WO2018006982A1

WO2018006982A1 - Method and apparatus for multi-antenna transmission

Info

Publication number: WO2018006982A1
Application number: PCT/EP2016/066328
Authority: WO
Inventors: Magnus ÅSTRÖM; Fredrik Nordström; Bo Lincoln
Original assignee: Telefonaktiebolaget LM Ericsson AB
Current assignee: Telefonaktiebolaget LM Ericsson AB
Priority date: 2016-07-08
Filing date: 2016-07-08
Publication date: 2018-01-11
Anticipated expiration: 2019-01-08

Abstract

When amplifying a set of signals for transmission, one or more of the signals may exceed the clipping threshold(s) of the corresponding power amplifiers (26). This disclosure proposes a technique that compensates for the clipping in a given signal or signals in the set by adding compensation values to one or more of the remaining signals in the set. The added compensation values reduce the effects of clipping at the transmitter (10) with respect to the corresponding received signal vector at a targeted receiver (12).

Description

METHOD AND APPARATUS FOR MULTI- ANTENNA TRANSMISSION

TECHNICAL FIELD

The present invention relates to wireless communications, and particularly relates to a method and apparatus for multi-antenna transmission. BACKGROUND

Coordinating transmissions across multiple antennas offers multiple benefits in wireless communication systems. Co-located antennas or antenna elements enable beamforming and spatial multiplexing, among other things.

With beamforming, a signal is transmitted from a first antenna, and phase-shifted versions of the signal are transmitted from adjacent antennas. The effect is that signal energy is interfering constructively in a certain direction, thus forming a beam, and mostly destructively in other directions. The beam direction depends on the phase differences of the signal between the different antennas.

Spatial multiplexing is a type of multi-layer transmission that exploits the spatial diversity provided by plural antennas. The technique permits the transmission of multiple data streams in parallel. Precoding the signals being transmitted allows the targeted receiver or receivers to resolve the multiple streams. With precoding

y = HWx + n,

where x is a vector including the multiple streams to be transmitted, H is the propagation channel, W is a precoder comprising a set of weights for weighting the signals for transmission, n represents noise, and y denotes the received streams.

Consider the case of Single User Multiple Input Multiple Output (SU-MIMO), and assume a case where two streams are being transmitted to the same user. The two streams are added with different temporal shift, to achieve different directivity. In an approach used in Long Term Evolution, LTE, networks, the different streams are constructed by using different precoding weights. Precoding allows the receiver to resolve the individual streams.

A slightly different approach is to direct the parallel streams to different users, which is then called multiuser MIMO (MU-MIMO). To achieve high beamforming abilities, a high antenna correlation is desirable. However, on the other hand, to achieve good spatial multiplexing abilities, a low correlation is desirable. Massive MIMO systems are likely to contain antennas with both high and low correlation.

Both LTE and IEEE 802.11 use Orthogonal Frequency Division Multiplexing, OFDM.

OFDM has several desirable properties in terms of robustness towards channel impairments, multi-path signal propagation, the possibility of removing Inter Symbol Interference, and allowing for straightforward demodulation. The preceding received-signal equation above may be expanded, to include the time- frequency transformation inherent in an OFDM system,

y = FHF^Wx + n.

Here, F is an orthonormal Fast Fourier Transform, FFT, matrix, and ^_1 is its inverse.

Considering the outputs at all antenna elements, the transmitted time domain signal z may be expressed as

z = FWx.

Unfortunately, because the time signal at the z:th antenna, z comprises multiple subcarriers, OFDM suffers from a high peak-to-average power ratio, PAPR. Consequently, the proper transmission of OFDM signals requires high linearity Power Amplifiers, PAs, to avoid amplifier "clipping" and the resulting signal distortions.

Notably, because F is orthonormal and for the cases where the precoder W also is orthonormal, clipping in z_t will be uncorrected from clipping in Z , where "z" and "j" denote different antennas involved in the multi-antenna transmission. Similarly, in the earlier introduced SU-MIMO transmission case, assuming that the multiple streams being transmitted are independent stochastic variables implies that clipping occurs independently with respect to the different amplified streams transmitted from the different antennas.

A linear PA, of the type used in the signals transmissions at issue, provides an amplified signal that is directly proportional to the input signal but significantly increases the signal power for radiated transmission from an antenna. PAs maintain linearity only over a limited range of input signal magnitude and input voltages or currents in excess of such limits "overdrive" the PA into a non-linear or saturated mode. The non-linearity effect is referred to as "clipping" because those portions of the input signal that push the PA into the saturation region are not reproduced, or at least not linearly reproduced, in the amplified signal output from the PA.

The need for linearity in signal amplification represents an ever current problem in wireless communications. It is desirable to have an as linear transmitter as possible, to introduce as little transmission noise as possible. However, linearity comes at a cost in terms of higher component quality and price, and higher current consumption. As often is the case, the preferred design is a compromise between cost and performance.

Because it is impractical to implement PAs that will not clip under all input signal conditions, communication transmitters use mitigation techniques that limit instances of clipping. It is known to operate PAs with a "power backoff." The power backoff may be understood as the ratio of the maximum saturation output power of the PA to the average output power of the PA. Larger backoff translates into less clipping and, hence, less non-linear distortion. A sufficiently large backoff can be used to limit the power of clipping-induced noise to a couple of percentage points of the total output power. For example, LTE requires that transmitter noise power be less than six percent of maximal output power. Further, to achieve higher order modulations, e.g., 256 QAM, it is necessary to reduce the noise power to less than three percent of maximal transmit power, and higher-rank transmissions also require low levels of non-linear distortion at the transmitter.

Unfortunately, lower power backoffs reduce amplifier efficiency, while increasing power consumption, and require PAs with high maximum powers to achieve reasonable average output powers. As such, the power backoff technique has significant practical limitations, and it is recognized herein that additional techniques are needed, to reduce non- linear distortions when using practical, affordable PAs.

SUMMARY

When amplifying a set of signals for transmission, one or more of the signals may exceed the clipping threshold(s) of the corresponding power amplifiers. This disclosure proposes a technique that compensates for the clipping in a given signal or signals in the set by adding compensation values to one or more of the remaining signals in the set. The added compensation values reduce the effects of clipping at the transmitter with respect to the corresponding received signal vector at a targeted receiver.

In one embodiment, a wireless communication apparatus implements a method of multi- antenna signal transmission. The method includes evaluating individual signals in a set of signals to identify signal components that will be clipped as a consequence of amplifier non-linearity. The identified signal components may be referred to as "residual signal components" to emphasize the fact that they represent parts of the original signal that will be lost or compressed in the corresponding amplified signal.

Here, each signal in the set of signals is to be amplified by a respective power amplifier in a corresponding set of power amplifiers, for transmission from a respective transmit antenna in a corresponding set of transmit antennas. Correspondingly, the method includes modifying the set of signals to obtain a modified set of signals and transmitting the modified set of signals, rather than the set of signals. The modified set of signals is obtained by, for a given residual signal component in a given signal in the set of signals, adding a compensation value to each of one or more other ones of the signals, to compensate signal reception at a targeted receiver for clipping of the given residual signal component in the given signal.

Transmitting the modified set of signals, rather than the set of signals, includes converting the modified set of signals into the analog domain, for amplification by the corresponding set of power amplifiers and transmission from the corresponding set of transmit antennas. It should be appreciated that the modified set of signals differs from the set of signals in that at least one of the signals in the original set is modified to compensate for clipping that will occur in at least one other one of the signals in the original set of signals. However, the modified set of signals may include one or more of the original signals. In general, the compensation is performed in recurring transmission intervals or with respect to corresponding segments across the set of signals. For example, the set of signals is processed for transmission as respective blocks of data samples. When one or more samples in a given block are identified as exceeding the clipping threshold, a corresponding sample or samples in one or more other blocks are modified, to compensate for the clipping.

In another embodiment, a wireless communication apparatus is configured for multi- antenna transmission. The wireless communication apparatus includes transmission circuitry, including a set of power amplifiers for amplifying signals for transmission from a corresponding set of antennas, and processing circuitry that is included in or is operatively associated with the transmission circuitry.

The processing circuitry is configured to evaluate individual signals in a set of signals to identify signal components, referred to as residual signal components, that will be clipped as a consequence of amplifier non-linearity. Each signal in the set of signals is to be amplified by a respective power amplifier in the corresponding set of power amplifiers, for transmission from a respective transmit antenna in the corresponding set of transmit antennas. The processing circuitry is further configured to modify the set of signals to obtain a modified set of signals. The processing circuitry obtains the modified set of signals, by, for a given residual signal component in a given signal in the set of signals, adding a compensation value to each of one or more other ones of the signals, to compensate signal reception at a targeted receiver for clipping of the given residual signal component in the given signal. Still further, the processing circuitry is configured to transmit the modified set of signals, rather than the set of signals, by converting the modified set of signals into the analog domain, for amplification by the corresponding set of power amplifiers and transmission from the corresponding set of transmit antennas.

Of course, the present invention is not limited to the above features and advantages. Indeed, those skilled in the art will recognize additional features and advantages upon reading the following detailed description, and upon viewing the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a block diagram of one embodiment of a wireless communication apparatus configured for multi-antenna transmission to another wireless communication apparatus as a targeted receiver.

Fig. 2 is a block diagram of a wireless communication apparatus performing a multi- antenna transmission to a targeted receiver.

Fig. 3 is a block diagram one embodiment of an antenna array, which may be used for multi-antenna transmission.

Fig. 4 is a logic flow diagram of one embodiment of a method of processing at a wireless communication apparatus.

Fig. 5 is a block diagram of one embodiment of transmitter circuitry configured for multi-antenna transmission.

Figs. 6A, 6B, 7, and 8 are pseudo-code listings for implementing and evaluating multi- signal, multi-antenna transmissions with compensation for transmitting clipping.

DETAILED DESCRIPTION

Fig. 1 illustrates a wireless communication apparatus, "WCA", 10 that is configured to carry out a multi-antenna transmission towards another WCA 12. As a non-limiting example, the WCA 10 is a radio base station in a cellular communication network, an access point in a Wireless Local Area Network (WLAN), or other communication network node configured for multi-antenna transmissions. By way of a corresponding example, the WCA 12 comprises a User Equipment, "UE", or other mobile terminal, such as a computer, tablet, network modem or wireless dongle, etc. In these roles, the WCA 10 performs downlink, "DL", multi-antenna transmissions towards the WCA 12 as a targeted receiver. Of course, the roles may be reversed, e.g., the WCA 10 comprises a terminal or other mobile device, and the WCA 12 comprises a base station or other radio network node. In that scenario, the WCA 10 performs uplink, "UL", multi-antenna transmissions towards the WCA 12 as a targeted receiver.

The WCA 10 includes transceiver circuitry 20, including transmitter circuitry 22 and receiver circuitry 24. As will be appreciated, the transmitter circuitry 22 and the receiver circuitry 24 may comprise a mix of digital and analog circuitry, e.g., where the transmit signal path begins in the digital baseband domain and continues into analog radio frequency domain. In mirror inverse, the receive signal path begins in the analog radio frequency domain and continues into the digital baseband domain.

The transmitter circuitry 22 is of particular interest in this discussion, and it includes a set of power amplifiers 26, e.g., power amplifiers or "PAs" 26-1, 26-2, ..., 26- , where Mis an integer number. Unless needed for clarity, the reference number "26" will be used without suffixing when referring to any one or more given PAs, in both the singular and plural forms. The set of PAs 26 is associated with a corresponding set of transmit antennas 28, e.g., transmit antennas 28-1, 28-2, ... 28- . As with the reference number "26", the reference number "28" will be used without suffixing when referring to any one or more given transmit antennas, in both the singular and plural forms.

The transmitter circuitry 22 further includes processing circuitry 30, which provides transmit-signal processing and control, e.g., forming or buffering the signal samples constituting the transmit signals, and processing them for transmission. The processing circuitry 30 may extend into or be implemented at least in part in the overall collection of processing circuitry 40 implemented within the WCA 10. In turn, the processing circuitry 40 includes or is associated with storage 42, which in one or more embodiments stores one or more computer programs 44, and, optionally, one or more items of configuration data 46.

While the WCA 10 may be larger, more powerful, or more complex than the WCA 12, or vice versa, the WCA 12 similarly includes transceiver circuitry 50, including a transmitter 52 and a receiver 54. In the illustrated example, the WCA 12 includes two or more receive/transmit antennas 56, e.g., to support Multiple Input Multiple Output, "MIMO", transmissions between the WCA 10 and the WCA 12.

The WCA 12 further includes processing circuitry 60, which in turn includes or is associated with storage 62. In one or more embodiments, the storage 62 stores one or more computer programs 64, and, optionally, one or more items of configuration data 66.

As for implementation details regarding the WCAs 10 and 12, it shall be understood that such nodes may comprise complex, computer-based processing systems. More generally, the processing circuitry 40/60 of the WCA 10/12 comprises fixed circuitry, programmed circuitry, or a mix of fixed and programmed circuitry. In at least one embodiment, the processing circuitry 40/60 comprises digital processing circuitry including one or more microprocessors, digital signal processors or DSPs, application specific integrated circuits or ASICs, field programmable gate arrays or FPGAs, etc., along with supporting memory or other storage.

Correspondingly, the storage 42/62 comprises one or more types of non-transitory computer-readable media, such as any one or more of electromagnetic disk storage, solid state disk storage, volatile memory circuits such as DRAM or SRAM, and non- volatile memory circuits, such as FLASH or EEPROM. "Non-transitory" does not necessarily mean permanent or unchanging, but does connote that the storage 42/62 provides storage of at least some persistence for information contained therein. In at least one embodiment, the storage 42/62 stores a computer program 44/64 comprising program instructions by which the processing circuitry 40/60 is specially adapted to carry out processing according to the teachings herein.

Similarly, the processing circuitry 30 of the transmitter circuitry 22 in the WCA 10 may, as previously noted, be implemented at least in part in the processing circuitry 40, and may comprise fixed circuitry, programmed circuitry, or a mix of both. In a particular example, the processing circuitry 30 comprises one or more Digital Signal Processors, "DSPs", or other digital processing circuits that are configured at least in part according to the execution of stored computer program instructions.

Within the above context, the WCA 10 in an example embodiment is configured for multi-antenna transmission and includes transmission circuitry 22, including a set of power amplifiers 26 for amplifying signals for transmission from a corresponding set of antennas 28. The transmission circuitry 22 further includes or is operatively associated with the processing circuitry 30, which is configured to evaluate individual signals in a set of signals to be transmitted, to identify signal components, referred to as residual signal components, that will be clipped as a consequence of amplifier non-linearity.

In particular, each signal in the set of signals is to be amplified by a respective power amplifier 26-/ in the corresponding set of power amplifiers 26, for transmission from a respective transmit antenna 28-z in the corresponding set of transmit antennas 28. The processing circuitry 30 is configured to modify the set of signals to obtain a modified set of signals, by, for a given residual signal component in a given signal in the set of signals, adding a compensation value to each of one or more other ones of the signals, to compensate signal reception at a targeted receiver— e.g., at the WCA 12— for clipping of the given residual signal component in the given signal. Still further, the processing circuitry 30 is configured to transmit the modified set of signals, rather than the set of signals, by converting the modified set of signals into the analog domain, for amplification by the corresponding set of power amplifiers 26 and transmission from the corresponding set of transmit antennas 28.

Thus, as part of its processing of the set of signals, the processing circuitry 30 detects signals components within respective ones of the signals that will be clipped when subsequently amplified by a respective one of the power amplifiers 26. Such detection may be based on known amplifier clipping thresholds, which may be known from the configuration data 46, or which may be dynamically computed based on relevant, "live" operational parameters of the transmitter circuitry 22. For any signal that will be clipped, the processing circuitry 30 modifies one or more other ones of the signals that will not be clipped— i.e., one or more other ones of the signals that have amplification "headroom" sufficient to allow for the addition of a compensating signal component. In an example embodiment, the compensation value added to each of the one or more other ones of the signals is based at least in part on the given residual signal component. In one such embodiment, the processing circuitry 30 is configured to modify the set of signals by, with respect to the given residual signal component in the given signal, identifying one or more of the remaining signals that will not be clipped at a time corresponding to the residual signal component and, for each identified signal, adding the residual signal component in whole or in part to the identified signal, as the compensation value for the identified signal.

As there may be multiple signals in the set that will experience clipping at the same time, the processing circuitry 30 in one or more embodiments may perform a fitting or other optimization process, to determine a set of compensation values to be added to the remaining signals.

For example, the compensation value added to each of the one or more other ones of the signals is computed as a function of a channel estimate matrix of the channel between the transmit antennas 28 and the involved receiver antenna 56 or antennas 56 of the targeted receiver 12. In at least one such embodiment, the compensation value or values are computed as a function of a reduced channel estimate matrix, a channel estimate matrix, and the difference between an ideal transmitted signal and a non-ideal transmitted signal, such as detailed immediately below.

Fig. 2 provides helpful references with respect to such details and provides a useful reference for the following mathematical exposition. For simplification, assume an ideal, one-tap propagation channel between the WCA 10 and the WCA 12, and assume that the WCA 10 transmits a set of M signals s from M transmit antennas 28, where the signals were amplified using ideal— non-clipping— PAs. Further, assume that the WCA 12 is operating with N receive antennas 56. With these assumptions, the received signal at the WCA 12 as the targeted receiver is

r = Hs,

in which s represents the M transmitted signals that would obtain with ideal power

amplification— i.e., in the absence of clipping. Further, r represents the N received signals, and H represents the N x M channel estimate matrix, which comprises a set of matrix elements given as

H =

h N-1,0 h N-1,M-1

Assume that the set of signals transmitted with ideal power amplification comprises

and, further, assume that the signal s₀ would have been clipped with real- world power amplification. We can denote the clipped version of s₀ as s₀ and, consequently, can represent the non-ideal set of transmitted signals as

Now, assume that z represents the baseband signal vector from which the transmitted signal vector s was derived. With clipping modeled as a known, nonlinear amplification function between z and s, we can express s as

S = (z) .

As a simple but useful example, it may be that any signal amplitude above a known threshold ^aciip will t>^e clipped during power amplification. Thus,

/ (z) = min (abs(z), a_clip) arg (z) .

Furthermore, the ideal amplification function with no clipping may be expressed using the amplification gain g as

s = gz.

Hence it is possible to know the clipping behavior of the power amplifiers 26 a priori with respect to any given set of signals to be transmitted. As such, the processing circuitry 30 can calculate or otherwise estimate what the deviation from ideal amplification will be and can do so in the baseband digital domain, in advance of passing the signals through the power amplifiers 26, for amplification and transmission. Thus, the processing circuitry 30 identifies which signal components in each signal within a set of signals to be amplified and transmitted that will be clipped during power amplification.

In one embodiment, to compensate for clipping, the processing circuitry 30 computes a set of one or more compensation values, denoted as δ. By way of example, assume that non- ideal power amplification results in s₀ being clipped, such that, absent the techniques taught herein, the WCA 10 would transmi s =

According to the teachings herein, however, the WCA 10 transmits a modified set of signals s, where

In other words, rather than simply transmitting the set of signals s subject to whatever clipping occurs, the WCA 10 adds one or more compensation values δ to one or more other ones of the signals that will not experience amplifier clipping. This technique results in the receiver side expression

r = Hs = Hs + H_!<S,

where H lacks the channel elements from the clipped output and is referred to herein as a "reduced channel estimate matrix" and is given as

Minimizing the distance between the ideal received signal, r, and the compensated received signal, , with respect to the Euclidean norm gives

min₅||r - ||₂ = min₅||Hs - Hs||₂ = min₅||Hs - Hs - Η₁δ\\₂.

The optimal compensation value(s) δ are found when the gradient is zero,

where, in the final step, the denominator may be discarded, yielding

H_!<S = H(S - S)

and finally

<S = H⁺H(s - s),

where HJ represents the pseudo inverse of H .

Thus, in at least one embodiment, the processing circuitry 30 is configured to

determining a vector of compensation values δ as the pseudoinverse of the reduced channel estimate matrix H_{1 ?} multiplied by the channel estimate matrix H multiplied by the difference between the ideal transmitted signal s and the clipped transmitted signal s. It is important to recognize here that s and s are not transmitted. Instead, s is the transmitted signal vector that would obtain with ideal power amplification— no clipping behavior by the PAs 26— and can be easily calculated or estimated by the processing circuitry, and s is the transmitted signal vector that would obtain with non-ideal power amplification— i.e., clipping— and without use of the compensation techniques taught herein. As with s, the processing circuitry 30 can easily calculate or estimate s, based on the known or estimated clipping characteristics of the power amplifiers 26. As a more general statement, the processing circuitry 30 in one or more embodiments is configured to calculate the compensation vector δ as a function of the reduced channel estimate matrix H_{1 ?} the channel estimate matrix H, and the difference between the ideal transmitted signal s and the clipped transmitted signal s. Further, in a related embodiment, the processing circuitry 30 is configured to formulate the problem as a constrained optimization problem. The constrained minimization is expressed as

\r - Hs\² + A\s - s\²,

in which λ controls the allowed amount of deviation of s from s. The processing circuitry 30 may be configured to solve the minimization problem iteratively, using a limited number of iterations whereby each iteration reduces the size of s by eliminating the element with the highest absolute value.

As a more general proposition of the teachings herein, the processing circuitry 30 in some embodiments is configured to modify the set of signals, by computing compensation values from the residual signal components and adding the compensation values to selected ones of the signals in the set of signals. Here, the compensation values are computed according to a minimization function that minimizes the distance between an ideal received signal vector corresponding to transmission of the set of signals in the absence of clipping and a compensated received signal vector corresponding to transmission of the modified set of signals in the presence of clipping.

When contemplated such operations, it is helpful to keep in mind that, in at least some embodiments, the identification of signal components that will be clipped and the corresponding signal modifications made to compensate for such clipping are done in advance of the actual power amplification and transmission— i.e., done before the clipping actually occurs. For example, a given set of baseband signals z is processed, converted to the analog domain, amplified and transmitted as a corresponding set of transmitted signals s, and the processing herein may be performed with respect to the baseband signals z.

In one example, each signal in the set of signals comprises a block of samples— e.g., digital samples in the baseband domain. The processing circuitry 30 according to this example is configured to evaluate the individual signals in the set of signals to identify the signal components that will be clipped as a consequence of amplifier non-linearity. The processing circuitry 30 makes this evaluation by, for each signal, evaluating the block of samples to identify samples that will be clipped, based on a known amplifier clipping threshold. The block of samples comprising each signal in the set of signals comprises, for example, a block of Fast Fourier Transform, FFT, coefficients. Concerning Fig. 3, one sees an array 70 of antenna elements 72. In general, proximate antenna elements 72 are likely to be correlated, while antenna elements 72 that are further away from each other are likely to be uncorrected, or at least to exhibit much lower correlation. Thus, the antenna array 70 may be operated with one or more defined sets 74 of correlated antenna elements 72, and such is the case in at least one embodiment contemplated herein. In particular, the processing circuitry 30 is configured to perform the signal evaluations, modifications, and transmissions described herein with respect to each of a plurality of defined sets 74 of correlated antenna elements 72. That is, the set of antennas 28 comprise an array 70 of antenna elements 72, and the WCA 10 performs the clipping compensation disclosed herein with respect to each of one or more sets 74 of correlated antenna elements 72.

In a simple example, the set of signals comprises a first signal and a second signal, e.g., a set z that includes first and second baseband signals z and z₂. The processing circuitry 30 in an example embodiment is configured to modify the set of signals by, for a given residual signal component identified in the first signal, adding, as the compensation value for the second signal, the given residual component in whole or in part to a corresponding segment of the second signal. The first and second signals are, for example, time aligned and the "corresponding segment" of the second signal is a segment that will be transmitted coincidentally with a segment of the first signal corresponding to the residual signal component, accounting for any temporal shifting applied for multi-antenna transmission of the modified set of signals. In other words, the compensated segment of the second signal is transmitted in conjunction with the clipped segment of the first signal.

Fig. 4 illustrates a method 400 that may be implemented using the WCA 10 introduced in Fig. 1. However, the method 400 is not limited to the particular circuit arrangements presented in the example of Fig. 1. Moreover, although presented as a sequence of steps, one or more of the steps may be performed in an order different than that suggested by the illustration. Additionally, one or more of the steps may be performed in conjunction with other steps, or other operations not illustrated, or may be performed on an ongoing, background, or repeating basis. Further, the illustrated processing may be carried out for multiple sets of signals to be transmitted.

The method 400 involves multi-antenna signal transmission, such as performed by the WCA 10. The method 400 includes evaluating (Block 402) individual signals in a set of signals to identify signal components, referred to as residual signal components, that will be clipped as a consequence of amplifier non-linearity, each signal in the set of signals to be amplified by a respective power amplifier 26-/ in a corresponding set of power amplifiers 26, for transmission from a respective transmit antenna 28-z in a corresponding set of transmit antennas 28. The method 400 further includes modifying (Block 404) the set of signals to obtain a modified set of signals. The modified signals are obtained, by, for a given residual signal component in a given signal in the set of signals, adding a compensation value to each of one or more other ones of the signals, to compensate signal reception at a targeted receiver 12 for clipping of the given residual signal component in the given signal. Still further, the method 400 includes transmitting (Block 406) the modified set of signals, rather than the set of signals, by converting the modified set of signals into the analog domain, for amplification by the corresponding set of power amplifiers 26 and transmission from the corresponding set of transmit antennas 28.

Fig. 5 provides an additional, non- limiting example of transmitter circuitry details useful for understanding the techniques disclosed herein. The transmitter circuitry 22 as illustrated includes two transmit signal processing branches 60-1 and 60-2, for processing two input information signals x and x₂ that comprise data for transmission. The processing circuitry 22 forms two transmission layers, Layer 1 or LI, and Layer 2 or L2, by precoding the two input streams in respective pairs of precoders— the LI and L2 precoders 62-1 and 62-2 in the first transmit processing branch 60-1, and the LI and L2 precoders 62-3 and 62-4 in the second transmit processing branch 60-2.

The weighted values of the information streams are summed in the respective branches 60-1 and 60-2, via summing or combining circuits 64-1 and 64-2. The combining circuit 64-1 outputs its resulting summed signal to an OFDM modulator 66-1, and the combining circuit 64-2 outputs its resulting summed signal to an OFDM modulator 66-2. The two modulators 66-1 and 66-2, in turn, output their modulated signals to respective clipping detectors 68-1 and 68-2. The modulated signals are denoted as z_x and z₂— i.e., they are an example of a set of two baseband digital signals operated on according to the techniques disclosed herein.

The clipping detectors 68-1 and 68-2 may be functionally implemented within DSPs or other digital processing circuitry comprising the processing circuitry 30, and they are configured to detect sample values in the signals z and z₂ that exceed the applicable clipping thresholds of the respective power amplifiers 26-1 and 26-2. In other words, they evaluate the individual signals in the set of signals z, to identify signal components that will be clipped during subsequent power amplification.

The clipping information is provided to a compensation circuit 74 that is configured to compute the compensation value or values, to be used to obtain the modified set of signals z' comprising z[ and z₂. Note when the samples in both z and z₂ are below the clipping threshold, then z[ = z-L and z₂ = z₂. However, to compensate for a sample value in z that will be clipped during power amplification, a compensation value δ will be added to the corresponding sample value in z₂ to obtain z₂. Conversely, to compensate for a sample value in z₂ that will be clipped during power amplification, a compensation value δ will be added to the corresponding sample value in z to obtain z[.

In some sense, the clipping compensation technique can be understood as using the "spare" power available in one or more of the power amplifiers 26 in the set of power amplifiers 26, to compensate for the effects of signal clipping in one or more other ones of the power amplifiers 26. In the context of the above example, if z and z₂ will both clip at the same time, it may be that neither of the corresponding power amplifiers 26-1 and 26-2 has spare power for adding compensating values. However, the question of whether or not a given amplifier 26 has "spare" power also depends on the phase relationship between the signal to be amplified by the given amplifier 26 and the compensation value to be added to the signal. For example, the given amplifier 26 may not have spare power for the case where the compensation value has the same phase as the signal to be amplified, but may have spare power for the case where the

compensation value and the signal to be amplified have opposing phase. Such assessments or determinations shall be understood as being part of the evaluation contemplated herein, in at least some embodiments.

Further, because in many practical scenarios, clipping events are uncorrected between the two signals z and z₂ and there is a high probability that clipping in one of them can be compensated by modifying the other one of them. The same holds true for larger sets of z. That is, for any one or more of the signals in z that will be clipped at a given sample time, there are likely to be one or more remaining signals in z that will, with regard to the amplification limits of their respective power amplifiers 26, accommodate the addition of a compensation value. The compensation circuit 74 outputs the compensation values, or otherwise controls respective signal modifiers 70-1 and 70-2 to apply the proper compensation values, to obtain the modified set of signals z', which are converted into the analogue domain via digital-to-analog converters 72-1 and 72-1, for power amplification by the respective PAs 26-1 and 26-2, and transmission from the respective antennas 28-1 and 28-2.

In one embodiment the modification of the second signal assumes equal channels from the two transmit antenna elements to the intended receiver. In another embodiment, the modification takes the antenna correlation into account, such that a second antenna with a higher correlation to the first antenna is given a higher weight of the residual signal component whose clipping is being compensated for, as compared to that of a second antenna with a lower correlation. In a further embodiment, the channel correlation is taken into account such that a higher channel correlation results in a larger modification than that of a lower channel correlation. One embodiment considers the superimposed streams' individual contributions to the clipped signal when determining which stream(s) to be modified. In other words, a stream with a large contribution to the clipped samples will be modified more on that stream's signal on another antenna. Another embodiment, in a MU-MIMO environment, determines the preferred stream to modify based on the respective transmission setups. For example, a stream with a more favorable transmission setup, e.g., a lower rank or one using beamforming, is modified compared to a spatially multiplexed, high-rank transmission setup. In alternative embodiments, the segment of the second signal may be shifted in time such that it is different from the first segment due to e.g., beamforming delays. This shift adjusts for the extra delay that one signal may have relative to another, as a consequence of a slightly longer pathway along a certain direction.

By using ά priori information about antenna properties, e.g., antenna spacing, and precoder information relating to the phase changes between neighboring antenna elements, it is possible to determine the temporal advance or delay that the modification should have. For example, in LTE for four transmit antenna elements, "precoder 1" rotates the different layers, π/2 radians (or a quarter of a period). Assuming a carrier frequency of 700 MHz gives (t=l/f), the corresponding time displacement is 0.36 ns between adjacent antenna elements.

Figs. 6A and 6B illustrate pseudocode for implementation of a "test bench" to assess the noise reduction obtained according to the techniques disclosed herein. Fig. 7 illustrates pseudocode for clipping-noise reduction based on a "minimum norm" approach, and Fig. 8 illustrates pseudocode for a constrained optimization implementation, such as described earlier herein.

Of course, the algorithms represented in the pseudocode may be varied or extended without departing from the fundamental aspects of this disclosure. Broadly, this disclosure contemplates a method and apparatus for reducing the effects of transmitter clipping in a multi- antenna system. An example approach includes detecting that transmitter clipping will occur in a first signal mapped to a first antenna element, estimating the amount of clipping, modifying a second signal mapped to a second antenna element, and transmitting the modified set of signals on the set of antenna elements.

Of course, more than two signals may be included in the set, and more than one signal may be modified at any given instance. However, in keeping with the simplified example of two signals, with the second signal modified to compensate for clipping of the first signal, the modification of the second signal may be based on the assumption of equal channels from the two transmit antenna elements to the intended receiver, or may be based on the assumption of correlated channels from the two transmit antenna elements to the intended receiver, or based on the known or estimated channel from each antenna element to the intended receiver antenna. In at least one embodiment, the modification is done proportionally to the clipping contribution of the different layers, where each signal in the set of signals represents a different transmission "layer". The technique may include estimating the channels to multiple receiver antennas, which are combined by the targeted receiver to demodulate the intended transmission layer. In any case, in one or more embodiments, the modification of the second signal is based on minimizing the effect of clipping at the intended receiver antenna while simultaneously minimizing the effect of interference on the remaining receiver antennas. For example, consider the min₅||r— f ||₂ optimization described herein, where r is a vector and where the first vector element is used to receive the intended transmission layer. The effect of the clipping and compensation at the transmitter will be minimized on all elements of r, which is equivalent to minimizing the effect of interference on the remaining receiver antennas.

Clipping detection involves, for example, comparing each of the signals with a clipping threshold, which may be same for all signals, or which may be different as between two or more of the signals. In at least one embodiment, the clipping analysis/estimation is done by table lookup, which provides for excellent computational efficiency.

Whether done by table lookup or computed on the fly, the modification may be done according to the Euclidean distance algorithm and, in general, only the signal or signals in the set that will not be clipped at the corresponding time or segment are modified, because they have spare PA power. As a further refinement, the processing circuitry 30 may define one or more thresholds below the clipping threshold, and identify the signal or signals that are the candidates for modification, or the ones that are most preferred for modification, by comparing the non- clipping signals to such thresholds. More generally, the processing circuitry may prioritize the non-clipping signals for modification according to the amount of spare power available for each one of them.

Notably, modifications and other embodiments of the disclosed invention(s) will come to mind to one skilled in the art having the benefit of the teachings presented in the preceding descriptions and the associated drawings. Therefore, it is to be understood that the invention(s) is/are not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of this disclosure. Although specific terms may be employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.

Claims

CLAIMS What is claimed is:

1. A method (400) of multi-antenna signal transmission performed by a wireless communication apparatus (10), the method (400) comprising:

evaluating (402) individual signals in a set of signals to identify signal components, referred to as residual signal components, that will be clipped as a consequence of amplifier non-linearity, each signal in the set of signals to be amplified by a respective power amplifier (26-/) in a corresponding set of power amplifiers (26), for transmission from a respective transmit antenna (28-/) in a corresponding set of transmit antennas (28);

modifying (404) the set of signals to obtain a modified set of signals, by, for a given residual signal component in a given signal in the set of signals, adding a compensation value to each of one or more other ones of the signals, to compensate signal reception at a targeted receiver (12) for clipping of the given residual signal component in the given signal; and

transmitting (406) the modified set of signals, rather than the set of signals, by converting the modified set of signals into the analog domain, for amplification by the corresponding set of power amplifiers (26) and transmission from the corresponding set of transmit antennas (28).

2. The method (400) of claim 1, wherein the compensation value added to each of the one or more other ones of the signals is based at least in part on the given residual signal component.

3. The method (400) of claim 1 or 2, wherein the compensation value added to each of the one or more other ones of the signals is computed as a function of a channel estimate matrix of the channel between the transmit antennas (28) and one or more receiver antennas (56) of the targeted receiver (12).

4. The method (400) of any of claims 1-3, wherein modifying (404) the set of signals comprises, with respect to the given residual signal component in the given signal, identifying one or more of the remaining signals that will not be clipped at a time corresponding to the residual signal component and, for each identified signal, adding the residual signal component in whole or in part to the identified signal, as the compensation value for the identified signal.

5. The method (400) of any of claims 1-4, wherein modifying (404) the set of signals comprises computing compensation values from the residual signal components and adding the compensation values to selected ones of the signals in the set of signals, the compensation values being computed according a minimization function that minimizes the distance between an ideal received signal vector corresponding to transmission of the set of signals in the absence of clipping and a compensated received signal vector corresponding to transmission of the modified set of signals in the presence of clipping.

6. The method (400) of claim 5, wherein the minimization function is:

min_e||r - ||₂ ,

where r is the ideal received signal vector, r is the compensated received signal vector, and where

r = Hs, and

r = Hs = Hs + Η δ,

in which H is an N x M channel matrix for N antennas in the set of transmit antennas (28) and M antennas in a set of receiver antennas (56), s represents a vector of transmitted signals corresponding to the set of signals without any clipping by the set of power amplifiers (26), s comprises a vector of transmitted signals corresponding to the modified set of signals with clipping by the set of power amplifiers (26), s comprises a vector of transmitted signals corresponding to the set of signals with clipping by the set of power amplifiers (26) and without the addition of the compensation values, δ comprises a vector of the compensation values, and H-L is a reduced channel matrix comprising a subset of H and corresponding to the signals within the set of signals to which the compensation values are added.

7. The method (400) of any of claims 1-6, wherein each signal in the set of signals comprises a block of samples, and wherein evaluating (402) the individual signals in the set of signals to identify the signal components that will be clipped as a consequence of amplifier non- linearity comprises, for each signal, evaluating the block of samples to identify samples that will be clipped, based on a known amplifier clipping threshold.

8. The method (400) of claim 7, wherein the block of samples comprising each signal in the set of signals comprises a block of Fast Fourier Transform, FFT, coefficients.

9. The method (400) of any of claims 1-9, wherein the set of antennas (28) comprise a defined set (74) of correlated antenna elements (72) within an array (70) of antenna elements (72) that comprises a plurality of defined sets (74) of correlated antenna elements (72), and further comprising performing the method (400) with respect to each of the plurality of defined sets (74) of correlated antenna elements (72).

10. The method (400) of any of claims 1, 2, and 7-9, wherein the set of signals comprises a first signal and a second signal, and wherein modifying the set of signals comprises, for a given residual signal component identified in the first signal, adding, as the compensation value for the second signal, the given residual component in whole or in part to a corresponding segment of the second signal.

11. The method (400) of claim 10, wherein the first and second signals are time aligned and wherein the corresponding segment of the second signal is a segment that will be transmitted coincidentally with a segment of the first signal corresponding to the residual signal component, accounting for any temporal shifting applied for multi-antenna transmission of the modified set of signals.

12. A wireless communication apparatus (10) configured for multi-antenna transmission, the wireless communication apparatus (10) comprising:

transmission circuitry (22), including a set of power amplifiers (26) for amplifying

signals for transmission from a corresponding set of antennas (28); and processing circuitry (30) included in or operatively associated with the transmission circuitry (22) and configured to:

evaluate individual signals in a set of signals to identify signal components, referred to as residual signal components, that will be clipped as a consequence of amplifier non-linearity, each signal in the set of signals to be amplified by a respective power amplifier (26-/) in the corresponding set of power amplifiers (26), for transmission from a respective transmit antenna (28-/) in the corresponding set of transmit antennas (28);

modify the set of signals to obtain a modified set of signals, by, for a given

residual signal component in a given signal in the set of signals, adding a compensation value to each of one or more other ones of the signals, to compensate signal reception at a targeted receiver (12) for clipping of the given residual signal component in the given signal; and

transmit the modified set of signals, rather than the set of signals, by converting the modified set of signals into the analog domain, for amplification by the corresponding set of power amplifiers (26) and transmission from the corresponding set of transmit antennas (28).

13. The wireless communication apparatus (10) of claim 12, wherein the compensation value added to each of the one or more other ones of the signals is based at least in part on the given residual signal component.

14. The wireless communication apparatus (10) of claim 12 or 13, wherein the compensation value added to each of the one or more other ones of the signals is computed as a function of a channel estimate matrix of the channel between the transmit antennas (28) and one or more receiver antennas (56) of the targeted receiver (12).

15. The wireless communication apparatus (10) of any of claims 12-14, wherein the processing circuitry (30) is configured to modify the set of signals by, with respect to the given residual signal component in the given signal, identifying one or more of the remaining signals that will not be clipped at a time corresponding to the residual signal component and, for each identified signal, adding the residual signal component in whole or in part to the identified signal, as the compensation value for the identified signal.

16. The wireless communication apparatus (10) of any of claims 12-15, wherein the processing circuitry (30) is configured to modify the set of signals by computing compensation values from the residual signal components and adding the compensation values to selected ones of the signals in the set of signals, the compensation values being computed according a minimization function that minimizes the distance between an ideal received signal vector corresponding to transmission of the set of signals in the absence of clipping and a compensated received signal vector corresponding to transmission of the modified set of signals in the presence of clipping.

17. The wireless communication apparatus (10) of claim 16, wherein the minimization function is:

min_e||r - ||₂ ,

r = Hs, and

r = Hs = Hs + Η δ, in which H is an N x M channel matrix for N antennas in the set of transmit antennas (28) and M antennas in a set of receiver antennas (56), s represents a vector of transmitted signals corresponding to the set of signals without any clipping by the set of power amplifiers (26), s comprises a vector of transmitted signals corresponding to the modified set of signals with clipping by the set of power amplifiers (26), s comprises a vector of transmitted signals corresponding to the set of signals with clipping by the set of power amplifiers (26) and without the addition of the compensation values, δ comprises a vector of the compensation values, and H-L is a reduced channel matrix comprising a subset of H and corresponding to the signals within the set of signals to which the compensation values are added.

18. The wireless communication apparatus (10) of any of claims 12-17, wherein each signal in the set of signals comprises a block of samples, and wherein the processing circuitry (30) is configured to evaluate the individual signals in the set of signals to identify the signal components that will be clipped as a consequence of amplifier non-linearity by, for each signal, evaluating the block of samples to identify samples that will be clipped, based on a known amplifier clipping threshold.

19. The wireless communication apparatus (10) of claim 18, wherein the block of samples comprising each signal in the set of signals comprises a block of Fast Fourier Transform, FFT, coefficients.

20. The wireless communication apparatus (10) of any of claims 12-20, wherein the set of antennas (28) comprise a defined set (74) of correlated antenna elements (72) within an array (70) of antenna elements (72) that comprises a plurality of defined sets (74) of correlated antenna elements (72), and wherein the processing circuitry (30) is configured to perform the

identification, evaluation, and modification operations with respect to each of the plurality of defined sets (74) of correlated antenna elements (72).

21. The wireless communication apparatus (10) of any of claims 12, 13, and 18-20, wherein the set of signals comprises a first signal and a second signal, and wherein the processing circuitry (30) is configured to modify the set of signals by, for a given residual signal component identified in the first signal, adding, as the compensation value for the second signal, the given residual component in whole or in part to a corresponding segment of the second signal.

22. The wireless communication apparatus (10) of claim 21, wherein the first and second signals are time aligned and wherein the corresponding segment of the second signal is a segment that will be transmitted coincidentally with a segment of the first signal corresponding to the residual signal component, accounting for any temporal shifting applied for multi-antenna transmission of the modified set of signals.