Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
RECEIVER AND PRECODING SYSTEM USING ASYMMETRIC IMPERFECT CHANNEL KNOWLEDGE
Document Type and Number:
WIPO Patent Application WO/2017/071731
Kind Code:
A1
Abstract:
The disclosure relates to a receiver (400), including: a decoder (305) comprising an adjustable receive filter (31 9) for decoding a receive signal (Y) received over a communication channel; and a first channel state information (CSI) handler (401) configured to provide the decoder (305) with receive-side channel state information (H̃) indicating a receive-side estimate of a channel state (H) of the communication channel and to provide the decoder (305) with transmit-side channel state information (Ĥ) indicating a transmit-side estimate of the channel state (Η) of the communication channel, wherein the decoder (305) is configured to adjust the receive filter (319) based on the transmit-side channel state information (Ĥ) if the transmit-side channel state information (Ĥ) is a less good estimate of the channel state (Η) of the communication channel than the receive-side channel state information (H̃). The disclosure further relates to a precoding system (500) including such receiver (400).

Inventors:
BENAMMAR MERYEM (DE)
ESTELLA AGUERRI INAKI (DE)
ZAIDI ABDELLATIF (DE)
Application Number:
PCT/EP2015/074744
Publication Date:
May 04, 2017
Filing Date:
October 26, 2015
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
HUAWEI TECH CO LTD (CN)
BENAMMAR MERYEM (DE)
ESTELLA AGUERRI INAKI (DE)
ZAIDI ABDELLATIF (DE)
International Classes:
H04L25/03; H04B7/06
Other References:
BERGEL ITSIK ET AL: "Dirty paper coding with partial channel state information", 2014 IEEE 15TH INTERNATIONAL WORKSHOP ON SIGNAL PROCESSING ADVANCES IN WIRELESS COMMUNICATIONS (SPAWC), IEEE, 22 June 2014 (2014-06-22), pages 334 - 338, XP032672241, DOI: 10.1109/SPAWC.2014.6941718
UGURLU UMUT ET AL: "A Novel Joint Precoder and Receiver Design with Imperfect CSI for Multi-User MIMO Downlink", 2014 IEEE 80TH VEHICULAR TECHNOLOGY CONFERENCE (VTC2014-FALL), IEEE, 14 September 2014 (2014-09-14), pages 1 - 5, XP032694788, DOI: 10.1109/VTCFALL.2014.6965827
DVORAKOVA INNA ET AL: "MMSE precoder for multipath channels with imperfectly known state information", 2015 38TH INTERNATIONAL CONFERENCE ON TELECOMMUNICATIONS AND SIGNAL PROCESSING (TSP), IEEE, 9 July 2015 (2015-07-09), pages 195 - 199, XP032792283, DOI: 10.1109/TSP.2015.7296251
FISCHER R F H ET AL: "Tomlinson-harashima precoding in space-time transmission for low-rate backward channel", BROADBAND COMMUNICATIONS, 2002. ACCESS, TRANSMISSION, NETWORKING. 2002 INTERNATIONAL ZURICH SEMINAR ON FEB. 19-21, 2002, PISCATAWAY, NJ, USA,IEEE, 19 February 2002 (2002-02-19), pages 7_1 - 7_6, XP010584387, ISBN: 978-0-7803-7257-3
MAX H. M. COSTA: "Writing on dirty paper", IEEE TRANSACTIONS ON INFORMATION THEORY, vol. 29, 1983, pages 439 - 441
URI EREZ; SHLOMO SHAMAI; RAM ZAMIR: "Capacity and Lattice Strategies for Canceling Known Interference", IEEE TRANS. ON INFO. THEORY, vol. 51, no. 11, November 2005 (2005-11-01), pages 3820 - 3833, XP011141518, DOI: doi:10.1109/TIT.2005.856935
M. TOMLINSON: "New automatic equalizer employing modulo arithmetic", IEEE ELECTRONIC LETTER, vol. 07, 1971, pages 138 - 139
H. HARASHIMA; H. MIYAKAWA: "Matched transmission technique for channels with intersymbol interference", IEEE TRANSACTIONS ON COMMUNICATIONS, vol. 20, 1972, pages 774 - 780, XP000990996, DOI: doi:10.1109/TCOM.1972.1091221
H. MIYAKAWA; H. HARASHIMA: "Information transmission rate in matched transmission systems with peak transmitting power limitation", NAT. CONF. REC., INSTITUTE OF ELECTRONIC AND INFORMATION COMMUNICATION ENGINEERING OF JAPAN, August 1969 (1969-08-01), pages 1268
Attorney, Agent or Firm:
KREUZ, Georg (DE)
Download PDF:
Claims:
CLAIMS:

1 . A receiver (400), comprising: a decoder (305) comprising an adjustable receive filter (31 9) for decoding a receive signal (Y) received over a communication channel; and a first channel state information (CSI) handler (401 ) configured to provide the decoder (305) with receive-side channel state information indicating a receive-side

estimate of a channel state (Η) of the communication channel and to provide the decoder

(305) with transmit-side channel state information indicating a transmit-side estimate

of the channel state (Η) of the communication channel, wherein the decoder (305) is configured to adjust the receive filter (319) based on the transmit-side channel state information if the transmit-side channel state

information is a less good estimate of the channel state (Η) of the communication

channel than the receive-side channel state information

2. The receiver (400) of claim 1 , wherein the receive-side channel state information and the transmit-side

channel state information are asymmetric imperfect estimates of the channel state (Η) of the communication channel.

3. The receiver (400) of claim 1 or 2, wherein the decoder (305) is configured to use a lattice-based decoding scheme to decode the receive signal (Y).

4. The receiver (400) of one of the preceding claims, wherein the decoder (305) is configured to change the adjusting of the receive filter

(31 9) from an adjustment based on the receive-side channel state information to an

adjustment based on the transmit-side channel state information based on a control

signal (402).

5. A precoding system (500) for precoding at least one message (W) intended for transmission over a communication channel (203) in the presence of interference (S), the precoding system (500) comprising: a transmitter (501 ), comprising: a precoder (201 ) comprising an adjustable transmit filter (209) for precoding the at least one message (W) and the interference (S) that is known to the transmitter (501 ) to generate at least one input signal (X) intended for transmission over the communication channel (203) ; and a second channel state information (CSI) handler (503) configured to provide the precoder (201 ) with the transmit-side channel state information

wherein the precoder (201 ) is configured to adjust the transmit filter (209) based on the transmit-side channel state information and

a receiver (400) according to one of claims 1 to 4.

6. The precoding system (500) of claim 5, wherein the precoder (201 ) is configured to use a lattice-based precoding scheme to precode the at least one message (W) and the interference (S).

7. The precoding system (500) of claim 5 or 6, wherein the transmitter (501 ) comprises a Signal-to-Noise Ratio (SNR) estimator configured to estimate an SNR (812) of the at least one input signal (X), wherein the precoder (201 ) is configured to adjust the transmit filter (209) based on the estimated SNR (812) of the at least one input signal (X), and wherein the decoder (305) is configured to adjust the receive filter (319) based on the estimated SNR (812) of the at least one input signal (X).

8. The precoding system (500) of claim 7, wherein the decoder (305) is configured to adjust the receive filter (319) based on the transmit-side channel state information ( H ) if the estimated SNR (812) of the at least one input signal (X) is below a threshold (71 0) and to adjust the receive filter (31 9) based on the receive-side channel state information ( H ) if the estimated SNR (81 2) of the at least one input signal (X) is above the threshold (71 0).

9. The precoding system (500) of claim 8, wherein the transmitter (501 ) comprises a threshold computation unit (700), configured to compute the threshold (71 0) based on a first transmission rate (R^ that is based on statistics of the receive-side channel state information and a second transmission rate (R2) that is based on statistics of the transmit-side channel state information ( H ).

10. The precoding system (500) of claim 9, wherein the threshold computation unit (700) is configured: to compute the first transmission rate (Ri) based on an interference covariance matrix (702), a noise covariance matrix (704), a receive-side estimation error covariance (706), and the transmit-side estimation error covariance (708) and to compute the second transmission rate (R2) based on the interference covariance matrix (702), the noise covariance matrix (704), the receive-side estimation error covariance (706) and the transmit-side estimation error covariance (708).

1 1 . The precoding system (500) of one of claims 5 to 10,

wherein the precoder (201 ) comprises a plurality of adjustable transmit filters (Η,) for precoding a respective plurality of messages (Wj) and the known interference (S) to generate a respective plurality of input signals (Xj) intended for transmission to a respective plurality of users (UE1 , UE2, UE3, UEK), wherein the precoder (201 ) is configured to precode a first input signal (Xk) of the plurality of input signals (Xj) against a second plurality of further input signals (Xj<k) of the plurality of input signals (Xj) forming the known interference (S).

12. The precoding system (500) of claim 1 1 , wherein the precoder (201 ) is configured to identify the second plurality of further input signals (Xj<k) based on their power levels; and wherein the precoder (201 ) is configured to treat the remaining input signals (Xj>k) of the plurality of input signals (Xj) as additional noise.

13. The precoding system (500) of one of claims 5 to 12, wherein the first channel state information (CSI) handler (401 ) is configured to provide the receive-side channel state information ( H ) based on channel estimation and to provide the transmit-side channel state information based on a quantization step,

and wherein the second channel state information (CSI) handler (503) provides the transmit-side channel state information based on feedback from the receiver.

14. A decoding method (1300), comprising: decoding (1301 ) a receive signal (Y) received over a communication channel by using an adjustable receive filter (319); providing (1 302) receive-side channel state information ( H ) indicating a receive- side estimate of a channel state (Η) of the communication channel and providing transmit- side channel state information indicating a transmit-side estimate of the channel state (Η) of the communication channel; and adjusting (1 303) the receive filter (31 9) based on the transmit-side channel state information if the transmit-side channel state information ) is a less good estimate

of the channel state (Η) of the communication channel than the receive-side channel state information

15. The decoding method (1300) of claim 14, comprising: changing the adjusting (1 303) of the receive filter (319) from an adjustment based on the receive-side channel state information ) to an adjustment based on the transmit-side channel state information based on a transmit-side condition, in particular a transmit-side SNR (812) crossing a threshold (71 0).

Description:
RECEIVER AND PRECODING SYSTEM USING ASYMMETRIC IMPERFECT

CHANNEL KNOWLEDGE

TECHNICAL FIELD

The present disclosure relates to a receiver and a precoding system that are using asymmetric imperfect channel knowledge. In particular, the disclosure relates to filter design techniques for lattice-based Dirty-Paper Coding (DPC) with asymmetric imperfect channel knowledge.

BACKGROUND

Interference is one of the most limiting factors of communication in networks. In certain cases, its effect can be mitigated through some precoding techniques, which can be categorized into linear and non-linear methods. Linear methods rely on some forms of Zero-Forcing or variants of it, and are generally suboptimal from a transmission rate or throughput view-point. Non-linear methods rely on Costa's well known dirty paper coding (DPC) according to "Max H. M. Costa, Writing on dirty paper, IEEE Transactions on Information Theory, Vol. 29, 1983, pp. 439-441 " and generally achieve better

performance, comparatively. The classical analysis of DPC requires perfect channel state information (CSI), at both transmitter and receiver sides, which is difficult to obtain in many scenarios. In the case of imperfect CSI, most existing works deal with the case in which the transmitter and receiver share the same imperfect CSI knowledge. In the case in which the transmitter and receiver have distinct imperfect CSI, referred hereinafter to as the case of "asymmetric imperfect channel knowledge", both the analysis and design of DPC precoders and decoders are generally challenging. Imperfections in the knowledge of the CSI can be caused by errors in the estimation or channel quantization over feedback links. SUMMARY

It is the object of the invention to provide a concept for an improved design of DPC pre- and decoders in the case of asymmetric imperfect channel knowledge. This object is achieved by the features of the independent claims. Further implementation forms are apparent from the dependent claims, the description and the figures. As mentioned above, the present disclosure relates to the scenario of asymmetric imperfect channel knowledge as illustrated in Fig. 1 where the underlying system model 100 is depicted. A basic idea of the invention is to use two lattice-based filter design techniques for DPC precoding in this case of asymmetric imperfect channel knowledge. The first method is based on the transmitter and receiver each utilizing the best CSI that they have. This method can be employed irrespective of the relative quality of the transmit CSI and the receive CSI. The second method is tailored specifically for the case in which the CSI at the receiver is of better quality than that at the transmitter and the receiver knows the actual channel CSI that the transmitter has (e.g., quantized feedback CSI over error-free feedback link). In this case, it is shown below that, for certain regimes, it is more beneficial that the receiver voluntarily ignores the better CSI that it has, and, instead, utilizes the less good transmit CSI for the design of its filter. Another idea is to selectively tune the filters according to the appropriate technique depending on the operating system parameters. The methods as presented below are based on lattice coding and prove to be more robust to channel uncertainty.

In order to describe the invention in detail, the following terms, abbreviations and notations will be used:

CSI: Channel State Information

THP: Tomlinson-Harashima Precodi

DPC: Dirty Paper Coding

Tx: Transmitter

Rx: Receiver

Systems, devices and methods according to the disclosure may be applied in MIMO communication systems that are based on a system model 100 as depicted in Fig. 1 and described in the following.

A transmitter (Tx) communicates with a receiver (Rx) in presence of some additive interference S. The interfering signal S is assumed to be Gaussian and is known beforehand, or non-causally, to Tx but not to Rx. W is the message to be transmitted from Tx to Rx; and X is the corresponding input signal. Tx is equipped with N t antennas, and Rx is equipped with N r antennas. The channel is assumed to be flat fading AWGN

(although more general cases are possible as detailed below). The channel matrix H is known only imperfectly, at both Tx and Rx. More precisely, Tx has only an estimate

8 is some estimation error with covariance matrix K £ . Similarly, Rx has only an estimate is some estimation error with covariance matrix The transmission is subjected to input power constraint where

K x is the input covariance matrix and Tr' is the trace operator. The interfering signal S has covariance matrix K s such that Tr (K s ) = Q. Z is some additive white Gaussian noise with covariance matrix K z = N I Nr . The SNR is defined as: The input-output

relationship is given by:

Note that Figure 1 illustrates a simple model in which the interference is affected by the same channel matrix as the input signal. This is in line with the main motivation which is precoding for downlink cellular systems (see Figure 12 below). The analysis here extends relatively easily to the case in which the interference is affected by a different channel.

Systems, devices and methods according to the disclosure may apply lattice-based precoding and decoding as illustrated in Figure 2 and described in the following. Figure 2 illustrates a system model for a communication system with implementation of Lattice- based DPC in the case of perfect CSI at Tx and Rx.

In the case of perfect CSI at both Tx and Rx, i.e., an efficient lattice- based implementation of DPC, which performs close to optimal for large dimensions, is shown in Figure 2. Lattice strategies for cancelling known interference in a single user channel were studied by Erez et al. in "Uri Erez, Shlomo Shamai (Shitz) and Ram Zamir, "Capacity and Lattice Strategies for Canceling Known Interference, IEEE Trans. On Info. Theory, Vol. 51 , No. 1 1 , Nov. 2005, pp. 3820-3833".

In the system of Fig. 2, the input signal X is obtained as:

where c w is a symbol or codeword that is associated through one-to-one mapping with the message w, A is a given lattice of dimension n (e.g., cubic lattice z n , Hexagonal lattice A 2 , the Checkerboard lattice D 4 and with normalized second moment equal to P x , the operation "mod" denotes the modulo reduction, d is a dither chosen to be uniformly distributed over the Voronoi cell of lattice Λ , and A is some filter whose choice is discussed below.

The receiver Rx decodes the message W from its output signal Y using standard modulo- lattice reduction operation, as: where A is used as the receive filter as well.

The performance evaluation of this lattice based DPC scheme is given by the following per-dimension throughput that is given by: where G(A) is the second moment of the lattice Λ and z eii is the effective noise at the receiver (composed of the standard noise z and some self noise), and given by:

The joint design of the transmit filter and the receive filter is pivotal for the overall performance of the system.

One possible choice, sometimes referred to as Zero-Forcing DPC (ZF-DPC) is obtained by setting A = I , where / denotes the identity matrix. ZF-DPC is asymptotically (in the dimension of the lattice) optimal at high signal-to-noise ratio (SNR). At small and moderate SNR, however, a better solution, sometimes referred to as minimum mean square error DPC (MMSE-DPC), is obtained by setting:

This choice removes almost completely the effect of the interference S on the

communication and minimizes thus the covariance of the equivalent noise that maximizes the throughput. The so-called Tomlinson-Harashima precoding (THP) as described in "M. Tomlinson, New automatic equalizer employing modulo arithmetic, IEEE Electronic Letter, Vol. 07, 1971 , pp. 138-139", "H. Harashima and H. Miyakawa, Matched transmission technique for channels with intersymbol interference, IEEE Transactions on Communications, Vol. 20, 1972, pp. 774-780" and "H. Miyakawa and H. Harashima, Information transmission rate in matched transmission systems with peak transmitting power limitation, Nat. Conf. Rec, Institute of Electronic and Information Communication Engineering of Japan, Aug. 1969, p. 1268" can be obtained as a special case by choosing Λ to be the cubic lattice z n and setting n = 1.

If the dimension n of the lattice is very large, i.e., with sphere lattices and

one recovers the performance of optimal DPC with random codes.

As seen from the above, in case of perfect CSI at both Tx and Rx, the optimal choice of the transmit and receive filters is such that they are identical.

Systems, devices and methods according to the disclosure may be applied in the case in which the transmitter Tx and receiver Rx have distinct imperfect CSI or asymmetric imperfect channel knowledge as illustrated in Figure 3 and described in the following.

The precoding and decoding are different from those of perfect CSI (see Figure 2) in that two distinct filters are utilized: a transmit filter A and a receiver filter B and in that the precoding is performed as follows: while the decoding is described by:

In the case of imperfect asymmetric CSI, the effective noise is given by:

- some self-noise , and

- some residual interference that did not intervene in the case of

perfect CSI. The optimal choice of the transmit and receive filters A H)and B H) is challenging since the effective noise is not Gaussian and is affected by the non-linear modulo-reduction operation, and also, since the filters are constrained to depend on two distinct channel estimates. In DPC-type precoding techniques, the optimal choice of the receive filter B depends heavily on that, at the transmitter, i.e. A.

As such, because of the asymmetric channel knowledge, an important dilemma or tension at the receiver arises between: i) choosing its filter B (H) in accordance with that, A(H) , of the transmitter, or ii) exploiting all the channel knowledge it has, which may be of better quality than that at the transmitter (for example in the case in which Rx has perfect CSI, i.e., H = H and Tx has only partial CSI).

In the case in which Tx and Rx share the same imperfect estimate of the channel, the Lattice-based DPC coding approach can be extended relatively easily to the vector multi- antenna case.

According to a first aspect, the invention relates to a receiver, comprising: a decoder comprising an adjustable receive filter for decoding a receive signal received over a communication channel; and a first channel state information (CSI) handler configured to provide the decoder with receive-side channel state information indicating a receive-side estimate of a channel state of the communication channel and to provide the decoder with transmit-side channel state information indicating a transmit-side estimate of the channel state of the communication channel, wherein the decoder is configured to adjust the receive filter based on the transmit-side channel state information if the transmit-side channel state information is a less good estimate of the channel state of the

communication channel than the receive-side channel state information.

Adjusting the receive filter based on the transmit-side channel state information if the transmit-side channel state information is a less good estimate of the channel state provides the advantage of better robustness to imperfections in the channel knowledge at transmit and receive sides; higher throughput in comparison with standard THP. No extra signaling between the transmitter and the receiver is required, thus complexity is reduced. Preferably, a measure for the quality of an estimate of the channel state H is the error in estimating the true channel H from the observed channel. That is, the transmit-side channel state information ) is a less good estimate of the channel state (H) of the communication channel than the receive-side channel state information if the error in estimating Η from is larger than that of estimating Η from , i.e.,

wherein is the minimum mean square value (MMSE) in estimating H from and is the minimum mean square value (MMSE) in estimating H from

In a first possible implementation form of the receiver according to the first aspect, the receive-side channel state information and the transmit-side channel state information are asymmetric imperfect estimates of the channel state of the communication channel.

This provides the advantage that an improved design of DPC pre- and decoders in the case of asymmetric imperfect channel knowledge can be implemented. In a second possible implementation form of the receiver according to the first aspect as such or according to the first implementation form of the first aspect, the decoder is configured to use a lattice-based decoding scheme to decode the receive signal.

This provides the advantage of low complexity implementation by using structured coding, i.e. lattices.

In a third possible implementation form of the receiver according to the first aspect as such or according to any of the preceding implementation forms of the first aspect, the decoder is configured to change the adjusting of the receive filter from an adjustment based on the receive-side channel state information to an adjustment based on the transmit-side channel state information based on a control signal.

This provides the advantage that the receive filter design can be flexibly changed based on the control signal.

According to a second aspect, the invention relates to a precoding system for precoding at least one message intended for transmission over a communication channel in the presence of interference, the precoding system comprising: a transmitter, comprising: a precoder comprising an adjustable transmit filter for precoding the at least one message and the interference that is known to the transmitter to generate at least one input signal intended for transmission over the communication channel; and a second channel state information handler configured to provide the precoder with the transmit-side channel state information), wherein the precoder is configured to adjust the transmit filter based on the transmit-side channel state information; and a receiver according to the first aspect as such or any of the preceding implementation forms of the first aspect. This provides the advantage of joint design of DPC transmit and receive filters, thereby reducing complexity.

In a first possible implementation form of the precoding system according to the second aspect, the precoder is configured to use a lattice-based precoding scheme to precode the at least one message and the interference.

This provides the advantage of low complexity implementation by using structured coding, i.e. lattices. In a second possible implementation form of the precoding system according to the second aspect as such or according to the first implementation form of the second aspect, the transmitter comprises a Signal-to-Noise Ratio (SNR) estimator configured to estimate an SNR of the at least one input signal, the precoder is configured to adjust the transmit filter based on the estimated SNR of the at least one input signal, and the decoder is configured to adjust the receive filter based on the estimated SNR of the at least one input signal.

This provides the advantage that a transmission rate through the system can be increased when switching the receive filter design method based on the SNR of the input signal.

In a third possible implementation form of the precoding system according to the second implementation form of the second aspect, the decoder is configured to adjust the receive filter based on the transmit-side channel state information if the estimated SNR of the at least one input signal is below a threshold and to adjust the receive filter based on the receive-side channel state information if the estimated SNR of the at least one input signal is above the threshold. This provides the advantage that the transmission rate can be optimized when the switching is based on an SNR threshold, see Figures 6 and 9 to 1 1 .

In a fourth possible implementation form of the precoding system according to the third implementation form of the second aspect, the transmitter comprises a threshold computation unit, configured to compute the threshold based on a first transmission rate that is based on statistics of the receive-side channel state information and a second transmission rate that is based on statistics of the transmit-side channel state information. This provides the advantage that an optimal transmission rate can be achieved.

In a fifth possible implementation form of the precoding system according to the fourth implementation form of the second aspect, the threshold computation unit is configured: to compute the first transmission rate based on an interference covariance matrix, a noise covariance matrix, a receive-side estimation error covariance, and the transmit-side estimation error covariance and to compute the second transmission rate based on the interference covariance matrix, the noise covariance matrix, the receive-side estimation error covariance and the transmit-side estimation error covariance. This provides the advantage that such statistical evaluation provides a high accuracy.

In a sixth possible implementation form of the precoding system according to the second aspect as such or according to any of the preceding implementation forms of the second aspect, the precoder comprises a plurality of adjustable transmit filters for precoding a respective plurality of messages and the known interference to generate a respective plurality of input signals intended for transmission to a respective plurality of users, the precoder is configured to precode a first input signal of the plurality of input signals against a second plurality of further input signals of the plurality of input signals forming the known interference.

This provides the advantage that the precoding system can be applied in cellular communication systems with multiple users, e.g. as described below with respect to Figure 12. In a seventh possible implementation form of the precoding system according to the sixth implementation form of the second aspect, the precoder is configured to identify the second plurality of further input signals based on their power levels; and the precoder is configured to treat the remaining input signals of the plurality of input signals as additional noise.

This provides the advantage that robust precoding can be implemented in multi-user communication systems.

In an eighth possible implementation form of the precoding system according to the second aspect as such or according to any of the preceding implementation forms of the second aspect, the first channel state information (CSI) handler is configured to provide the receive-side channel state information based on channel estimation and to provide the transmit-side channel state information based on a quantization step, and the second channel state information (CSI) handler provides the transmit-side channel state information based on feedback from the receiver. This provides the advantage that the first CSI handler can easily acquire both, receive- side and transmit-side CSI.

According to a third aspect, the invention relates to a decoding method, comprising:

decoding a receive signal received over a communication channel by using an adjustable receive filter; providing receive-side channel state information indicating a receive-side estimate of a channel state of the communication channel and providing transmit-side channel state information indicating a transmit-side estimate of the channel state of the communication channel; adjusting the receive filter based on the transmit-side channel state information if the transmit-side channel state information is a less good estimate of the channel state of the communication channel than the receive-side channel state information.

Adjusting the receive filter based on the transmit-side channel state information if the transmit-side channel state information is a less good estimate of the channel state provides the advantage of better robustness to asymmetry in the channel knowledge at transmit and receive sides; higher throughput in comparison with standard THP. No extra signaling between the transmitter and the receiver is required, thus complexity is reduced.

In a first possible implementation form of the decoding method according to the third aspect, the decoding method comprises: changing the adjusting of the receive filter from an adjustment based on the receive-side channel state information to an adjustment based on the transmit-side channel state information based on a transmit-side condition, in particular a transmit-side SNR crossing a threshold.

This provides the advantage of high flexibility by choosing between separate and joint design of DPC transmit and receive filters.

According to a fourth aspect, the invention relates to a Lattice-based transmit- and receive-filter design method for DPC with asymmetric imperfect channel knowledge that is based on each side utilizing the best CSI that it has.

This provides the advantage that this method can be employed irrespective to the relative quality of the transmit and receive CSI.

In a first possible implementation form of the filter design method according to the fourth aspect, the method comprises: utilizing the less good transmit CSI for the receive-filter design.

This provides an improvement of the Lattice-based transmit- and receive-filter design technique that is based on the receiver voluntarily ignoring part of the CSI that it has. This method is tailored specifically for scenarios in which the CSI at Rx is of better quality than that at Tx and Rx knows the actual channel CSI that Tx has.

In a second possible implementation form of the filter design method according to the fourth aspect as such or the first implementation form of the fourth aspect, the method comprises computing the SNR threshold and selecting the appropriate filter design technique among the two methods presented below based on a comparison with the operating SNR value.

This provides the advantage of flexibility in the filter design method, the solution allows for separate (Methodi described below) and joint (Method 2 described below) design of DPC transmit- and receive-filters.

BRIEF DESCRIPTION OF THE DRAWINGS Further embodiments of the invention will be described with respect to the following figures, in which: Fig. 1 shows a schematic diagram illustrating the system model 100 of a MIMO

communication system;

Fig. 2 shows a block diagram illustrating the system model 200 of a communication system with implementation of Lattice-based DPC in the case of perfect CSI at transmitter and receiver;

Fig. 3 shows a block diagram illustrating the system model 300 of a communication system with implementation of Lattice-based DPC in the case of asymmetric imperfect channel knowledge at transmitter and receiver;

Fig. 4 shows a block diagram illustrating a receiver 400 using asymmetric imperfect channel knowledge according to an implementation form; Fig. 5 shows a block diagram illustrating a precoding system 500 using asymmetric imperfect channel knowledge according to an implementation form;

Fig. 6 shows an exemplary performance diagram 600 illustrating transmission rates of a precoding system according to Figure 5 when using first and second filter design methods;

Fig. 7 shows a block diagram illustrating a threshold computation unit 700 according to an implementation form; Fig. 8 shows a functional diagram illustrating a selection of the appropriate filter design method of a precoding system according to Figure 5 according to an implementation form;

Fig. 9 shows an exemplary performance diagram 900 illustrating comparison of the two filter design methods when using sphere lattices;

Fig. 10 shows an exemplary performance diagram 1000 illustrating comparison of the two filter design methods when using Gosset lattices;

Fig. 1 1 shows an exemplary performance diagram 1 100 illustrating comparison of the two filter design methods when using Cubic lattices; Fig. 12 shows a schematic diagram illustrating the downlink communication channel of a cellular communication system 1200; and

Fig. 13 shows a schematic diagram illustrating a decoding method 1300 using asymmetric imperfect channel knowledge according to an implementation form.

DETAILED DESCRIPTION OF EMBODIMENTS

In the following detailed description, reference is made to the accompanying drawings, which form a part thereof, and in which is shown by way of illustration specific aspects in which the disclosure may be practiced. It is understood that other aspects may be utilized and structural or logical changes may be made without departing from the scope of the present disclosure. The following detailed description, therefore, is not to be taken in a limiting sense, and the scope of the present disclosure is defined by the appended claims.

It is understood that comments made in connection with a described method may also hold true for a corresponding device or system configured to perform the method and vice versa. For example, if a specific method step is described, a corresponding device may include a unit to perform the described method step, even if such unit is not explicitly described or illustrated in the figures. Further, it is understood that the features of the various exemplary aspects described herein may be combined with each other, unless specifically noted otherwise.

Fig. 4 shows a block diagram illustrating a receiver 400 using asymmetric imperfect channel knowledge according to an implementation form.

The receiver 400 includes a decoder 305, e.g. a decoder 305 as described above with respect to Figure 3, and a channel state information (CSI) handler 401 . The decoder 305 includes an adjustable receive filter 31 9 for decoding a receive signal Y received over a communication channel, e.g. a communication channel 203 as described above with respect to Figures 2 and 3. The channel state information handler 401 provides the decoder 305 with receive-side channel state information H indicating a receive-side estimate of a channel state Η of the communication channel and further provides the decoder 305 with transmit-side channel state information H indicating a transmit-side estimate of the channel state Η of the communication channel. The channel state Η may correspond to the perfect channel state information as described above with respect to Figure 2. The decoder 305 adjusts the receive filter 31 9 based on the transmit-side channel state information H if the transmit-side channel state information is a less good estimate of the channel state Η of the communication channel than the receive-side channel state information . This adjusting technique is also referred hereinafter as "Method 2" or "Filter Design Method 2".

The channel state information (CSI) handler 401 may decide which of the transmit-side channel state information and the receive-side channel state information is a better

estimate depending on a quality of both CSI values and . For example, a quality of

both estimates may be checked based on the statistics of both CSI values and

e.g. based on interference covariance, noise covariance, receiver estimation covariance and transmit estimation covariance values as described below with respect to Figures 7 and 8. This decision may be performed without an explicit knowledge of the perfect channel state information Η as described above with respect to Figure 2.

The receive-side channel state information and the transmit-side channel state

information may be asymmetric imperfect estimates of the channel state Η of the communication channel. The decoder 305 may use a lattice-based decoding scheme to decode the receive signal Y, e.g. as described above with respect to Figures 1 to 3.

The decoder 305 may change the adjusting the receive filter 319 from an adjustment based on the receive-side channel state information to an adjustment based on the

transmit-side channel state information based on a control signal 402. The control signal 402 may provide the decoder 305 with the respective CSI values .

Fig. 5 shows a block diagram illustrating a precoding system 500 using asymmetric imperfect channel knowledge according to an implementation form. The precoding system 500 may be used for precoding at least one message W intended for transmission over a communication channel 203 in the presence of interference S, e.g. according to the description of Figure 3.

The precoding system 500 includes a transmitter 501 and a receiver 400, e.g. a receiver 400 as described above with respect to Figure 4, which are coupled over a communication channel 203, e.g. a communication channel 203 as described above with respect to Figures 2 and 3.

The transmitter 501 includes a precoder 201 , e.g. a precoder 201 as described above with respect to Figures 2 and 3, and a second CSI handler 503. The precoder 201 includes an adjustable transmit filter 209 for precoding the at least one message W and the interference S that is known to the transmitter 501 to generate at least one input signal X intended for transmission over the communication channel 203. The second CSI handler

503 provides the precoder 201 with the transmit-side channel state information The precoder 201 adjusts the transmit filter 209 based on the transmit-side channel state information

The precoder 201 may use a lattice-based precoding scheme to precode the at least one message W and the interference S, e.g. as described above with respect to Figures 1 to 3.

The transmitter 501 may include a Signal-to-Noise Ratio (SNR) estimator for estimating an SNR of the input signal X. The precoder 201 may adjust the transmit filter 209 based on the estimated SNR of the input signal X, e.g. as described below with respect to Figure 8. The decoder 305 may adjust the receive filter 319 based on the estimated SNR of the input signal X.

The decoder 305 may adjust the receive filter 319 based on the transmit-side channel state information if the estimated SNR of the input signal X is below a threshold, e.g. a

threshold 71 0 as computed according to the description of Figures 7 and 8 shown below. This adjusting technique is also referred hereinafter as "Method 2" or "Filter Design Method 2".

In the opposite case, i.e., if the estimated SNR of the input signal X, e.g. the estimated SNR 81 2 as depicted in Figure 8, is above the threshold 710, the decoder 305 may adjust the receive filter 31 9 based on the receive-side channel state information This

adjusting technique is also referred hereinafter as "Method 1 " or "Filter Design Method 1 ".

The transmitter 501 may include a threshold computation unit 700, e.g. as described below with respect to Figure 7, for computing the threshold 71 0 based on a first transmission rate that is based on statistics of the receive-side channel state

information and a second transmission rate R 2 that is based on statistics of the transmit-side channel state information H .

The threshold computation unit 700 may compute both, the first transmission rate Ri and the second transmission rate R 2 based on an interference covariance matrix 702, a noise covariance matrix 704, a receive-side estimation error covariance 706, and the transmit- side estimation error covariance 708, e.g. as described below with respect to Figure 7.

The precoder 201 may include a plurality of adjustable transmit filters Η,, e.g. as described below with respect to Figure 1 2, for precoding a respective plurality of messages W j and the known interference S to generate a respective plurality of input signals X j intended for transmission to a respective plurality of users UE1 , UE2, UE3, UEK, see description below with respect to Figure 12. The precoder 201 may precode a first input signal X k of the plurality of input signals X j against a second plurality of further input signals of the plurality of input signals X j forming the known interference S, e.g. as described below with respect to Figure 1 2.

The precoder 201 may identify the second plurality of further input signals X j<k based on their power levels; and may treat the remaining input signals X j>k of the plurality of input signals X j as additional noise, e.g. as described below with respect to Figure 12.

The first CSI handler 401 may provide the receive-side channel state information

based on channel estimation and may provide the transmit-side channel state information based on a quantization step. The second CSI handler 503 may provide the transmit- side channel state information based on feedback from the receiver 400.

In an exemplary implementation of the precoding system 500, the receiver quantizes the CSI it has ( based on the estimation it performed) and then feeds it back to the transmitter. As such, the CSI that is acquired at the transmitter is a quantized version of the CSI at the receiver, and thus, it is known to the receiver (since it is the receiver that performs this quantization). This means, that the transmitter does not use channel reciprocity, and thus the precoding system may be mainly used with FDD systems.

In the following, two filter design methods are presented that are referred to as "Method 1 " and "Method 2" as mentioned above. Both methods take into account the asymmetric channel knowledge by allowing the transmit 209 and receive 319 filters to be distinct. In the first method ("Method 1 " below), each of the Tx 501 and Rx 400 designs its filter 209, 319 based on the best CSI that is available to it. That is, Tx 501 designs its filter 209 as a function of the transmit-side channel estimate and Rx 400 designs its filter 319 as a function of the receive-side channel estimate

The second method ("Method 2" below) is tailored specifically for the case in which the channel estimate at Rx 400 is of better quality than that at Tx 501 and the receiver 400 knows the actual channel CSI that Tx 501 has. In this case, for some regimes, it may be more beneficial that Rx 400 voluntarily ignores the good channel estimate that it has, and, instead, utilizes the less good transmit-side for the design of its filter.

Evaluating and comparing the transmission rates that are offered by these two methods, it has been shown (see the performance diagrams below with respect to Figures 9 to 1 1 ) that none of the two methods outperforms the other in general. Which method leads to better rates depends on the system parameters, i.e., the input power, interference power, noise power, and channel statistics. Based on this observation, a selection for choosing the DPC filters may be based either on Method 1 or Method 2, depending on the system parameters, as described below with respect to Figure 8. In the following sections, Method 1 is described which performs precoding by using the best available CSI.

In Method 1 , the precoding is performed as follows:

While the decoding is described by:

With this Method 1 , each of the Tx 501 and Rx 400 designs its filter based on the best channel estimate that it has. That is, Tx 501 designs its filter A 209 as if the true channel were i.e., as:

Similarly, Rx 400 designs its filter B 319 as if as if the true channel were H, i.e., as:

In the above equations, is the covariance of the equivalent noise with this Method 1 ,

given by:

This Method 1 leads to the following transmission rate:

where is the covariance matrix of the equivalent noise at the receiver 400. This covariance matrix is given by:

With

and

In the following sections, Method 2 is described which performs precoding by using the transmit CSI.

In Method 2, the precoding may be performed as follows:

While the decoding is described by: As mentioned previously, this Method 2 is tailored specifically for the case in which the channel estimate at Rx 400 is of better quality than that at Tx 501 and the receiver 400 knows the actual channel CSI that Tx 501 has. This is the case when the CSI is measured at Rx 400 and then quantized and sent over essentially error-free feedback links from Rx 400 to Tx 501 . In what follows it is assumed that the CSI at Tx 501 and Rx 400 are such that: where is the estimation error at Rx 400 and £ w is some quantization noise (e.g., over some feedback link). The covariance matrices of these noises are related through:

With this Method 2, the receiver 400 designs its receive filter 319 in accordance with that of the transmitter 501 and ignores his better channel knowledge:

In the above equations is the covariance of the equivalent noise with this Method 2, given by:

This Method 2 leads to the following transmission rate:

where:

With

and

The asymmetric imperfect CSI knowledge at Tx 501 and Rx results in an effective noise which is given by:

The MMSE-DPC transmit and receive filters are "matched" in the perfect CSI case, in the sense that: This choice also minimizes the covariance of the effective noise

In the case with asymmetric imperfect CSI, minimizing the covariance of this effective noise under the constraint that the receive filter 319 is constrained to depend only on the CS that is available at Rx 400 and the transmit filter 209 is constrained to depend only on the CSI H that is available at Tx 501 is not easy. One consequence is that

the optimal filters designed this way are not matched in the above sense. Although the above presented two methods for choosing the filters are suboptimal, i.e. they do not satisfy the relationship (*), they reduce the effect of the above-indicated effective noise In each case, this leads to a double-filter policy, with different ad hoc regularization factors. Fig. 6 shows an exemplary performance diagram 600 illustrating transmission rates of a precoding system according to Figure 5 when using first and second filter design methods. As can be seen from the numerical results depicted in Fig. 6, none of the above presented two methods (Method 1 , 601 and Method 2, 602) strictly outperforms the other in general. Which method leads to better rates depends on the systems parameters, i.e., input power, interference power, noise power, and channel statistics. Figure 6 shows typical rate curves associated with the two methods as function of the SNR.

From Figure 6 it can be seen that selecting the filters 209, 319 may be performed based on the operating point. That is, the filters may be selected according to the one among the above two methods that performs better, depending on the operating point, e.g. SNR, noise and interference covariance matrices, and channel estimation error covariance matrices. Fig. 7 shows a block diagram illustrating a threshold computation unit 700 according to an implementation form.

The threshold computation unit 700 may compute the desired SNR threshold 710 essentially based on the comparison of the performance of the two filter design methods as described above with respect to Fig. 5. The threshold computation unit 700 may evaluate noise 704 and interference 702 covariance matrices, and channel estimation error covariance matrices of receiver 706 and transmitter 708, for example by solving the equation R 1 (x)-R 2 (x)=0. Fig. 8 shows a functional diagram illustrating a selection of the appropriate filter design method of a precoding system according to Figure 5 according to an implementation form.

The threshold 710 may be computed by the threshold computation unit 700 as described above with respect to Figure 7. In a comparison block 803 this threshold 710 may be compared against a determined SNR 812 of the input signal X as described above with respect to Figures 4 and 5. If the threshold 710 is greater than the computed SNR 812, Method 1 may be used 807. In the opposite case that the threshold 710 is less than the computed SNR 812, Method 2 may be used 805. The above means that the selection of the filter design technique that offers the best throughput (among the presented two methods) is essentially based on the comparison 803 of the actual SNR 812 with the computed SNR threshold 710, e.g. as shown in Figure 6 (which itself depends on the system parameters, and in particular the interference level).

Fig. 9 shows an exemplary performance diagram 900 illustrating comparison of the two filter design methods when using sphere lattices and an interference-to-noise ratio of 10 times the signal-to-noise ratio. In Fig. 9 there can be seen a crossing point where the curve according to Method 1 crosses the curve according to Method 2.

Figure 9 illustrates the performance of the above described "Method 1 " and "Method 2" for an example system in which the following parameters are applied: Number of transmit / receive antennas: Additive Noise power N = 1 ; Interference to Signal

^ ; Estimation error at the receiver: Quantization noise on

the feedback link: ( Signal to Noise

It is assumed here that the quantization and estimation noise both scale inversely with the SNR and this is due to the fact that the better the link between the two users, the better the estimate at the receiver. As for the quantization noise, it is assumed that the feedback link is of enough quality when the direct link is of good quality. Figure 9 depicts the evolution of the rates that are offered by Method 1 and Method 2 as a function of the SNR = ^ (in dB). The input covariance matrix is chosen

As it can be seen from the figure, Method 1 outperforms Method 2 at SNRs that are larger than some SNR threshold of about 6 dB. Below that threshold, however, Method 2 performs better, comparatively. This means that for SNRs that are smaller than 6dB, it is more beneficial for Rx to design its filter as in Method 2 (i.e., using the noisier transmit CSI, instead of its own CSI which is of better quality). The interpretation is that utilizing its own CSI would lead to a big discrepancy with the transmit filter, possibly yielding bigger residual noise and interference terms.

Fig. 10 shows an exemplary performance diagram 1000 illustrating comparison of the two filter design methods when using Gosset lattices. Fig. 1 1 shows an exemplary

performance diagram 1 100 illustrating comparison of the two filter design methods when using Cubic lattices. Figure 10 and Figure 1 1 show similar curves obtained with the two methods as described above for several choices of finite dimension lattices (cubic lattice and Gosset lattice for the same parameters as in Figure 9.

In both Figures 10 and 1 1 a crossing point can be observed where the curve according to Method 1 crosses the curve according to Method 2.

Fig. 12 shows a schematic diagram illustrating the downlink communication channel of a cellular communication system 1200.

In cellular networks, the CSI that is available at the transmitter depends on the way it is obtained. In frequency-division duplex (FDD) systems, the CSI at Tx is typically obtained through feedback from Rx. In time-division-duplex (TDD) systems, the Tx typically estimates the channel directly from its received signal. In addition to that the estimation is performed independently, possible errors due to calibration, generally lead to distinct channel estimations at the Tx and Rx.

The base station 1201 (BS) transmits distinct messages or information to distinct mobile terminals (UE1 , UE2, UE3, UEK). It is assumed that there are K users in the cell. The channel to user k, 1≤ k≤ K, is denoted by H k . The system operates in FDD mode; and, due to possible calibration and estimation errors, the base station 1 201 has only an estimate of the channel H k . Similarly, user k has only an estimate k of its channel H k . (Note that the above described Method 1 and Method 2 also apply if

In this case, the information X k that is sent by the BS 1201 to user k constitutes interference in the eyes of the other users. Assuming the precoding order k = 1, 2, . . , K , an efficient precoding technique consists in precoding the information X k to be sent to user k against all those sent to users . That is, is to be precoded against the

interference which is known at the BS 1201 . The remaining users' signals j > k are treated as additional noise.

More precisely, the signal received by user k is given by:

where is the interference part that is known at the BS 1 201 and Z k =

k is the total noise.

In the following an exemplary implementation form is described. Given a power allocation strategy, each user is allocated P k as useful power and a covariance matrix For each user k , the source will compute the threshold

based on the interference covariance:

on the noise covariance:

and on each of the estimation covariance matrices

As such, if the operating point is below the threshold: then the source will choose Method 2 for this user. Otherwise, it chooses Method 1 .

For this embodiment, considering an LTE scenario with BSs and users all equipped with single antennas, a bandwidth of 20 MHz, a number of users of 12 inside the cell, equal power transmission from the BS to the users, at SNR of 10 dB Method 1 provides 8 Mb/s improvement over Method 2, whereas at SNR of 3 dB, it is Method 2 which improves upon Method 1 by almost 2.4 Mb/s.

Fig. 13 shows a schematic diagram illustrating a decoding method 1300 using asymmetric imperfect channel knowledge according to an implementation form.

The decoding method 1300 includes decoding 1301 a receive signal Y received over a communication channel by using an adjustable receive filter 319, e.g. as described above with respect to Figures 4 and 5. The decoding method 1300 includes providing 1302 receive-side channel state information H indicating a receive-side estimate of a channel state H of the communication channel and providing transmit-side channel state information H indicating a transmit-side estimate of the channel state H of the

communication channel, e.g. as described above with respect to Figures 4 and 5. The decoding method 1300 further includes adjusting 1303 the receive filter 319 based on the transmit-side channel state information H if the transmit-side channel state information H is a less good estimate of the channel state Η of the communication channel than the receive-side channel state information H , e.g. as described above with respect to Figures 4 and 5. The decoding method 1 300 may further include changing the adjusting 1303 of the receive filter 319 from an adjustment based on the receive-side channel state information

H to an adjustment based on the transmit-side channel state information H based on a transmit-side condition, in particular a transmit-side SNR 81 2 crossing a threshold 710, e.g. as described above with respect to Figures 4 and 5.

The present disclosure also supports a computer program product including computer executable code or computer executable instructions that, when executed, causes at least one computer to execute the performing and computing steps described herein, in particular the method 1300 described above with respect to Fig. 1 3 or the filter design method 1 and filter design method 2 as described above with respect to Figures 4 to 12. Such a computer program product may include a readable non-transitory storage medium storing program code thereon for use by a computer. The program code may perform the method 1 300 described above with respect to Fig. 13 or the filter design method 1 and filter design method 2 as described above with respect to Figures 4 to 1 2.

While a particular feature or aspect of the disclosure may have been disclosed with respect to only one of several implementations, such feature or aspect may be combined with one or more other features or aspects of the other implementations as may be desired and advantageous for any given or particular application. Furthermore, to the extent that the terms "include", "have", "with", or other variants thereof are used in either the detailed description or the claims, such terms are intended to be inclusive in a manner similar to the term "comprise". Also, the terms "exemplary", "for example" and "e.g." are merely meant as an example, rather than the best or optimal. The terms "coupled" and "connected", along with derivatives may have been used. It should be understood that these terms may have been used to indicate that two elements cooperate or interact with each other regardless whether they are in direct physical or electrical contact, or they are not in direct contact with each other.

Although specific aspects have been illustrated and described herein, it will be

appreciated by those of ordinary skill in the art that a variety of alternate and/or equivalent implementations may be substituted for the specific aspects shown and described without departing from the scope of the present disclosure. This application is intended to cover any adaptations or variations of the specific aspects discussed herein.

Although the elements in the following claims are recited in a particular sequence with corresponding labeling, unless the claim recitations otherwise imply a particular sequence for implementing some or all of those elements, those elements are not necessarily intended to be limited to being implemented in that particular sequence.

Many alternatives, modifications, and variations will be apparent to those skilled in the art in light of the above teachings. Of course, those skilled in the art readily recognize that there are numerous applications of the invention beyond those described herein. While the present invention has been described with reference to one or more particular embodiments, those skilled in the art recognize that many changes may be made thereto without departing from the scope of the present invention. It is therefore to be understood that within the scope of the appended claims and their equivalents, the invention may be practiced otherwise than as specifically described herein.




 
Previous Patent: LENGTH CONTROL FOR PACKET HEADER SAMPLING

Next Patent: WHISK