WO2011103809A1 - 一种信号波峰削除的方法及设备 - Google Patents

一种信号波峰削除的方法及设备 Download PDF

Info

Publication number
WO2011103809A1
WO2011103809A1 PCT/CN2011/071277 CN2011071277W WO2011103809A1 WO 2011103809 A1 WO2011103809 A1 WO 2011103809A1 CN 2011071277 W CN2011071277 W CN 2011071277W WO 2011103809 A1 WO2011103809 A1 WO 2011103809A1
Authority
WO
WIPO (PCT)
Prior art keywords
peak
signal
clipping
peaking
peak clipping
Prior art date
Application number
PCT/CN2011/071277
Other languages
English (en)
French (fr)
Inventor
肖鹏
熊军
Original Assignee
电信科学技术研究院
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 电信科学技术研究院 filed Critical 电信科学技术研究院
Publication of WO2011103809A1 publication Critical patent/WO2011103809A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L27/00Modulated-carrier systems
    • H04L27/26Systems using multi-frequency codes
    • H04L27/2601Multicarrier modulation systems
    • H04L27/2614Peak power aspects

Definitions

  • the present invention relates to the field of mobile communications, and in particular to a method and apparatus for erasing signal peaks. Background technique
  • the linearity of the transmitting end is very high.
  • PAPR Peak to Average Power Ratio
  • the power is too large. It will cause a lot of waste.
  • the peak-to-average ratio will cause the performance of the system to deteriorate sharply. Too much peak-to-average ratio will directly affect the operating cost and efficiency of the whole system. Therefore, it is necessary to find a way to reduce the peak-to-average power ratio.
  • the signal when the signal is composed of multiple subcarriers, if the peak-to-average ratio of the subcarrier is too high, the nonlinear distortion of the signal will be caused, the orthogonality between the subcarriers will be destroyed, and the out-of-band interference will be increased, thereby deteriorating the performance of the system. How to properly and effectively remove the peak signal power point becomes a key issue.
  • OFDM Orthogonal Frequency Division Multiple
  • OFDM Orthogonal Frequency Division Multiple
  • ITI Inter-Symbol Interference
  • MCM Multi-carrier Modulation
  • the OFDM system subcarriers have 1/2 overlap but remain orthogonal to each other. Information is separated at the receiving end by correlation demodulation techniques.
  • FIG. 1 it is a superposition process of observing 4 carriers from the time domain.
  • Subcarrier superposition each time is a carrier superposition of 4 different frequencies. The transit time of a symbol becomes NT, which is absolutely oversampled for the lowest frequency. But for the highest frequency signal, the Nyquist sampling theorem is satisfied.
  • the PAR of the signal increases sharply.
  • OFDM-TDD Time Division Duplexing
  • a radio frame is 10ms, consisting of 2 fields, each half frame is 5mS in length, and the half frame is divided into 5 subframes (subframes). 4 regular subframes and one special subframe.
  • Each regular subframe is divided into 2 slots, so the regular slot length is 0.5 mS, and the special subframe is composed of three special slots of DwPTS, GP, and UpPTS.
  • the DwPTS and UpPTS lengths are configurable, but the total length of DwPTS, GP, and UpPTS is required to be lmS. There are 15360 sample points per time slot.
  • a sub-frame has two slots, each slot has ⁇ (7) symbols, one subframe has 14 symbols, each symbol has ⁇ a RB (Resource Block, resource blocks).
  • One subcarrier under one symbol is uniquely identified as RE under one time slot.
  • the division of different symbol reference signals Different cloths have different reference signal distributions for different antennas. Since not all symbols have reference signals, and the RBs allocated by the service are randomly allocated, after the symbols are respectively subjected to IFFT (Inverse Fast Fourier Transform) processing, some symbols exceed the peak point of the threshold. Rarely, there are many peak points where some symbols exceed the threshold. This results in a large probability of statistical oscillations in signal amplitude. Each of the peak points with fewer peak points has a large number of sample points that exceed the threshold. The number of sampling points where each peak point exceeds the threshold in the symbol with more peak points will be less.
  • IFFT Inverse Fast Fourier Transform
  • the PC-CFR Puleak Cancellation Crest Factor Reduction
  • PC-CFR Peak Cancellation Crest Factor Reduction
  • CFR is a digital peak clipping module. It mainly performs peak clipping on the signal above the threshold, reduces the average power of the signal, and further suppresses the click.
  • the CFR module is mainly divided into six sub-modules, and its functions are as follows:
  • DELAY Module Since the link that produces the peak clipping sequence has a fixed delay, the module completes the fixed delay of the input data, maintaining the same time as the peak clipping sequence generated, and finally synchronizing the input subtraction module.
  • PEAK_DETECT module The detection module of the peak point, all the parts above the threshold are subjected to peak discrimination.
  • is the detected peak sequence; ⁇ is the threshold, angle(x).
  • PEAK_ALLOCATOR module The peak distribution module assigns a peak clipping generator (CPG) module to each peak detected.
  • CPG peak clipping generator
  • the peak clipping pulse generation module mainly stores a fixed length peak clipping sequence. Each CPG module can only process one peak point at a time. Once in the idle state, the PEAK_ALLOCATOR module assigns a peak to the idle CPG module. Processing, if all CPG modules are not idle, no peaking will be performed for the next detected peak point.
  • a CPG requires a complex multiplier with 4 multipliers. If the clock rate is 4 of the signal processing rate, then the 4 multipliers are time-shared. Reuse. So a complex multiplier can complete the operation of 4 CPGs.
  • SUM module Adds the outputs of all CPG modules to form a peak clipping sequence.
  • SUBTRACT module After passing the original signal through the DELAY module, subtract the peak clipping sequence formed by the SUM module. The structure is the final output of the CFR module.
  • NS-CFR Noise Shaping Crest Factor Reduction
  • Amplitude and angle calculation module Calculate ( ⁇ ⁇ ⁇ - ⁇ ), where X is the input signal amplitude,
  • is the peak clipping threshold and ⁇ is the angle of the input signal. Since the sampling rate of the input signal is generally lower than the FPGA operating clock, the I and Q channels can be time-multiplexed with one multiplier.
  • Peak Find Module Detects if there is a signal that exceeds the peak window.
  • the filter performs noise shaping: filtering and shaping the signal in the peak window.
  • This part of I, Q is time-division multiplexed with one filter. Since the filter can be generally designed as a symmetric filter, in order to save the multiplier, symmetric pulsation is adopted. structure.
  • Data Delay Module Delays the input data to ensure that the input signal and the peak clipping signal are synchronized.
  • NS-CFR FFT (Fast Fourier Transform) transform
  • IFFT IFFT
  • the above algorithms have their own defects, which are difficult to implement.
  • the PC-CFR algorithm is difficult to effectively cut the peak for the discontinuous allocation of subcarriers;
  • the NS-CFR algorithm consumes a lot of resources.
  • the cyclic limiting filtering algorithm has large resources and long delay, which is difficult to implement by FPGA.
  • An object of the present invention is to solve at least one of the above technical drawbacks, and in particular to solve the problem of peak clipping of an input signal and reduction of a peak-to-average ratio of a signal.
  • an embodiment of the present invention provides a method for signal peak clipping, which includes the following steps:
  • the sidelobe compression peaking RC1 operation is performed; and the DUC (Digital Up Conversion) intermediate frequency filtering is performed on the signal subjected to the sidelobe compression peaking RC1 operation;
  • Another embodiment of the present invention further provides a signal peak clipping apparatus, including an RC1 peak clipping module, an intermediate frequency filtering module, and an RC2 peak clipping module.
  • the RC1 peak clipping module is configured to perform side-edge compression peak clipping RC1 operation after performing IFFT transformation on the physical layer signal;
  • the intermediate frequency filtering module is configured to operate the side-clip compression peaking RC1
  • the signal is subjected to digital up-converter DUC intermediate frequency filtering;
  • the RC2 peak clipping module is configured to perform a sidelobe compression peaking RC2 operation on the DUC intermediate frequency filtered signal.
  • the above technical solution proposed by the present invention can eliminate the peak power in any subcarrier configuration; at the same time, the processing delay can be guaranteed to be within 20 us, and the processing delay is small.
  • the above technical solution proposed by the present invention consumes less hardware equipment resources, especially the hardware multiplier resources are less than 30, and the system performance indicators such as ACPR and EVM deteriorate to a lesser extent than the existing algorithms.
  • 1 is a schematic diagram of superposition of 4 subcarriers in a time domain
  • FIG. 2 is a schematic diagram of a frame structure type 2 in an LTE-TDD downlink slot structure
  • FIG. 3 is a schematic diagram of a reference signal of a downlink map
  • Figure 4 is a schematic diagram of PC-CFR peak clipping
  • Figure 5 is a comparison of amplitudes before and after peak-cutting of PC-CFRR
  • Figure 6 is a general block diagram of the FPGA implementation of NS-CFR
  • Figure 7 is a comparison of amplitudes before and after NS-CFR peaking
  • FIG. 8 is a flowchart of a method for implementing signal peak clipping according to an embodiment of the present invention
  • FIG. 9 is a general block diagram of an algorithm for implementing RC-CFR in two steps according to an embodiment of the present invention
  • FIG. 10 is a schematic diagram of a signal full-lobe intention according to an embodiment of the present invention
  • FIG. 11 is a schematic diagram of signal main lobe compression according to an embodiment of the present invention.
  • Figure 13 is a spectrum diagram of the intermediate frequency filter
  • FIG. 14 is a spectrum diagram of a PRB processed by a DUC after random clipping of a PRB;
  • FIG. 15 is a schematic diagram of oversampling;
  • 16 is a schematic diagram of an emulation link for verifying a peak RC-CFR re-start;
  • FIG. 17 is a schematic diagram showing peak-to-average ratio rise of different digital intermediate frequency rate signals after passing through a DAC;
  • Figure 19 is a spectrum diagram of a prototype filter with different peak clipping bandwidths
  • Figure 20 is a schematic diagram of filtering after peak clipping in 100RB configuration
  • Figure 21 is a comparison of amplitudes before and after peak cutting of 100 PRB;
  • Figure 22 is a comparison of amplitudes before and after 40 PRB peak clipping
  • Figure 23 is a comparison of amplitudes before and after 6 PRB peak clipping
  • FIG. 24 is a schematic structural diagram of an apparatus for signal peak clipping according to an embodiment of the present invention
  • FIG. 25 is a spectrum diagram before and after peak-shaping filtering after in-phase compression.
  • the embodiment of the invention provides a method for signal peak clipping, which comprises the following steps:
  • a flow chart of a method for implementing signal peak clipping according to an embodiment of the present invention includes the following steps:
  • step S101 After performing an IFFT transformation on the physical layer signal, performing a sidelobe compression peaking RC1 operation.
  • step S101 is a general block diagram of a two-step implementation of the RC-CFR (Reduced Peak and PC-CFR, reduced peak and peak cancellation peak factor reduction) embodiment of the present invention.
  • the core idea of the RC algorithm is to use a multi-stage sidelobe compression peak clipping algorithm.
  • the design idea of the sidelobe compression peak clipping algorithm is as follows:
  • Block physical resource block
  • the number of corresponding subcarriers to determine the number of sampling points corresponding to one PRB in the core clipping vector. Then, according to the PRB index sent by the physical layer, the frequency domain response of the core clipping vector of the time slot is determined, and the core clipping peak vector is obtained according to the obtained frequency domain response. For the achievability, the physical layer is not transmitted.
  • the PRB index uses a uniform core clipping vector for peak clipping filtering.
  • the time domain expression of the core clipping vector is obtained as follows, and the frequency domain of this core clipping vector is the subcarrier full configuration:
  • the core clipping vector is not adaptively adjusted with the configuration of the subcarriers, and a core clipping vector is fixed. Different subcarrier configurations use the same core clipping vector.
  • the sidelobe suppression algorithm is that the main lobe remains unchanged, and the adjacent sidelobes are compressed. In general, the main lobe is at the center of the clipping vector, and the sidelobe compression operations on both sides of the main lobe are:
  • O ⁇ compress_gaiii is a compression factor and a positive number less than 1.
  • the complexity of this compression algorithm is still relatively high, and the design of the core clipping vector is difficult, and there are some defects in the leakage.
  • the peak cutter has only one main lobe, so as long as the peak signal exceeding the threshold is detected, it can be removed one by one. There is no leakage in the advantage, and the peak value is not affected when the peak is removed. Therefore, the EVM deterioration caused by clipping is minimized, and the peak clipping effect is most obvious. This is similar to the beamforming of smart antennas.
  • the signals after multi-antenna shaping are all received by the user. The power received by the user is the largest and the interference to other users is minimal.
  • FIG. 10 is a schematic diagram of the full-valve of the signal, and it can be seen that there is no compression of the sidelobe signal substantially coincident;
  • FIG. 11 is a schematic diagram of the compression of the main lobe of the signal, and it can be seen that only one signal is retained after the sidelobe compression.
  • the sidelobe compression peaking RC1 operation includes a main-valve peaking operation, the main-valve peaking operation is an in-phase hard-cut peak operation, and the main-valve clipping operation retains only the maximum signal in the main lobe of the signal.
  • the main-lobe clipping algorithm after the IFFT is the physical layer in-phase hard-cut peak operation, and the in-phase clipping algorithm is performed before the DUC processing is as follows:
  • the main work of the above algorithm is the calculation of the amplitude and phase of the signal.
  • the above algorithm is modified as follows:
  • ⁇ ( ⁇ ) is the signal after the IFFT transform of the physical layer signal, and A is a predetermined peak clipping threshold.
  • Normal filtering is then performed at the intermediate frequency, and normal DUC processing after peak clipping can effectively suppress spurious caused by out-of-band clipping. Therefore, no additional IFFT/FFT operations are needed at this time, which greatly reduces the amount of computation and shortens the delay.
  • peak clipping processing can be performed by time division multiplexing, as shown in Fig. 12, which is a schematic diagram of peak clipping processing for multi-antenna multiplexing.
  • Fig. 12 is a schematic diagram of peak clipping processing for multi-antenna multiplexing.
  • the signals of multiple antennas can be time division multiplexed.
  • the physical layer hard-cutting module is input, so that the resources used for peak clipping of multiple antennas are the same as the resources consumed by one antenna clipping, which saves the use of hardware resources.
  • S102 Perform digital up-converter DUC intermediate frequency filtering on the signal subjected to side-lobe compression peaking RC1 operation.
  • the IF filter needs to have a steep transition band to remove the effects of sideband distortion.
  • the intermediate frequency filter is implemented as a multi-stage filter combination. As shown in Figure 13, for the IF filter spectrum, the steep transition zone can be used to match the effects of the physical layer clipping.
  • In-phase limiting peak clipping is a nonlinear process that will cause some in-band noise and out-of-band interference.
  • the out-of-band interference is removed by the subsequent filtering.
  • the spectrum of the PRB is randomly configured after the peak clipping is performed. After the filtering is completed, no out-of-band interference is observed.
  • the physical layer After the physical layer completes the peak clipping, it is processed by the DUC filter interpolation process and reaches the intermediate frequency.
  • the IF signal is subjected to peak clipping operation.
  • IF side-cutting Compression peak clipping algorithm such as PC-CFR algorithm.
  • the reason for the DUC peak regeneration is that after the physical layer signal is subjected to interpolation filtering, since the digital filter order used in the hardware is limited, it is not an ideal filter, so a certain passband ripple jitter and a stopband attenuation are stored. It is also limited, and the amount of data in the physical layer is multiplied, so that after the physical layer signal passes through the intermediate frequency, there will be a peak re-start. Similarly, if there is an interpolation filter after PC-CFR, the peak regeneration problem will also occur, because the signal after clipping is passed through the DAC (Digital to Analog Converter), and the DAC internally increases the number of signals.
  • DAC Digital to Analog Converter
  • the input clipping signal is again interpolated and the input signal is oversampled.
  • Sampling at a rate higher than twice the signal bandwidth is called oversampling.
  • Oversampling is a very important function because it gives gain to the signal-to-noise ratio of the received signal in the digital domain.
  • the faster the sampling rate the lower the quantization noise level. Since the quantization noise is distributed over a wider frequency band, that is, within the sampling clock frequency, and the total noise is constant, the oversampling schematic is shown in FIG.
  • Df s the rate at which the signal is sampled, so as long as
  • SNR is the signal-to-noise ratio
  • N is the number of bits of the ADC (Analog to Digit Converter) / DAC
  • Fs is the frequency of the ADC/DAC sampling clock
  • f BW is the signal bandwidth.
  • the signal after the peak clipping enters the D AC signal will usually be interpolated again, usually using a multi-stage HB (Half Band) filter.
  • the simulation platform is verified according to the following simulation platform, as shown in Figure 16, to verify the simulation link of the RC-CFR peak re-start.
  • Multi-carrier signals with different intermediate frequency rates fs are subjected to PC-CFR clipping and then processed by three cascaded HB filters (HB1, HB2, HB3) to observe the A point after CFR and C and D after passing through the HB filter. The peak-to-average ratio of points.
  • the signal rate after the intermediate frequency filtering conforms to f s ⁇ f s . , where the rate f s is at f s .
  • the PAR rise of the signal after interpolation filtering at a rate is less than or equal to a predetermined threshold LdBc such that PAR(f s ) - PAR(f s0 ) ⁇ IdBc.
  • the digital IF signal does have a certain rise after three HBs, but the higher the data rate, the smaller the PAR rise, and a minimum rate f s is specified .
  • the PAR rise of the signal is less than 0.3dBc: APAR(f s0 ) ⁇ 0.3dBc, and the signal rate after PC-CFR is f s ⁇ f s . Yes, that is PAR( f s ) - PAR( f s .) ⁇ 0.3dBc.
  • the signal rate of the PC-CFR output is only greater than f s ⁇ 92.16MHZ.
  • the peak-to-average ratio is raised below 0.2dBc to meet the system requirements. That is, in the above embodiment, LdBc takes a value of 0.3 dBc.
  • the physical layer is interpolated and filtered, and the peak is re-started. At this time, the peak is limited by the full-cut peak.
  • the sidelobe compression peaking RC2 operation includes full-valve peaking operation, the full-valve peaking operation does not compress the sidelobes, and the full-valve peaking operation is the peak cancellation peak factor reduction PC-CFR. Peak clipping operation.
  • the intermediate frequency uses the side-lobe compression algorithm to remove the remaining peak power.
  • the sidelobe compression algorithm only compresses the sidelobes of the distal end, and the side lobes of the proximal end remain unchanged, thus ensuring that the overall ACPR does not deteriorate much. If the intermediate frequency ACPR is not large enough, the side lobes are not compressed at this time. In this case, the full-lobe peaking algorithm is used to remove the remaining peaks. Since most of the peaks have been removed before entering the CFR, the full lobed at this time
  • the peak device can be designed in a single tube.
  • the main-edge peak-cut + IF full-valve peak in the physical layer has the following advantages: Regardless of how the physical layer sub-carriers are configured, the peak can be effectively removed, although the peak is re-started, but the peak re-start is generally PAR. Will exceed 7.0dBc.
  • PC-CFR One of the important settings of PC-CFR is the peak-cut prototype filter, which matches the spectrum of the OFDM signal after DUC.
  • Filter_f firls(cfr_ntaps-l, [0 fl f2 1], [1 1 0 0], [1 10])...
  • the parameter beta determines the attenuation of the side lobes as follows:
  • the peak-to-peak filtering scheme is configured for 100RB, as shown in Figure 21, which is the amplitude comparison graph before and after the peak clipping of 100 PRB.
  • Fig. 23 is the comparison chart of the amplitude before and after the 6 PRB peak clipping.
  • the purpose and difference of RC-CFR peak clipping algorithm for sidelobe compression before and after DUC is that the main peak clipping before DUC is to remove most of the peak power points.
  • the full-lobe peak after DUC filtering is mainly to remove DUC filtering.
  • the recurrence peak caused.
  • the above method for signal peak clipping disclosed in the embodiment of the present invention can eliminate the peak power in any subcarrier configuration; at the same time, the processing delay can be guaranteed to be within 20 us, and the processing delay is small.
  • the method for signal peak clipping mentioned in the embodiment of the present invention consumes less hardware equipment resources, especially the hardware multiplier resource is less than 30, and the system performance indicators such as ACPR and EVM deteriorate to a lesser extent than the existing algorithms.
  • the above method for signal peak clipping proposed by the embodiment of the present invention has little change to the existing system, does not affect the compatibility of the system, and achieves a single and high efficiency.
  • FIG. 24 it is a schematic structural diagram of an apparatus 100 for signal peak clipping according to an embodiment of the present invention, which includes an RC1 peak clipping module 110, an intermediate frequency filtering module 120, and an RC2 peak clipping module 130.
  • the RCl peak clipping module 110 is configured to perform a sidelobe compression peaking RC1 operation after performing an IFFT transformation on the physical layer signal.
  • the RC1 peak clipping module 110 performs sidelobe compression peak clipping.
  • the RC1 operation includes a main lobe clipping operation.
  • the main lobe clipping operation is an in-phase hard cutting peak operation.
  • the main lobe clipping operation only retains the maximum signal in the main lobe of the signal.
  • the in-phase hard cutting peak operation performed by the RC1 peak clipping module 110 includes,
  • the output signal obtained by the RC1 peak clipping module 110 after the input signal is subjected to the in-phase hard cutting peak operation is:
  • x(n) is the signal after the IFFT transform of the physical layer signal
  • A is the predetermined peak clipping threshold
  • the ⁇ ( ⁇ ) signal is combined with the data of the multiple antennas before the peak clipping, and the signals of the multiple antennas are input into the physical layer hard-cutting module in a time division multiplexed manner to complete the in-phase hard cutting peak operation.
  • the intermediate frequency filtering module 120 is configured to perform digital up-converter DUC intermediate frequency filtering on the signal subjected to the side-lobe compression peaking RC1 operation.
  • the DUC intermediate frequency filtering in the intermediate frequency filtering module 120 is implemented as a multi-stage filter combination.
  • the signal rate f s in the intermediate frequency filtering module 120 conforms to f s ⁇ f s . , where the rate f s is at f s .
  • the PAR rise of the signal is less than or equal to the predetermined threshold LdBc, such that PAR(f s )- PAR(f s0 ) ⁇ MBc.
  • the RC2 peak clipping module 130 is configured to perform sidelobe compression peaking RC2 operation on the signal filtered by the DUC intermediate frequency.
  • the sidelobe compression peaking RC2 operation performed by the RC2 peak clipping module 130 includes a full-lobe peaking operation, the full-valve peaking operation does not compress the side lobes, and the full-valve peaking operation is a peak cancellation peak factor reduction PC. -CFR peak clipping operation.
  • the PC-CFR peak clipping operation performed by the RC2 peak clipping module 130 is After looking for the peak power point of the signal, the peak power point is pulse-cut.
  • the RC2 peak clipping module 130 performs a PC-CFR peak clipping operation, and the PC-CFR peak clipping prototype coefficient passband bandwidth is less than or equal to the DUC intermediate frequency filtered signal band.
  • the device for eliminating the signal peaks disclosed in the embodiment of the present invention can eliminate the peak power in any subcarrier configuration; and the processing delay can be guaranteed to be within 20 us, and the processing delay is small.
  • the above-mentioned signal peak clipping device proposed by the embodiment of the present invention consumes less hardware equipment resources, especially the hardware multiplier resource is less than 30, and the system performance indicators such as ACPR and EVM deteriorate to a lesser extent than the existing algorithms.
  • the above-mentioned signal peak clipping device proposed by the embodiment of the present invention has small changes to the existing system, does not affect the compatibility of the system, and realizes the single and high efficiency.
  • RC-CFR is the best, PAR It can be stably suppressed to below 6.5dBc, and EVM can be controlled below 7%.
  • Figure 25 it is a schematic diagram of the peak-cutting performance curve of RC-CFR for various RB configurations.
  • RC-CFR not only has good performance, but also has less resources and short delay, which is easy to implement in FPGA hardware.
  • the RC-CFR algorithm is very close to the PC-CFR algorithm, and the added resources are few.
  • NS-CFR occupies a lot of hardware multipliers. From the peak clipping effect, the signal at other locations is less affected, except for the peak, which is due to the physical layer completing most of the peak clipping.
  • the NS-CFR algorithm In addition to the disadvantages of large resource consumption, the NS-CFR algorithm often suffers from insufficient peak clipping.
  • the peak window threshold is set at 7.5dB, and the peak clipping threshold is changed separately. Three sets of cases are simulated. When the PRB is assigned to 100 and 4, the signal after peak clipping can converge to about 7.5dB. PRB allocation At 40 and 12, the peak cut-off threshold is greatly affected, and the probability of peak re-occurrence is also large.
  • the biggest drawback of the PC-CFR peak clipping algorithm is that the peak clipping effect is unstable, especially in the case of few subcarrier configurations.
  • the stability of PAR reduction is generally good. So in summary, the RC-CFR algorithm is optimal or suboptimal in terms of resource saving, delay, and effectiveness of erasure. Therefore, LTE-CFR selects RC-CFR.
  • the PAR of the signal can be stably controlled below 7dBC.
  • each functional unit in each embodiment of the present invention may be integrated into one processing module, or each unit may exist physically separately, or two or more units may be integrated into one module.
  • the above integrated modules can be implemented in the form of hardware or in the form of software functional modules.
  • the integrated modules, if implemented in the form of software functional modules and sold or used as stand-alone products, may also be stored in a computer readable storage medium.
  • the above-mentioned storage medium may be a read only memory, a magnetic disk or an optical disk or the like.
  • the above description is only a preferred embodiment of the present invention, and it should be noted that those skilled in the art can also make several improvements and retouchings without departing from the principles of the present invention. Should be regarded as the scope of protection of the present invention

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Transmitters (AREA)

Description

种信号波峰削除的方法及设备 本申请要求于 2010年 2 月 25 日提交中国专利局, 申请号为 201010114656.8, 发明名称为 "一种信号波峰削除的方法及设备" 的 中国专利申请的优先权, 其全部内容通过引用结合在本申请中。 技术领域
本发明涉及移动通信领域, 具体而言, 本发明涉及一种信号波峰 的削除方法及设备。 背景技术
在移动通信系统中, 为了不失真地传输无线信号, 对发射端的线 性度要求很高, 当系统发射的信号的 PAPR ( Peak to Average Power Ratio, 峰均比)太大时, 首先过大的功率会造成很大浪费, 其次太大 的峰均比使得系统的性能急剧恶化,太大的峰均比将直接影响整个系 统的运行成本和效率, 因此需要寻找降低峰值平均功率比的方法。 此 外, 当信号是有多个子载波组成时, 如果子载波的峰均比过高会导致 信号的非线性失真, 破坏子载波之间的正交性, 增加带外干扰, 使系 统性能恶化, 因此如何合理有效地削除峰值信号功率点成为关键问 题。
OFDM ( Orthogonal Frequency Division Multiple , 正交频分多址 ) 多载波系统采用了正交频分信道,所以能够在不需要复杂的均衡技术 的情况下支持高速无线数据传输, 并具有很强的抗衰落和抗 ISI ( Inter-Symbol Interference, 符号间干扰 )的能力, 但 OFDM系统最 主要的缺点是具有较大的 PAPR, 它直接影响着整个系统的运行成本 和效率。峰均比问题是 MCM ( Multi-carrier Modulation, 多载波调制 ) 中一个普遍存在的问题。
OFDM系统子载波有 1/2重叠,但保持相互正交。在接收端通过相 关解调技术分离出信息来。任何一个子载波的正弦和余弦和另一个子 载波的正弦和余弦都是正交的。即当 n≠m时 jcos(niyft)cos(m»ft)dt = 0 , 这一点可以保证,每一个子载波正弦和余弦的 °振荡频率都是基频的整 数倍。如图 1所示,是从时域观测 4个载波的叠加过程。子载波叠加, 每一个时刻是 4个不同频率的载波叠加。一个符号的传递时间变成了 NT, 对于最低频率来说绝对是过采样。 但是对于最高频率的信号来 说是满足 Nyquist采样定理。
随着载波数量的不断增加, 例如子载波增加到 1200个时, 信号 的 PAR急剧增加。
OFDM-TDD(Time Division Duplexing, 时分双工)帧结构相对复 杂, 一个无线帧是 10ms, 由 2个半帧组成, 每个半帧长度为 5mS, 半帧分为 5个子帧 (subframes ),有 4个常规子帧和一个特殊子帧。 每一个常规子帧分为 2个时隙, 所以常规时隙长度 0.5mS, 特殊子帧 由 DwPTS、 GP、 UpPTS这 3个特殊时隙构成。 DwPTS和 UpPTS长 度是可配置的, 但要求 DwPTS、 GP、 UpPTS总长度为 lmS。 每一个 时隙有 15360个采样点。 下行链路的一个时隙有 N b =7个符号, 第 一个符号的 CP长度为 160,其余的是 144,每一个符号的长度是 2048, 所以一个时隙的时间 ( 2048x7+144x6+160 ) *Ts=15360*Ts=5ms , 如 图 2所示。
一个子帧有 2个时隙, 每一个时隙有^ ( 7 ) 个符号, 所以一 个子帧有 14个符号, 每一个符号有 Ν 个 RB ( Resource Block, 资源 块)。 一个符号下的一个子载波再一个时隙下被 ( )唯一标识为 RE
( Resource Element , 资 源 单元 ) , 其 中 k = 0,···,Ν^Ν -1 1 = 0,···,Νδ -1分别指示为时域和频域。 如果再考虑多天线可以写成 aw , 其中 P 指明是哪个天线。 如图 3所示, 由于信道估计是基于时 隙来完成的,所以参考信号的分布按照时隙来设计,横轴就是符号数, 一个子帧有 14个符号即^ 2。纵轴是一个 RB的子载波分布情况, 每一个!^有^^ ( 12 )个子载波。 所以下图中的每一个模块有 12 X 14个 RE。 并且时隙上每个 RB的参考符号图样都是一样的。
从下行链路映射的参考信号可以看出来,不同符号参考信号的分 布不同, 不同天线的参考信号分布不同。 由于不是所有符号上都有参 考信号, 而业务分配的 RB 会随机分配, 这样基于符号分别来进行 IFFT ( Inverse Fast Fourier Transform, 快速傅里叶逆变换 )处理运算 后, 有些符号超过门限的峰值点很少, 有些符号超过门限的峰值点会 很多。 这样导致信号幅度的概率统计振荡会比较大。峰值点少的符号 中每一个峰值点都有大量超过门限的采样点。峰值点多的符号中每一 个峰值点超过门限的采样点会比较少。
PC-CFR ( Peak Cancellation Crest Factor Reduction, 峰值对消波 峰因子降低)算法是通过寻找峰值功率点, 对峰值功率点进行脉沖对 削。 如图 4所示, 是找到峰值功率点, 然后对峰值功率点进行峰值削 除示意图, 其中:
CFR是数字削峰模块,主要完成对内部高于门限值的信号进行削 峰处理, 降低信号的平均功率, 进一步对躁声进行抑制。 CFR模块内 部主要分为六个子模块, 其功能介绍如下:
DELAY模块: 由于产生削峰序列的链路有一个固定的延时, 此 模块就完成输入数据的固定延时,保持和产生的削峰序列所需要的时 间相同, 最后同步输入减法模块。
PEAK_DETECT模块: 峰值点的检测模块, 所有高于门限值的 部分都进行峰值判别。
PEAK_SCALING模块: = (| - 'exp(j |χ| 为检测到的峰值 序列; ^为门限值, = angle(x)。
PEAK_ALLOCATOR模块: 峰值分配模块, 将检测到的每一个 峰值分配一个削峰脉沖产生器(CPG )模块。
CPG模块: 削峰脉沖产生模块, 主要存储着固定长度的削峰序 列, 每一个 CPG模块一次只能处理一个峰值点, 一旦处于空闲状态 时, PEAK_ALLOCATOR模块就会分配一个峰值给空闲着的 CPG模 块处理, 如果所有 CPG模块都无空闲, 那么对于接下来检测到的峰 值点都不会进行消峰处理。 一个 CPG需要一个复数乘法器既 4个乘 法器, 时钟速率如果是信号处理速率的 4被, 那么这 4个乘法器分时 复用。 所以一个复数乘法器就可以完成 4个 CPG的操作。
SUM模块: 将所有的 CPG模块的输出都相加, 形成削峰序列。 SUBTRACT模块: 用原始信号经过 DELAY模块后, 减去 SUM 模块形成的削峰序列, 所的结构就是 CFR模块的最终输出结果。
上述技术必须要找到每一次超过门限的峰值功率点,通过对一个 峰值功率点的削除, 使得信号功率全都下降到削峰门限之下。但是这 种算法最大的一种缺陷是当 OFDM子载波分配比较少时 -这种情况 出现在没有参考信号的符号上同时业务使用的 RB也很少, 这个时候 就出现了在连续的一段时间内存在许多超过门限的采样点,这个时候 在连续的一段时间内超过门限的采样点数量众多。 由于 PC-CFR适合 削除尖锐的峰值点, 对于连续的众多的大功率信号 PC-CFR就会失 效, 如图 5所示。 因此直接应用 PC-CFR对于 LTE-TDD系统不是 4艮 适合。
另外一个算法叫做 NS-CFR ( Noise Shaping Crest Factor Reduction, 噪声成型削峰)算法, 算法原理如下:
FPGA ( Field - Programmable Gate Array, 现场可编程门阵列)实 现的总体框图如图 6所示, 其中:
幅度和角度计算模块: 计算 (ΙχΙ— Α) , 其中 X为输入信号幅度,
Α为削峰阈值, Θ为输入信号的角度。 由于输入信号的采样率为一般 低于 FPGA工作时钟, 所以 I, Q两路可以时分复用一个乘法器。
峰值查找模块: 检测是否有超过峰值窗的信号。
滤波器进行噪声成型:对峰值窗内的信号进行滤波成型, 此部分 I,Q两路时分复用一个滤波器, 由于滤波器一般可以设计为对称滤波 器, 为了节省乘法器, 采用对称式脉动结构。
数据延时模块: 对输入的数据进行延时, 以保证, 输入信号和削 峰信号保持同步。
除此以外还有循环限幅滤波算法, 循环限幅滤波算法分为四部 分: 限幅、 FFT ( Fast Fourier Transform, 快速傅里叶变换)变换、 去 除频域外噪声、 IFFT。 NS-CFR的问题是由于有一个噪声成型滤波器, 此滤波器的阶数 一般都需要 120阶,这样就需要 120个乘法器,如此多的乘法器 FPGA 很难实现。 另外 NS-CFR会出现峰值再生的问题。 NS-CFR削峰前后 幅度对比图如图 7所示。
在上述现有技术中,上述算法都有各自的缺陷,难以很好的实现, 例如: PC-CFR 算法对于子载波非连续分配的情况很难有效削峰; NS-CFR算法耗费的资源多, 并且容易峰值再起; 循环限幅滤波算法 资源大, 延时长, 难以 FPGA实现。
所以, 有必要提出一种有效的信号波峰的削除方案, 能够削除任 意子载波配置下的峰值功率, 同时处理时延较小, 方案所使用的资源 适中 , ACPR( Adjacent Channel PowerRatio )、 EVM( ErrorVector Magni rude )等系统性能指标也符合系统要求。 发明内容
本发明的目的旨在至少解决上述技术缺陷之一,特别是解决对输 入信号削峰、 降低信号峰均比的问题。
为达到上述目的,本发明实施例一方面提出了一种信号波峰削除 的方法, 包括以下步骤:
对物理层信号进行 IFFT变换之后,进行旁瓣压缩削峰 RC1操作; 对所述经过旁瓣压缩削峰 RC1操作的信号进行 DUC ( Digital Up Conversion, 数字上变频器) 中频滤波;
对所述经过 DUC 中频滤波后的信号进行旁瓣压缩削峰 RC2操 作。
本发明实施例另一方面还提出了一种信号波峰削除的装置, 包括 RC1削峰模块、 中频滤波模块以及 RC2削峰模块,
所述 RC1削峰模块, 用于对物理层信号进行 IFFT变换之后, 进 行旁瓣压缩削峰 RC1操作;
所述中频滤波模块, 用于对所述经过旁瓣压缩削峰 RC1操作的 信号进行数字上变频器 DUC中频滤波;
所述 RC2削峰模块,用于对所述经过 DUC中频滤波后的信号进 行旁瓣压缩削峰 RC2操作。
本发明提出的上述技术方案,能够削除任意子载波配置下的峰值 功率; 同时处理时延能够保证在 20us以内, 处理时延小。 此外, 本 发明提出的上述技术方案, 消耗较少的硬件设备资源, 尤其是硬件乘 法器资源少于 30, ACPR、 EVM等系统性能指标恶化的程度要低于 现有算法。
本发明附加的方面和优点将在下面的描述中部分给出,部分将从 下面的描述中变得明显, 或通过本发明的实践了解到。 附图说明
本发明上述的和 /或附加的方面和优点从下面结合附图对实施例 的描述中将变得明显和容易理解, 其中:
图 1为 4个子载波在时域的叠加的示意图;
图 2为 LTE-TDD 下行链路时隙结构中帧结构类型 2的示意图; 图 3为下行链路映射的参考信号的示意图;
图 4为 PC-CFR削峰的示意图;
图 5为 PC-CFRR削峰前后幅度对比图;
图 6为 NS-CFR的 FPGA实现的总体框图示意图;
图 7为 NS-CFR削峰前后幅度对比图;
图 8为本发明实施例实现信号波峰削除的方法流程图; 图 9为本发明实施例 RC-CFR两步实现的算法总体框图; 图 10为本发明实施例信号全瓣意图示意图;
图 11为本发明实施例信号主瓣压缩示意图;
图 12为多天线复用的削峰处理示意图;
图 13为中频滤波器频谱图;
图 14为 PRB随机配置削峰后经过 DUC处理的频谱图; 图 15为过采样示意图; 图 16为验证 RC-CFR峰值再起的仿真链路示意图; 图 17为不同数字中频速率信号经过 DAC后峰均比抬升的示意 图;
图 18为物理层内插滤波后峰值再起的示意图;
图 19为不同削峰带宽的原型滤波器频谱图;
图 20为 100RB配置削峰后滤波示意图;
图 21为 100 PRB削峰前后幅度对比图;
图 22为 40 PRB削峰前后幅度对比图;
图 23为 6 PRB削峰前后幅度对比图;
图 24为本发明实施例信号波峰削除的装置的结构示意图; 图 25为同相压缩后削峰滤波前后频谱图。 具体实施方式 下面详细描述本发明的实施例, 所述实施例的示例在附图中 示出, 其中自始至终相同或类似的标号表示相同或类似的元件或 具有相同或类似功能的元件。 下面通过参考附图描述的实施例是 示例性的, 仅用于解释本发明, 而不能解释为对本发明的限制。
本发明实施例提出了一种信号波峰削除的方法, 包括以下步 骤:
对物理层信号进行 IFFT 变换之后, 进行旁瓣压缩削峰 RC1 操作; 对所述经过旁瓣压缩削峰 RC1操作的信号进行数字上变频 器 DUC中频滤波; 对所述经过 DUC中频滤波后的信号进行旁瓣 压缩削峰 RC2操作。
如图 8 所示, 为本发明实施例实现信号波峰削除的方法流程 图, 包括以下步骤:
S101 : 对物理层信号进行 IFFT变换之后, 进行旁瓣压缩削峰 RC1操作。 在步骤 S101 中, 如图 9 上半部分所示, 为本发明实施例 RC-CFR ( Reduce Peak and PC-CFR, 降低峰值和峰值对消波峰因 子降低) 两步实现的算法总体框图。 具体而言, RC算法的核心思 想是采用多级旁瓣压缩削峰算法, 旁瓣压缩削峰算法的设计思想 如下:
根据系统带宽确定核心削峰向量的长度 N ,同时依据中频的内 插倍数 L, 物理层 FFT的采样点数和一个 PRB ( Physical Resource
Block, 物理资源块)对应的子载波个数, 来确定核心削峰向量中 一个 PRB对应的采样点数。然后再根据物理层发送过来的 PRB索 引来确定本时隙的核心削峰向量的频域响应, 根据得到的频域响 应进行 IFFT得到核心削峰向量,为了可实现性,考虑物理层不传递
PRB 索引, 使用一个统一的核心削峰向量进行削峰滤波处理。 此 时就得到核心削峰向量的时域表达式如下, 并且此核心削峰向量 的频域是子载波满配置:
p = IFFT(Ik)
考虑系统计算的复杂度, 核心削峰向量并不随着子载波的配 置而自适应调整, 固定一个核心削峰向量, 不同的子载波配置都 使用这同一个核心削峰向量。 旁瓣抑制算法是主瓣维持不变, 邻 近的旁瓣压缩, 一般情况下主瓣在削峰向量的中心, 主瓣两边的 旁瓣压缩操作为:
left—lob = P(l: left_point)*compress_gain;
right_lob= P(right_point: end)* compress—gain;
主瓣的取值如下:
main_lob=P(left_point: right—point)
其中 O≤compress_gaiii 为压缩因子, 小于 1的正数。 这种压缩 算法复杂度还是比较高, 并且核心削峰向量的设计难度大, 存在 一定漏削的缺陷。
为了得到一个切实可行的削峰算法, 旁瓣压缩算法本身可采 取演变算法如下: 主瓣削峰算法: 削峰器只留主瓣中的最大信号, 即压缩所有 的旁瓣, compress_gain=0, 此时削峰滤波器对信号的削峰就相当 于同相硬切操作。 此时削峰器只有一个主瓣, 所以只要检测到超 过门限的峰值信号, 都可以逐个削除。 优点时既不会存在漏削, 并且削除本峰值时不会影响相邻信号, 所以此时削峰造成的 EVM 恶化最小, 削峰效果最明显。 这点类似于智能天线的波束赋型, 多天线赋型后的信号全都被本用户接收, 用户接收到的功率最大, 并且对其它用户的干扰最小。
全瓣削峰算法: 削峰器保留了所有的核心削峰向量, 没有对 旁瓣进行任何抑制, compress_gain=l , 所以削峰之后系统的 ACPR 不会有恶化。
采用这两级削峰的优点是, 信号的峰值能够最大限度的削除, EVM和 ACPR的恶化都相对小。
如附图所示, 其中, 图 10为信号全瓣意图示意图, 可见没有 压缩旁瓣信号基本重合; 图 11为信号主瓣压缩示意图, 可见旁瓣 压缩后只保留了一个,、的信号。
如图 9下半部分所示, 旁瓣压缩削峰 RC1操作包括主瓣削峰 操作, 主瓣削峰操作为同相硬切削峰操作, 主瓣削峰操作只保留 信号主瓣中的最大信号。
具体而言, IFFT之后的主瓣削峰算法即物理层同相硬切削峰 操作, 进行 DUC处理之前进行同相削峰算法如下:
Figure imgf000011_0001
上述算法的主要工作是信号的幅度和相位的计算, 优选地, 上述算法的修正如下:
x(n) = A* (cos6>(n) + j sin6>(n)), 6>(n) = angle(x(n)),当 |x(n)| > A
Figure imgf000011_0002
x(n), 当 |x(n)|≤A , 其中, χ(η)为物理层信号进行 IFFT 变换之后的信号, A为预定 削峰门限值。
随后在中频完成正常的滤波, 削峰后的正常 DUC处理, 可以 有效抑制带外的削峰造成的杂散。 所以此时不再需要额外的 IFFT/FFT运算, 使得运算量大大降低, 时延也变得很短。
一个峰值的削除需要的运算除去一个正弦运算, 一个余弦运 算外, 只需要两个乘法运算就可以完成。 同相限幅削峰是一个非 线性过程, 它将导致一定的带内噪声和带外干扰。 带外干扰被随 后的中频滤波削除。
由于在物理层进行削峰操作, 物理层速率很低, 所以可以通 过时分复用方式进行削峰处理, 如图 12所示, 为多天线复用的削 峰处理示意图。 在削峰之前对多天线的数据进行合并, 本实施例 中为 4根天线, 只要每一根天线的 I/Q两路在时域上严格对齐, 多 天线的信号就可以时分复用的方式输入物理层硬切模块, 这样多 根天线削峰使用的资源和一根天线削峰消耗的资源一样, 节省了 硬件资源的使用。
S102: 对经过旁瓣压缩削峰 RC1操作的信号进行数字上变频 器 DUC中频滤波。
虽然旁瓣压缩降低峰值算法虽然可以有效削除峰值, 但是频 谱会有一定的恶化。 所以中频滤波器需要有陡峭的过渡带, 削除 边带失真的影响。 优选地, 中频滤波器为多级滤波器组合实现。 如图 13所示, 为中频滤波器频谱图, 陡峭的过渡带可以 ^艮好的配 合物理层削峰造成的影响。
同相限幅削峰是一个非线性过程, 它将导致一定的带内噪声 和带外干扰。 带外干扰被随后的滤波削除, 如图 14所示, 为 PRB 随机配置削峰后经过 DUC处理的频谱图, 滤波完成之后则没有看 到带外干扰。
物理层完成削峰以后, 通过 DUC滤波内插处理以后, 到达中 频。 中频信号进行削峰操作。 中频的削除操作可以使用中频旁瓣 压缩削峰算法, 例如 PC-CFR算法。
通过 DUC峰值再生的原因是:物理层信号经过内插滤波之后, 由于现在硬件上使用的数字滤波器阶数受限, 都不是理想滤波器, 故此存一定的通带波纹抖动, 同时阻带衰减也受限, 还有物理层 的数据量成倍增加, 故此物理层信号经过中频之后, 会有峰值再 起。同理在 PC-CFR之后如果还有内插滤波器同样会出现峰值再生 的问题, 因为削峰后的信号通过 DAC ( Digit to Analog Converter , 数摸变换器 ) , DAC内部为了提高信号在进行数摸变换时的 SNR, 会对输入的削峰信号再次进行内插滤波, 对输入信号过采样。 以 高于信号带宽两倍的速率进行采样称为过采样。 过采样是一个非 常重要的功能, 因为它能在数字域内给接收到的信号的信噪比带 来增益。 在数摸转换的过程中, 采样速率越快, 量化噪声电平就 会越低。 因为量化噪声分布在更宽的频带内, 即采样的时钟频率 内, 而总的噪声是一定的, 过采样示意图如图 15所示。
过采样噪声功率谱密度:
= - + 101g Dfs , 其中, Dfs为采样信号的速率, 所以只要
Figure imgf000013_0001
提高 Dfs , 采样的信噪比就会抬高。
具体到 A/D,D/A变换器的信噪比, 可以由下式来计算: SNR = 6.02 - N+1.76+ 101og10 (Fs /2 fBW) ,
其中, SNR为信噪比, N为 ADC ( Analog to Digit Converter, 摸数变换器) /DAC的位数, Fs为 ADC/DAC采样时钟的频率, fBW 为信号带宽。 从公式可以看出, 如果采样速率提高一倍的话, ADC/DAC的 SNR将会提高 3dB。
所以削峰以后的信号进入 D AC之后的信号一般会再次进行内 插滤波, 一般是使用多级 HB ( Half Band, 半带)滤波器完成。 为 了验证数据经过中频削峰以后再内插以后 PAR的变化, 按下述仿 真平台进行验证, 如图 16所示, 为验证 RC-CFR峰值再起的仿真 链路示意图。 采用不同中频速率 fs的多载波信号经过 PC-CFR削峰后再经 过三个级联 HB滤波器( HB 1、 HB2、 HB3 )处理, 观察 CFR后 A 点, 以及经过 HB滤波器后 C、 D点的峰均比。
DAC对峰均比的影响(单位: dB)
Figure imgf000014_0001
在本发明实施例中, 中频滤波后的信号速率 符合 fs≥ fs。, 其 中, 速率 fs为在 fs。速率下再进行内插滤波后信号的 PAR抬升小于 等于预定门限 LdBc , 使得 PAR( fs ) - PAR(fs0 )≤ IdBc。
如图 17所示, 为不同数字中频速率信号经过 DAC后峰均比 抬升的示意图。 从上述结果分析可以得到:
从图 17可以看出,数字中频信号经过三个 HB后 PAR确实都 有一定的抬升, 但是数据速率越高, PAR抬升越小, 规定一个最 低速率 fs。, 在这个速率下信号的 PAR 抬升小于 0.3dBc : APAR( fs0) < 0.3dBc , PC-CFR 后的信号速率 fs≥ fs。 即可, 即 PAR( fs) - PAR( fs。)≤0.3dBc。 从本次仿真可以看出来, PC-CFR输出的 信号速率只要大于 fs≥92.16MHZ即可, 此时再进入 DAC后, 峰均 比抬升再在 0.2dBc之下,满足系统要求。即在上述实施例中, LdBc 取值为 0.3dBc。
如图 18所示, 为物理层内插滤波后峰值再起的示意图, 此时 全瓣削峰限制后再起峰值。
S103: 对经过 DUC中频滤波后的信号进行旁瓣压缩削峰 RC2 操作。
如图 9下半部分所示, 旁瓣压缩削峰 RC2操作包括全瓣削峰 操作, 全瓣削峰操作对旁瓣不进行压缩, 全瓣削峰操作为峰值对 消波峰因子降低 PC-CFR削峰操作。
通过 DUC 以后有峰值再起, 此时中频采用旁瓣压缩算法削 除剩余的峰值功率。 旁瓣压缩算法仅仅对远端的旁瓣压缩, 近端 的旁瓣保持不变, 这样可以保证整体的 ACPR不会有太大的恶化。 如果考虑中频 ACPR不够大, 此时对旁瓣不进行压缩, 此时采用 的就是全瓣削峰算法对剩余峰值进行削除, 由于在进入 CFR之前 大部分峰值都已经削除, 此时的全瓣削峰器可以筒单设计。
相对于单一的 PC-CFR, 在物理层主瓣削峰 +中频全瓣削峰有 如下好处: 无论物理层子载波如何配置, 峰值能够很有效的削除, 虽然峰值再起, 但是峰值再起一般 PAR不会超过 7.0dBc。
其中 PC-CFR—个重要的设置是削峰原型滤波器,该原型滤波 器与 DUC后 OFDM信号的频谱匹配,
设原型滤波器为 filter_f ( η ) , η = 0,· · ·, Ν - 1 , Ν为削峰系数长 度, 其可预先生成存储。 由 Matlab的 firls函数, 并使用 Kaiser加 窗得到。 其生成方法如下式所示, 式中 为通带带宽, 4为阻带衰 减, fs为采样频率, beta:
cfr_ntaps = 255;
fs_mhz = 92.16;
beta = 5; fl = 18.0/fs_mhz;
f2 = 19.2/fs_mhz;
filter_f = firls(cfr_ntaps-l, [0 fl f2 1], [1 1 0 0], [1 10])...
.*kaiser(cfr_ntaps, beta)';
其中参数 beta决定了旁瓣的衰减, 如下所示:
0.1102 ί at - 8.7), κ > 50
β 0.584≥ f a - 21 ) 0 4 + 0, Q7SS6 ϊ Q: - SI) , 50 a 21
0, K S1 为了使得削峰后的频谱没有恶化, 原型滤波器的通道带宽一 般稍稍小于或者接近 DUC 出来的信号频谱带宽, 如图 19所示, 不同削峰带宽的原型滤波器频谱图。 从图中可以看出, 选择通带 带宽 16MHZ比较合适, fl = 16.5/fs_MHz, f2 = 18.5/fs_MHz。
如图 20所示, 为 100RB配置削峰后滤波示意图, 如图 21所 示, 为 100 PRB削峰前后幅度对比图。
如图 22所示,为 40 PRB削峰前后幅度对比图,如图 23所示, 为 6 PRB削峰前后幅度对比图。
RC-CFR 削峰算法在 DUC 前后进行旁瓣压缩的目的和区别 是, DUC之前的主瓣削峰是为了削除绝大多数的峰值功率点, DUC 滤波之后的全瓣削峰主要是削除 DUC滤波造成的再起峰值。
本发明实施例公开的上述信号波峰削除的方法, 能够削除任 意子载波配置下的峰值功率;同时处理时延能够保证在 20us以内, 处理时延小。 此外, 本发明实施例提出的上述信号波峰削除的方 法, 消耗较少的硬件设备资源, 尤其是硬件乘法器资源少于 30, ACPR、 EVM 等系统性能指标恶化的程度要低于现有算法。 本发 明实施例提出的上述信号波峰削除的方法, 对现有系统的改动很 小, 不会影响系统的兼容性, 而且实现筒单、 高效。
相应于上述方法, 如图 24所示, 为本发明实施例信号波峰削 除的装置 100的结构示意图, 包括 RC1 削峰模块 110、 中频滤波 模块 120以及 RC2削峰模块 130。 其中, RCl削峰模块 110用于对物理层信号进行 IFFT变换之 后, 进行旁瓣压缩削峰 RC1操作。
RC1削峰模块 110进行的旁瓣压缩削峰 RC1操作包括主瓣削 峰操作, 主瓣削峰操作为同相硬切削峰操作, 主瓣削峰操作只保 留信号主瓣中的最大信号。
具体而言, RC1削峰模块 110进行的同相硬切削峰操作包括,
RC1 削峰模块 110对输入信号经过同相硬切削峰操作后得到的输 出信号为:
= A* (cos6>(n) + j sin6>(n)), 6>(n) = angle(x(n)),当 |x(n)| > A y(n)
Figure imgf000017_0001
x(n), 当 |x(n)|≤A
, 其中, x(n)为物理层信号进行 IFFT 变换之后的信号, A为预定 削峰门限值。
具体而言, χ(η)信号在削峰之前的多天线的数据进行合并, 多 天线的信号以时分复用的方式输入物理层硬切模块, 完成同相硬 切削峰操作。
中频滤波模块 120用于对经过旁瓣压缩削峰 RC1操作的信号 进行数字上变频器 DUC中频滤波。
具体而言, 中频滤波模块 120中的 DUC中频滤波为多级滤波 器组合实现。
具体而言, 中频滤波模块 120中的信号速率 fs符合 fs≥ fs。, 其 中, 速率 fs为在 fs。速率下再进行内插滤波后信号的 PAR抬升小于 等于预定门限 LdBc, 使得 PAR( fs )- PAR(fs0 )< MBc。
RC2削峰模块 130, 用于对经过 DUC中频滤波后的信号进行 旁瓣压缩削峰 RC2操作。
具体而言, RC2削峰模块 130进行的旁瓣压缩削峰 RC2操作 包括全瓣削峰操作, 全瓣削峰操作对旁瓣不进行压缩, 全瓣削峰 操作为峰值对消波峰因子降低 PC-CFR削峰操作。
具体而言, RC2削峰模块 130进行的 PC-CFR削峰操作为通 过寻找信号的峰值功率点, 对峰值功率点进行脉沖对削。
具体而言, RC2削峰模块 130在执行 PC-CFR削峰操作中, PC-CFR的削峰原型系数通带带宽小于等于 DUC中频滤波后的信 号频带。
本发明实施例公开的上述信号波峰削除的装置, 能够削除任 意子载波配置下的峰值功率;同时处理时延能够保证在 20us以内, 处理时延小。 此外, 本发明实施例提出的上述信号波峰削除的装 置, 消耗较少的硬件设备资源, 尤其是硬件乘法器资源少于 30, ACPR、 EVM 等系统性能指标恶化的程度要低于现有算法。 本发 明实施例提出的上述信号波峰削除的装置, 对现有系统的改动很 小, 不会影响系统的兼容性, 而且实现筒单、 高效。
为了说明本发明实施例公开的上述方案的技术效果, 下面将 本发明实施例提出的方案与现有的其它方案进行试验仿真, 对比 说明如下:
Figure imgf000018_0001
因此, 从链路仿真性能来看, RC-CFR 效果是最好的, PAR 都可以稳定的抑制到 6.5dBc以下, EVM可以控制在 7%以下, 如 图 25所示, 为 RC-CFR对各种 RB配置下的削峰效果性能曲线示 意图。
从上面的仿真结果看出: RC-CFR的削峰效果是 EVM恶化最 小, PAR削峰效果最明显。
RC-CFR不仅性能好, 并且资源少, 时延短, 易于 FPGA硬件 实现
( 3 ) 不同削峰算法资源比较
Figure imgf000019_0001
从资源上来看 RC-CFR算法非常接近 PC-CFR算法,增加的资 源很少。 但是 NS-CFR占用的硬件乘法器很多。 从削峰效果来看, 除去峰值以外, 其它位置的信号受的影响很小, 这都得益于物理 层完成了大部分的峰值削除。
除上述 3个算法之外还有循环限幅算法, 即 IFFT/FFT, 循环 限幅算法的资源消耗大, 时延长, 削峰效果不太明显。
NS-CFR算法除了资源消耗大的缺点外,经常出现削峰不足的 现象。 将峰值窗的阈值都设在 7.5dB处, 分别更改削峰阈值, 仿真 了三组情况, 其中 PRB分配为 100和 4时, 削峰后信号都能很好 的收敛到 7.5dB左右; PRB分配为 40和 12时, 受削峰阈值的影 响较大, 并且存在峰值再起的概率也很大。 PC-CFR削峰算法的最 大缺点削峰效果不稳定, 尤其在子载波配置很少的情况下。
( 4 ) 不同削峰算法综合比较
Figure imgf000019_0002
FPGA资源消耗 大 小 小
PAR降低效果 一般 好 好
(子载波多)
PAR降低效果(子 一般 好
载波少)
PAR降低的稳定性 一般 好 所以综上所述, RC-CFR算法无论在资源节省, 时延还是削除 的有效性, EVM 的恶化程度都是最优或者次优。 故此 LTE-CFR 选择 RC-CFR。 信号的 PAR可以稳定的控制在 7dBC以下。
本领域普通技术人员可以理解实现上述实施例方法携带的全 部或部分步骤是可以通过程序来指令相关的硬件完成, 所述的程 序可以存储于一种计算机可读存储介质中, 该程序在执行时, 包 括方法实施例的步骤之一或其组合。
另外, 在本发明各个实施例中的各功能单元可以集成在一个 处理模块中, 也可以是各个单元单独物理存在, 也可以两个或两 个以上单元集成在一个模块中。 上述集成的模块既可以采用硬件 的形式实现, 也可以采用软件功能模块的形式实现。 所述集成的 模块如果以软件功能模块的形式实现并作为独立的产品销售或使 用时, 也可以存储在一个计算机可读取存储介质中。
上述提到的存储介质可以是只读存储器, 磁盘或光盘等。 以上所述仅是本发明的优选实施方式, 应当指出, 对于本技 术领域的普通技术人员来说, 在不脱离本发明原理的前提下, 还 可以做出若干改进和润饰, 这些改进和润饰也应视为本发明的保 护范围

Claims

权利要求
1、 一种信号波峰削除的方法, 其特征在于, 包括以下步骤: 对物理层信号进行快速傅里叶逆变换 IFFT之后, 对所述经过
IFFT操作后的信号进行旁瓣压缩削峰 RC1操作;
对所述经过旁瓣压缩削峰 RC1操作的信号进行数字上变频器 DUC中频滤波;
对所述经过 DUC 中频滤波后的信号进行旁瓣压缩削峰 RC2 操作。
2、 如权利要求 1所述的信号波峰削除的方法, 其特征在于, 所述旁瓣压缩削峰 RC1操作包括主瓣削峰操作, 所述主瓣削峰操 作为同相硬切削峰操作, 所述主瓣削峰操作只保留信号主瓣中的 最大信号; 所述旁瓣压缩削峰 RC2操作包括全瓣削峰操作, 所述 全瓣削峰操作对旁瓣不进行压缩, 所述全瓣削峰操作为峰值对消 波峰因子降低 PC-CFR削峰操作。
3、 如权利要求 2所述的信号波峰削除的方法, 其特征在于, 所述同相硬切削峰操作包括,输入信号 χ(η)经过同相硬切削峰操作 后得到的输出信号 y(n)为:
x(n) = A* (cos 6>(n) + j sin 6>(n)), θ(η) = angle(x(n)),当 |x(n)| > A y(n)
x(n), 当 |x(n)|≤ A
, 其中, x(n)为物理层信号进行 IFFT 变换之后的信号, A为预定 削峰门限值。
4、 如权利要求 3所述的信号波峰削除的方法, 其特征在于, 所述 x(n)信号在削峰之前对多天线的数据进行合并,所述多天线的 信号以时分复用的方式输入物理层硬切模块, 完成所述同相硬切 削峰操作。
5、 如权利要求 2所述的信号波峰削除的方法, 其特征在于, 所述 DUC中频滤波为多级滤波器组合实现。
6、 如权利要求 5所述的信号波峰削除的方法, 其特征在于, 所述 DUC 中频滤波后的信号速率 符合 fs≥ fs。, 其中, 速率 为 在 fs。速率下再进行内插滤波后信号的 PAR抬升小于等于预定门限 LdBc, 使得 PAR( fs )- PAR(fs。)≤ IdBc。
7、 如权利要求 2所述的信号波峰削除的方法, 其特征在于, 所述 PC-CFR削峰操作为通过寻找信号的峰值功率点 ,对峰值功率 点进行脉沖对削。
8、 如权利要求 7所述的信号波峰削除的方法, 其特征在于, 所述 PC-CFR削峰操作中, PC-CFR的削峰原型系数通带带宽小于 等于所述 DUC中频滤波后的信号频带。
9、 一种信号波峰削除的装置, 其特征在于, 包括 RC1削峰模 块、 中频滤波模块以及 RC2削峰模块,
所述 RC1削峰模块, 用于对物理层信号进行 IFFT变换之后, 对所述经过 IFFT变换后的信号进行旁瓣压缩削峰 RC1操作;
所述中频滤波模块, 用于对所述经过旁瓣压缩削峰 RC1操作 的信号进行数字上变频器 DUC中频滤波;
所述 RC2削峰模块,用于对所述经过 DUC中频滤波后的信号 进行旁瓣压缩削峰 RC2操作。
10、 如权利要求 9所述的信号波峰削除的装置, 其特征在于, 所述 RC1削峰模块进行的所述旁瓣压缩削峰 RC1操作包括主瓣削 峰操作, 所述主瓣削峰操作为同相硬切削峰操作, 所述主瓣削峰 操作只保留信号主瓣中的最大信号; 所述 RC2削峰模块进行的所 述旁瓣压缩削峰 RC2操作包括全瓣削峰操作, 所述全瓣削峰操作 对旁瓣不进行压缩, 所述全瓣削峰操作为峰值对消波峰因子降低 PC-CFR削峰操作。
11、如权利要求 10所述的信号波峰削除的装置,其特征在于, 所述 RC1削峰模块进行的所述同相硬切削峰操作包括, 所述 RC1 削峰模块对输入信号 x(n)经过同相硬切削峰操作后得到的输出信 号 y(n)为: x(n) = A* (cos 6>(n) + j sin 6>(n)), θ(η) = angle(x(n)),当 | x(n)| > A x(n
Figure imgf000023_0001
), 当 | x(n)|≤ A
, 其中, x(n)为物理层信号进行 IFFT 变换之后的信号, A为预定 削峰门限值。
12、如权利要求 11所述的信号波峰削除的装置,其特征在于, 所述 x(n)信号在削峰之前对多天线的数据进行合并,所述多天线的 信号以时分复用的方式输入物理层硬切模块, 完成所述同相硬切 削峰操作。
13、如权利要求 10所述的信号波峰削除的装置,其特征在于, 所述中频滤波模块中的所述 DUC 中频滤波为多级滤波器组合实 现。
14、如权利要求 13所述的信号波峰削除的装置,其特征在于, 所述中频滤波模块中的信号速率 fs符合 fs≥ fs。, 其中, 速率 fs为在 fs。速率下再进行内插滤波后信号的 PAR 抬升小于等于预定门限
LdBc, 使得 PAR( fs ) - PAR(fs。)≤ IdBc。
15、如权利要求 10所述的信号波峰削除的装置,其特征在于, 所述 RC2削峰模块进行的所述 PC-CFR削峰操作为通过寻找信号 的峰值功率点, 对峰值功率点进行脉沖对削。
16、如权利要求 15所述的信号波峰削除的装置,其特征在于, 所述 RC2削峰模块在执行所述 PC-CFR削峰操作中, PC-CFR的削 峰原型系数通带带宽小于等于所述 DUC中频滤波后的信号频带。
PCT/CN2011/071277 2010-02-25 2011-02-24 一种信号波峰削除的方法及设备 WO2011103809A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN 201010114656 CN102111368B (zh) 2010-02-25 2010-02-25 一种信号波峰削除的方法及设备
CN201010114656.8 2010-02-25

Publications (1)

Publication Number Publication Date
WO2011103809A1 true WO2011103809A1 (zh) 2011-09-01

Family

ID=44175400

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2011/071277 WO2011103809A1 (zh) 2010-02-25 2011-02-24 一种信号波峰削除的方法及设备

Country Status (2)

Country Link
CN (1) CN102111368B (zh)
WO (1) WO2011103809A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016043359A1 (ko) * 2014-09-15 2016-03-24 주식회사 쏠리드 Pc-cfr 장치 및 papr 감소 방법, 및 피크 값 결정 장치

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103414482B (zh) * 2013-06-17 2015-08-19 电子科技大学 模数结合的闭环旁瓣对消方法及系统
CN103401831A (zh) * 2013-08-02 2013-11-20 武汉邮电科学研究院 一种基于fpga的lte数字中频削峰方法及装置
CN103491045B (zh) * 2013-09-03 2016-08-24 大唐移动通信设备有限公司 多载波削峰处理方法与装置
US9614499B2 (en) * 2014-03-21 2017-04-04 Huawei Technologies Co., Ltd. System and method for filter design and usage thereof
US10044543B2 (en) * 2015-07-30 2018-08-07 Futurewei Technologies, Inc. Reducing crest factors
US10367670B2 (en) 2017-01-25 2019-07-30 Telefonaktiebolaget Lm Ericsson (Publ) Method and apparatus for crest factor reduction
EP3598708B1 (en) 2017-04-21 2021-07-28 Huawei Technologies Co., Ltd. Signal quality control method and base station
CN108650206A (zh) * 2018-05-03 2018-10-12 南京理工大学 基于新型算法的正交频分复用系统高峰均功率比优化方法
US10594269B2 (en) * 2018-08-08 2020-03-17 Shenzhen GOODIX Technology Co., Ltd. Crest factor reduction in power amplifier circuits
CN113691479B (zh) * 2021-07-15 2023-05-23 中国电子科技集团公司第二十九研究所 一种信号消除旁瓣影响的方法、装置及可读存储介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080292014A1 (en) * 2007-05-22 2008-11-27 Masashi Naito Communication equipment
CN101350646A (zh) * 2007-07-20 2009-01-21 中兴通讯股份有限公司 一种用于智能天线系统降低峰值功率的装置与方法
CN101378377A (zh) * 2007-08-29 2009-03-04 株式会社日立国际电气 正交多路复用信号的峰值抑制方法、峰值抑制电路及发送装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080292014A1 (en) * 2007-05-22 2008-11-27 Masashi Naito Communication equipment
CN101350646A (zh) * 2007-07-20 2009-01-21 中兴通讯股份有限公司 一种用于智能天线系统降低峰值功率的装置与方法
CN101378377A (zh) * 2007-08-29 2009-03-04 株式会社日立国际电气 正交多路复用信号的峰值抑制方法、峰值抑制电路及发送装置

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016043359A1 (ko) * 2014-09-15 2016-03-24 주식회사 쏠리드 Pc-cfr 장치 및 papr 감소 방법, 및 피크 값 결정 장치

Also Published As

Publication number Publication date
CN102111368B (zh) 2013-07-24
CN102111368A (zh) 2011-06-29

Similar Documents

Publication Publication Date Title
WO2011103809A1 (zh) 一种信号波峰削除的方法及设备
Yli-Kaakinen et al. Efficient fast-convolution-based waveform processing for 5G physical layer
EP2115985B1 (en) Apparatuses and a method for reducing peak power in a transmitter of telecommunications systems
Ihalainen et al. Filter bank based multi-mode multiple access scheme for wireless uplink
US8165539B2 (en) Apparatuses and a method for reducing peak power in telecommunications systems
KR102269498B1 (ko) 다중 반송파 통신 시스템에서 전력을 제어하는 방법 및 장치
Khan et al. Multicarrier waveform candidates for beyond 5G
Taher et al. Enhanced F-OFDM candidate for 5G applications
CA2657091A1 (en) Apparatuses and a method for reducing peak power in a transmitter of telecommunications systems
Gökceli et al. PAPR reduction with mixed-numerology OFDM
JP4615404B2 (ja) マルチキャリア無線通信システム、送信機及び受信機並びにマルチキャリア無線通信方法
Yadav et al. PAPR reduction using clipping and filtering technique for nonlinear communication systems
Mowla et al. Performance improvement of PAPR reduction for OFDM signal in LTE system
Yli-Kaakinen et al. Optimized reconfigurable fast convolution-based transmultiplexers for flexible radio access
CN111865858A (zh) 一种基于部分传输序列技术的边信息传输方法和装置
Kaur et al. Review and analysis of Simulink based OFDM
US8085890B2 (en) Apparatus and method for base band filtering
EP2489164B1 (en) EVM pooling for multi-standard and multi-carrier systems
Loulou et al. Enhanced OFDM techniques for fragmented spectrum use
Dewangan A detailed Study of 4G in Wireless Communication: Looking insight in issues in OFDM
GB2426420A (en) Reducing peak to average power ratio (PAPR) in an orthogonal frequency division multiplexing transmitter
CN107483379B (zh) 一种针对tdd-ofdm系统优化定标装置及方法
Montalvo et al. Comparison and analysis of PAPR reduction techniques in OFDMA and SC-FDMA systems
Chintala et al. Performance evaluation of hybrid transform for universal filtered multi‐carrier system
Wang et al. Nonuniform subband superposed OFDM with variable granularity spectrum allocation for 5G

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 11746851

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 11746851

Country of ref document: EP

Kind code of ref document: A1