Method and apparatus for crest factor reduction

• US 8,724,721 B2
• Filed: 02/22/2011
• Issued: 05/13/2014
• Est. Priority Date: 02/22/2010
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Soler Garrido, Josep Vicent
• Primary Examiner(s): Fan, Chieh M
• Assistant Examiner(s): Koeth, Michelle M
• Application Number:
  US13/032,347
• Publication Number:
  US 20110228872A1
• Time in Patent Office: 1,176 Days
• Field of Search: 455/501, 375/260
• US Class Current: 375/260
• CPC Class Codes: H04L 27/2623 Reduction thereof by clipping

Abstract

According to embodiments there is provided a method for reducing the crest factor of an input signal. The method is performed in a part of a transmitter and comprises creating a clipped signal by clipping the input signal using a predetermined clipping threshold and creating a peak signal from the input signal. The peak signal comprises the difference between the input signal and the clipped signal. At least part of the peak signal is convoluted with an inverse Fourier transform of a filtering mask defined in the frequency domain. In this way a filtered signal is created. A subtraction signal is generated by subtracting the filtered signal from the input signal.

Claims

1. A method for reducing the crest factor of an input signal related to a symbol, the method performed in a part of a transmitter and comprising:

creating a clipped signal by clipping the input signal using a predetermined clipping threshold;

subtracting the clipped signal from the input signal to create a peak signal;

selecting only a predetermined number of data points of the peak signal, wherein the predetermined number of data points are the predetermined number of data points of the peak signal with the highest magnitudes;

convoluting the selected data points of the peak signal with the inverse Fourier transform of a filtering mask defined in the frequency domain to create a filtered signal; and

creating a subtraction signal by subtracting the filtered signal from the input signal;

wherein an initial peak signal is created prior to receipt of all data points of the input signal relating to the symbol and wherein said convolution is started based on the initial peak signal; and

wherein further data points are added to the initial peak signal as the further data points are received as part of the input signal relating to the said symbol, thereby creating an updated initial peak signal and wherein said convolution is continued based on the updated initial peak signal.

2. A method according to claim 1, wherein the inverse Fourier transform of the filtering mask is generated once and stored in a memory.

3. A method according to claim 1, further comprising computing the inverse Fourier transform of the filtering mask using a processor of the transmitter, wherein the processor is also adapted to generate the input signal by computing the inverse Fourier transform of signals that are to be transmitted by the transmitter.

4. A method according to claim 1, wherein only the largest values of the inverse Fourier transform of the filtering mask representing at least 80%, at least 90% or at least 99% of the energy of the inverse Fourier transform of the filtering mask are used in convoluting the at least a part of the peak signal.

5. A method according to claim 1, wherein at least 80%, at least 90% or at least 99% of the energy of the inverse Fourier transform of the filtering mask is concentrated within a first number of data points of the inverse Fourier transform of the filtering mask, centred on the origin of the inverse Fourier transform of the filtering mask;

and

wherein the input signal is a part of the inverse Fourier transform of a symbol of an OFDM signal extending from a first data point of the inverse Fourier transform and comprising at least half of the first number of data points.

6. A method according to claim 5, wherein the method is consecutively applied to a plurality of input signals, the plurality of input signals relating to the same symbol of an OFDM signal, wherein a first input signal of the plurality of input signals is part of the inverse Fourier transform of the symbol, the part comprising at least a number of data points that is equal to half of the first number of data points, the first data point of the part being the first data point of the inverse Fourier transform;

and

wherein subsequent input signals are generated by receiving further parts of the inverse Fourier transform of the symbol and appending the received further parts to a previous input signal.

7. A method according to claim 1, further comprising determining the error vector magnitude of the subtraction signal and, if the error vector magnitude exceeds a maximum acceptable error vector magnitude, setting a data point of the peak signal that has the smallest magnitude of the data points of the peak signal to zero.

8. A method for reducing the crest factor of a time domain signal relating to a symbol, the method performed in a transmitter and comprising:

creating a clipped signal by clipping the time domain signal using a predetermined clipping threshold;

subtracting the clipped signal from the signal to create a peak signal;

selecting only a predetermined number of data points of the peak signal, wherein the predetermined number of data points are the predetermined number of data points of the peak signal with the highest magnitudes;

generating a filtered peak signal by transforming at least a part of the peak signal into the frequency domain, creating a filtered signal by applying a frequency domain filtering mask to the transformed peak signal and transforming the filtered signal into the time domain;

subtracting the transformed filtered signal from the signal;

wherein an initial peak signal is created prior to receipt of all data points of the time domain signal relating to the symbol and wherein said convolution is started based on the initial peak signal; and

wherein further data points are added to the initial peak signal as the further data points are received as part of the time domain signal relating to the said symbol, thereby creating an updated initial peak signal and wherein said convolution is continued based on the updated initial peak signal.

9. A method according to claim 8, further comprising applying complex weights to the data values in the at least part of the peak signal prior to convolution or transformation.

10. An apparatus comprising an input for receiving a time domain input signal, a processor arranged to create a clipped signal by clipping the input signal using a predetermined clipping threshold, to create a peak signal subtracting the clipped signal from the input signal, to select only a predetermined number of data points of the peak signal, wherein the predetermined number of data points are the predetermined number of data points of the peak signal with the highest magnitudes'"'"' to convolute at least a part of the peak signal with the inverse Fourier transform of a filtering mask defined in the frequency domain to create a filtered signal, and to create a subtraction signal by subtracting the filtered signal from the input signal;

wherein the input signal relates to a symbol and wherein an initial peak signal is created prior to receipt of all data points of the time domain input signal relating to the symbol and wherein said convolution is started based on the initial peak signal; and

wherein further data points are added to the initial peak signal as the further data points are received as part of the time domain input signal relating to the said symbol, thereby creating an updated initial peak signal and wherein said convolution is continued based on the updated initial peak signal.

11. An apparatus according to claim 10, wherein the processor comprises a number of multipliers, the processor arranged to input data values of the peak signal into respective multipliers, the processor further arranged to input values of the inverse Fourier transform of the filter mask into the said respective multipliers, each of the input values of the inverse Fourier transform of the filter mask associated with a respective one of the data values of the peak signal.

12. An apparatus according to claim 10, further comprising a memory storing at least part of the inverse Fourier transform of the filtering mask.

13. An apparatus according to claim 12, further comprising an IFFT processor arranged to perform an inverse Fourier transform of a signal to create the input signal, the IFFT processor further arranged to calculate the inverse Fourier transform of the filtering mask during an idle period and store the inverse Fourier transform of the filtering mask in the memory.

14. An apparatus according to claim 10, further comprising a peak sorting arrangement, arranged to sort the peaks of the input signal in order of magnitude and to select a predetermined number of peaks for convolution.

15. An apparatus according to claim 10, further arranged to convolute said peak signal in real time after a predetermined number of peak signal values have become available for convolution.

16. An apparatus according to claim 10, wherein the processor is further arranged to determine whether the error vector magnitude of the subtraction signal is above a predetermined threshold and to set a smallest non-zero value of the peak signal that has been used or that would be used in creating the subtraction signal to zero prior to the convoluting.

17. An apparatus for reducing the crest factor of a time domain signal comprising a signal input for receiving the time domain signal and a processor arranged to:

create a clipped signal by clipping the signal using a predetermined clipping threshold;

create a peak signal by subtracting the clipped signal from the input signal;

select only a predetermined number of data points of the peak signal, wherein the predetermined number of data points are the predetermined number of data points of the peak signal with the highest magnitudes;

transform the at least a part of the peak signal into the frequency domain;

create a filtered signal by applying a frequency domain filtering mask to the transformed peak signal;

transform the filtered signal into the time domain; and

subtract the transformed filtered signal from the signal;

wherein the time domain signal relates to a symbol and wherein an initial peak signal is created prior to receipt of all data points of the time domain signal relating to the symbol and wherein said convolution is started based on the initial peak signal; and

wherein further data points are added to the initial peak signal as the further data points are received as part of the time domain signal relating to the said symbol, thereby creating an updated initial peak signal and wherein said convolution is continued based on the updated initial peak signal.

18. An apparatus according to claim 17, the processor further arranged to apply complex weights to the at least a part of the peak signal prior to convolution or transformation.