WO2011127730A1 - 一种频率偏差的估计方法及装置 - Google Patents

一种频率偏差的估计方法及装置 Download PDF

Info

Publication number
WO2011127730A1
WO2011127730A1 PCT/CN2010/078830 CN2010078830W WO2011127730A1 WO 2011127730 A1 WO2011127730 A1 WO 2011127730A1 CN 2010078830 W CN2010078830 W CN 2010078830W WO 2011127730 A1 WO2011127730 A1 WO 2011127730A1
Authority
WO
WIPO (PCT)
Prior art keywords
synchronization code
frequency domain
frequency
downlink synchronization
signal
Prior art date
Application number
PCT/CN2010/078830
Other languages
English (en)
French (fr)
Inventor
邱宁
李强
曾文琪
于天昆
刘中伟
邢艳楠
梁立宏
李立文
林峰
褚金涛
Original Assignee
中兴通讯股份有限公司
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 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Priority to US13/258,448 priority Critical patent/US8867443B2/en
Priority to KR1020127000919A priority patent/KR101351496B1/ko
Publication of WO2011127730A1 publication Critical patent/WO2011127730A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L7/00Arrangements for synchronising receiver with transmitter
    • H04L7/02Speed or phase control by the received code signals, the signals containing no special synchronisation information
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04BTRANSMISSION
    • H04B1/00Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
    • H04B1/69Spread spectrum techniques
    • H04B1/707Spread spectrum techniques using direct sequence modulation
    • H04B1/7073Synchronisation aspects
    • H04B1/7087Carrier synchronisation aspects
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04JMULTIPLEX COMMUNICATION
    • H04J13/00Code division multiplex systems

Definitions

  • the present invention relates to a Time Division-Synchronized Code Division Multiple Access (TD-SCDMA) system, and more particularly to a method and apparatus for estimating a frequency offset.
  • TD-SCDMA Time Division-Synchronized Code Division Multiple Access
  • Time Division-Synchronized Code Division Multiple Access (TD-SCDMA) is one of the three mainstream standards of 3G (third-generation mobile communication technology) and has broad application prospects.
  • BS base stations
  • terminals User Equipment
  • the UEs are both transmitting and receiving at a nominal carrier frequency. Due to device level limitations, there is a certain bias between the actual transmit and receive frequencies.
  • the TD-SCDMA system requires that the carrier frequency error of the base station is less than 0.05 PPM, and the carrier frequency error of the terminal is required to be less than 0.1 PPM.
  • the frequency accuracy of the oscillator can be satisfied due to the small limitations of temperature, volume, power consumption and cost.
  • the frequency accuracy of the selected crystal oscillator is usually not Meet the standard requirements.
  • the role of AFC Automatic Frequency Control
  • AFC Automatic Frequency Control
  • the existing frequency offset estimation mostly utilizes the downlink pilot time slot in the beacon channel (Downlink Pilot).
  • TimeSlot, DwPTS DwPTS
  • Training sequence Minamble
  • the invention provides a method and a device for estimating frequency deviation, which solves the problems of performance and stability reduction caused by techniques such as precise timing and multipath position estimation, and realizes frequency offset estimation for precise timing and multipath position estimation. Dependence, improving the overall stability of the system.
  • a method for estimating a frequency offset includes: after receiving a subframe, converting a downlink synchronization code of the subframe to a frequency domain, and converting the converted downlink synchronization code to a local frequency domain.
  • the synchronization code performs a correlation operation to obtain a conjugate downlink synchronization code sequence; and determines a bit of the maximum value in the conjugate downlink synchronization code sequence, and calculates a frequency deviation value according to the bit position of the maximum value.
  • the local frequency domain synchronization code can be a set of synchronization code sequences that are affected by different frequency offsets.
  • the method may further include: constructing a local frequency domain synchronization code by: performing a frequency offset operation on the pre-configured synchronization code; and, after completing the frequency offset operation, eliminating timing offset and multipath pair pre-configured synchronization code Influence, get the local frequency domain synchronization code.
  • the step of performing the frequency offset operation may include: generating signals of different frequencies; multiplying each sample point in the pre-configured synchronization code by ( ⁇ , where j is a complex unit, and i is a pre-configured synchronization of each sample point The number of bits in the code; and, the synchronization code after the completion of the ( ) operation is multiplied by the generated signal.
  • the steps of eliminating the timing offset and the effect of multipath on the pre-configured synchronization code may include: The configured synchronization code is converted to the frequency domain, and the synchronization code in the frequency domain is subjected to differential operation to obtain a local frequency domain synchronization code.
  • the method may further include: compensating the pre-configured synchronization code to adapt to the baseband signal prior to converting the pre-configured synchronization code to the frequency domain. The compensation can be compensated for the raised cosine.
  • the method may further include: before performing the difference operation on the synchronization code in the frequency domain, reordering the samples of the synchronization code in the frequency domain according to the frequency corresponding to each sample point.
  • the method may further include: searching for the location of the downlink synchronization code from the received subframe, and intercepting the downlink synchronization code from the subframe according to the length of the data window. And if the number of samples of the downlink synchronization code signal is not an integer multiple of the number of samples of the local frequency domain synchronization code, processing the downlink synchronization code signal such that the number of samples of the downlink synchronization code signal is a local frequency domain synchronization code An integer multiple of the number of points.
  • the length of the data window may be the length of the chip of the downlink synchronization code plus the length of each m chip of the downlink synchronization code, where m > 0.
  • the method may further include: performing differential operation on the downlink synchronization code signal when the number of samples of the downlink synchronization code signal is an integer multiple of the number of samples of the local frequency domain synchronization code.
  • the method may further include: performing differential operation on the downlink synchronization code signal to perform compensation for eliminating the influence of the fixed time domain timing offset.
  • the step of obtaining the conjugate downlink synchronization code sequence may include: after performing the correlation operation between the converted downlink synchronization code and the local frequency domain synchronization code, discarding the imaginary part of each correlation operation result, and retaining the result of each correlation operation The real part obtains a conjugated downlink synchronization code sequence.
  • an apparatus for estimating a frequency deviation includes: an interconnected local frequency domain synchronization code construction module and a frequency deviation estimation module, wherein
  • the local frequency domain synchronization code construction module is configured to construct a local frequency domain synchronization code, and send the constructed local frequency domain synchronization code to the frequency deviation estimation module;
  • the frequency deviation estimation module is configured to: after receiving the subframe, convert the downlink synchronization code of the subframe to the frequency domain, and perform correlation operation between the converted downlink synchronization code and the received local frequency domain synchronization code to obtain Conjugated downlink synchronization code sequence, determining the maximum value in the conjugate downlink synchronization code sequence Bit, the frequency deviation is calculated based on the position of the maximum value.
  • the local frequency domain synchronization code construction module may be configured to construct a local frequency domain synchronization code by: performing a frequency offset operation on the pre-configured synchronization code; and, after completing the frequency offset operation, eliminating timing offset and multipath pair pre- The effect of the configured synchronization code is obtained by the local frequency domain synchronization code.
  • the estimation method provided by the present invention does not need to achieve accurate sampling position, quasi-determination synchronization, and even stable operation without knowing multipath distribution and position information.
  • FIG. 2 is a characteristic diagram of a raised cosine compensation sequence used in the present invention.
  • FIG. 3 is a flowchart of a frequency offset estimation process in an embodiment of the present invention.
  • FIG. 4 is a schematic diagram of a data window used in the present invention.
  • FIG. 5 is a schematic diagram showing the influence of the position deviation and the timing deviation on the estimated value in the present invention
  • FIG. 6 is a schematic diagram showing the influence of the random multipath distribution on the estimated value in the present invention
  • FIG. 7 is a schematic diagram of a standard deviation of a residual frequency offset of an AWGN channel according to the present invention.
  • Figure 8 is a schematic view showing the influence of the position deviation and the timing deviation of the sample value on the residual frequency offset performance in the present invention.
  • FIG. 10 is a schematic diagram of the effect of a random multipath channel on residual frequency offset performance in the present invention.
  • Figure 11 is a block diagram showing the apparatus for estimating the frequency deviation of the present invention. Preferred embodiment of the invention
  • the existing frequency offset estimation method Due to the short length of the synchronization code and the midable sequence used in TD-SCDMA, the existing frequency offset estimation method has serious performance degradation in the environment of sample position deviation and dense multipath distribution, and in some scenarios. It may turn into a biased estimate, causing the system to not work properly.
  • One solution is to ensure that the working environment of the existing estimation method meets the requirements through techniques such as precise timing synchronization, accurate multipath position estimation and inter-path interference cancellation, but often these techniques are complicated and difficult to be in harsh scenes. The accuracy is guaranteed, and most of these technologies rely on the control accuracy of the AFC itself, making the system sensitive and not robust enough to the working scene.
  • the basic principle of the present embodiment is to construct a local frequency domain synchronization code as a reference, and transform the received downlink synchronization code into a frequency domain and then perform a correlation operation on the local frequency domain synchronization code, and use the position of the maximum correlation value as the frequency offset estimation result.
  • the local frequency domain synchronization code is a set of synchronization code sequences ⁇
  • the rules of the correlation operation are as follows. For example, the sequences al, a2, ... an are related to the sequences bl, b2, ... bn, and each element of the b sequence is conjugated to the element corresponding to the position of the a sequence.
  • the embodiment is divided into a pre-processing process and a frequency offset estimation process, where the pre-processing process generates a local frequency domain synchronization code, which needs to be executed when the terminal is powered on or the handover service cell causes the synchronization code sequence number to change, and the frequency offset estimation process is received at the terminal.
  • the downlink synchronization code of each subframe is executed afterwards.
  • the synchronization code number refers to an identifier that is pre-assigned to each cell, and that distinguishes the cells when the distance between the two cells is relatively close and the same frequency is used.
  • the downlink synchronization code is a sequence that is sent by the base station to the terminal for timing synchronization and frame synchronization. In fact, the downlink synchronization code to be sent by the base station is pre-configured in the terminal.
  • the pre-processing process can also be performed on the base station side, that is, the local frequency domain synchronization code is generated by the base station, and then the base station transmits the generated local frequency domain synchronization code to the terminal.
  • FIG. 1 is a pre-processing process in the present embodiment.
  • the terminal locally generates N signals of different frequencies, where N > 3;
  • the above signals are best selected with a single tone.
  • N is a minimum of 3, and when N is 3, it contains a 0-frequency tone, a positive frequency tone, and a negative frequency tone.
  • the length of the tone can be selected with reference to the length of the sync code to select 64 samples, and the single chip (cut spectrum) is selected.
  • the method of generating a tone can be any known method available.
  • N single tones can contain one zero frequency tone, and each of the positive and negative frequency tones (Nl)/2.
  • the frequency interval between single tones can be selected as 156.25Hz, and the frequency estimation range can be selected from -9843.75Hz to +9843.75Hz.
  • the frequency estimation range of -9843.75Hz ⁇ +9843.75Hz is based on the 3ppm crystal oscillator used in the terminal.
  • the maximum frequency offset is positive and negative 6kHz under the 2GHz carrier, taking into account the frequency offset and channel Doppler effect of the base station. After a certain redundancy is selected, different choices can be made for different crystal oscillators.
  • the frequency interval of 156.25 Hz is similar to the frequency estimation range, and adaptive selection is possible.
  • the number of positive and negative frequency tones will be asymmetrical. In this case, you can choose to generate more positive frequency tones, or you can choose to generate more negative frequency tones.
  • the terminal multiplies each point of the pre-configured synchronization code (the synchronization code to be sent by the base station) by (7), where j is a complex unit, where i is the number of bits of the synchronization code sample point;
  • the terminal completes (the pre-configured synchronization code after the ⁇ operation is multiplied by the N tones respectively, to perform frequency offset operation on the pre-configured synchronization code to obtain N single tone synchronization codes;
  • each sequence in the N tone synchronization codes is 64 samples, and for the sequence of N 64 samples, 64 zeros are added to the data tail of each group of sequences. Forming a sequence of 128 samples of N groups;
  • the FFT transform increases the number of points in the sequence to 128 in the time domain. After the FFT transform, the resolution of the high frequency domain can be improved.
  • the terminal separately compensates the N single tone synchronization codes in the obtained frequency domain to be compatible with the baseband signal;
  • the raised cosine compensation is used first, and the best compensation effect can be achieved, and the root number raised cosine compensation can also be performed.
  • the base station signal on the terminal side is obtained by the base station performing the root number raised cosine filter shaping on the base station signal and the terminal performing the root number raised cosine filter matching
  • the base station signal is formed by the root number raised cosine filter, and the frequency range is -800 kHz to +800 kHz.
  • 0 ⁇ 480kHz has no attenuation
  • 480kHz ⁇ 640 kHz power attenuation is half
  • power is attenuated to 0 at 800kHz
  • -800kHz ⁇ 0 is symmetric with this, that is, from -640kHz to 480kHz and 480kHz ⁇
  • the signal has entered the roll-off bandwidth of the raised cosine filter (gradually attenuated bandwidth), so in order to more accurately match the received frequency domain signal, the cosine compensation is performed on the N sets of sequences respectively (two root numbers)
  • the raised cosine filter is equivalent to a raised cosine filter), and the characteristics of the compensation sequence are shown in Figure 2.
  • the abscissa indicates the sample point, and the ordinate indicates the amplitude.
  • the frequency corresponding to the sample point of the sequence is multiplied by the amplitude corresponding to the sample point in Fig. 2.
  • the terminal reorders each tone synchronization code according to a frequency from low to high, to obtain a new frequency domain sequence, and intercepts a frequency range of -640 kHz to 640 kHz;
  • the FFT transform Since the FFT transform is performed, with 0 as the boundary, the first half of the sequence corresponds to the positive frequency, and the second half of the sample corresponds to the negative frequency, which is not monotonically increasing.
  • the frequency of the sequence after reordering satisfies the monotonous increase relationship. Subsequent differential operations provide convenience.
  • the actual bandwidth of the TD signal is -800kHz ⁇ 800kHz, but the ratio of signal power to total power outside the 640kHz bandwidth is very small. Therefore, in order to simplify the operation, only the signals in the range of -640kHz ⁇ 640kHz are reserved, and of course, according to the requirements. Intercepting different frequency ranges is not limited here.
  • the terminal separately performs a differential operation on the reordered sequence to obtain a differential frequency domain synchronization code; the difference operation process is: multiplying each data in the sequence by a conjugate of the data with the sequence number plus one in the sequence, to generate A new set of sequences.
  • the timing deviation causes the phase of the frequency domain to increase linearly with the frequency point
  • the multipath effect of the wireless channel It should cause the frequency selective effect in the frequency domain of the received signal.
  • the differential operation here can eliminate the influence of timing deviation and greatly suppress the influence of multipath (frequency selectivity).
  • PtsF PtsF(l: end - l,:). * conj(PtsF(2: end,:)), where PtsF is the frequency domain sequence generated in step 107, each data in the sequence The conjugate of the data with the sequence number plus one in the sequence is multiplied to generate a new set of sequences.
  • step 107 If the reordering of step 107 is not performed, the adjacent samples in the samples 0 to 63, the adjacent samples in the samples 67-127 are separately subjected to the difference operation, and the samples 0 and 127 are further performed. Differential operation, to obtain a differential frequency domain synchronization code.
  • the terminal performs power normalization on each of the N differential frequency domain synchronization codes, that is, the power of each differential frequency domain synchronization code is the same, and serves as a reference for the local frequency domain synchronization code.
  • the frequency of the above-mentioned N sets of tones is evenly divided. Considering the application of AFC, it is only necessary to have a higher estimation accuracy near the 0 frequency. The accuracy far away from the 0 frequency only slightly affects the convergence speed. Therefore, It is also possible to set the frequency point by means of successive two partitions.
  • the method of successive two partitions refers to the highest frequency to be estimated, 1/2, 1/4, 1/8, ... of the highest frequency as the actual frequency point, for example, the estimation range.
  • the frequency of 8kHz, -8kHz, 4kHz, -4kHz, 2kHz, -2kHz, 1kHz, -1kHz is selected as the frequency point, which can effectively reduce the local storage space and computation.
  • N is 127
  • the local frequency domain reference between the two partitions is 13 groups.
  • FIG. 3 is a frequency offset estimation process in the embodiment, including:
  • the terminal After receiving the subframe, the terminal searches for the location of the downlink synchronization code, and determines the starting position of the subframe.
  • Each sub-frame has a downlink synchronization code, and the start position of the sub-frame is found by finding the downlink synchronization code.
  • the terminal intercepts the downlink synchronization code signal from the subframe according to the length of the data window, and the length of the data window is predetermined, for example, the length of the downlink synchronization code and the samples of each of the m chips, m > 0 ;
  • the length of the intercepted signal is the downlink synchronization code plus the length of each m chip
  • the downlink synchronization code is 64 chip
  • m chip samples are added on both sides of the downlink synchronization code, which can be guaranteed to exist.
  • the timing deviation of m chips is not exceeded, the entire downlink can be completely intercepted. Step code.
  • the terminal adds 0 to the tail of the downlink synchronization code signal, so that the number of samples of the intercepted signal is an integer multiple of the local synchronization code, and transforms it into the frequency domain;
  • the chip rate is 1.6MHz. Since the signal bandwidth is 1.6MHz, considering that the sample rate needs to be greater than the signal bandwidth, the chip rate is
  • twice the chip rate is the sample rate, and the data window is 64chip+16chi.
  • the sync code plus the guard strips at both ends has a total of 160 samples, which can intercept the signal.
  • the tail is complemented by 96 0s, which constitutes a 256-point time domain sequence, 256 is an integer multiple of the local synchronization code, and is an integer multiple of 2, which facilitates FFT conversion.
  • SignalF FFT([Signal; zeros(96, 1 )])), which adds 96 zeros and performs FFT transformation.
  • Signal is the received signal in the time domain, and SignalF is the transformed frequency domain signal.
  • the intercepted signal may not be complemented by 0, and directly converted to the frequency domain.
  • the data window may also be determined as 64 chip, m is taken as 0, and then the intercepted signal may not be complemented by 0. , directly convert the frequency domain.
  • the frequency domain range of the local synchronization code is -640 kHz to 640 kHz, and the frequency domain range of the intercepted signal may be the same as the local synchronization code, or Different, as long as there is an intersection.
  • SignalF SignalF([l 93: 256,1: 64])
  • SignalF is the transformed frequency domain signal variable generated in step 303, a total of 256 points. This step will be the first 64 points (1 ⁇ 64) and the last 64 points ( 193 ⁇ 256) interchangeable.
  • the terminal performs a differential operation on the downlink synchronization code signal.
  • SignalF SignalF(l: end - 1). * conj(SignalF(2: end)), where SignalF is the frequency domain signal variable generated in step 304.
  • Each data in the sequence corresponds to the sequence number plus 1 data.
  • the yoke performs a multiplication operation to generate a new set of sequences.
  • the data window has a known offset of 8 chips from the downlink synchronization code, and the fixed position offset is pre-compensated in order to improve the estimated signal-to-noise ratio;
  • SignalF SignalF * exp(-2 * pi * j/16) , where SignalF is the frequency domain signal variable generated in step 305, and the exponential phase of exp is the phase corresponding to l/8 pi; Linearly increasing phase rotation. After the difference, the rotation is a constant phase.
  • the compensation method in this step is the effect of the frequency domain rotation compensation fixed time domain timing deviation.
  • the terminal separately performs a conjugate-related operation on the differential downlink synchronization code and the local frequency domain synchronization code reference.
  • CorrC real(SignalF*PtsF)' , where SignalF is the frequency domain signal obtained in step 306,
  • PtsF is the local frequency domain synchronization code
  • real is the complex real part operation
  • CorrC is the generated correlation result, that is, the conjugate downlink synchronization code sequence, and each value in the sequence corresponds to each reference frequency of the local code.
  • the terminal may perform smoothing filtering on the downlink synchronization code sequence to improve estimation accuracy;
  • a 1st order IIR filter can be used.
  • the typical filter coefficient FltCoef is 1/16.
  • CorrR (1 - FltCoef) * CorrR + FltCoef * CorrC , where is the result of multiple observations, where CorrC is the current single correlation result, FltCoef is the filter coefficient, CorrR is the correlation cumulant of multiple results, FltCoef It determines that the current single correlation result accounts for the weight in the relevant cumulant. The smaller the FltCoef is, the stronger the effect of smoothing filtering, and the weaker the smoothing filtering effect.
  • the terminal determines a bit of the maximum value in the conjugate downlink synchronization code sequence, obtains a frequency offset estimation result, and calculates a frequency offset.
  • CorrM, FreqEsti max(CorrR)
  • CorrR is a set of correlation cumulants corresponding to the frequency offset, where the frequency offset corresponding to the maximum cumulant is the frequency offset estimate that needs to be solved, and max is the maximum value.
  • CorrM is the maximum correlation cumulant value
  • FreqEsti is the index of the largest correlation cumulant, for example, CorrR is 65, 78, 33, 58; then CorrM calculates 78 and FreqEsti evaluates to 2.
  • FreqEsti (FreqEsti - 64) * 156.25 , which converts the index value of the frequency offset estimate to the actual
  • the frequency offset is determined by the frequency of the tone signal in step 101.
  • N is 127
  • the tone frequency interval is 156.25 Hz.
  • the frequency is one by one.
  • the 156.25 Hz is added, and the actual frequency of the No. 64 tone is 0 Hz, so there is a correspondence described in this equation.
  • the influence of the sample position deviation and the timing deviation is further analyzed to make the effect of the embodiment more obvious.
  • the sample position deviation and the timing deviation cause a linear phase rotation which increases with the increase of the frequency in the frequency domain, and the differential frequency
  • This rotation is converted to a constant phase value.
  • the amplitude of the projection value is only slightly affected, but even when the timing deviation is up to 4 chips, the projection angle is pi/16, the projection value. It also only drops to the original value of 0.9808, and the effect on the signal-to-noise ratio is almost negligible.
  • Figure 5 simulates the effect of the sample position deviation and timing deviation on the estimated value.
  • the excitation signal is the no-frequency offset synchronization code.
  • the sample position deviation is gradually changed from -4chip to +4chip, and each time the transition is l/4chip.
  • the estimation method of the present embodiment is independent of the timing deviation except for the position of the sample value, that is, This method does not rely on the timing synchronization accuracy of the system, nor does it need to provide an accurate sync code position.
  • the estimation method does not use multipath position information, that is, this implementation.
  • the method of the method does not require multipath position estimation to work well.
  • the following is a TD-SCDMA automatic frequency control based on the method of the present embodiment.
  • the (AFC) loop is taken as an example to explain the use of the method.
  • AFC AFC Integrated Circuit
  • TotalFreqEsti TotalFreqEsti + AdjCoef * FreqEsti , where FreqEsti is the current frequency offset estimate, AdjCoef is the adjustment compensation, and TotalFreqEsti is the total frequency offset estimate.
  • the physical meaning of this equation is that each residual residual frequency estimate is constant. The discount is included in the total frequency offset to reduce the impact of a single estimation error on the frequency offset adjustment.
  • the standard deviation performance of the residual frequency offset of the AFC loop under the AWGN channel is analyzed. Please refer to FIG. 7.
  • the simulation number of each sample in FIG. 7 is 10,000 subframes. It can be seen that the AFC constructed based on the method is The loop can achieve better performance under the AWGN channel, and the standard deviations of 0dB, 5dB and 10dB SNR are better than 50Hz, 30Hz and 20Hz respectively.
  • Figure 9 compares the performance of static and other strong 2 and equal 4 environments. These paths are separated by lchip, energy and phase are the same. It can be seen that static multipath has little effect on performance.
  • the simulation results show that the AFC loop can still work stably without significant degradation in performance in the presence of dense multipath and completely unknown multipath information.
  • FIG. 11 is a diagram showing an apparatus for estimating a frequency offset according to an embodiment of the present invention, including: a local frequency domain synchronization code construction module and a frequency deviation estimation module connected to each other, wherein:
  • the local frequency domain synchronization code construction module is configured to construct a local frequency domain synchronization code, and send the constructed local frequency domain synchronization code to the frequency deviation estimation module; wherein, the process of constructing the local frequency domain synchronization code
  • the method includes: performing a frequency offset operation on the pre-configured synchronization code; after completing the frequency offset operation, eliminating the influence of the timing offset and the multipath on the pre-configured synchronization code, and obtaining the local frequency domain synchronization code.
  • the frequency deviation estimation module is configured to: after receiving the subframe, convert the downlink synchronization code of the subframe to the frequency domain, and perform correlation operation between the converted downlink synchronization code and the received local frequency domain synchronization code to obtain
  • the conjugate downlink synchronization code sequence determines the bit of the maximum value in the conjugate downlink synchronization code sequence, and calculates the frequency deviation according to the bit position of the maximum value.
  • the present invention does not need to achieve accurate sample position, accurate timing synchronization, and even stable operation without knowing the multipath distribution and position information.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Synchronisation In Digital Transmission Systems (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本发明公开了一种频率偏差的估计方法及装置,包括:接收到子帧后,将该子帧的下行同步码转换到频域,并将转换后的下行同步码与本地频域同步码进行相关运算,得到共轭下行同步码序列;判断共轭下行同步码序列中的最大值所在位,根据该最大值所在位折算出频率偏差值。本发明提供的估计方法不需要实现精确采样值位置、准确定时同步,甚至不需要获知多径分布和位置信息就可以较为稳定的工作。

Description

一种频率偏差的估计方法及装置
技术领域
本发明涉及时分同步码分多址接入 ( Time Division-Synchronized Code Division Multiple Access, TD-SCDMA ) 系统, 尤其涉及一种频率偏差的估计 方法及装置。
背景技术
时分同步码分多址 ( Time Division- Synchronized Code Division Multiple Access, TD-SCDMA )是 3G (第三代移动通信技术) 的三大主流标准之一, 具有广泛的应用前景。
在 TD-SCDMA系统中,基站( Base Station, BS )和终端( User Equipment,
UE )均以标称的载波频率进行发送和接收。 由于器件水平限制, 实际发送和 接收的频率之间会有一定的偏差。 TD-SCDMA系统要求基站的载波频率误差 小于 0.05PPM,要求终端的载波频率误差小于 0.1PPM。在基站侧, 由于温度、 体积、 功耗和成本等的限制比较小, 振荡器的频率精度可以满足要求, 而在 终端侧, 受各种因素限制, 所选用的晶体振荡器的频率精度通常不满足标准 要求。 AFC (自动频率控制) 的作用是纠正 BS和 UE之间的载波频率误差, 以保证后续解调解码的工作性能。
现有的频偏估计大都是利用信标信道中的下行导频时隙 (Downlink Pilot
TimeSlot, DwPTS )和训练序列 ( Midamble )进行。 原理是将数据分为前后 两部分各自进行信道估计, 以两段估计值相位旋转的差距来推算载波频偏。
由于同步码较短的自身特性所限, 釆样位置偏差和多径分布都会导致现 有方法在 0频率附近明显退化为有偏估计,且这种偏差难以通过 AFC控制策 略解决, 只能通过尽力维护工作场景的方式来降低这一影响。 例如:
( 1 ) 实现更加精确的定时, 以保证釆样值位置偏差远小于 l/4chip (切 谱, 时间单位, 1/1.28M秒) 。
( 2 )釆用径间干扰抵消, 多径信道重构等多径位置估计技术抑制多径干 扰的负面效应。
但是, 上述技术的引入又给系统带来了一系列新的问题:
( 1 )上述估计和抵消技术自身的误差引入了新的干扰, 降低了性能和稳 定性。
( 2 )这些技术大都需要较高的频率精度作为前提, 当频偏估计性能下降 时, 这些估计和抵消效果也会降低, 进一步引起频偏估计性能的下降, 会陷 入恶性循环, 系统不够稳健。
( 3 )在恶劣环境下,如何高精度的实现上述技术往往本身就是一个难题, 增加了实现复杂度和开发周期。
发明内容
本发明提供一种频率偏差的估计方法及装置, 以解决釆用精确定时和多 径位置估计等技术带来的性能和稳定性降低等问题, 实现消除频偏估计对精 确定时和多径位置估计的依赖, 提高系统的整体稳定性。
为解决上述技术问题, 本发明的一种频率偏差的估计方法, 包括: 接收到子帧后, 将该子帧的下行同步码转换到频域, 并将转换后的下行 同步码与本地频域同步码进行相关运算, 得到共轭下行同步码序列; 以及 判断共轭下行同步码序列中的最大值所在位, 根据该最大值所在位折算 出频率偏差值。
本地频域同步码可以为一组受到不同频偏影响的同步码序列。 该方法还 可包括: 通过如下方式构造本地频域同步码: 对预先配置的同步码进行频率 偏移操作; 以及, 完成频率偏移操作后, 消除定时偏差及多径对预先配置的 同步码的影响, 得到本地频域同步码。
进行频率偏移操作的步骤可包括: 产生不同频率的信号; 将预先配置的 同步码中的各样点分别乘以(γ , 其中, j为复数单位, i为各样点在预先配置 的同步码中所在的位数; 以及, 将完成 ( )运算后的同步码与所产生的信号分 别相乘。
消除定时偏差及多径对预先配置的同步码的影响的步骤可包括: 将预先 配置的同步码转换至频域, 并对频域上的该同步码进行差分运算, 得到本地 频域同步码。 该方法还可包括: 在将预先配置的同步码转换至频域前, 对该 预先配置的同步码进行补偿, 以与基带信号相适应。 补偿可以为升余弦补偿。
该方法还可包括: 在对频域上的该同步码进行差分运算前, 按照各样点 所对应的频率的高低对频域上的该同步码的样点进行重排序。
该方法在将子帧的下行同步码转换到频域的步骤前还可包括: 从所接收 到的子帧中查找下行同步码的位置, 按照数据窗的长度从子帧中截取出下行 同步码信号; 以及若下行同步码信号的样点数不是本地频域同步码的样点数 的整数倍, 则对下行同步码信号进行处理, 使该下行同步码信号的样点数为 本地频域同步码的样点数的整数倍。 数据窗的长度可以为下行同步码的切谱 ( chip )长度加下行同步码左右各 m个 chip长度, 其中, m > 0。
该方法还可包括: 在下行同步码信号的样点数为本地频域同步码的样点 数的整数倍时, 对下行同步码信号进行差分运算。 该方法还可包括: 对下行 同步码信号进行差分运算后, 进行消除固定时域定时偏差影响的补偿。
得到共轭下行同步码序列的步骤可包括: 在将转换后的下行同步码与本 地频域同步码进行相关运算后, 将每个相关运算结果的虚部丟弃, 保留每个 相关运算结果的实部, 得到共轭下行同步码序列。
不同频率的信号中可包括 0赫兹频率的信号; 根据最大值所在位折算出 频率偏差值的步骤可包括: 将共轭下行同步码序列中的最大值所在位与不同 频率的信号中 0赫兹频率信号所在位之间的位数差, 乘以该不同频率的信号 的频率间隔, 得到频率偏差值。
进一步地, 本发明的一种频率偏差的估计装置, 包括: 相互连接的本地 频域同步码构造模块和频率偏差估计模块, 其中,
本地频域同步码构造模块设置成构造本地频域同步码, 并将所构造的本 地频域同步码发送给频率偏差估计模块;
频率偏差估计模块设置成: 在接收到子帧后, 将该子帧的下行同步码转 换到频域, 并将转换后的下行同步码与所接收到的本地频域同步码进行相关 运算, 得到共轭下行同步码序列, 判断共轭下行同步码序列中的最大值所在 位, 根据该最大值所在位折算出频率偏差。
本地频域同步码构造模块可设置成通过如下方式构造本地频域同步码: 对预先配置的同步码进行频率偏移操作; 以及, 在完成频率偏移操作后, 消 除定时偏差及多径对预先配置的同步码的影响, 得到本地频域同步码。
综上所述, 本发明提供的估计方法不需要实现精确釆样值位置、 准确定 时同步, 甚至不需要获知多径分布和位置信息就可以较为稳定的工作。 附图概述
图 1为本发明实施方式中的预处理过程的流程图;
图 2为本发明中釆用的升余弦补偿序列的特性图;
图 3为本发明实施方式中的频偏估计过程的流程图;
图 4为本发明中釆用的数据窗的示意图;
图 5为本发明中釆样位置偏差和定时偏差对估计值影响的示意图; 图 6为本发明中随机多径分布对估计值影响的示意图;
图 7为本发明中 AWGN信道残余频偏标准偏差的示意图;
图 8为本发明中釆样值位置偏差和定时偏差对残余频偏性能影响的示意 图;
图 9为本发明中静态多径对残余频偏性能影响的示意图;
图 10为本发明中随机多径信道对残余频偏性能影响的示意图;
图 11为本发明频率偏差的估计装置的结构图。 本发明的较佳实施方式
由于 TD-SCDMA釆用的同步码和 midable序列长度都较短, 因此现有的 频偏估计方法在釆样位置偏差、 密集多径分布等环境中, 性能退化较为严重, 且在某些场景下可能转变为有偏估计, 致使系统不能正常工作。 一种解决方 式是通过精确定时同步、 精确多径位置估计和径间干扰抵消等技术来保障现 有估计方法的工作环境满足要求, 但往往这些技术既复杂又难以在恶劣场景 中保证精度, 此外大多数此类技术本身还依赖于 AFC的控制精度, 致使系统 对于工作场景颇为敏感, 不够稳健。
本实施方式的基本原理是构建本地频域同步码作为基准, 将接收到的下 行同步码变换到频域后本地频域同步码进行相关运算, 以最大相关值所在位 置作为频偏估计结果。 本地频域同步码为一组受到不同频偏影响的同步码序 歹 |J。 相关运算的规则如下, 例如, 序列 al,a2,...an与序列 bl,b2,...bn进行相 关运算, 将 b序列每个元素取共轭后与 a序列的对应位置的元素相乘产生序 歹' J al*conj(bl),a2*conj(b2),...an*conj(bn), 其中 conj为复数共轭运算, 最后对 该序歹 'J累力口求和 sum(al*conj(bl),a2*conj(b2),...an*conj(bn)), 其中 , sum为求 和运算。
本实施方案分为预处理过程和频偏估计过程, 其中, 预处理过程生成本 地频域同步码,在终端开机或切换服务小区导致同步码序号改变时需要执行, 频偏估计过程在终端接收到每个子帧的下行同步码后均要执行。 同步码序号 是指预先分配给每个小区、 在两个小区距离较近且釆用相同频率的情况下对 小区进行区分的标识。 下行同步码是指基站发送给终端, 供终端进行定时同 步和帧同步的一个序列, 实际上终端中已预先配置了基站所要下发的下行同 步码。
预处理过程也可以在基站侧进行, 即, 由基站生成本地频域同步码, 再 由基站将所生成的本地频域同步码下发给终端。
下面结合附图对本发明的具体实施方式进行详细说明。
图 1为本实施方式中的预处理过程, 下面以终端侧进行预处理过程为例 进行说明, 包括:
101 : 终端本地产生 N个不同频率的信号, 其中, N > 3;
上述信号最佳选用单音。
N最小为 3 , 当 N为 3时, 包含一个 0频率单音, 一个正频率单音和一 个负频率单音。
釆用单音时, 单音的长度可参照同步码的长度选择 64样点, 且选择单倍 chip (切谱) 。 产生单音的方法可选用现有任何已知方法。 N个单音中可以包含 1个 0频率单音,正频率和负频率单音各 (N-l)/2个。 如, 当 N=127时, 包含正频率和负频率单音各 63个, 单音之间的频率间隔 可选择 156.25Hz , 频率估计范围可选择 -9843.75Hz~+9843.75Hz 。 -9843.75Hz~+9843.75Hz 的频率估计范围是基于终端釆用 3ppm 的晶振, 在 2GHz载波下最大频偏为正负 6kHz, 考虑到基站的频偏和信道多普勒效应, 并在此基础上留有一定的冗余后所选, 对于不同的晶振可进行不同的选择, 156.25Hz 的频率间隔与频率估计范围类似, 可进行适应性选择。
当 N为偶数时, 除去一个 0频率单音, 正频率和负频率单音的数量将不 对称, 此时可选择多生成一个正频率单音, 也可以选择多生成一个负频率单 音。
102: 终端对预先配置的同步码(基站将会下发的同步码)的各样点乘以 (7) , 其中, j为复数单位, i为同步码样点所在的位数;
如对于同步码的第 0位样点,将其乘以 1 ;将同步码的第 1位样点乘以 j; 将同步码的第 2位样点乘以 j2
103: 终端将完成 ( γ运算后的预先配置的同步码与 N个单音分别相乘, 以对预先配置的同步码进行频率偏移操作, 得到 N个单音同步码;
104: 在单音为 64样点时, N个单音同步码中的每个序列均为 64样点, 对该 N个 64样点的序列 ,在每组序列的数据尾部补上 64个 0,构成 N组 128 样点序列;
由于下面需要对 N组序列进行快速傅立叶变换(Fast Fourier Transform,
FFT )变换, 在时域上将序列的点数增加到 128, 在进行 FFT变换后, 可以提 高频域的分辨率。
105: 对 N组序列分别进行 FFT变换, 将时域的序列转换至频域; 上述操作可表示为: PtsF = FFT(PtsT. * SftF,128) , 其中, PtsT为完成 /y运算后的预先配置的同步码, SftF为单音, PtsF为 FFT 变换后得到的频 域上的序列, 128表示进行 128点 FFT变换。
106: 终端对所得到的频域上的 N个单音同步码分别进行补偿, 以与基 带信号相适应; 此处优先釆用升余弦补偿, 可以到达最佳的补偿效果, 也可以进行根号 升余弦补偿。
由于终端侧的基带信号是基站对基站信号进行根号升余弦滤波成型以及 终端进行根号升余弦滤波匹配后得到的,基站信号经根号升余弦滤波成型后, 频率范围为 -800kHz~+800kHz, 其中, 0~480kHz无衰减, 480kHz~640 kHz功 率衰减到一半, 功率在 800kHz的频率上已衰减到 0, -800kHz~0的情况与此 对称, 也就是说从 -640kHz— 480kHz以及 480kHz~640kHz开始, 信号已经进 入升余弦滤波器的滚降带宽 (逐渐衰减的带宽) , 因此, 为了更准确地与接 收到的频域信号匹配, 分别对 N组序列进行升余弦补偿(两次根号升余弦滤 波相当于一次升余弦滤波) , 其中, 补偿序列的特性请参考图 2。
在图 2中横坐标表示样点, 纵坐标表示幅度, 对序列进行补偿时, 将序 列的样点对应的频率乘以图 2中该样点对应的幅度。
107: 终端对每个单音同步码按照频率的从低到高的顺序进行重新排序, 得到新的频域序列, 并截取频率范围为 -640kHz~640kHz;
由于进行 FFT变换后, 以 0为界, 序列的前半部分样点对应正频率, 后 半部分样点对应负频率, 并不是单调增加的关系, 重排序后序列的频率满足 单调增加的关系, 为后续的差分运算提供了方便。
TD信号的实际带宽为 -800kHz~800kHz ,但 640kHz带宽以外信号功率占 总功率的比例非常微小, 因此, 为简化运算此处仅仅保留了 -640kHz~640kHz 内的信号参与运算, 当然也可以根据需求截取不同的频率范围, 在此并不限 定。
例如, 序列的重新排序可表示为: PtsF = PtsF([65 : 128,1 : 64],:) , 其中, PtsF为频域同步码变量, 共 128点, 本步骤运算将前 64点 ( 1 ~ 64与后 64 点 (65 ~ 128 )位置互换。
108: 终端对重排序后的序列分别进行差分运算, 获得差分频域同步码; 差分运算的过程为: 将序列中每个数据与该序列中序号加 1的数据的共 轭进行相乘, 生成一组新的序列。
考虑到定时偏差会引起频域的相位随频点线性增大, 无线信道的多径效 应会引起接收信号频域上的频率选择性效果, 此处的差分运算能够消除定时 偏差影响, 大幅抑制多径(频率选择性) 的影响。
差分运算可表示为: PtsF = PtsF(l : end - l,:). * conj(PtsF(2 : end,:)),其中, PtsF为步骤 107生成的频域序列, 序列中每个数据与该序列中序号加 1的数 据的共轭相乘, 生成一组新的序列。
如果不进行步骤 107的重新排序, 则此处对样点 0~63中的相邻样点, 样 点 67-127中的相邻样点分别进行差分运算,并对样点 0和 127再进行差分运 算, 得到差分频域同步码。
109: 终端对 N个差分频域同步码各自进行功率归一化, 也就是使每个 差分频域同步码的功率均相同, 作为本地频域同步码基准。
上述 N组单音的频点是均匀划分的 ,考虑到 AFC应用的场合实际只需要 在 0频率附近有较高的估计精度, 远离 0频率处的精度只会对收敛速度略有 影响, 因此, 也可以釆用逐次二分区间的方式设置频点, 逐次二分区间法是 指将需要估计的最高频率、 最高频率的 1/2, 1/4, 1/8, …作为实际频点, 例 如估计范围选取 -8kHz ~ +8kHz时, 选取频点 8kHz, -8kHz, 4kHz, -4kHz, 2kHz, -2kHz, 1kHz, -1kHz作为频点, 可以有效地降低本地存储空间和运算 量。 N取 127时, 二分区间方式本地频域基准为 13组。
图 3为本实施方式中的频偏估计过程, 包括:
301 : 终端在接收到子帧后, 查找下行同步码的位置, 确定子帧的起始位 置;
每个子帧有一个下行同步码, 查找到下行同步码就找到了子帧的起始位 置。
302: 终端按照数据窗的长度从子帧中截取出下行同步码信号, 数据窗的 长度为预先确定,如取定为下行同步码及左右各 m个 chip的样点构成的长度, m > 0;
如图 4所示,截取的信号的长度为下行同步码加上左右各 m个 chip的长 度, 下行同步码为 64chip , 在下行同步码的两边各增加 m个 chip的样点, 可 以保证当存在不超过 m个 chip的定时偏差时,可以完整的截取到整个下行同 步码。
考虑到最大可能的定时偏移和时延扩展, 这里 m取值为 8。
303: 终端在下行同步码信号的尾部补 0, 使截取信号的样点数为本地同 步码的整数倍, 并将其变换至频域;
由于信号带宽为 1.6MHz, 考虑到釆样率需要大于信号带宽, chip速率是
1.28MHz , 可取两倍 chip速率为釆样率 , 而数据窗为 64chip+ 16chi , 按照两 倍 chip速率的釆样率, 则同步码加上两端保护带共 160个样点, 可以在截取 信号的尾部补上 96个 0, 构成 256点时域序列, 256为本地同步码的整数倍, 且是 2的整数倍, 方便进行 FFT转换。
SignalF = FFT([Signal; zeros(96, 1 )]) , 补充 96个零点后将其进行 FFT变 换, 其中 Signal是时域的接收信号, SignalF是变换后的频域信号。
如果截取信号的点数为本地同步码的整数倍, 也可以不对截取信号进行 补 0, 直接转换到频域, 如数据窗也可以确定为 64chip, m取 0, 此时就可以 不对截取信号补 0, 直接进行频域的转换。
304: 对应于本地同步码的频域范围, 选取下行同步码信号的频域范围; 本地同步码的频域范围为 -640kHz~640kHz , 截取信号的频域范围可以与 本地同步码相同, 也可以不同, 只要有一个交集即可。
SignalF = SignalF([l 93: 256,1: 64]) , SignalF为步骤 303生成的变换后频 域信号变量, 共 256点, 本步骤运算将前 64点 ( 1 ~ 64 )和后 64点 ( 193 ~ 256 )互换。
305: 终端对下行同步码信号进行差分运算;
SignalF = SignalF(l: end - 1). * conj(SignalF(2: end)), 其中, SignalF为步 骤 304生成的频域信号变量, 序列中每个数据对应与序列中序号加 1数据的 共轭进行相乘运算, 生成一组新的序列。
306: 考虑到数据窗选取时, 数据窗较下行同步码有 8chip的已知偏移, 为了提高估计信噪比, 对此固定位置偏移进行预补偿;
SignalF = SignalF * exp(-2 * pi * j/16) , 其中, SignalF为步骤 305生成的 频域信号变量, exp的指数相为 l/8pi对应的相位; 由于定时偏差导致随频率 线性增加的相位旋转, 差分后该旋转是常数相位, 本步骤补偿方式为频域旋 转补偿固定时域定时偏差的影响, 除了本步骤釆用的方法外也可以釆用时域 上的定时平移补偿的方式, 即步骤 303 中 的表达式变更为 SignalF = FFT([Signal(17: end);zeros(96,l);Signal(l: 16)]) , 时域信号向左平 移 16个样点 ( 8个 chip ) 。
307:终端将差分下行同步码与本地频域同步码基准分别进行共轭相关运 算;
308: 为了提高信噪比和平滑滤波增益并降低实现复杂度, 仅保留每组共 轭相关运算结果的实部, 将虚部丟弃, 得到共轭下行同步码序列;
CorrC = real(SignalF*PtsF)' ,其中, SignalF为步骤 306得到的频域信号,
PtsF为本地频域同步码, real为复数取实部运算, CorrC为生成的相关结果, 即共轭下行同步码序列,序列中的每个值与本地码的每个基准频率——对应。
309: 由于下行同步码的实际频偏在 5ms的观测周期上是一个緩变过程, 终端可以对下行同步码序列进行平滑滤波, 以提高估计精度;
简单起见,可以釆用了一个 1阶的 IIR滤波器,典型的滤波器系数 FltCoef 取值为 1/16。
CorrR = (1 - FltCoef) * CorrR + FltCoef * CorrC ,此处为多次观测的结果, 其中, CorrC为当前单次的相关结果, FltCoef 为滤波系数, CorrR为多次结 果的相关累积量, FltCoef决定了当前单次相关结果占相关累积量中的权重, FltCoef越小平滑滤波的效果越强, 反之平滑滤波的效果越弱。
310: 终端判断共轭下行同步码序列中最大值所在的位, 获得频偏估计结 果, 并折算出频偏。
[CorrM, FreqEsti] = max(CorrR) , CorrR为一组与频偏——对应的相关累 积量, 其中最大累积量所对应的频偏为需要求解的频偏估计值, max为求取 最大值运算, CorrM为最大的相关累积量的数值, FreqEsti为最大的相关累积 量的索引,例如, CorrR为 65, 78, 33 , 58; 则 CorrM计算结果为 78, FreqEsti 计算结果为 2。
FreqEsti = (FreqEsti - 64) * 156.25 , 该式将频偏估计的索引值转换为实际 频偏大小, 其对应关系取决于步骤 101 中的单音信号频率的构建方式, 在本 实施方式的上述步骤中 N取 127, 单音频率间隔为 156.25Hz, 随着索引的增 大, 频率逐一增加 156.25Hz, 第 64号单音实际频率为 0Hz, 因此有本式所描 述的对应关系。
下面进一步对釆样位置偏差和定时偏差的影响进行分析, 以使本实施方 式的效果更加明显, 釆样位置偏差和定时偏差会导致频域上随频率增加而增 加的线性相位旋转, 对于差分频域序列, 这种旋转转变为常数相位值, 釆用 实数最大值的方式时, 仅会略微影响投影值的幅度, 但即便在定时偏差高达 4个 chip时, 投影角度为 pi/16, 投影值也仅仅下降为原值的 0.9808, 对信噪 比的影响几乎可以忽略。
图 5仿真了釆样位置偏差和定时偏差对估计值的影响, 其中, 激励信号 为无频偏同步码,釆样位置偏差从 -4chip逐步变化到 +4chip,每次跳变 l/4chip。
考虑到 2倍 chip速率釆样的最大釆样值位置偏差为 l/4chip,与理论分析 一致, 本实施方式的估计方法除了与釆样值位置无关外, 也与定时偏差无关, 也就是说, 该方法既不依赖于系统的定时同步精度, 也不需要提供精确的同 步码径位置。
另外, 由于在频域进行了差分运算, 多径导致的频率选择性被大幅抑制, 与釆样值位置偏差和分析定时偏差影响时一致, 此处也考虑最恶劣的场景, 仿真一种随机 4径信道, 这 4条径的到达时刻为 0~3chip , 信道系数均为独立 的复高斯随机数, 每个子帧重新产生且不进行功率归一化, 当不存在实际频 偏和信道噪声的情况下, 估计结果如图 6所示。
由此可见, 即便是在完全随机的信道条件下, 多径分布对频偏估计的影 响也非常微小, 值得注意的是, 这里的估计方法没有釆用多径位置信息, 也 就是说, 本实施方式的方法并不需要进行多径位置估计就可以良好地工作。
下面以釆用本实施方式的方法为基础构建的 TD-SCDMA 自动频率控制
( AFC )环路为例, 对本方法的使用情况进行说明。
考虑到 AFC的目的是纠正本地晶振频偏, 而本地晶振频偏是一个非常緩 变的参数, 因此从提高系统稳定性的角度出发, 没有必要将当次的频偏估计 结果完全送入 AFC控制回路进行补偿,此处对估计值乘以一个较小的因子作 为实际补偿的频偏值调整量, 其典型值选取 1/256。
TotalFreqEsti = TotalFreqEsti + AdjCoef * FreqEsti , 其中, FreqEsti为当 前的频偏估计值, AdjCoef为调整补偿, TotalFreqEsti为总频偏估计值, 本式 的物理意义为, 每次的残余频偏估计值以一定的折扣计入总频偏, 以降低单 次估计误差对频偏调整的影响。
釆用典型参数 FltCoef=l/16, AdjCoef=l/256对本实施方式构建的 AFC环 路在各种场景下的性能进行仿真比较。
首先, 对在 AWGN信道下该 AFC环路残余频偏的标准偏差性能进行分 析, 请参考图 7, 图 7 中每个样点的仿真数量为一万个子帧, 可见, 基于本 方法构建的 AFC环路在 AWGN信道下可以获得较好的性能, 0dB、 5dB和 10dB信噪比时标准偏差分别优于 50Hz、 30Hz和 20Hz。
其次,仿真釆样值位置偏差和定时偏差对 AFC环路残余频偏性能的影响。 图 8中四条曲线分别对应理想定时, 1/4釆样值位置偏差, lchip定时偏差和 4chip定时偏差, 与理论分析一致, 基于本方法构建的 AFC环路与釆样值位 置偏差和定时偏差基本无关。
在此基础上进一步评估多径信道对 AFC环路性能的影响。 图 9比较了静 态等强 2径和等强 4径环境下的性能, 这些径相互间隔 lchip, 能量和相位相 同, 可见, 静态多径对性能影响微小。
此外, 再来考虑最恶劣多径的场景, 仿真前述的随机 2径信道和随机 4 径信道, 时延扩展分别为 2chip和 4chip,信道系数均为独立的复高斯随机数, 每个子帧重新产生且不进行功率归一化。 请参考图 10, 仿真结果表明在存在 密集多径且完全未知多径信息的情况下, AFC环路依然能够稳定的工作, 且 性能没有出现明显劣化。
图 11所示为本实施方式频率偏差的估计装置, 包括: 相互连接的本地频 域同步码构造模块和频率偏差估计模块, 其中:
本地频域同步码构造模块设置成构造本地频域同步码, 并将所构造的本 地频域同步码发送给频率偏差估计模块; 其中, 构造本地频域同步码的过程 包括: 对预先配置的同步码进行频率偏移操作; 完成频率偏移操作后, 消除 定时偏差及多径对预先配置的同步码的影响, 得到本地频域同步码。
频率偏差估计模块设置成: 在接收到子帧后, 将该子帧的下行同步码转 换到频域, 并将转换后的下行同步码与所接收到的本地频域同步码进行相关 运算, 得到共轭下行同步码序列, 判断共轭下行同步码序列中的最大值所在 位, 根据该最大值所在位折算出频率偏差。
上述频率偏差的估计装置的各模块的其它功能请参考方法内容的描述。 以上所述仅为本发明的优选实施例而已, 并不用于限制本发明, 对于本 领域的技术人员来说, 本发明可以有各种更改和变化。 凡在本发明的精神和 原则之内, 所作的任何修改、 等同替换、 改进等, 均应包含在本发明的保护 范围之内。
工业实用性
与现有技术相比, 本发明不需要实现精确釆样值位置、 准确定时同步, 甚至不需要获知多径分布和位置信息就可以较为稳定的工作。

Claims

权 利 要 求 书
1、 一种频率偏差的估计方法, 包括:
接收到子帧后, 将该子帧的下行同步码转换到频域, 并将转换后的下行 同步码与本地频域同步码进行相关运算, 得到共轭下行同步码序列; 以及 判断所述共轭下行同步码序列中的最大值所在位, 根据该最大值所在位 折算出频率偏差值。
2、 如权利要求 1所述的方法, 其中, 所述本地频域同步码为一组受到不 同频偏影响的同步码序列。
3、 如权利要求 2所述的方法, 还包括: 通过如下方式构造所述本地频域 同步码:
对预先配置的同步码进行频率偏移操作; 以及
完成所述频率偏移操作后, 消除定时偏差及多径对所述预先配置的同步 码的影响, 得到所述本地频域同步码。
4、 如权利要求 3所述的方法, 其中, 对预先配置的同步码进行频率偏移 操作的步骤包括:
产生不同频率的信号;
将所述预先配置的同步码中的各样点分别乘以 /y , 其中, j为复数单位, i为所述各样点在所述预先配置的同步码中所在的位数; 以及 将完成 / 运算后的同步码与所产生的信号分别相乘。
5、 如权利要求 3所述的方法, 其中, 消除定时偏差及多径对所述预先配 置的同步码的影响的步骤包括:
将所述预先配置的同步码转换至频域, 并对频域上的该同步码进行差分 运算, 得到所述本地频域同步码。
6、 如权利要求 5所述的方法, 还包括: 在将所述预先配置的同步码转换 至频域前, 对该预先配置的同步码进行补偿, 以与基带信号相适应。
7、 如权利要求 6所述的方法, 其中, 所述补偿为升余弦补偿。
8、 如权利要求 5所述的方法, 还包括: 在对频域上的该同步码进行差分 运算前, 按照各样点所对应的频率的高低对频域上的该同步码的各样点进行 重排序。
9、如权利要求 1所述的方法, 其在将该子帧的下行同步码转换到频域的 步骤前还包括:
从所接收到的子帧中查找下行同步码的位置, 按照数据窗的长度从所述 子帧中截取出下行同步码信号; 以及
若所述下行同步码信号的样点数不是所述本地频域同步码的样点数整数 倍, 则对所述下行同步码信号进行处理, 使该下行同步码信号的样点数为所 述本地频域同步码的样点数的整数倍。
10、 如权利要求 9所述的方法, 其中, 所述数据窗的长度为所述下行同 步码的切谱(chip )长度加所述下行同步码左右各 m个 chip长度, 其中, m > 0o
11、 如权利要求 9所述的方法, 还包括: 若所述下行同步码信号的样点 数为所述本地频域同步码的样点数的整数倍时, 对所述下行同步码信号进行 差分运算。
12、 如权利要求 11所述的方法, 还包括: 对所述下行同步码信号进行差 分运算后, 进行消除固定时域定时偏差影响的补偿。
13、 如权利要求 1所述的方法, 其中, 将转换后的下行同步码与本地频 域同步码进行相关运算, 得到共轭下行同步码序列的步骤包括: 将每个相关 运算结果的虚部丟弃, 保留每个相关运算结果的实部, 得到所述共轭下行同 步码序列。
14、 如权利要求 4所述的方法, 其中, 所述不同频率的信号中包括 0赫 兹频率的信号;
根据该最大值所在位折算出频率偏差值的步骤包括:
将所述共轭下行同步码序列中的最大值所在位与所述不同频率的信号中
0 赫兹频率信号所在位之间的位数差, 乘以所述信号的频率间隔, 得到所述 频率偏差值。
15、 一种频率偏差的估计装置, 包括: 相互连接的本地频域同步码构造 模块和频率偏差估计模块, 其中,
所述本地频域同步码构造模块设置成构造本地频域同步码, 并将所构造 的本地频域同步码发送给所述频率偏差估计模块;
所述频率偏差估计模块设置成: 在接收到子帧后, 将该子帧的下行同步 码转换到频域, 并将转换后的下行同步码与所接收到的本地频域同步码进行 相关运算, 得到共轭下行同步码序列, 判断所述共轭下行同步码序列中的最 大值所在位, 根据该最大值所在位折算出频率偏差。
16、 如权利要求 15所述的装置, 其中, 所述本地频域同步码构造模块是 设置成通过如下方式构造所述本地频域同步码:
对预先配置的同步码进行频率偏移操作;
完成所述频率偏移操作后, 消除定时偏差及多径对所述预先配置的同步 码的影响, 得到所述本地频域同步码。
PCT/CN2010/078830 2010-04-15 2010-11-17 一种频率偏差的估计方法及装置 WO2011127730A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US13/258,448 US8867443B2 (en) 2010-04-15 2010-11-17 Method and apparatus for estimating frequency deviation
KR1020127000919A KR101351496B1 (ko) 2010-04-15 2010-11-17 주파수 편차의 추정 방법 및 장치

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201010148719.1 2010-04-15
CN201010148719.1A CN102223322B (zh) 2010-04-15 2010-04-15 一种频率偏差的估计方法及装置

Publications (1)

Publication Number Publication Date
WO2011127730A1 true WO2011127730A1 (zh) 2011-10-20

Family

ID=44779754

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2010/078830 WO2011127730A1 (zh) 2010-04-15 2010-11-17 一种频率偏差的估计方法及装置

Country Status (4)

Country Link
US (1) US8867443B2 (zh)
KR (1) KR101351496B1 (zh)
CN (1) CN102223322B (zh)
WO (1) WO2011127730A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117319161A (zh) * 2023-11-24 2023-12-29 深圳市国电科技通信有限公司 Chirp信号的解调方法、装置、计算机设备和存储介质

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI462539B (zh) * 2012-03-06 2014-11-21 Mstar Semiconductor Inc 頻率校正方法
CN105991495B (zh) * 2015-01-26 2019-04-30 上海数字电视国家工程研究中心有限公司 物理层中的定时同步方法
CN105957324B (zh) * 2016-06-07 2019-10-01 Tcl海外电子(惠州)有限公司 遥控信号的解码方法、装置及系统
WO2018041362A1 (en) * 2016-09-02 2018-03-08 Sony Mobile Communications Inc. Downlink synchronization signals
CN106998237B (zh) * 2017-04-05 2019-02-22 大唐联诚信息系统技术有限公司 一种时频同步方法及装置
CN107743052A (zh) * 2017-09-15 2018-02-27 江西洪都航空工业集团有限责任公司 一种调制度测试方法
CN110109342B (zh) * 2018-02-01 2024-05-14 深圳市英特瑞半导体科技有限公司 一种晶振频率守时的方法及守时设备
CN110536405B (zh) * 2018-05-25 2021-08-03 华为技术有限公司 一种同步方法及接入点
CN110324097A (zh) * 2019-06-21 2019-10-11 天津理工大学 一种非接触式ais船台检测仪的载波频偏检测方法
CN110311747B (zh) * 2019-07-16 2020-09-01 京信通信系统(中国)有限公司 抗大频偏帧同步方法、装置及系统
CN113890804B (zh) * 2021-10-21 2023-09-15 成都中科微信息技术研究院有限公司 一种适用于大频偏场景下的高性能同步方法
CN114531327B (zh) * 2022-01-26 2022-11-04 小唐科技(上海)有限公司 一种频偏自跟踪装置

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005053258A2 (en) * 2003-11-28 2005-06-09 Koninklijke Philips Electronics N.V. Method and apparatus of frequency estimation for the downlink of td-scdma systems
CN1917408A (zh) * 2005-08-18 2007-02-21 中兴通讯股份有限公司 一种同步码分多址系统的小区码资源规划方法
CN101075845A (zh) * 2006-05-15 2007-11-21 大唐移动通信设备有限公司 一种小区初搜中实现下行同步的方法和装置

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7912137B2 (en) * 2006-01-11 2011-03-22 Amicus Wireless Technology Ltd. OFDMA device and method of correcting frequency offset in OFDMA signals
TW200913592A (en) * 2007-05-25 2009-03-16 Amicus Wireless Technology Ltd OFDM-based device and method for performing synchronization in the presence of interference signals

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005053258A2 (en) * 2003-11-28 2005-06-09 Koninklijke Philips Electronics N.V. Method and apparatus of frequency estimation for the downlink of td-scdma systems
CN1917408A (zh) * 2005-08-18 2007-02-21 中兴通讯股份有限公司 一种同步码分多址系统的小区码资源规划方法
CN101075845A (zh) * 2006-05-15 2007-11-21 大唐移动通信设备有限公司 一种小区初搜中实现下行同步的方法和装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117319161A (zh) * 2023-11-24 2023-12-29 深圳市国电科技通信有限公司 Chirp信号的解调方法、装置、计算机设备和存储介质
CN117319161B (zh) * 2023-11-24 2024-01-30 深圳市国电科技通信有限公司 Chirp信号的解调方法、装置、计算机设备和存储介质

Also Published As

Publication number Publication date
CN102223322A (zh) 2011-10-19
KR101351496B1 (ko) 2014-01-15
CN102223322B (zh) 2014-11-05
KR20120020202A (ko) 2012-03-07
US8867443B2 (en) 2014-10-21
US20130021966A1 (en) 2013-01-24

Similar Documents

Publication Publication Date Title
WO2011127730A1 (zh) 一种频率偏差的估计方法及装置
CN101553028B (zh) Td-scdma通信系统接收同步中基于差分相位的频偏与相位估计方法
EP2876837B1 (en) Method and device of frame synchronization of wireless system and wireless system
CN1886957A (zh) 一种用于对td-scdma系统下行链路进行频率估测的装置和方法
JP2004532587A (ja) 時間ずれ補償を備えたcdmaサーチャ
WO2007003071A1 (en) A method of uplink synchronization establishment based on the united transmission technology and the transmission matrix thereof
KR20070049832A (ko) 이동통신 시스템에서의 주파수 옵셋 추정 방법 및 그 장치
CN112039816A (zh) 一种窄带物联网系统下行同步方法
JP2009514487A (ja) 無線通信システムにおけるタイミング決定の方法および装置
CN102098239B (zh) 一种信道估计方法、装置和vamos系统
CN100556011C (zh) 自适应滤波器
CN109561042B (zh) 一种ofdm系统接收机的定时频率同步方法
CN100438399C (zh) 定时跟踪装置,接收装置,定时跟踪和调整方法
CN111555856A (zh) 一种基于多路差分加权相关的导引辅助同步方法
US7286593B1 (en) System and method for channel estimation in a radio frequency receiver
CN101630961B (zh) 频偏估计设备、方法和通信装置
WO2009155864A1 (zh) 一种下行同步跟踪方法及装置
CN107276654B (zh) 信号处理方法和系统
CN107276953B (zh) 定时同步方法、装置和系统
Manzoor et al. Novel SNR estimation technique in Wireless OFDM systems
CN107276943B (zh) 定时同步方法、装置和系统
Syed et al. Efficient cross-correlation algorithm and architecture for robust synchronization in frame-based communication systems
CN107277913B (zh) 定时同步方法、装置和系统
CN107276940B (zh) 定时同步方法、装置和系统
CN112688889B (zh) 一种无人机测控系统的信道估计方法和装置

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 13258448

Country of ref document: US

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

Ref document number: 10849739

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 20127000919

Country of ref document: KR

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 10849739

Country of ref document: EP

Kind code of ref document: A1