WO2017093740A2

WO2017093740A2 - Apparatus, method and computer program for an interference-aware receiver

Info

Publication number: WO2017093740A2
Application number: PCT/GB2016/053782
Authority: WO
Inventors: Alexandr Kuzminskiy; Pei XIAO; Rahim Tafazolli
Original assignee: University Of Surrey
Priority date: 2015-12-02
Filing date: 2016-12-01
Publication date: 2017-06-08
Also published as: GB201521252D0; WO2017093740A3

Abstract

Apparatus, methods, and computer programs for an interference-aware receiver are disclosed. The apparatus comprises: a multi-antenna array comprising a plurality of antennas; a covariance matrix estimator configured to obtain an estimated covariance matrix based on known pilot symbols and a plurality of antenna signals received by the multi-antenna array; a covariance matrix adapter configured to determine a value of a scenario-independent metric based on the estimated covariance matrix, compare the determined value to a target value of the metric for a current configuration of the signal receiving apparatus, and determine an adapted covariance matrix based on the result of the comparison; and a signal processing unit configured to perform further signal processing using the adapted covariance matrix. In some embodiments, the metric is a likelihood ratio and the target value is the median value of a probability distribution function associated with the current receiver configuration.

Description

Apparatus, Method and Computer Program for an Interference- Aware Receiver

Technical Field

The present invention relates to interference-aware receivers in wireless

communication networks.

Background

Future communications networks are expected to become increasingly dense due to the proliferation of wireless communication devices. Consequently, networks will become interference limited as the number of interference sources increases. Interference Rejection Combining (IRC) is an example of one technology which has been developed to allow interference-aware receivers to operate effectively in an interference-limited environment. IRC can be used as an alternative to conventional Maximum Ratio Combining (MRC).

An important step in IRC is estimation of the interference plus noise covariance matrix, which is normally performed over some sets of pilot symbols. Direct application of the conventional sample matrix estimation for a limited number of samples can lead to significant performance degradation, depending on the interference scenario.

Furthermore, application of the IRC in noise-limited scenarios may lead to

performance degradation compared to conventional MRC. Known solutions to these problems include regularization (diagonal loading) of the sample covariance matrix and/or complete switching to its diagonal version (equivalent to MRC). Different switching and diagonal loading factor selection rules are defined for different interference scenarios, and stored in a Scenario Data Bank. A Scenario Analyser uses performance metrics and thresholds from the Scenario Data Bank to define a diagonal loading factor and/or switching conditions. Then, the defined diagonal loading factor is used to estimate the regularized covariance matrix.

A drawback of conventional scene dependent and empirical regularization techniques is that the diagonal loading factor selection rules are adjusted to particular scenarios, and usually are not appropriate for other scenarios. The rules for each scenario are typically obtained based on simulation results. Future network will be characterized by increasing number of possible transmission modes and receiver configurations, leading to an ever-increasing variety of scenarios. Empirical, simulation-based estimates of the interference-plus-noise covariance matrix may not be suitable for interference-aware receivers in such networks, due to the very large number of scenarios that would need to be considered when setting up the Scenario Data Bank.

The invention is made in this context. Summary of the Invention

According to a first aspect of the present invention, there is provided signal receiving apparatus for use in a wireless communication network, the apparatus comprising: a multi-antenna array comprising a plurality of antennas; a covariance matrix estimator configured to obtain an estimated covariance matrix based on known pilot symbols and a plurality of antenna signals received by the multi-antenna array; a covariance matrix adapter configured to determine a value of a scenario-independent metric based on the estimated covariance matrix, compare the determined value to a target value of the metric for a current configuration of the signal receiving apparatus, and determine an adapted covariance matrix based on the result of the comparison; and a signal processing unit configured to perform further signal processing using the adapted covariance matrix.

In some embodiments according to the first aspect, the covariance matrix adapter is configured to determine a value of the scenario-independent metric for each one of a plurality of different regularized versions of the estimated covariance matrix, and select the regularized version of the estimated covariance matrix which gives the closest value of the metric to the target value as the adapted covariance matrix.

In some embodiments according to the first aspect, the metric is a likelihood ratio and the current configuration of the signal receiving apparatus is defined by the number of antennas in the multi-antenna array and by the number of pilot symbols currently in use. The target value may comprise a statistical measure related to a probability distribution function of the likelihood ratio. In some embodiments, the statistical measure is a median value of the probability distribution function.

The covariance matrix adapter can be configured to retrieve the target value of the metric from a plurality of stored pre-calculated optimum values each associated with a different configuration of the signal receiving apparatus. In some embodiments according to the first aspect, the covariance matrix adapter is configured to determine a value of the scenario-independent metric for the estimated covariance matrix, wherein in response to the value of the metric for the estimated covariance matrix being greater than a threshold, the covariance matrix adapter is configured to set a diagonal loading factor of the adapted covariance matrix equal to one, and wherein in response to the value of the metric for the estimated covariance matrix being greater than the threshold, the covariance matrix adapter is configured to proceed to determine the value of the metric for each one of a plurality of different regularized versions of the estimated covariance matrix, and select the adapted covariance matrix according to the determined metrics. When the diagonal loading factor is equal to one, the adapted covariance matrix is a pure diagonal matrix.

The signal processing unit can be configured to perform further signal processing by selectively performing Interference Rejection Combining or Maximum Ratio

Combining according to a value of the adapted covariance matrix, and/or can be further configured to perform maximum likelihood sequence detection and/or adaptive coding modulation using the adapted covariance matrix. In other embodiments according to the first aspect, the signal receiving apparatus can further comprise any of the apparatus features and components which are described herein and/or illustrated in the accompanying drawings.

According to a second aspect of the present invention, there is provided a method comprising: obtaining an estimated covariance matrix based on known pilot symbols and a plurality of antenna signals received by a multi-antenna array; determining a value of a scenario-independent metric based on the estimated covariance matrix; comparing the determined value to a target value of the metric for a current

configuration of the signal receiving apparatus; determining an adapted covariance matrix based on the result of the comparison; and performing further signal processing using the adapted covariance matrix.

In some embodiments according to the second aspect, the method further comprises determining a value of the scenario-independent metric for each one of a plurality of different regularized versions of the estimated covariance matrix, wherein the regularized version of the estimated covariance matrix which gives the closest value of the metric to the target value is selected as the adapted covariance matrix.

In some embodiments according to the second aspect, the metric is a likelihood ratio, and the current configuration of the signal receiving apparatus is defined by the number of antennas in the multi-antenna array and by the number of pilot symbols currently in use. The target value may comprise a statistical measure related to a probability distribution function of the likelihood ratio. In some embodiments, the statistical measure is a median value of the probability distribution function.

In other embodiments according to the second aspect, the method can further comprise any of the method steps which are described herein and/ or illustrated in the accompanying drawings. According to a third aspect of the present invention, there is provided a non-transitory computer-readable storage medium arranged to store computer program instructions which, when executed, perform a method according to the second aspect.

Brief Description of the Drawings

Embodiments of the present invention will now be described, by way of example only, with reference to the accompanying drawings, in which:

Figure ι schematically illustrates a signal receiving apparatus for use in a wireless communication network, according to an embodiment of the present invention;

Figure 2 schematically illustrates an example of a structure of the covariance matrix adapter, according to an embodiment of the present invention;

Figure 3 is a flowchart showing a method of determining an adapted covariance matrix, according to an embodiment of the present invention;

Figure 4 is a flowchart showing an alternative method of determining an adapted covariance matrix, according to an embodiment of the present invention;

Figure 5 is a graph plotting the probability distribution functions of the likelihood ratio for different receiver configurations, according to an embodiment of the present invention;

Figure 6 illustrates a series of graphs plotting the bit error rate (BER) as a function of signal to interference ratio (SIR) for different interference scenarios and different receiver configurations, according to embodiments of the present invention; and Figure 7 illustrates a series of graphs plotting the BER as a function of signal to noise ratio (SNR) for different interference scenarios and different receiver configurations, according to embodiments of the present invention. Detailed Description

Figure 1 schematically illustrates a signal receiving apparatus for use in a wireless communication network, according to an embodiment of the present invention. The apparatus can be included in any suitable device, for example, a network Access Point (AP) or User Equipment (UE).

The apparatus comprises a multi-antenna array 101 comprising a plurality of antennas yi to and further comprises a channel estimator 102 and covariance matrix estimator 103 configured to receive antenna signals from the multi-antenna array 101. In addition, both the channel estimator 102 and covariance matrix estimator 103 have access to information regarding pilot signals that are used during channel estimation. Specifically, the as-transmitted pilot symbols are known to both channel estimator 102 and covariance matrix estimator 103. The channel estimator 102 is configured to perform channel estimation to obtain an estimate of the wireless channel, h, through which the multi-antenna array 101 receives the pilot signals. The channel estimator 102 may perform channel estimation in a conventional manner.

The covariance matrix estimator 103 is configured to obtain an estimated covariance matrix based on the known pilot symbols and the antenna signals received by the multi-antenna array 101. The covariance matrix estimator 103 can calculate the estimated covariance matrix over a set of pilots for a fixed interference scenario. In an LTE downlink example this can be a single resource block (RB) containing 16 cell- specific reference signals (CRSs) for 2 transmit antenna modes, with 12 of the CRSs overlapping with the data symbols of the desired signal in the synchronous LTE network. Then, the estimated interference plus noise sample covariance matrix, R , for K receive antenna elements can be calculated as follows:

R = L-'∑Z _/ Z ; ,

x = s + z , where x, h, and z are the (K ^χ ι) vectors of the received signal, propagation channel and interference plus noise signal, s is the desired signal, R is the (K x K) actual

interference plus noise covariance matrix, L is the number of CRS pilots pi, h/ is the channel estimate, and (.)* denotes the complex conjugate transpose operation.

The apparatus further comprises a covariance matrix adapter no for adapting the estimated interference-plus-noise covariance matrix to the current receiver

configuration. The covariance matrix adapter no is configured to receive the estimated covariance matrix and determine a value of a scenario-independent metric based on the estimated covariance matrix. In the present embodiment the metric is the Likelihood Ratio (LR), which is a function of the number of antenna elements, K, and the number of pilot signals, L, in the current receiver configuration, and is independent of the current interference scenario. The LR is defined as follows:

LR(R-R) ₌ ^det(^R| '†)»P'f > ,

exp trfR-'RjJ where det(A) and tr(A) are the determinant and trace of matrix A, respectively.

However, in other embodiments a different form of metric may be used.

The covariance matrix adapter 110 is further configured to compare the determined value to a target value of the metric for a current configuration of the signal receiving apparatus, and determine an adapted interference-plus-noise covariance matrix based on the result of the comparison.

In the present embodiment the target value is a median value of the probability distribution function (p.d.f.) of the LR for the current receiver configuration. The optimum diagonal loading factor for the current receiver configuration is found when the LR is equal to the median p.d.f. value, and hence in the present embodiment the target value can also be referred to as the optimum value. However, the use of the median value is not essential and in other embodiments a different value may be chosen as the target value. The adapted covariance matrix, R , is obtained by selecting a suitable diagonal loading factor, δ, for regularizing the estimated covariance matrix obtained by the covariance matrix estimator 103. In some embodiments, a diagonal loading factor of zero or one may be selected. The adapted covariance matrix may also be referred to as a regularized interference-plus-noise covariance matrix, and can be defined as:

R = (l - J)R + <SD , where R is the estimated covariance matrix and D is a diagonal matrix, for example, the diagonal sub-matrix of R , diag(R). For example, in some embodiments R is estimated over one resource block and D is averaged across the whole frequency band. The boundary cases of δ = 1 and δ = o correspond respectively to MRC (δ = ι), and to IRC without regularization (δ = o).

In embodiments of the invention, various different methods can be used to determine the adapted covariance matrix. For example, in some embodiments the adapted covariance matrix is obtained by selecting a regularized version of the covariance matrix that gives a value of the metric close to the target value. In other embodiments the adapted covariance matrix can be obtained differently, for example, by selecting a predefined diagonal loading factor when the calculated value of the metric is above or below a certain threshold.

A scenario-independent metric is one for which the optimum value depends only on the receiver configuration, and not on the current interference scenario. For example, the receiver configuration can be defined in terms of the number of antenna elements and the number of pilot signals, as described above. In other embodiments, different parameters may be used to define the receiver configuration, according to the particular metric used. Since a scenario-independent metric is used in embodiments of the present invention, it is not necessary to store pre-calculated rules for different scenarios, and the Scenario Data Bank that is conventionally used in interference-aware receivers can be omitted. Instead, in the present embodiment it is only necessary to store a target value of the metric for each receiver configuration. The number of possible receiver configurations will generally be significantly lower than the number of possible interference scenarios. The apparatus further comprises one or more signal processing units configured to perform further signal processing using the adapted covariance matrix, such as a weights calculator 104 and a detector 105 as shown in Fig. 1. In the present embodiment the weights calculator 104 is configured to perform weights calculation as follows:

where h_m and w_m are the (K ^χ ι) vectors of the channel estimate and antenna weights for the m* data symbol in the given Resource Block (RB), and R is the adapted covariance matrix. In some embodiments the weights calculator 104 can perform IRC or MRC depending on the form of the adapted covariance matrix selected by the covariance matrix adapter 110. Additionally, in some embodiments, further signal processing functions including maximum likelihood sequence detection and/ or adaptive coding modulation can be performed using the adapted covariance matrix.

Referring now to Fig. 2, an example of a structure of the covariance matrix adapter is illustrated, according to an embodiment of the present invention. Figure 3 is a flowchart showing a method performed by the covariance matrix adapter of Fig. 2.

In the embodiment of Fig. 2, the covariance matrix adapter comprises N regularized covariance matrix estimators 211a and N corresponding metric calculators 211b. First, in step S301 the covariance matrix adapter receives the estimated interference-plus- noise covariance matrix from the covariance matrix estimator 103. In step S302, each of the N regularized covariance matrix estimators 211a obtains a different regularized version of the estimated covariance matrix, based on a different one of a predefined set of iV diagonal loading factors. Then, in step S303, each metric calculator 211b determines a value of the scenario-independent metric for the regularized version of the estimated covariance matrix received from the corresponding regularized covariance matrix estimator 211a. In the present embodiment the scenario- independent metric is the LR, as described above. The covariance matrix adapter of the present embodiment further comprises a covariance matrix selector 212, which receives the metric values from the metric calculators 211b. The covariance matrix selector 212 is further configured to retrieve the target value of the metric from a receiver configuration data bank 220. The receiver configuration data bank 220 contains parameters of the scenario independent probability distributions of the likelihood ratios (LR) of the actual covariance matrices. The p.d.f. can be precalculated for any i^and L, for example by using simple

simulations for any R, e.g. R=I where I is the unit matrix. An example of a suitable method for calculating the p.d.f. is disclosed in "Modified GLRT and AMF framework for adaptive detectors" by Y. I. Abramovich, N. K. Spencer and A. Y. Gorokhov, IEEE Trans. AES-43, no.3, pp. 1017-1051, July 2007, as follows:

where B₀=i, v= V Mi +i), Nis the training sample size, and Mis the adaptive antenna (filter) dimension. However, this is merely one example, and other methods of calculating the p.d.f. maybe used in other embodiments.

Examples of the p.d.f. of the LR for three different receiver configurations ( =2; K=4; K=6) are illustrated in Fig. 5, based on 10⁶ trials with L=i2 (Kxi) independent Gaussian vectors with unit variance. In the present embodiment, the receiver configuration data bank 220 is configured to store statistical measures of the p.d.f. of the LR for the different receiver configurations, as shown below in Table 1. In the present

embodiment, the stored parameters include the median value, μ, and parameters βι and β₂ which define the LR distribution concentration interval, where

Prob[LR(R, R)< β ] = Prob[LR(R, R) > β₂ \ = 5%

Table 1 Next, in step S304 the covariance matrix selector 212 compares each of the values received from the metric calculators 211b to the target value for the current receiver configuration, which in the present embodiment is the median μ( ,£). Then, in step S305 the covariance matrix selector 212 selects the regularized version of the estimated covariance matrix which gives the closest value of the metric to μ(Κ,Σ) as the adapted covariance matrix. That is, the covariance matrix selector 212 sets the diagonal loading factor of the adapted covariance matrix to be equal to the diagonal loading factor used by the regularized matrix estimator 211a which produced the closest value of the metric to the target value, and outputs the adapted covariance matrix.

As an example, the covariance matrix adapter may use predefined diagonal loading factors δ = o, 0.2, 0.4, 0.6, 0.8, 1 for N = 6. The η^Λ regularized covariance matrix estimator 211a calculates the n^th regularized covariance matrix as follows: R„ = (1 - ¾ )R + ¾D , and the n^th metric calculator 211b estimates the η^Λ LR value, η_η, as follows:

7. = LR[R_B ] for n = 1, N. The covariance matrix selector 212 then selects the one of the N regularized adapted covariance matrices which gives the closest η_η to μ(Κ,Σ) as the adapted covariance matrix, i.e.: n_Q = arg min[/7„ - μ{κ, L}f

n

Referring now to Fig. 4, a flowchart showing an alternative method of determining an adapted covariance matrix is illustrated, according to an embodiment of the present invention. In the embodiment shown in Fig. 4, after receiving the estimated covariance matrix in step S401 the covariance matrix adapter is configured to determine the value of the scenario-independent metric for the diagonal sub-matrix of the estimated covariance matrix, LR[DJ, in step S402. In step S403 the covariance matrix adapter compares the determined value to the threshold βι for the current receiver

configuration. If LR[D] is greater than β! then in step S404 the covariance matrix adapter sets the diagonal loading factor δ = 1, which is equivalent to MRC with a pure diagonal matrix.

On the other hand, if LR[D] is less than or equal to βι, then the covariance matrix adapter proceeds to calculate metric values for different candidate versions of the regularized covariance matrix and determine the adapted covariance matrix in steps S405 to S408. The functions performed at steps S405, S406, S407 and S408 are similar to those described above in steps S302, S303, S304 and S305 of Fig. 3, respectively, and a detailed description will not be repeated here. However, step S407 differs from step S304 in that the second threshold, β₂, is used as the target value in the present embodiment, as opposed to the median value, μ.

The method illustrated in Fig. 4 gives priority to MRC and IRC without regularization solutions in the pure noise and interference limited scenarios, and can be advantageous in applications where demonstration of the maximum possible performance is required.

Figures 6 and 7 illustrate a series of graphs plotting the bit error rate (BER) as a function of signal to interference ratio (SIR) and signal to noise ratio (SNR) for different interference scenarios and different receiver configurations, according to embodiments of the present invention. The graphs in Figs. 6 and 7 compare the performance of the methods described above with reference to Figs. 3 and 4 to conventional MRC, IRC without regularization, and IRC with empirical regularization. In Figs. 6 and 7, curves plotted using data obtained for the method of Fig. 3 are labelled as "IRC, EL median", and curves plotted using data obtained for the method of Fig. 4 are labelled as "IRC, EL bias". The results in Figs. 6 and 7 were obtained from a simulation of a simplified synchronous LTE downlink scenario with 10MHz bandwidth, TM6 transmission mode with QPSK signal and full band allocation for the serving and interfering cells with the same SIR, VA5 propagation channels, conventional two- dimensional CRS-based channel estimation, 1 RB based IRC with L=12 CRSs per RB (only the CRSs overlapping with the data symbols are used for interference plus noise covariance matrix estimation), and i<=2,4,6 low correlation received antennas. As shown in Figs. 6 and 7, the methods described above with reference to Figs. 3 and 4 provide improved performance over the conventional MRC and IRC-based approaches, offering a reduced BER for a given SIR or SNR level. Whilst certain embodiments of the invention have been described herein with reference to the drawings, it will be understood that many variations and modifications will be possible without departing from the scope of the invention as defined in the

accompanying claims.

Claims

1. Signal receiving apparatus for use in a wireless communication network, the apparatus comprising:

a multi-antenna array comprising a plurality of antennas;

a covariance matrix estimator configured to obtain an estimated covariance matrix based on known pilot symbols and a plurality of antenna signals received by the multi-antenna array;

a covariance matrix adapter configured to determine a value of a scenario- independent metric based on the estimated covariance matrix, compare the determined value to a target value of the metric for a current configuration of the signal receiving apparatus, and determine an adapted covariance matrix based on the result of the comparison; and

a signal processing unit configured to perform further signal processing using the adapted covariance matrix.

2. The apparatus of claim 1, wherein the covariance matrix adapter is configured to determine a value of the scenario-independent metric for each one of a plurality of different regularized versions of the estimated covariance matrix, and select the regularized version of the estimated covariance matrix which gives the closest value of the metric to the target value as the adapted covariance matrix.

3. The apparatus of claim 2, wherein the metric is a likelihood ratio, and the current configuration of the signal receiving apparatus is defined by the number of antennas in the multi-antenna array and by the number of pilot symbols currently in use.

4. The apparatus of claim 1, 2 or 3, wherein the target value comprises a statistical measure related to a probability distribution function of the likelihood ratio.

5. The apparatus of claim 4, wherein the statistical measure is a median value of the probability distribution function.

6. The apparatus of any one of claims 2 to 5, wherein the covariance matrix adapter is configured to retrieve the target value of the metric from a plurality of stored pre-calculated target values each associated with a different configuration of the signal receiving apparatus.

7. The apparatus of claim 1, wherein the covariance matrix adapter is configured to determine a value of the scenario-independent metric for the estimated covariance matrix,

wherein in response to the value of the metric for the estimated covariance matrix being greater than a threshold, the covariance matrix adapter is configured to set a diagonal loading factor of the adapted covariance matrix equal to one, and

wherein in response to the value of the metric for the estimated covariance matrix being greater than the threshold, the covariance matrix adapter is configured to proceed to determine the value of the metric for each one of a plurality of different regularized versions of the estimated covariance matrix, and select the adapted covariance matrix according to the determined metrics.

8. The apparatus of any one of the preceding claims, wherein the signal processing unit is configured to perform further signal processing by selectively performing Interference Rejection Combining or Maximum Ratio Combining according to a value of the adapted covariance matrix.

9. The apparatus of any one of the preceding claims, wherein the signal processing unit is further configured to perform maximum likelihood sequence detection and/ or adaptive coding modulation using the adapted covariance matrix.

10. A method comprising:

obtaining an estimated covariance matrix based on known pilot symbols and a plurality of antenna signals received by a multi-antenna array;

determining a value of a scenario-independent metric based on the estimated covariance matrix;

comparing the determined value to a target value of the metric for a current configuration of the signal receiving apparatus;

determining an adapted covariance matrix based on the result of the

comparison; and

performing further signal processing using the adapted covariance matrix.

11. The method of claim 10, further comprising: determining a value of the scenario-independent metric for each one of a plurality of different regularized versions of the estimated covariance matrix,

wherein the regularized version of the estimated covariance matrix which gives the closest value of the metric to the target value is selected as the adapted covariance matrix.

12. The method of claim 10 or n, wherein the metric is a likelihood ratio, and the current configuration of the signal receiving apparatus is defined by the number of antennas in the multi-antenna array and by the number of pilot symbols currently in use.

13. The method of claim 10, 11 or 12, wherein the target value comprises a statistical measure related to a probability distribution function of the likelihood ratio.

14. The method of claim 13, wherein the statistical measure is a median value of the probability distribution function.

15. A non-transitory computer-readable storage medium arranged to store computer program instructions which, when executed, perform the method of any one of claims 10 to 14.