by improving the overall error rate performance and avoiding the multichannel Viterbi equalizer. Built-in automatic switching between JSTOF-based receivers and conventional receivers reduces losses in the AWGN case.

Suitable receiver structures can be used to meet the DARP requirements. An interference cancellation match filter (ICMP) may use the example of virtual antenna and beamforming as described to eliminate interference. The circuit detects an estimation error of the channel impulse response (CIR) of the desired signal. Joint demodulation (JD) has shown good performance for various cases. In addition to the difficulty of removing the asynchronous interferer, there may be excessive computational complexity involved in finding the CIR of the interferer.

In one embodiment, the virtual antenna 24 is operable with adaptive space-time filtering, such that a joint space-time optimal filter (JSTOF) circuit 30 may be used. One difference from the ICMF is that the space-time filter weights used to suppress the interference and the CIR estimation of the desired signal are jointly estimated and optimized in the JSTOF, but these are estimated separately in the ICMF. The JSTOF circuit 30 may be a MIMO circuit that takes advantage of the rank deficiency feature of the desired CIR matrix depending on the space time setting. The simulations showed satisfactory performance for various DARP cases. If fixed-point Cholesky factorization and EVD / SVD are feasible, the computational burden is considered acceptable.

This method has some simplicity and little computational complexity. The system is also robust because it makes very few assumptions about the source of interference. Also, because the solution is integrated as a preprocessing step for input data, the system can continue to use existing equalizer structures. This allows the system to use the HW equalizer accelerator if available.

To support the evaluation of this technique, a system level block error rate (BLER) simulator has been extended to support all interferor models / scenarios being used by the 3GPP DARP specification.

This is followed by a description of the simulation performance for the DARP case using the JSTOF circuit. It is to be understood that space-time processing and channel estimation for joint interference reduction are used in base stations where an array of M antennas is available. Assuming that the equivalent channel response for a single desired user can be modeled as an L-tap finite impulse response (FIR) filter, a snapshot sample of the received baseband signal can be expressed as follows.

에 축적함으로써 [수학식 1]에 대한 시공간 확장을 다음과 같이 얻을 수 있다. Where x (k) is an M x 1 vector representing the output from the antenna, H is an M x L matrix containing the channel response for the antenna array, and s (k) is L x for the corresponding symbol transmitted. 1 vector, and v (k) is an M × 1 vector containing AWGN and interference. Longer MN x 1 vector with N time-delayed versions of x (k)

By accumulating in Equation 1, the space-time extension to Equation 1 can be obtained as follows.

인 MN x (L + N - 1) 행렬은 H의 블록 토플리츠 행렬 형태이고,

이다. 트레이닝 시퀀스에 대응하는 샘플은 다음과 같이 수집될 수 있다. here,

MN x (L + N-1) matrix is in the form of a block topple matrix of H,

to be. Samples corresponding to the training sequence may be collected as follows.

는 MN x p 행렬이고,

는 트레이닝 심볼의 (L + N - 1) x p 컨벌루션(convolution) 행렬이다. 공동 최적화는 필터의 출력 간섭 잔류가 최소화되도록, 시공간 필터에 대한 비자명(non-trivial) MN x 1 가중치 벡터 w 및 필터 후의 비자명 (L + N - 1) x 1 채널 추정 벡터 h를 찾기 위한 것으로, 즉 다음의 최적화 문제를 푸는 것이다. Where p = P-L-N + 2, P is the number of symbols in the training sequence,

Is the MN xp matrix,

Is the (L + N-1) xp convolution matrix of training symbols. The joint optimization is to find a non-trivial MN x 1 weight vector w for the space-time filter and an unnamed (L + N-1) x 1 channel estimate vector h for the space-time filter so that the output interference residuals of the filter are minimized. That is, to solve the following optimization problem.

The optimal weight can be found as follows:

의 최소 고유값(eigenvalue)에 대응하는 고유벡터(eigenvector)이다. Optimal Channel Estimation h _opt is a Matrix

Is the eigenvector corresponding to the minimum eigenvalue.

, (MN x MN)

, (MN x MN)

, ((L+N-1) x (L+N-1))

, ((L + N-1) x (L + N-1))

, ((MN) x (L+N-1))

, ((MN) x (L + N-1))

가 더 이상 백색은 아니지만 미지의 공분산(covariance) 행렬

에 의해 근사하게 가우시안 분포된다고 하면, 채널에 대한 최적 추정

은 최대 우도(ML) 추정이며, 이는 다음 양의 최소한이다. Noise Plus Interference Components in the Space-Time Model of Equation (3)

Is no longer white, but an unknown covariance matrix

Given an approximation of Gaussian distribution by

Is the maximum likelihood (ML) estimate, which is the minimum of:

는 보통 랭크 부족(rank deficient)이고, 즉

이다. 랭크 부족 ML 문제는 시공간 필터의 랭크-1 근사화에 사용될 수 있다. In this non-limiting space-time model, the number of independent channels is always less than or equal to M,

Is usually rank deficient, i.e.

to be. The rank deficient ML problem can be used to rank-1 approximation of the space-time filter.

의 ML 추정을 찾는 것이 가능하다. 추정은 다음과 같이 분해될 수 있다. In one embodiment, the JSTOF circuit may use another approach to find a joint optimal solution for filter weights and channel estimation.

It is possible to find the ML estimate of. Estimation can be resolved as follows.

는

의 공간 행렬의 추정이고,

는

의 시간 행렬의 추정이다. 그것들은 다음에 의해 얻을 수 있다.here,

Is

Is an estimate of the spatial matrix of

Is

Is an estimate of the time matrix. They can be obtained by

는 촐레스키 인수분해이고,

은 행렬 D의 상위 M개의 고유값에 대응하는 M 고유벡터로 구성되고, D는 다음과 같다. here,

Is the Cholesky factorization,

Is composed of M eigenvectors corresponding to the upper M eigenvalues of matrix D, where D is as follows.

In the next step, the optimal weight for the space-time filter can be obtained by

, (MN x M)

, (MN x M)

The optimal channel estimation is as follows.

, (M x (L+N-1))

, (M x (L + N-1))

The optimal space-time filter of Equation 14 can then be applied to samples from the antenna array 24. Certainly the output of filter 30 still has the M channel, which is a MIMO system. The optimal channel estimate of Equation 15 may be used in the multichannel matched filter 32. The output of the matched filter is then combined (summed) and rescaled to the level desired to be changed in the rescaling circuit 34. The final output is a single channel sample stream and can be fed to a Viterbi equalizer 38. Note that the number of channel taps after JSTOF has been changed to L + N-1 compared to L of modeled channel taps before JSTOF.

It has been observed by simulation that the JSTOF receiver results in more than 1 dB loss in pure AWGN cases compared to conventional receivers using conventional filters. To reduce this loss, an automatic switching scheme between JSTOF and conventional receivers has been developed. This conversion is based on the measurement of the difference between the input and output SINRs of the JSTOF. If the difference is lower than a predefined threshold, the JSTOF receiver is off and the conventional receiver is on.

의 추정이 행해지면 입력 SINR은 다음과 같이 용이하게 계산될 수 있고, In Equation 10 above

With the estimation of, the input SINR can be easily calculated as

The output SINR can be calculated from [Equation 14] and [Equation 15].

For the mobile side, the virtual antenna array can be set by a combination of oversampling and real and imaginary separation as shown in FIG.

According to various embodiments, the joint optimal MIMO space-time filter and channel estimation described in Equations 14 and 15 increase the interference suppression performance. The MISO multichannel matched filter 32 based on channel estimation of Equation 15 improves the error rate performance while reducing the complexity of the Viterbi equalizer 38. The automatic switching between JSTOF and conventional receivers reduces the loss in pure AWGN cases.

의 역이 계산되는 방식과 채널

이 감소된 랭크를 이용하여 추정되는 방식이다. JSTOF defined by Equations 6 to 17 can be implemented in various ways with respect to numerical stability and computational complexity. The main difference is the autocorrelation matrix

How and inverse of are calculated

This is estimated using the reduced rank.

의 촐레스키 분해 기반의 행렬 역 및 [수학식 13]에서 행렬

의 고유값 분해이다. 구체적으로,

는 대칭 양정치(symmetric positive definite)이므로, 촐레스키 분해는 다음과 같이 나타난다:One such embodiment is

Matrix inverse based on Cholesky decomposition of and matrix in [Equation 13]

Is the eigenvalue decomposition of. Specifically,

Since is a symmetric positive definite, Cholesky decomposition looks like this:

D can be rewritten as:

은 다음과 같다.here,

Is as follows.

의 제곱근을 이용하여 수행되고, 역치환(back-substitution)에 의해 역의 명시적인 계산을 피할 수 있음을 주목하여야 한다. 또한,

는 그 상호 소거의 구조로 인해 수치적으로 안정하다. 이는

의 조건수가 300보다 큰 경우가 거의 없다는 것을 보여준 시뮬레이션에 의해 증명되었다. 당해 기술 분야에 숙련된 자들이라면 인식할 수 있듯이, 이는

에 대한 고유값 분해가 통상적인 적용에 대해 과도하게 복잡한 알고리즘을 요구하지 않는다는 것을 내포한다. 사실, 이 접근법은 잠재적으로 여기서 개괄된 접근법들 중 가장 적은 계산 복잡도를 갖는 것일 수 있다. The station actually

It should be noted that it is performed using the square root of, and that explicit calculation of the inverse can be avoided by back-substitution. Also,

Is numerically stable due to the structure of mutual cancellation. this is

This is demonstrated by a simulation that shows that the number of conditions of is rarely greater than 300. As will be appreciated by those skilled in the art,

The eigenvalue decomposition for implies that it does not require an overly complex algorithm for a typical application. In fact, this approach could potentially be one with the least computational complexity of the approaches outlined here.

에 대한 촐레스키 분해인데, 이는 그 조건수가 잠재적으로 상대적으로 높을 수 있고, 그 양정치 특성이 반올림 오 차(round-off error)에 의해 어느 정도 오프셋될 수 있기 때문이다. 그러나 시뮬레이션에서는 매우 높고 매우 낮은 캐리어 대 잡음(C/I) 비와 같은 일부 극단적인 시나리오에서조차

의 조건수가 10⁷보다 작다는 것을 보여주었다. One potential numerical concern

The Cholesky decomposition for is because its condition number can potentially be relatively high, and its stoichiometric properties can be offset to some extent by round-off errors. However, in simulations, even in some extreme scenarios such as very high and very low carrier-to-noise (C / I) ratios.

Showed that the condition number of was less than 10 ⁷ .

의 역의 직접 계산을 피하기 위해 샘플 도메인에서 QR 분해가 사용될 수 있다. [수학식 3]에서

는 전 컬럼 랭크(full column rank)를 갖기 때문에, 고유의 QR 분해를 갖는다. According to an alternative embodiment,

QR decomposition can be used in the sample domain to avoid direct computation of In [Equation 3]

Since has a full column rank, it has its own QR decomposition.

는 직교 컬럼을 갖는 p x MN 행렬이고,

은 전 랭크 MN x MN 상위의 삼각 행렬이다. 이는 다음과 같이 볼 수 있고, here,

Is a px MN matrix with orthogonal columns,

Silver Rank MN x MN The upper triangular matrix. This can be seen as

을 이용하여 [수학식 19]의 형태로 쓸 수 있다. D in Equation 13 is redefined as

Can be written in the form of [Equation 19].

에 대한 고유값 분해를 이용하여 감소된 랭크 채널 추정이 수행될 수 있고, [수학식 14]의 최적 필터 가중치 행렬이 다음과 같이 감소될 수 있다.As in the previous approach

Reduced rank channel estimation can be performed using the eigenvalue decomposition for < RTI ID = 0.0 >,< / RTI >

라는 것을 보일 수 있기 때문에 기본적으로 샘플 도메인에서의 촐레스키 분해의 동등 버전이다. 이는 QR 분해의 더 큰 복잡도(소정 크기의 행렬에 대하여 대략 두 배만큼의 많은 동작을 요구함) 및 더 큰 샘플 행렬(M=4, N=2 및 L=5인 예시적인 경우에 대략 3배만큼의 많은 로우(row)를 가짐)의 부담에서 개선된 수치적 안정성을 갖는다. This approach

It is basically an equivalent version of Cholesky decomposition in the sample domain, as it can be seen. This is about 3 times greater complexity of QR decomposition (requires approximately twice as much operation for a certain sized matrix) and a larger sample matrix (M = 4, N = 2 and L = 5 in the example case). Has a number of rows) of improved numerical stability.

The two approaches described above still require calculation of the triangular matrix inverse, but this can be done by inverse substitution. Moving now to another alternative approach, the singular value decomposition (SVD) approach, matrix inversion can be avoided and in some applications, the numerical stability can be further improved. This approach starts with SVD for the sample matrix in Equation 3.

는 직교 컬럼을 갖는 p x MN 행렬이고,

는 MN x MN 직교 행렬이고,

는 그 대각선에 특이값을 갖는 MN x MN 대각 행렬,

이다. 이는 다음으로 볼 수 있다. here

Is a px MN matrix with orthogonal columns,

Is an MN x MN orthogonal matrix,

Is an MN x MN diagonal matrix with singular values on its diagonal,

to be. This can be seen as follows.

는 다음에 의해 정의되는

을 가지며 여전히 [수학식 19]의 형태를 갖는다. Of Equation 13

Is defined by

And still have the form of Equation 19.

에 대한 SVD에 의해 채널 추정을 얻을 수 있고, 필터 가중치 행렬은 다음과 같이 쓸 수 있다.

The channel estimate can be obtained by SVD for, and the filter weight matrix can be written as

은

의 상위 M 개의 우측 특이 벡터를 포함한다. 이 접근법에서의 SVD는 이전의 두 개의 접근법에서 사용된 촐레스키 및 QR 분해보다 더 많은 계산을 요구할 수 있다. here,

silver

Contains the top M right singular vectors of. SVD in this approach may require more computation than Cholesky and QR decomposition used in the previous two approaches.

As a comparison of the three approaches outlined here (ie, Cholesky, QR and SVD), the table of FIG. 9 lists step-by-step calculations for the example M = 4, N = 2 and L = 5. In order to find the optimal timing of the burst, JSTOF searches through a number of timing hypotheses, and the one corresponding to the minimum output residual is selected as the optimal timing. The output residual is defined by

The search process basically repeats the operations listed in the table for each hypothesis, but the input sample matrices from successive timing hypotheses are changed slightly by adding and deleting columns. Updating and downdating algorithms are potentially applicable to some of the operations, and the overall computational burden can potentially be reduced.

이 시각 k에서의 샘플 행렬을 나타낸다고 하자. 이는 [수학식 3]으로부터 다음과 같이 분배될 수 있다.

Suppose we represent the sample matrix at this time k . This can be distributed from Equation 3 as follows.

는 다음과 같다.here,

Is as follows.

The sample matrix at time k + 1 can be expressed as follows.

At time k + 1 the autocorrelation matrix is of the form

This is a combination of rank-1 downdate and rank-1 update. One hyperbolic rotation based algorithm for updating / downdating Cholesky factorization is described in Matrix Computations, 3rd edition, 1996 by Golub et al.

Another applicable update / downdate algorithm disclosed in the text of Golub et al. Is for QR decomposition based on Givens rotation. Of course, as one of ordinary skill in the art will appreciate, the predetermined approach to be used for a particular application will depend on factors such as available processing resources, computational complexity, and the like. As will be appreciated by those skilled in the art, other approaches may also be used.

The performance of the JSTOF-based receivers was evaluated by Matlab simulation using the extended BLER simulation engine. Parameters for the JSTOF-based receiver may be set in various aspects. Examples of values are as follows:

1) An oversampling ratio (OSR) of 2 can be chosen, which maps to the number of virtual antennas (M) of 4 in this non-limiting example, and reducing the OSR to 1 in the simulation results in significant performance degradation. Shows that

2) The number N of time delayed samples may be selected as two. However, increasing the number does not always improve performance.

3) The reduced rank for the channel response matrix may be selected as M. Increasing or decreasing the rank does not necessarily improve performance.

4) The auto-switch threshold may be 4.75 dB.

5) The soft decision output can be quantized to 5 bits wide. By increasing the width to 8 bits, the performance can be easily improved for the DTS-5. Soft decision correction may be possible.

An AMR speech channel, TCH-AFS12.2, can be used to evaluate the performance of JSTOF with respect to FER. The propagation condition TU50km / h-1950MHz can be assumed throughout the simulation. The simulation ran 1000 trials (blocks) for each case.

The FER of the receiver against the carrier to interference (C / I) ratio is shown in the graph of FIG. 3. Margins for the specified baseline performance are listed in the table below.

Receiver performance under pure AWGN and DTS-5 cases with and without automatic switching is shown in the graphs of FIGS. 4 and 5, respectively. This approach reduced the loss by about 1dB in AWGN (FER = 10%) and caused little loss for the DTS-5.

The JSTOF receiver may include a number of Viterbi equalizers and then may include a multichannel matched filter that combines the soft decisions after the equalizer. The results are shown in the graph of FIG. 6 and compared with the original case.

Performance can be evaluated using a modified case of the DTS-5R, in which the delay of an asynchronous interferer can be configured. Performance at burst lengths of 1, 1/4, 1/2 and 3/4 is shown in the graph of FIG. As a result, the performance of the JSTOF receiver is "slowly" degraded by the severe delay of the interferer.

The receiver described above can be advantageously used for cellular base stations as well as for example mobile wireless devices (e. G. Cellular devices). An example of a mobile wireless communication device 1000 that can be used is further described in the example below with reference to FIG. 8. Device 1000 is shown to include a housing 1200, keypad 1400, and output device 1600. The output device shown is a display 1600, which is preferably an all-graphics LCD. Other types of output devices may alternatively be used. The processing device 1800 is contained within the housing 1200 and is coupled between the keypad 1400 and the display 1600. The processing device 1800 controls the overall operation of the mobile device 1000 as well as the operation of the display 1600 in response to the action of a key on the keypad 1400 by the user.

The housing 1200 may extend vertically or may take other sizes and shapes (including a clamshell housing structure). The keypad may include a mode selection key or other hardware or software for switching between text input and telephone input.

In addition to the processing device 1800, other portions of the mobile device 1000 are shown schematically in FIG. 8. This includes communication subsystem 1001, short-range communication subsystem 1020, keypad 1400, and display 1600 and includes memory devices 1160, as well as other input / output devices 1060, 1080, 1100, and 1120. 1180 and various other device subsystems 1201 together. Mobile device 1000 is preferably a two-way RF communication device having voice and data communication capabilities. Mobile device 1000 also preferably has the ability to communicate with other computer systems via the Internet.

Operating system software executed by processing device 1800 may preferably be stored in permanent storage, such as flash memory 1160, but may be stored in other types of memory devices, such as read-only memory (ROM) or similar storage elements. May be stored. In addition, system software, specific device applications, or portions thereof may be temporarily loaded into volatile storage, such as random access memory (RAM) 1180. Communication signals received by the mobile device may also be stored in the RAM 1180.

In addition to its operating system functionality, processing device 1800 enables execution of software applications 1300a-1300n on device 1000. A predetermined set of applications that control basic device operations, such as data and voice communications 1300a and 1300b, may be installed on device 1000 during manufacturing. In addition, a personal information manager (PIM) application can be installed during manufacturing. The PIM can preferably organize and manage data items such as emails, calendar events, voice mails, schedules, and work items. The PIM application may also send and receive data items, preferably via the wireless network 1401. Preferably, the PIM data items are integrated, synchronized, and updated seamlessly through the wireless network 1401 with corresponding data items stored in or associated with the host computer system.

Communication functions, including data and voice communications, are performed through the communication subsystem 1001 and possibly through the short-range communication subsystem. The communication subsystem 1001 includes a receiver 1500, a transmitter 1520, and one or more antennas 1540 and 1560. In addition, communication subsystem 1001 also includes a processing module, such as a digital signal processor (DSP) 1580, and a local oscillator (LO) 1601. The specific design and implementation of the communication subsystem 1001 depends on the communication network in which the mobile device 1000 intends to operate. For example, mobile device 1000 is designed to operate in accordance with Mobitex ^™ , Data TAC ^™ or General Packet Radio Service (GPRS) mobile data communications networks, and may also be used in AMPS, TDMA, CDMA, WCDMA, PCS, GSM, EDGE, and the like. And communication subsystem 1001 designed to operate in accordance with any of a variety of such voice communication networks. Other types of data and voice networks may also be used separately and integrated for mobile device 1000. Mobile device 1000 may also be compatible according to other communication standards such as 3GSM, 3GPP, UMTS, and the like.

Network access requirements vary depending on the type of communication system. For example, in Mobitex and DataTAC networks, mobile devices are registered with the network using a unique personal identification number or PIN associated with each device. However, in GPRS networks, network access is associated with the subscriber or the user of the device. Thus, GPRS devices require a subscriber identity module, commonly called a SIM card, to operate on a GPRS network.

If the required network registration or activation procedure is completed, the mobile device 1000 may transmit and receive a communication signal through the communication network 1401. The signal received by the antenna 1540 from the communication network 1401 is routed to the receiver 1500, which provides signal amplification, frequency down conversion, filtering, channel selection, and the like. Can provide. Analog-to-digital conversion of the received signal allows the DSP 1580 to perform more complex communication functions such as demodulation and decoding. In a similar manner, the signal to be transmitted to the network 1401 is processed (e.g., modulated and encoded) by the DSP 1580, followed by digital to analog conversion, frequency up conversion, filtering, amplification, and antenna ( Provided to transmitter 1520 for transmission to communication network 1401 (or networks) via 1560.

In addition to processing the communication signals, the DSP 1580 provides control of the receiver 1500 and transmitter 1520. For example, the gain applied to the communication signal at the receiver 1500 and the transmitter 1520 may be adaptively controlled through an automatic gain control algorithm implemented in the DSP 1580.

In the data communication mode, received signals, such as text messages or web page downloads, are processed by the communication subsystem 1001 and input to the processing device 1800. The received signal is then further processed by the processing device 1800 for output to the display 1600, or alternatively for output to some other auxiliary I / O device 1060. Device users can also configure data items such as email messages using the keypad 1400 and / or some other secondary I / O device 1060, such as a touchpad, rocker switch, thumbwheel, or some other type of input device. have. The configured data item may then be transmitted via communication network 1401 via communication subsystem 1001.

In the voice communication mode, the overall operation of the device is substantially similar to the data communication mode, except that the received signal is output to the speaker 1100 and a signal for transmission is generated by the microphone 1120. Alternative voice or audio I / O subsystems, such as voice message recording subsystems, may also be implemented on device 1000. In addition, display 1600 may also be used in a voice communication mode, for example, to display the calling party's identity, duration of a voice call, or other voice call related information.

The short range communication system enables communication between the mobile system 1000 and other nearby systems or devices that are not necessarily similar devices. For example, the short-range communication subsystem may include an infrared device and associated circuits and components, or a Bluetooth ^™ communication module, to provide communication with similarly capable systems and devices.

Those skilled in the art will be able to contemplate many variations and other embodiments of the present invention that would benefit from the teachings presented in the foregoing descriptions and the associated drawings. Accordingly, it is to be understood that the invention is not limited to the specific embodiments disclosed and that many modifications and embodiments are intended to be included within the scope of the invention.

Claims (23)

A wireless communication device, Including a wireless transmitter and a wireless receiver, The wireless receiver includes a filter for reducing co-channel interference in a communication receiver, the filter comprising: A multichannel space-time filter circuit for filtering n signal portions separated from the communication signal by jointly estimating the space-time filter weight and the multichannel impulse response (CIR) based on Cholesky and eigenvalue decompositions; And A multichannel matched filter circuit receiving a multichannel signal from the multichannel space-time filter circuit and having a filter response provided by a channel impulse response estimate from the space-time filter circuit And a wireless communication device. The method according to claim 1, The filter further comprises a virtual antenna circuit connected to the multichannel space-time filter circuit that separates the communication signal into odd and even sampled real and imaginary n signal parts. The method according to claim 1, And the multichannel space-time filter circuit comprises at least one multiplier that multiplies each space-time filter weight by each signal portion. The method according to claim 3, The at least one multiplier includes a multiplier pair connected in parallel, and the multichannel space-time filter circuit further comprises a respective delay circuit for each signal portion connected to an input of one of the multiplier pairs. device. The method according to claim 4, The communication signal comprises a plurality of symbols, and wherein the multiplier and the delay circuit each have a delay of about one symbol associated therewith. The method according to claim 3, Wherein the filter further comprises respective summer circuitry for summing output of the multiplier for each channel. The method according to claim 1, The filter further receives a joint sequence filter weight and channel estimator that receives training sequence symbols and timing uncertainty data and generates a space-time filter weight for the multichannel space-time filter circuit and a multichannel impulse response for the multichannel matched filter circuit. And a wireless communication device. The method according to claim 1, And the filter further comprises an equalizer circuit downward from the multichannel matched filter circuit. A wireless communication device, Including a wireless transmitter and a wireless receiver, The wireless receiver includes a filter system for reducing cochannel interference, wherein the filter system includes: A multichannel space-time filter circuit for filtering n signal portions separated from the communication signal by jointly estimating the space-time filter weight and the multichannel impulse response (CIR) based on Cholesky and eigenvalue decomposition, and from the multichannel space-time filter circuit A joint space-time filter comprising a multi-channel matched filter circuit for receiving a multi-channel signal and having a filter response provided by channel impulse response estimate from the space-time filter circuit; And An alternative filter that operates when the interference level is lower than a predetermined threshold, and includes a matched filter, a cross-correlation circuit, and a switching mechanism for switching the n signal parts to the matched filter and the cross-correlation circuit. And a wireless communication device. The method according to claim 9, And the filter system further comprises a virtual antenna circuit connected to the multichannel space-time filter circuit that separates the communication signal into odd and even sampled real and imaginary n signal parts. The method according to claim 9, And wherein the multi-channel space-time filter circuit comprises at least one multiplier that multiplies each space-time filter weight by each signal portion. The method according to claim 11, The at least one multiplier includes a multiplier pair connected in parallel, and the multichannel space-time filter circuit further comprises a respective delay circuit for each signal portion connected to an input of one of the multiplier pairs. device. The method according to claim 12, The communication signal comprises a plurality of symbols, and wherein the multiplier and the delay circuit each have a delay of about one symbol associated therewith. The method according to claim 11, And the filter system further includes respective summer circuitry for summing output of the multiplier for each channel. The method according to claim 9, The filter system further receives a joint sequence filter weight and channel estimator that receives training sequence symbols and timing uncertainty data and generates a space-time filter weight for the multichannel space-time filter circuit and a multichannel impulse response for the multichannel matched filter circuit. And a wireless communication device. The method according to claim 9, And the filter system further comprises an equalizer circuit downward from the multichannel matched filter circuit. A method of reducing cochannel interference in a wireless communication device, comprising: Providing a wireless receiver for a wireless communication device comprising a multichannel space-time filter circuit and a multichannel matched filter circuit; Separating the communication signal into n signal parts; Filtering the n signal parts in a multichannel space-time filter circuit and jointly estimating the space-time filter weight and the multichannel channel impulse response (CIR) based on Cholesky and eigenvalue decompositions; And Receiving a multi-channel signal from the space-time filter circuit in a multi-channel matched filter circuit having a filter response provided by the channel impulse response estimate from the space-time filter circuit. The method according to claim 17, And said separating comprises sampling the communication signal into even and odd samples, and separating the even and odd samples into real and imaginary signal parts. The method according to claim 17, Summing the outputs of the matched filters and rescaling to a desired level. The method according to claim 19, Equalizing a single channel signal after rescaling to the desired level. The method according to claim 17, Filtering the n signal parts in an alternate filter if the interference level is lower than a threshold. The method according to claim 17, And multiplying the respective signal parts based on the space-time filter weights. The method according to claim 22, And summing the signal parts for each channel after the multiplying step.

Images