WO2017012671A1

WO2017012671A1 - Receiver device and method thereof

Info

Publication number: WO2017012671A1
Application number: PCT/EP2015/066878
Authority: WO
Inventors: Jianjun Chen; Fredrik RUSEK; Junshi Chen
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2015-07-23
Filing date: 2015-07-23
Publication date: 2017-01-26
Also published as: CN107750445A

Abstract

The present invention relates to a receiver device and a method thereof. The present method 200 comprises the major steps of: Receiving a communication signal CS; Identifying a plurality of interferers n = 1,...,Ν in the communication signal CS; At iteration i = 1 for interferer n = 1 obtain a covariance estimate for interferer n = 1 (See Fig.2B); At iteration i = 1 for interferers n = 2,...,N: obtain a covariance estimate for interferer n = 2,...,N (see Fig.2C); At iteration i = 2,...,I for the plurality of interferers n = 1,.,.,Ν: obtain at least one covariance estimate for interferer n = 1,...,N at iteration i = 2,..., I (see Fig. 2D); Computing (212) the total covariance estimate Ryy for the communication signal CS based on the covariance estimates for the plurality of interferers n = 1,.,.,Ν at iteration i = I > 1. Furthermore, the present invention also relates to a computer program, and a computer program product.

Description

RECEIVER DEVICE AND METHOD THEREOF Technical Field

The present invention relates to a receiver device. Furthermore, the present invention also relates to a corresponding method, a computer program, and a computer program product.

Background

3rd Generation Partnership Project (3GPP) proposes a frequency reuse factor of 1 in Long Term Evolution (LTE) systems. Therefore, interference will generally be high in such LTE systems. For the downlink, a User Equipment (UE) will suffer from interference from neighbouring e-NodeB(s) (eNB). The downlink interference will degrade the receiver performance of the UE.

Hence, a widely recognized problem of the LTE downlink is that the UE is not only receiving signals from the eNB in its own cell, but also from a number of neighbouring cells which are referred to as interfering cells. A common way to tackle this interference is through Interference Rejection Combining (IRC) techniques. In IRC, the algorithm is not trying to decode the interfering signals, but merely trying to whiten the noise plus interference signal. In this respect the UE must estimate the covariance matrix of the interfering signals.

Covariance matrix estimation has a long and rich history within wireless communications, and in the LTE Transmission Modes (TMs) 3 and 4, the problem is somewhat simplified as some parts of the covariance matrix are known. Since the physical channels from the interfering cells are highly correlated over time and frequency, they can in fact be fairly well estimated by the UE. The eNBs are unfortunately changing the Power Amplification (PA) levels and the Precoding Matrices (PM) at a rapid pace, so that even with perfect knowledge of the physical channel, the covariance matrix of the interference is not at hand.

Thus, the problem of estimating the interference covariance matrix is replaced by the problem of detecting the PAs and the PMs of the interfering cells. A further problem is that the eNBs do not apply the PMs at the training data locations and further that the PAs at the training data may differ from the PAs applied at payload data positions. Thus, the training data positions cannot be used for PA and PM detection. Besides from using IRC, one can also try to actually decode the signals from the interfering cells and then cancel them, e.g. in the context of the recent LTE feature Network Assisted Interference Cancellation and Supression (NAICS). This is especially tempting when at least one of the interfering cells is stronger than the serving cell. In this case, one does not need the interfering covariance matrix but instead the effective channel matrix.

This however boils down into the very same problem, i.e. to build a hard-decision covariance matrix by detecting the interfering cells' PAs and PMs, or build a soft estimate of the covariance matrix by weighted combining of a set of possible values of the interfering cells' PAs and PMs.

There are some previous conventional solutions about detecting the PA, PM and covariance matrix of one colliding Common Reference Signal (CRS) interfering cell. However, when there are multiple colliding CRS interfering cells, these conventional solutions will not work. The approaches in these conventional solutions are either use exhaustive search through all possible PA and PM values for all ranks, or assuming that all the interfering signals from multiple interfering cells have the same PA values or the same PM values, such that greatly reduce the search space of parameter combinations.

There are mainly two drawbacks of the mentioned conventional solutions. The first drawback is the high computing complexity. When there are multiple interferers, the complexity is very high for the exhaustive search of the PA and PM parameters. The second drawback is performance since assuming that all interfering signals have the same PA level is unrealistic meaning degraded performance.

Summary

An objective of embodiments of the present invention is to provide a solution which mitigates or solves the drawbacks and problems of conventional solutions.

Especially, embodiments of the present invention aim to provide improved covariance estimates for improved receiver performance in wireless communication systems. An "or" in this description and the corresponding claims is to be understood as a mathematical OR which covers "and" and "or", and is not to be understood as an XOR (exclusive OR).

The above objectives and further objectives are solved by the subject matter of the independent claims. Further advantageous implementation forms of the present invention can be found in the dependent claims. According to a first aspect of the invention, the above mentioned and other objectives are achieved with receiver device for a wireless communication system, the receiver device comprising:

a receiver configured to

receive a communication signal;

a processor configured to

identify a plurality of interferers n = 1, ... , N in the communication signal;

at iteration i = 1 for interferer n = 1:

- compute an error covariance and a corresponding covariance estimate for each possible combination of power amplification, PA, and Precoding Matrix, PM, pairs, wherein the error covariance is computed based on the communication signal,

- compute a metric for each error covariance,

- form a set of computed metrics and corresponding covariance estimates,

- select a first subset among the set of computed metrics and corresponding covariance estimates,

- compute a likelihood for each one of the computed metrics in the first subset,

- combine the covariance estimates in the first subset with their respective likelihoods so as to obtain a covariance estimate for interferer n = 1 at iteration i = 1;

at iteration i = 1 for interferers n = 2, ... , N:

- compute an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal and the covariance estimates for previous interferers m = Ι, . , . , η - 1 at iteration i = 1,

- compute a metric for each error covariance,

- form a set of computed metrics and corresponding covariance estimates,

- compute the likelihood for each one of the computed metrics in the first subset,

- combine the covariance estimates in the first subset with their respective likelihoods so as to obtain a covariance estimate for interferer n = 2, ... , N

at iteration i = 2, .. . , 1 for the plurality of interferers n = 1, ... , N:

- compute an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal and the covariance estimates for the previous interferers m = l, ... , n - 1 of the current iteration i and the covariance estimates for the subsequent interferers m = n + 1, ... , N of the previous iteration i - 1,

- compute a metric for each computed error covariance, - form a set of computed metrics and corresponding covariance estimates,

- combine the covariance estimates in the first subset with their respective likelihoods so as to obtain at least one covariance estimate for interferer n = Ι, .,. , Ν at iteration i =

2 /;

compute the total interference covariance estimate R for the communication signal based on the covariance estimates for the plurality of interferers n = 1, ... , N at iteration i = I > 1.

The present receiver device provides a number of advantages over conventional solutions. By providing an iterative solution for estimating the interference covariance (matrix) lower computational complexity is possible compared to conventional solutions. Moreover, the present solution also provides a more accurate estimate of the covariance compared to conventional solutions. Therefore, improved receiver performance is possible since interference can be better mitigated or reduced.

In a first possible implementation form of a receiver device according to the first aspect, the processor, before performing iteration i = 1, further is configured to

sort the plurality of interferers n = l, ... , N in a descending order according to their respective received power.

An advantage with the first possible implementation form is that by sorting the plurality of interferers n = 1, ... , N in a descending order according to their respective received power the covariance estimate can be estimated with higher precision.

In a second possible implementation form of a receiver device according to the first possible implementation form of the first aspect or to the first aspect as such, the processor further is configured to

sort the plurality of interferers n = l, ... , N in a descending order based on the

Frobenius norms of their respective channel estimates.

An advantage with the second possible implementation form is that the second possible implementation form is easy to implement, since the Frobenius norm can be achieved by simply adding the power of all the elements in a matrix together. In a third possible implementation form of a receiver device according to any of the preceding possible implementation forms of the first aspect or to the first aspect as such, the likelihood for each computed metric is based on the energy content of the computed metric. An advantage with the third possible implementation form is that the third possible implementation form is easy to implement and also has good performance. Since, by using the likelihood based on the energy content of the computed metric matrix inversion is not needed. In a fourth possible implementation form of a receiver device according to any of the preceding possible implementation forms of the first aspect or to the first aspect as such, the computed metric is the Frobenius norm F of the error covariance.

An advantage with the fourth possible implementation form is that the fourth possible implementation form is easy to implement and also has good performance, since the Frobenius norm can be achieved by simply adding the power of all the elements in a matrix together.

In a fifth possible implementation form of a receiver device according to the fourth possible implementation form of the first aspect, the Frobenius norm F of the error covariance for interferer n for the PA an PM pair _n, W_n is computed according to the equation:

F(a_n, W_n) = ||R^ - R_yy(a_n,W_n) ||²

where R^_y is the covariance of the communication signal, and R_yy( _n, W_n) is the total covariance estimate. An advantage with the fifth possible implementation form is that by using this equation the distance between the real covariance and the estimated covariance can easily and quickly be found.

In a sixth possible implementation form of a receiver device according to any of the preceding possible implementation forms of the first aspect or to the first aspect as such, each first subset comprises the K smallest computed metrics in the set of computed metrics and corresponding covariance estimates.

K is smaller than the overall number of computed metrics. An advantage with the sixth possible implementation form is that by only using K number of smallest computed metrics the performance is improved. Further, the computational complexity is also reduced with this possible implementation form. In a seventh possible implementation form of a receiver device according to any of the preceding possible implementation forms of the first aspect or to the first aspect as such, the processor further is configured to compute the total interference covariance estimate R for the communication signal by

summing the covariance estimates for the plurality of interferers n = 1, ... , N at iteration /, serving cell covariance and noise covariance.

An advantage with the seventh possible implementation form is that the computational complexity is low compared to conventional algorithms. Also performance is increased gradually by iterative estimation.

In an eight possible implementation form of a receiver device according to the fifth possible implementation forms of the first aspect, the processor further is configured to

form a common set comprising all the K metrics and their corresponding covariance estimates for the plurality of interferers n = Ι, .,. , Ν at iteration /;

compute the total interference covariance estimate R for the communication signal based on at least one metric and its corresponding covariance estimate in the common set of metrics and corresponding covariance estimates.

An advantage with the eight possible implementation form is that the complexity can be reduced by taking only K^N metrics instead of all the possible combination of M^N , where M is the number of possible combination of PA and PM for one interferes

In a ninth possible implementation form of a receiver device according to the eight possible implementation forms of the first aspect, the processor further is configured to

select a second subset of the common set of metrics and corresponding covariance estimates;

compute the total interference covariance estimate R for the communication signal based on the metrics and the corresponding covariance estimates in the second subset. An advantage with the ninth possible implementation form is that the complexity can be further reduced due to the selection of a second subset of metrics. In a tenth possible implementation form of a receiver device according to the ninth possible implementation forms of the first aspect, the processor further is configured to

compute the likelihood for each one of the metrics in the second subset, wherein the likelihood is based on the energy content of the metric,

compute the total interference covariance estimate R for the communication signal by combine the covariance estimates in the second subset with their respective likelihoods.

An advantage with the tenth possible implementation form is that by soft combining the soft likelihoods a better covariance estimate is provided.

In a eleventh possible implementation form of a receiver device according to the ninth possible implementation forms of the first aspect, the processor further is configured to compute the likelihood for each one of the metrics in the second subset, wherein the likelihood is the Maximum Likelihood, ML, of the metric,

An advantage with the eleventh possible implementation form is that with the ML method better performance can be provided, e.g. compared to the energy based method. However, the complexity is higher with the ML method.

In a twelfth possible implementation form of a receiver device according to any of the preceding possible implementation forms of the first aspect or to the first aspect as such, the covariance estimate for interferer n = 1 at iteration i = 1 is computed according to the equation:

where ∑ ₌₁ | ^, wJ⁵) iH₁wJ^<:(Wj^<:)^HHi is weighted covariance estimate, and

is the normalization factor, and wherein the covariance estimate for interferer n = 2, ... , N at iteration i = 1 plus serving cell covariance and noise covariance is computed according to the equation:

Ryy(a_n,W_n) = a₀HoW₀Wo^HHi? + a_nH_nW_nW_n ^HH" + ^ R^ + R_UU

m<n

where

is the covariance for the serving cell, _NH_NW_NW„ H„ is the covariance estimate of interferer n of the current iteration, ∑_m<„R^ is the covariance estimate of previous interferers m = 1, . . . , n— l of the current iteration, and R_uu is the covariance of noise. An advantage with the twelfth possible implementation form is that an initial covariance estimate can be provided without prior information of other interferers at the initiation of the iterative method.

In a thirteenth possible implementation form of a receiver device according to the twelfth possible implementation forms of the first aspect, the total covariance estimate Ryy for the communication signal at iteration i = I > 1 is computed according to the equation:

Ryy (a_n, W_n) = a₀H₀W₀W₀ ^HH + a_nH_nW_nW_n ^HH" + ^ R + ^ R^_m + R_uu

m<n m>n

where ocoHoWoW^Ho is the covariance of the serving cell, _nH_nW_nW„ H„ is the covariance estimate of interferer n of current iteration,∑_m<„ R^ is the covariance estimate of interferers m = l, . . . , n - l of current iteration, ∑_m>n R^m is the covariance estimate of interferers m = n + 1, ... , N of the previous iteration, and R_uu is the covariance of noise.

An advantage with the thirteenth possible implementation form by updating the covariance estimate iteratively according to this equation an improved covariance estimate can be provided.

According to a second aspect of the invention, the above mentioned and other objectives are achieved with a method for a wireless communication system, the method comprising:

receiving a communication signal,

identifying a plurality of interferers n = 1, ... , N in the communication signal;

at iteration i = 1 for interferer n = 1:

- computing an error covariance and a corresponding covariance estimate for each possible combination of power amplification, PA, and Precoding Matrix, PM, pairs, wherein the error covariance is computed based on the communication signal,

- computing a metric for each error covariance,

- forming a set of computed metrics and corresponding covariance estimates,

- selecting a first subset among the set of computed metrics and corresponding covariance estimates,

- computing a likelihood for each one of the computed metrics in the first subset,

- combining the covariance estimates in the first subset with their respective likelihoods so as to obtain a covariance estimate for interferer n = 1 at iteration i = 1;

at iteration i = 1 for interferers n = 2, ... , N:

- computing an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal and the covariance estimates for previous interferers m = 1, n - 1 at iteration i = 1,

- computing a metric for each error covariance,

- forming a set of computed metrics and corresponding covariance estimates,

- computing the likelihood for each one of the computed metrics in the first subset,

- combining the covariance estimates in the first subset with their respective likelihoods so as to obtain a covariance estimate for interferer n = 2, ... , N

at iteration i = 2, . . . , 1 for the plurality of interferers n = 1, ... , N:

- computing an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal and the covariance estimates for the previous interferers m = Ι, . , . , η - 1 of the current iteration i and the covariance estimates for the subsequent interferers m = n + Ι, .,. , Ν of the previous iteration i - 1,

- computing a metric for each computed error covariance,

- forming a set of computed metrics and corresponding covariance estimates,

- combining the covariance estimates in the first subset with their respective likelihoods so as to obtain at least one covariance estimate for interferer n = Ι, .,. , Ν at iteration i =

2 /;

computing the total interference covariance estimate R for the communication signal based on the covariance estimates for the plurality of interferers n = 1, ... , N at iteration i = I > 1.

In a first possible implementation form of a method according to the second aspect, the method, before performing iteration i = 1, further comprises

sorting the plurality of interferers n = l, ... , N in a descending order according to their respective received power.

In a second possible implementation form of a method according to the first possible implementation form of the second aspect or to the second aspect as such, the method further comprises

sorting the plurality of interferers n = l, ... , N in a descending order based on the Frobenius norms of their respective channel estimates. In a third possible implementation form of a method according to any of the preceding possible implementation forms of the second aspect or to the second aspect as such, the likelihood for each computed metric is based on the energy content of the computed metric. In a fourth possible implementation form of a method according to any of the preceding possible implementation forms of the second aspect or to the second aspect as such, the computed metric is the Frobenius norm F of the error covariance.

In a fifth possible implementation form of a method according to the fourth possible implementation form of the second aspect, the Frobenius norm F of the error covariance for interferer n for the PA an PM pair _n, W_n is computed according to the equation:

F(a_n, W_n) =

- R_yy(a_n,W_n) ||²

where R^_y is the covariance of the communication signal, and R_yy( _n, W_n) is the total covariance estimate. In a sixth possible implementation form of a method according to any of the preceding possible implementation forms of the second aspect or to the second aspect as such, each first subset comprises K number of smallest computed metrics in the set of computed metrics and corresponding covariance estimates. In a seventh possible implementation form of a method according to any of the preceding possible implementation forms of the second aspect or to the second aspect as such, the total covariance estimate R for the communication signal is computed by

In an eight possible implementation form of a method according to the fifth possible implementation forms of the second aspect, the method further comprises

forming a common set comprising all the K metrics and their corresponding covariance estimates for the plurality of interferers n = Ι, .,. , Ν at iteration /;

computing the total interference covariance estimate R for the communication signal based on at least one metric and its corresponding covariance estimate in the common set of metrics and corresponding covariance estimates.

In a ninth possible implementation form of a method according to the eight possible implementation forms of the second aspect, the method further comprises selecting a second subset of the common set of metrics and corresponding covariance estimates;

computing the total interference covariance estimate R for the communication signal based on the metrics and the corresponding covariance estimates in the second subset.

In a tenth possible implementation form of a method according to the ninth possible implementation forms of the second aspect, the method further comprises

computing the likelihood for each one of the metrics in the second subset, wherein the likelihood is based on the energy content of the metric,

computing the total interference covariance estimate R for the communication signal by combine the covariance estimates in the second subset with their respective likelihoods.

In a eleventh possible implementation form of a method according to the ninth possible implementation forms of the second aspect, the method further comprises

computing the likelihood for each one of the metrics in the second subset, wherein the likelihood is the Maximum Likelihood, ML, of the metric,

computing the total interference covariance estimate R for the communication signal by combine the covariance estimates in the second subset with their respective likelihoods. In a twelfth possible implementation form of a method according to any of the preceding possible implementation forms of the second aspect or to the second aspect as such, the covariance estimate for interferer n = 1 at iteration i = 1 is computed according to the equation:

∑_k ^K ₌₁i({¾} l wi)

where

is weighted covariance estimate, and ∑k₌₁i({¾} | αι, wj⁵) is the normalization factor, and wherein the covariance estimate for interferer n = 2, ... , N at iteration i = 1 plus serving cell covariance and noise covariance is computed according to the equation:

Ryy(a_n, W_n) = a₀HoW₀Wo^H Hi? + a_nH_nW_nW_n ^H H" + ^ R^ + R_uu

m<n

where OCOHOW_QW^HO is the covariance for the serving cell, _NH_NW_NW„ H„ is the covariance estimate of interferer n of the current iteration, ∑_m<„R^ is the covariance estimate of previous interferers m = 1, . . . , n— l of the current iteration, and R_uu is the covariance of noise. In a thirteenth possible implementation form of a method according to the twelfth possible implementation forms of the second aspect, the total covariance estimate Ryy for the communication signal at iteration i = I > 1 is computed according to the equation:

Ryy («„, W_n) = a₀H₀W₀W₀ ^HH" + a_nH_nW_nW_n ^HH" + ^ R + ^ R^_m + R_uu

m<n m>n

The advantages of the methods according to the second aspect are the same as those for the receiver device according to the first aspect.

The present invention also relates to a computer program, characterized in code means, which when run by processing means causes said processing means to execute any method according to the present invention. Further, the invention also relates to a computer program product comprising a computer readable medium and said mentioned computer program, wherein said computer program is included in the computer readable medium, and comprises of one or more from the group: ROM (Read-Only Memory), PROM (Programmable ROM), EPROM (Erasable PROM), Flash memory, EEPROM (Electrically EPROM) and hard disk drive.

Further applications and advantages of the present invention will be apparent from the following detailed description.

Brief Description of the Drawings

The appended drawings are intended to clarify and explain different embodiments of the present invention, in which:

- Fig. 1 shows a receiver device according to an embodiment of the present invention;

- Figs. 2A-2D show a method according to an embodiment of the present invention;

- Fig. 3 shows a wireless communication system according to an embodiment of the present invention;

- Figs. 4-6 shows performance results for embodiments of the present invention. Detailed Description

Fig. 1 shows a receiver device 100 according to an embodiment of the present invention. The receiver device 100 comprises a processor 102 which is communicably coupled with communication means 108 to a receiver 104 in this exemplary embodiment. The communication means 108 are illustrated as dotted arrows between the processor 102 and the receiver 104 in Fig. 1 . The communication means 108 are according to techniques well known in the art. The communication means 108 may e.g. be used for transfer of data or control signalling between the processor 102 and the receiver 104. The user device 100 in this particular embodiment further comprises control means 1 10 by which the processor 102 operates (or controls) the receiver 104. The control means are illustrated with the arrow from the processor 102 to the receiver 104. The user device 100 also comprises antenna means 106 coupled to the receiver 104 for reception in the wireless communication system 300. Optionally, the receiver 104 may be part of a transceiver for reception and transmission in the wireless communication system 300.

According to the present solution, the receiver 104 is configured to receive a communication signal CS transmitted in the wireless communication system 300 as illustrated in Fig. 1 . The processor 102 is configured to identify a plurality of interferers n = l, ... , N in the communication signal CS. The plurality of interferers n = l, ... , N may according to an embodiment be interfering cells in a cellular system, such as LTE. Different interfering cells have different cell Identities (IDs), which is unique for each cell. The cell ID is, in LTE systems, contained in the Synchronous Channel (SCH). Therefore, by detecting the SCH, the cell ID can be derived, and after the cell ID has been derived, the Common Reference Signal (CRS) and corresponding channel estimation of the interfering cell(s).

The processor 102 is further configured to perform an iterative method for providing the total interference covariance estimate R for the communication signal CS. The provided total interference covariance estimate R can be used for interference reduction, mitigation or suppression and thereby improve the system performance. For example, the present solution can be used in an Interference Rejection Combining (IRC) algorithm. Therefore, according to an embodiment of the present invention, the processor is further configured to use the total interference covariance estimate R as an input in an interference reduction, mitigation or suppression algorithm. Therefore, the processor 102 of the receiving device 100 is further configured to, at iteration i = 1 for interferer n = 1: - compute an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal CS,

- compute a metric for each error covariance,

- form a set of computed metrics and corresponding covariance estimates,

Therefore, the processor 102 of the receiving device 100 is further configured to, at iteration i = 1 for interferers n = 2, ... , N:

- compute an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal CS and the covariance estimates for previous interferers m = Ι, . , . , η - 1 at iteration i = 1,

- compute a metric for each error covariance,

- form a set of computed metrics and corresponding covariance estimates,

Therefore, the processor 102 of the receiving device 100 is further configured to, at iteration i = 2, ... , 1 for the plurality of interferers n = 1, ... , N:

- compute an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal CS and the covariance estimates for the previous interferers m = 1, ... , n - 1 of the current iteration i and the covariance estimates for the subsequent interferers m = n + 1, ... , N of the previous iteration i - 1,

- compute a metric for each computed error covariance,

- form a set of computed metrics and corresponding covariance estimates,

- compute the likelihood for each one of the computed metrics in the first subset, - combine the covariance estimates in the first subset with their respective likelihoods so as to obtain at least one covariance estimate for interferer n = Ι, .,. , Ν at iteration i =

2 /; Finally, the processor 102 of the receiving device 100 is further configured to compute the total interference covariance estimate R for the communication signal CS based on the covariance estimates for the plurality of interferers n = Ι, .,. , Ν at iteration i = I > 1.

Figs. 2A-2D shows a corresponding method 200. The method 200 may be executed in a receiver device 100, such as the one shown in Fig. 1. The present iterative method may be divided into some major steps including sub-steps. That is, step 206 - obtain covariance estimate for interferer n = l at iteration i = 1, see Fig. 2B. Thereafter, step 208 - obtain covariance estimate for interferer n = 2, . . . , N at iteration i = 1, see Fig. 2C. Thereafter, step 210 - obtain covariance estimate for interferer n = Ι, . , . , Ν at iteration i = 2, . . . , I, see Fig. 2D. Finally, the total interference covariance estimate R for the communication signal CS is computed. According to an embodiment of the present invention the total interference covariance estimate R is computed by applying one of three termination algorithms more described in the following disclosure. Therefore the present method 200 comprises the major steps of:

• Receiving a communication signal CS;

• Identifying a plurality of interferers n = Ι, .,. , Ν in the communication signal CS;

• At iteration i = 1 for interferer n = 1 obtain a covariance estimate for interferer n = 1 (See Fig. 2B);

· At iteration i = 1 for interferers n = 2, ... , N: obtain a covariance estimate for interferer n = 2, ... , N (see Fig. 2C);

• At iteration i = 2, ... J for the plurality of interferers n = Ι, .,. , Ν: obtain at least one covariance estimate for interferer n = 1, ... , N at iteration i = 2, . .. J (see Fig. 2D);

• Computing (212) the total covariance estimate R_yy for the communication signal CS based on the covariance estimates for the plurality of interferers n = l, ... , N at iteration i = I > 1.

With reference to Fig. 2B the method 200 further comprises the steps of:

• Computing an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal CS; • Computing a metric for each error covariance;

• Forming a set of computed metrics and corresponding covariance estimates;

• Selecting a first subset among the set of computed metrics and corresponding covariance estimates;

· Computing a likelihood for each one of the computed metrics in the first subset;

• Combining the covariance estimates in the first subset with their respective likelihoods so as to obtain a covariance estimate for interferer n = 1 at iteration i = 1.

With reference to Fig. 2C the method 200 further comprises the steps of:

· Computing an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal CS and the covariance estimates for previous interferers m = Ι, . , . , η - 1 at iteration i = 1;

• Computing a metric for each error covariance;

· Forming a set of computed metrics and corresponding covariance estimates;

• Computing the likelihood for each one of the computed metrics in the first subset;

• Combining the covariance estimates in the first subset with their respective likelihoods so as to obtain a covariance estimate for interferer n = 2, ... , N at iteration i = 1.

With reference to Fig. 2D the method 200 further comprises the steps of:

• Computing an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal CS and the covariance estimates for the previous interferers m = 1, . . . , n - 1 of the current iteration i and the covariance estimates for the subsequent interferers m = n + 1, ... , N of the previous iteration i - 1;

• Computing a metric for each computed error covariance;

• Forming a set of computed metrics and corresponding covariance estimates;

· Selecting a first subset among the set of computed metrics and corresponding covariance estimates;

• Combining the covariance estimates in the first subset with their respective likelihoods so as to obtain at least one covariance estimate for interferer n = 1, ... , N at iteration i = 2.. . ..1. In one embodiment of the present invention, the receiver device 100 may be any of a User Equipment (UE), mobile station (MS), wireless terminal or mobile terminal being enabled to communicate wirelessly in a wireless communication system, sometimes also referred to as a cellular radio system. The UE may further be referred to as mobile telephones, cellular telephones, computer tablets or laptops with wireless capability. The UEs in the present context may be, for example, portable, pocket-storable, hand-held, computer-comprised, or vehicle-mounted mobile devices, enabled to communicate voice or data, via the radio access network, with another entity, such as another receiver or a server. The UE can be a Station (STA), which is any device that contains an IEEE 802.1 1 -conformant Media Access Control (MAC) and Physical Layer (PHY) interface to the Wireless Medium (WM).

Fig. 3 shows a wireless communication system 500 according to an embodiment of the present invention. The receiver device 100 is in Fig. 3 illustrated as a UE 100. The UE 100 receives a communication signal CS comprising a wanted signal from a serving base station 400a of a serving cell. However, the communication signal CS also comprises interfering signals from neighbouring interfering base stations 400b and 400c, respectively, in this particular example. Due to the interferers the receiver performance will be degraded in the UE 100. However, by applying the iterative method in the UE 100 according to the present solution improved receiver performance is provided.

The base station(s) 400 may be a (radio) network node or an access node or an access point or a base station, e.g. a Radio Base Station (RBS), which in some networks may be referred to as transmitter, "eNB", "eNodeB", "NodeB" or "B node", depending on the technology and terminology used. The radio network nodes may be of different classes such as e.g. macro eNodeB, home eNodeB or pico base station, based on transmission power and thereby also cell size. The radio network node can be a Station (STA), which is any device that contains an IEEE 802.1 1 -conformant Media Access Control (MAC) and Physical Layer (PHY) interface to the Wireless Medium (WM). Moreover, in the following disclosure embodiments of the present invention are described in a LTE context. Therefore, LTE terminology, system concepts, etc. are used. It should however be understood that the present solution is not limited to such LTE system but can be applied in any suitable wireless communication system. Further, in the following described embodiments the present receiver device 100 is represented as a UE.

We consider a scenario in which a serving cell is impaired by N + M interfering cells. N of those cells are colliding interferers, meaning that the training symbols of those cells are overlapped in time and frequency with the training symbols of the serving cell. M interfering cells are non-colliding, meaning that the CRS training symbols of those cells do not overlap the training symbols of the serving cell. The received signal at the UE of the serving cell, at any Resource Element (RE), can be described as

In equation (1 ), the index pair {k, /) denotes the k-Vn RE in time and Z-th in frequency within a given Pair of Resource Blocks (PRB). The variables a₀, a_n, and a_m denote the PA values used by the eNBs for the serving cell, the n-th colliding cell, and the m-th non-colliding cell, respectively. The matrices H denote the channel matrices from the eNBs to the serving cell UE, the matrices W denote the selected PM at the eNBs, the vectors s denote the transmitted vectors, and w is complex Gaussian noise with covariance matrix N₀I.

Moreover, we have assumed in equation (1 ) that all channel matrices remain constant across the entire PRB. Due to LTE specifications, the PMs and the PAs are constant by definition within one PRB. For simplicity, we shall only consider a scenario where all eNBs of the wireless communication system are equipped with 4 antennas. However, the proposed solution can be applied to other antenna configurations as well which is recognized by the skilled person in the art. We denote the noise plus the non-colliding interference as

"k,l a_mH_mW_ms_{m k l} + w_k>1

so that we can write equation (1 ) as

yk,i = V°o^Ho^woSo,k,i + ^ °nH_nW_nS_{n ki}i + W_{k l}

(2)

It can be assumed that the covariance matrix of w_{k l}, here denoted by R_uu, is perfectly known and also that the channel matrices H_n are known for 0 ≤ n ≤ N, i.e. the channel matrix is known for the serving cell as well as for all colliding interfering cells. It is assumed that the interfering channels and the interference part of the interference covariance matrix R_uu are known since the channel matrices are changing slowly, so that interpolation methods can be used across several P Bs. At the training symbol positions, since all training symbols are known to the UE, an iterative interference cancellation technique can be carried out according to embodiments of the present invention. Such an iterative method is capable of estimating all channels H _{n=0 1 N} with sufficiently high precision so that this is not the bottleneck for estimating the interference covariance matrix R_uu. When all channels are estimated, their influence on the training data positions can be cancelled, which enables an estimation of the covariance matrix R_uu. This estimation is of high precision since after cancelation the only remaining signal is w^'.

Although the interfering channel matrices can be assumed to be known, the PAs and the PMs are not. This is so since these can change abruptly at the PRB borders, which limits the UE to a single PRB for their detection. The PA and the PM values for the serving cell, i.e. a₀ and W₀, are known as they are embedded in the control data.

Embodiments of the present solution provide devices and methods to detect the PAs and the PMs of the interfering cells, i.e., a_n and W_n for 1 ≤ n ≤ N. As mentioned earlier, at the training data positions the PAs can differ from the PAs at the payload data positions, and no PM has been applied for the training data. Therefore, we are limited to the payload data positions only for the detection of the PAs and the PMs.

In IRC, one is not interested in the PAs and the PMs themselves, but rather in the covariance matrix of the interfering signals, i.e.

To estimate the covariance matrix is not the same problem as estimating the PAs and the PMs as the best estimate of the covariance matrix R may not correspond to any valid set of PAs and PMs. We shall investigate methods to combine the most probable PAs and PMs into an estimate of the covariance matrix R.

According to the 3GPP specifications, the PA value can take one out of 8 values. However, the interfering cell can be idle, so that a_n = 0. On the training symbols, the interfering cell is never idle, but must transmit with a known non-zero PA value. Hence, it is not a contradiction to say that the value of N is known, but that it is unknown if a_n = 0 or not. The number of interfering cells will be stable over time, but within a single PRB, the colliding interfering cells may not transmit any payload data. Therefore, one must include the value o„ = 0 as a possibility which give us the choices:

a e {0, -6dB, -4.77dB, -3dB, -1.77dB, OdB, ldB, 2dB, 3dB}

For TM4 4x4, there are 64 PMs, but since for all the rank 4 PMIs W_nW" = 1/4, therefore there are only 49 different W_nW„ . Considering the number of possible PAs and the case when certain cell is not transmitting signal, then the total number of possible combinations of PA and PM is:

(8 x 49 + 1) which are 393 for one colliding cell and 154449 for two colliding cells.

Energy based method forN = 1

In this section we discuss an energy based method for the case N = 1 for simplicity. The energy based method is later used for calculating metrics in the present iterative algorithm. Note that the energy based method can also be used for N > 1.

On the REs containing payload data, the covariance matrix, as a function of the unknowns o₁ and W_b equals

where QOHOW_QW^ HO is the covariance matrix of serving cell.

The sample mean covariance matrix based on payload data ^'

where the sum is over the T_d payload data carrying REs.

The energy based method is given by

where

F(Q₁,W₁) = llRyy -

It should be observed that for full rank precoders Wj, the quantity

= I by construction, hence, these PMs are indistinguishable from each other. It therefore suffices to exhaust 8 ^χ 49 + 1 = 393 metrics. This is feasible, and we therefore do not pursue any complexity reduction method. There is indeed great potential to re-use computations between these 393 metric evaluations, and in particular, for a given PM the best PA value can be efficiently found. From above, one can directly reach an estimate of the covariance matrix of the interfering signal as

This estimate can be improved by using soft combination of the most probable PAs and PMs, as is explained next. The norm Fio^W^ can be written as

where ¾ are the elements of the error covariance matrix (R^_y - Ryy^o^W^). It can be shown that if the diagonal elements of the underlying covariance matrix R_yy are identical, then all elements of R^_y have the same variance. In order to obtain a mathematically tractable problem, we assume that this still holds.

Moreover, in our case is unknown since the PA and PM are unknown. In order to proceed forward we make the assumptions that all diagonal elements are the same, and that the error terms ¾ are Gaussian distributed. Then the likelihood of the errors {¾} given can be expressed through {e_i}j} as

where δ² is the variance of the diagonal elements and 7 is a normalization constant, which can be thrown away in the later steps for reaching a soft combined covariance matrix estimate. Due to the rapid decay of the exponential function, it suffices to include only a few terms with the smallest F^ metrics. Suppose that we include T smallest metrics, then the soft-combined estimate of the covariance matrix is formed as

where (a' , W') is the combination of PA and PM values that produces the f-th smallest metric

It remains to discuss how to choose an appropriate value of the variance δ². We select this value as twice the mean value of the diagonal elements of R^_y, divided by the number of data RE T_d .

Maximum likelihood method for N = 1

In this section we discuss the Maximum Likelihood (ML) method for the case N = 1 for simplicity. The ML method is later used for calculating metrics in the present iterative algorithm. Note that the ML method can also be used for N > 1.

The likelihood in energy based method is not the likelihood of receiving the signals {yk,i}, only the likelihood of getting the errors {¾}. Basing the detection on the true likelihood will produce better results, but as we shall see also lead to higher complexity. If we approximate the transmitted signals as complex Gaussian then the likelihood of receiving the signals {yk,i}, given a pair of PA and PM equals

p({y_k,i}|ai,W₁) = _K ^exP (-yk,l^Ry (^ai' ^Wi)yk,l)

where κ is a normalization constant. Equivalently, the log-likelihood function equals atty_k.iJ w = Wjyk,!] + log(_K)

= -T_d log(det[R_yy(a₁, W₁)]) - Td

WjRyy] + log(«)

Given the likelihood function, an ML detection of the PA and the PM can be reached as

For the K best PA and PM pairs, the interference covariance matrix estimate can be formed as

As already mentioned, for a single interferes i.e., N = 1, the total number of possible PA and PM pairs is 393, a number that is feasible to exhaust. However, in the ML based approach, each pair requires a matrix inverse. For practical implementation, the complexity of the ML approach cannot be considered small. We therefore propose to first compute the energy based metrics FCa^W for the 393 pairs, choose the K best, and then run the ML- based PA and PM detection (or covariance estimation) using the K pairs.

The present iterative algorithm

We next turn our attention to the case N > 1 using the present iterative solution. As previously stated, an exhaustive search needs to test (8 * 49 + l)^w combinations. Already for N = 2, this means 154449 combinations, clearly an infeasible number in practice. In what follows we propose an iterative algorithm and we describe this for the energy based method. The same iterative algorithm can be applied verbatim with the likelihood cost function. At the end we combine the two methods to save complexity.

As we can see, the parameters of interferers with strong power can be estimated more accurately than those of interferers with weak power, the first step of the algorithm is to sort the interferers according to their power. Since their PA values are not at hand, this needs to be done based on the Frobenius norms of the channel matrices H_n. We assume that this is already taken care of, so that HHJ > ||H₂ || > ··· > ||H_N|| .

Iteration i = 1

As before, we let R^_n denote the estimated covariance matrix corresponding to the n-th interferer in the first iteration step. In the first iteration, these covariance matrix are unknown, and we therefore assume that they are all zeros at the beginning and initialize them with R^n = 0, 1 ≤ n ≤ N (since we don^'t know anything about the interferers at this stage). Let us now redefine R_yy(a_n, W_n) as R_yy(Q_n,W_n) = a₀H₀W₀W₀ ^HH + a_nH_nW_nW_n ^HH" + ^ R_1>p + R_uu

p≠n (5) where a₀H₀W₀W^ Ho is for the serving cell and a_nH_nW_nW"H" +∑_p≠n R^_p + R_uu for interferers.

Then, we start by exhausting all the first interferer's 393 possible combinations of the PA and PM pairs (a^W_!) and compute the 393 metrics based on Frobenius norm,

F(a₁,W₁) =

- R_yy(a₁,W₁) \\²

In one embodiment of the present invention, the likelihood for each computed metric is based on the energy content of the computed metric.

In one further embodiment of the present invention, the computed metric is the Frobenius norm F of the error covariance.

We then sort these metrics and get the K number of smallest metrics which correspond to the most possible combination of PA and PM pairs (C^ WL), and update the covariance matrix ing to equation (3) based on the K best pairs, i.e. the pairs having the least metric

is

∑ξ_{= 1} L({_¾} |c^k W₁ ^k)a?H₁W₁ ^k(W₁ ^k)^HHi^I

δ² is the variance of the diagonal elements of

Ryy - R_yy(a^k, W^k), and {¾} is the element of F(a^k, W^k). Then, we repeat this for the remaining N - 1 interferers in the communication signal CS. For the n-th interferes we update the R_yy(a_n,W_n) as shown below firstly

RyyCc W = a₀H₀W₀W₀ ^HH₀ ^H + a_nH_nW_nW_n ^HH» + Y Rl,m + Ruu where a₀H₀W₀W"Ho is for the serving cell and a_nH_nW_nW"H" +∑_m<„ + R_uu is for interferers. Note that, the covariance matrices for m < n have been achieved and do not equal to zero. Then we exhaust all the n-th interferer's 393 possible combination of the PA and PM pairs (a_n, W_n), get the K smallest metric of F(a_n, W_n) , and then update the covariance matrix R^_n with the same way as we do for in 1 ^st process. (a„,W_n)

For calculation convenience, we point out here that for each W_n, we can very efficiently compute the metrics for all a_n since a_n is a scalar.

The above described process requires the computation of 393Λ/ metrics F(a_n, W_n), and the results in N estimated covariance matrices R^_n, 1 < n < J for each interferes In one embodiment of the present invention, the processor 102, before performing iteration i = 1, further is configured to sort the plurality of interferers n = 1, ... , N in a descending order according to their respective received power. In an alternative, the plurality of interferers n = 1, ... , N are sorted in descending order based on the Frobenius norms of their respective channel estimates.

Iteration steps i = 2, ... , N

At iteration step i > 1, the process of iteration 1 of above section is repeated, with the difference that when we exhaust the interferer n, the interference covariance matrix of interferer m > n are taken from the previous iteration i - 1. Here the updated R_yy(a_n,W_n) is expressed as follow

R_yy(a_n,W_n) = a₀H₀W₀W₀ ^HH» + a_nH_nW_nW_n ^HH" +∑ RTm +∑ i Tm + u

m<n m>n where a₀H₀W₀W₀ ^HH for the serving cell and a_nH_nW_nW_n ^HH^ +∑_m<„ R^ +∑_m>n R^;_m + R_uu is for the interfering cells.

Generally, the iterative method has to be terminated at some stage and the total interference covariance estimate R for the communication signal CS outputted or used in further applications. In the following disclosure three different algorithm terminations are provided according to embodiments of the present invention. Algorithm Termination 1

When we reach the end of the final iteration /, we have 393 PA and PM pairs for each interferer n = l, ... , N. To conclude the iterative algorithm, many paths forward can be envisioned. One simple termination is to estimate the total interference covariance matrix as sum of the current estimates, i.e.,

Algorithm Termination 2

Another method to terminate the present iterative algorithm is to exhaust the best pairs of all the interferers jointly. Let us choose the best K PA and PM pairs for each interferer n. In total there are K^N combinations of the PA and PM pairs. Denote the total covariance matrix constructed from the Z-th such combination as

N

R_yy ({a<^l),W<^l>}) = a₀H₀W₀W₀ ^HH₀ ^H + a«H_nW«(W«)^HH" + R_uu

n=l We can then compute the K^N metrics as

F ({a« W«}) = || R~ - R_yy ({a« W«}) || ² and save the best K PA and PM pairs. With a straightforward extension of the construction in equation (3), we can now make a soft combination of these K PA and PM pairs in order to form an estimate of the total interference covariance matrix as

Algorithm Termination 3

Yet another method to terminate the present iterative algorithm is to use the ML framework, similar to what was done for N = 1 described above. The likelihood in this case reads

Based on these likelihoods, one can form a soft estimate of the total interference covariance matrix R . Note that since each combination involves a matrix inversion, the number K must be fairly small.

Equivalently, the log-likelihood function equals

Given the likelihood function, an ML detection of the PA and PM pairs can be reached as

Numerical results for the case of two colliding interfering cells, i.e. , N = 2 are presented in the following disclosure with reference to Figs. 4-6. The x-axis in mentioned Figs. 4-6 represents the Signal-to-Noise Ratio (SNR) of the serving cell. The y-axis represents PM detection probability, PA detection probability, and Mean Square Error (MSE) of the error matrix norm (between the genuine covariance matrix and the estimated covariance matrix) for Figs. 4, 5 and 6, respectively. It is assumed a covariance matrix R_uu that is a scaled identity matrix, so that we avoid specifying the number of non-colliding interferers M. Moreover, we concentrate on the case of N = 2, i.e., two interfering cells. The two interfering cells are normalized to average mean SNR of 12 and 10 dB, respectively, and the channel profiles are assumed to be Extended Vehicle A (EVA) at pedestrian speeds. In Fig. 4 and Fig. 5 we show the detection probabilities for the PM and the PA of the stronger of the two interfering cells. In all cases, rank 2 transmissions have been used at all cells including the serving cell and the interfering cells, and the interfering PA values are chosen randomly and uniformly over the possible set of 9 values. We point out that the detection results get worse with increasing SNRs of the serving cell. This is so since the interfering cells "drown" in the serving cell signal. Although HO, WO, and aO of the serving cell are known, the payload data is not known and the serving cell signal will act as noise for the estimation of the interference parameters. The elements of the total covariance matrix R_yy increase in magnitude when the signal and the interference power increase. Therefore, when the serving signal's power is significantly larger than the interfering signal, it becomes more difficult to estimate the statistics of the interfering signal from R_yy.

As mentioned, Fig. 4 shows the PM detection results for two colliding interfering cells. The thick lines are exhaustive searches based on the energy based method and the ML method, respectively. The dashed lines shows the result for one iteration i = 1 and the solid lines the results for three iterations i = 3. Fig. 5 shows the PA detection result with the same simulation setup as in Fig. 1. Fig. 6 shows the MSE of estimated interference covariance matrix. The simulation setup is the same as the one in Fig. 1. Further, from Figs. 4-6 we can see that an exhaustive ML search performs (with the highest complexity) significantly better than an energy based method (around 4-6 dB). In all cases, the value K has been chosen as K = 8, i.e., we compute K² = 64 combinations of the most promising PM and PA pairs for the two interfering cells. As can be seen, the iterative method with termination algorithm 3 (ML based for the K² = 64 best combinations) performs close to optimal.

From Figs. 4-6, one can also see the impact of the different algorithm terminations. It can be seen that algorithm termination 3 (ML based) is the best. However, it is also more complex as it involves 64 matrix inversions. Algorithm termination 2 (energy based) improves over algorithm termination 1 (sum) since it considers the best K = 8 pairs for each interferes and then outputs the best joint combination within the 64 possible combinations. In fact, with algorithm termination 1 , 3 iterations perform worse than 1 iteration with algorithm termination 2. From this we can conclude that it is important to terminate by evaluating the interferers jointly.

While Fig. 4 and Fig. 5 considered the hard output PM and PA detection results, we consider the estimation of the total interference covariance matrix R in Fig. 6. The metric we present is the MSE, namely E||R - R|| . All parameters for this test case are identical to those in Fig. 4 and 5. Also for this metric, an exhaustive ML search (with the highest complexity) performs much better than an exhaustive energy method. Further, the iterative method with algorithm termination 3 (ML based) can outperform the exhaustive energy based search, this even for a single iteration, 1 = 1. From this we can conclude that in order to obtain a low MSE, it is important to utilize the ML framework and not only the energy based method.

Furthermore, any method according to the present invention may be implemented in a computer program, having code means, which when run by processing means causes the processing means to execute the steps of the method. The computer program is included in a computer readable medium of a computer program product. The computer readable medium may comprises of essentially any memory, such as a ROM (Read-Only Memory), a PROM (Programmable Read-Only Memory), an EPROM (Erasable PROM), a Flash memory, an EEPROM (Electrically Erasable PROM), or a hard disk drive.

Moreover, it is realized by the skilled person that the present receiver device comprises the necessary communication capabilities in the form of e.g., functions, means, units, elements, etc., for performing the present solution. Examples of other such means, units, elements and functions are: processors, memory, buffers, control logic, encoders, decoders, rate matchers, de-rate matchers, mapping units, multipliers, decision units, selecting units, switches, interleavers, de-interleavers, modulators, demodulators, inputs, outputs, antennas, amplifiers, receiver units, transmitter units, DSPs, MSDs, TCM encoder, TCM decoder, power supply units, power feeders, communication interfaces, communication protocols, etc. which are suitably arranged together for performing the present solution.

Especially, the processors of the present receiver device comprises, e.g., one or more instances of a Central Processing Unit (CPU), a processing unit, a processing circuit, a processor, an Application Specific Integrated Circuit (ASIC), a microprocessor, or other processing logic that may interpret and execute instructions. The expression "processor" may thus represent a processing circuitry comprising a plurality of processing circuits, such as, e.g., any, some or all of the ones mentioned above. The processing circuitry may further perform data processing functions for inputting, outputting, and processing of data comprising data buffering and device control functions, such as call processing control, user interface control, or the like.

Finally, it should be understood that the present invention is not limited to the embodiments described above, but also relates to and incorporates all embodiments within the scope of the appended independent claims.

Claims

1 . Receiver device for a wireless communication system (300), the receiver device (100) comprising:

a receiver (104) configured to

receive a communication signal (CS);

a processor (102) configured to

identify a plurality of interferers n = Ι, .,. , Ν in the communication signal (CS);

at iteration i = 1 for interferer n = 1:

- compute an error covariance and a corresponding covariance estimate for each possible combination of power amplification, PA, and Precoding Matrix, PM, pairs, wherein the error covariance is computed based on the communication signal (CS),

- compute a metric for each error covariance,

- form a set of computed metrics and corresponding covariance estimates,

at iteration i = 1 for interferers n = 2, ... , N:

- compute an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal (CS) and the covariance estimates for previous interferers m = Ι, . , . , η - 1 at iteration i = 1,

- compute a metric for each error covariance,

- form a set of computed metrics and corresponding covariance estimates,

at iteration i = 2, .. . , 1 for the plurality of interferers n = 1, ... , N:

- compute an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal (CS) and the covariance estimates for the previous interferers m = 1, ... , n - 1 of the current iteration i and the covariance estimates for the subsequent interferers m = n + 1, ... , N of the previous iteration i - 1, - compute a metric for each computed error covariance,

- form a set of computed metrics and corresponding covariance estimates,

2 /;

compute the total interference covariance estimate R for the communication signal (CS) based on the covariance estimates for the plurality of interferers n = l, ... , N at iteration i = I > 1.

2. Receiver device (100) according to claim 1 , wherein the processor (102), before performing iteration i = 1, further is configured to

sort the plurality of interferers n = Ι, .,. , Ν in a descending order according to their respective received power.

3. Receiver device (100) according to claim 1 or 2, wherein the processor (102) further is configured to

sort the plurality of interferers n = 1, ... , N in a descending order based on the Frobenius norms of their respective channel estimates.

4. Receiver device (100) according to any of the preceding claims, wherein the likelihood for each computed metric is based on the energy content of the computed metric.

5. Receiver device (100) according to any of the preceding claims, wherein the computed metric is the Frobenius norm F of the error covariance.

6. Receiver device (100) according to claim 5, wherein the Frobenius norm F of the error covariance for interferer n for the PA and PM pair a_n, W_n is computed according to the equation:

F(a_n, W_n) = ||R^ - R_yy(a_n,W_n) ||²

where R^_y is the covariance of the communication signal (CS), and R_yy(a_n, W_n) is the total covariance estimate.

7. Receiver device (100) according to any of the preceding claims, wherein each first subset comprises the K smallest computed metrics in the set of computed metrics and corresponding covariance estimates.

8. Receiver device (100) according to any of the preceding claims, wherein the processor (102) further is configured to compute the total interference covariance estimate R for the communication signal (CS) by

9. Receiver device (100) according to claim 6, wherein the processor (102) further is configured to

compute the total interference covariance estimate R for the communication signal (CS) based on at least one metric and its corresponding covariance estimate in the common set of metrics and corresponding covariance estimates.

10. Receiver device (100) according to claim 9, wherein the processor (102) further is configured to

compute the total interference covariance estimated for the communication signal (CS) based on the metrics and the corresponding covariance estimates in the second subset.

1 1. Receiver device (100) according to claim 10, wherein the processor (102) further is configured to

compute the total interference covariance estimate R for the communication signal (CS) by combine the covariance estimates in the second subset with their respective likelihoods.

12. Receiver device (100) according to claim 10, wherein the processor (102) further is configured to

compute the likelihood for each one of the metrics in the second subset, wherein the likelihood is the Maximum Likelihood, ML, of the metric, compute the total interference covariance estimate R for the communication signal (CS) by combining the covariance estimates in the second subset with their respective likelihoods.

13. Receiver device (100) according to any of the preceding claims, wherein the covariance estimate for interferer n = 1 at iteration i = 1 is computed according to the equation:

∑_k ^K _{= 1} i({¾} l wi)

where

is weighted covariance estimate, and

Ryy(a_n,W_n) = a₀HoW₀Wo^HHi? + a_nH_nW_nW_n ^HH" + ^ R^ + R_UU

m<n

where

is the covariance for the serving cell, _NH_NW_NW„ H„ is the covariance estimate of interferer n of the current iteration, ∑_m<„R^ is the covariance estimate of previous interferers m = 1, . . . , n - l of the current iteration, and R_uu is the covariance of noise.

14. Receiver device (100) according to claim 13, wherein the total covariance estimate R_yy for the communication signal (CS) at iteration i = I > 1 is computed according to the equation:

Ryy (a_n, W_n) = a₀H₀W₀W₀ ^HH + a_nH_nW_nW_n ^HH" + ^ R + ^ R^_m + R_uu

m<n m>n

where ocoHoWoW^Ho is the covariance of the serving cell, _nH_nW_nW„ H„ is the covariance estimate of interferer n of current iteration,∑_m<„R^ is the covariance estimate of interferers m = Ι, . , . , η— 1 of current iteration, ∑_m>_n R^m is the covariance estimate of interferers m = n + 1, ... , N of the previous iteration, and R_uu is the covariance of noise.

15. Method for a wireless communication system (300), the method (200) comprising:

receiving (202) a communication signal (CS),

identifying (204) a plurality of interferers n = 1, ... , N in the communication signal (CS);

at iteration i = 1 for interferer n = 1:

- computing (206a) an error covariance and a corresponding covariance estimate for each possible combination of power amplification, PA, and Precoding Matrix, PM, pairs, wherein the error covariance is computed based on the communication signal (CS), - computing (206b) a metric for each error covariance, - forming (206c) a set of computed metrics and corresponding covariance estimates,

- selecting (206d) a first subset among the set of computed metrics and corresponding covariance estimates,

- computing (206e) a likelihood for each one of the computed metrics in the first subset, - combining (206f) the covariance estimates in the first subset with their respective likelihoods so as to obtain a covariance estimate for interferer n = 1 at iteration i = 1; at iteration i = 1 for interferers n = 2, ... , N:

- computing (208a) an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal (CS) and the covariance estimates for previous interferers m = 1, n - 1 at iteration i = 1,

- computing (208b) a metric for each error covariance,

- forming (208c) a set of computed metrics and corresponding covariance estimates,

- selecting (208d) a first subset among the set of computed metrics and corresponding covariance estimates,

- computing (208e) the likelihood for each one of the computed metrics in the first subset,

- combining (208f) the covariance estimates in the first subset with their respective likelihoods so as to obtain a covariance estimate for interferer n = 2, ... , N;

at iteration i = 2, . . . , 1 for the plurality of interferers n = 1, ... , N:

- computing (210a) an error covariance and a corresponding covariance estimate for each possible combination of PA and PM pairs, wherein the error covariance is computed based on the communication signal (CS) and the covariance estimates for the previous interferers m = Ι, . , . , η - 1 of the current iteration i and the covariance estimates for the subsequent interferers m = n + 1, ... , N of the previous iteration i - 1,

- computing (210b) a metric for each computed error covariance,

- forming (210c) a set of computed metrics and corresponding covariance estimates,

- selecting (21 Od) a first subset among the set of computed metrics and corresponding covariance estimates,

- computing (21 Oe) the likelihood for each one of the computed metrics in the first subset, - combining (21 Of) the covariance estimates in the first subset with their respective likelihoods so as to obtain at least one covariance estimate for interferer n = Ι, .,. , Ν at iteration i = 2, . . . , I;

computing (212) the total interference covariance estimate R for the communication signal (CS) based on the covariance estimates for the plurality of interferers n = 1, ... , N at iteration i = I > l.

16. Computer program with a program code for performing a method according to claim 15 when the computer program runs on a computer.