JP5988622B2 - Method for improving synchronization and information transmission in a communication system - Google Patents

Description

  The present invention relates to a method for synchronization and information transmission in a communication system, and more particularly to a wireless communication system, a transmitter unit and a receiver unit.

  Several different proposals for EUTRA synchronization channel (SCH) intended for use within the cell search procedure have been proposed in the 3rd Generation Partnership Project RAN1 to date. See, for example, Motorola, “Cell Search and Initial Acquisition for OFDM Downlink”, R1-051329, Seoul, Korea, Nov. 7-11, 2005 (this paper is referred to below as Document 1).

  Compared to solutions that exist within the WCDMA standard, the Motorola® proposal has made progress towards concurrent initial timing acquisition and cell identification. In this way, it is assumed that the time required for the entire cell search procedure, resulting in complete timing acquisition and cell identification results, will be reduced.

According to this proposal, the synchronization channel is unique for two concatenated identical cells, which are preceded by a cyclic prefix of L _CP samples (identical to the last L _CP samples of the OFDM waveform). It has an OFDM waveform. Such an SCH is designed to support initial timing acquisition by blind differential correlation detection in the receiver. T.A. M.M. Shimidl and D.C. C. Cox, “Robust Frequency and Timing Synchronization for OFDM”, IEEE Trans. On Communications, Vol. 45, pp. 1613-1621, Dec. See 1997 (this paper is referred to below as Document 5).

  Cell identification is performed by detecting a cell-specific OFDM waveform obtained by modulating a subcarrier with components of a prime-length cell-specific Zadoff-Chu sequence after initial timing acquisition. (The Zadoff-Chu sequence is the basis for generating a much broader family of so-called GCL sequences. B. M. Popovic, “Generalized chirp-like polyphases with optimum correlation E.T. vol.38, pp.1406-1409, July 1992. (This paper is referred to below as document 6.) The cell-specific index of the GCL sequence is the differential code of the received block of signal samples. After conversion, it may be detected by using an inverse discrete Fourier transform (IDFT).

  Although the above sync channel solution seems very promising in terms of reduced overall cell search time, the timing acquisition is nevertheless due to the wide triangular shape of the differential correlation function, Very sensitive to noise / interference.

The SCH signal from document 1 comprises a basic cell-specific OFDM waveform W (l), l = 0, 1,..., N / 2-1 repeated twice, and a synchronization signal s (k), k = It has a cyclic prefix followed by 0, 1, ..., N-1. Here, N is the number of samples in the OFDM signal acquired after IDFT in the transmitter. The timing of the SCH can be detected in the receiver by a subsequent algorithm.
Find the delay of the block of N samples that yields the maximum correlation amplitude in the received signal and select such delay as the initial timing for OFDM symbol demodulation.

  The differential correlation C (p) of the received signals r (k), k = 0, 1,..., N−1 is mathematically

It can be expressed as Here, p represents the delay of the first sample in the block of N received samples with respect to the correct position of the first sample of the synchronization signal, and “ ^* ” represents the complex conjugate. If the received signal comprises only the repetitive waveform W (k) (without a cyclic prefix), then the differential correlation of the received signal, where N is an even number,

Equal to the differential correlation function C _W (p) of the waveform W (k), which exists only for

Follow what is given by.

  The differential correlation function of the synchronization signal from document 1 with a 10-sample cyclic prefix generated with N = 128 samples of IFFT is shown in FIG.

  Equation (2) describes the wide triangular shape of the differential correlation function in FIG. The small distortion of the triangle shape comes from fluctuations in the signal envelope. In this way, it can be seen from (2) that the differential correlation depends only on the envelope of the synchronization signal, so that different synchronization signals with a stationary envelope will produce the same differential correlation. The differential correlation function in FIG. 1 achieves a plateau type with a length equal to the length of the cyclic prefix (document 5).

  Differential correlation peak detection can be done, for example, by finding the maximum value of the correlation function calculated within the (10 ms) frame of the received sample. However, there can be synchronization signals from multiple cells that can be received simultaneously in the user equipment (UE) and all of them must be detected in the cell search procedure. Therefore, the peak detection of the differential correlation within the received sample frame is not sufficient. This is because the peaks coming from different cells cannot be distinguished.

  Alternatively or additionally, some kind of threshold-based selection must be applied. For example, to find the exact arrival time of each synchronization signal, all correlation values that are greater than a certain percentage of signal energy within the corresponding correlation window are selected for further processing by peak detection. Thus, the amplitude of each differential correlation value is compared to an adaptive threshold proportional to the signal energy within the correlation window of N / 2 samples used to calculate the observed correlation value. You may do it.

  The comparison with the adaptive threshold above compares the normalized differential correlation (normalized by the received energy in the second half symbol) defined in document 5, equation (8), 0 and 1 Is equivalent to comparing with a fixed threshold value between Since timing acquisition performance is basically determined by the nature of the differential correlation, we will not discuss normalization with signal energy any more.

  If the differential correlation function has an impulsive shape with a narrow center correlation peak corresponding to zero delay and low correlation sidelobes for other delays, similar to the aperiodic autocorrelation function of a pseudorandom signal If so, much better timing acquisition properties are obtained.

Impulse differential correlation function Park et al., “A Novel Timing Estimation Method for OFDM Systems”, IEEE Communication Letters, Vol. 7, NO. 5, pp. In equation (10) of 239-241, May 2003, s (k) = [W (k) Z (k) W ^* (k) Z ^* (k)] (3)
(This paper is referred to below as document 7). Here, a waveform W (k) of N / 4 sample length is generated by IFFT of a pseudo-noise sequence, while the waveform Z (k) is designed to be symmetric with respect to W (k). . Synchronization signal (3) is (Document 7)

Detected by modified differential correlation, defined as

  Since the signal (3) is explicitly and exclusively defined as an OFDM signal, as generated by the inverse fast Fourier transform (IFFT), document 7 describes other types of signals such as discrete spectrum direct sequence signals. A centrally symmetric synchronization signal is not assumed.

  If we ignore the complex conjugate in signal (3), we are basically a repetitive signal whose basic repetition waveform of its repetition length N / 2 samples is centrosymmetric I can understand that. Such signals have an impulse-like differential correlation function, but the repetitive structure is always constant regardless of the nature of the pseudo-noise sequence used to modulate the subcarrier within the OFDM signal. The result is a highly correlated sidelobe equal to one quarter of the signal energy. Since highly correlated side lobes can cause an increased probability of timing acquisition errors, it is preferable to reduce them as much as possible.

  In addition, the shorter length (N / 2) of the basic waveform repeated in the synchronization signal (3) suggests a smaller number of different synchronization signals that can be generated. Synchronous signals should not be used not only for timing acquisition, but also for information transmission (not considered in document 7) in applications of interest, such as cell search in cellular systems, with low cross-correlation. The smaller number of different synchronization signals that can occur in turn implies a smaller amount of information that can be carried by the synchronization signal.

  Further, the complex conjugate of the basic repetitive waveform in the second half of the signal can complicate the implementation of the signal generator and demodulator, especially if the signal is assumed to be acquired by an IDFT of a complex pseudo-noise sequence.

  Also, since the centrally symmetric part (3) of the synchronization signal has two symmetrical waveforms, N / 2 is an even number. However, in some situations it is preferable to have an odd length N / 2 single centrosymmetric waveform, which can be repeated many times with a synchronization signal.

  Zhang et al., “Joint Frame Synchronization and Frequency Offset Estimate OFDM Systems”, IEEE Trans. On Broadcasting, vol. 51, no. 3, September 2005 describes joint frame synchronization and carrier frequency offset estimation structure. This paper seems to concentrate mainly on improving frequency error estimation. It is not explained how exactly the arrival time of the training symbol is estimated.

JP 2001-333441 A JP 2006-166436 A JP 2007-13982 A

Zhongshan Zhang et al., 'Frequency Offset Estimation With Fast Acquisition in OFDM System', Communications Letters, IEEE, March 2004, pp. 171-173 Motorola, 'Cell Search and Initial Acquisition for OFDM Downlink', R1-051329, 3GPP, November 7, 2005, URL, http://www.3gpp.org/ftp/tsg_ran/WG1_RL1/TSGR1_43/Docs/R1- 051329.zip

  It is an object of the present invention to propose a solution or reduction to one or more problems of the prior art. The main objective is therefore to propose a method that allows the synchronization of a communication system with reduced sensitivity to noise / interference and also allows the simultaneous transmission of information.

Thus, according to the present invention, a signal that improves synchronization and information transmission in a communication system is generated with a centrally symmetric part s (k), which is in the form of its absolute value. Symmetric and centrally symmetric part s (k) is of arbitrary length N and is based on a uniquely identifiable sequence c (l) from a set of sequences. According to the invention, the signal is transmitted on the communication channel and then received. Find the delay of a block of N samples of the received signal that yields the maximum correlation amplitude and select such delay as the initial timing for demodulation, then the only sequence c (from the set of sequences l) is detected and the transmitted information is extracted accordingly.
The centrally symmetric part s (k) can be, for example, one of the following.

The centrally symmetric part of the signal may be preceded by a cyclic prefix of L _CP samples that is identical to the last L _CP samples of the centrally symmetric part of the signal.
The centrally symmetric part s (k) of the signal is a spectrum of N subcarrier weights H (H (n) = H (N−n), n = 0, 1, 2,..., N−1. It may be generated as an OFDM signal acquired as the IDFT of n). Here, H (N) = H (0) is established by the periodicity of the DFT.
The spectrum H (n) is obtained as a Fourier coefficient of the occupied subcarrier frequency by using the components of the pseudo-random sequence c (l), l = 0, 1,..., L−1, L = N.

As may be obtained. Here, c (l), l = 0, 1,..., L−1 is an odd-length L centrally symmetric sequence.

  The method of the present invention may be implemented through a transmitter unit in a communication system and in a receiver unit. Together they form part of a wireless communication system comprising at least one transmitter unit and at least one receiver unit. The transmitter unit and the receiver unit are implemented as follows.

  The present invention is applied in applications where the synchronization signal transmitted to support and mitigate timing acquisition within the receiver must also carry some information such as the identification number of the transmitter. Substantially improving the timing acquisition performance in the receiver. One such application is a cell search procedure in a cellular system. In addition, it makes it possible to carry an increased amount of information compared to the prior art in document 1 with a synchronization signal.

  Additional features and advantages of the present invention will become apparent from the following description.

  Embodiments illustrating the present invention are described herein with reference to the accompanying drawings.

FIG. 1 illustrates a correlation function according to the prior art. FIG. 2 illustrates the inverse correlation function. FIG. 3 is a graph showing various accurate timing acquisition probabilities. FIG. 4 is a graph showing various accurate timing acquisition probabilities. FIG. 5 is a graph showing various accurate timing acquisition probabilities. FIG. 6 is a graph showing various accurate timing acquisition probabilities. FIG. 7 illustrates a wireless communication system according to an embodiment of the present invention.

  To achieve an impulsive differential correlation function, we first modify the definition of differential correlation so that as many different products as possible are included in the sum corresponding to the different delays. In this way, differential correlation values corresponding to unsynchronized delays are randomized.

  One way to achieve a differential correlation value that is not randomly synchronized is to reverse the order of the samples in one of the blocks of samples used in (1). We have the so-called inverted differential correlation D (p)

Define as
Where p represents the delay of the first sample in the block of N received samples with respect to the correct position of the first sample of the synchronization signal, and

Represents the ceiling function of x, ie the smallest integer greater than or equal to x.

In order to obtain the maximum possible correlation value (5) at p = 0, which is equal to the signal energy within the correlation window of samples, the synchronization signal s (k), k = 0, 1,. Is centrally symmetric, ie

Should be. Where N is any positive integer and we have the signal energy

Assume that it is evenly distributed between the first and second blocks of samples.

From (5) and (6), the inverted differential correlation D _s (p) of the synchronization signal s (k) is

Exist only against

Follow what is given by.

  Equation (7) is

Very similar to the aperiodic autocorrelation function R (p) of the synchronization signal s (k), defined as

As will be appreciated, the only difference between D _s (p) and R (p) is in the reduced number of total components. In this way, if s (k) comprises an impulsive aperiodic autocorrelation function, its inverted differential correlation function has a very good chance of becoming impulsive as well.

  Equation (7) shows that, in general, a pseudorandom signal that is non-repetitive but centrally symmetric produces a lower correlated sidelobe than the repetitive signal.

  In an alternative to the centrosymmetric synchronization signal defined by (6),

In this case, the inverted differential correlation is

Needs to be redefined.

  The prior art, OFDM synchronization signal (3) proposed in equation (10) of document 7, can be considered as a special case of signal (9). (3) is defined only for N mod 4 = 0, while (9) is defined for any length N, so that (9) is more general pay attention to.

  The same maximum absolute value of the inverted differential correlation is that if the signal is distortion symmetric, i.e.

Can be obtained.

  Similarly, the absolute value of (10) is

If defined as, it will not change.

Embodiment 1
To illustrate the design of the centrosymmetric synchronization signal (6) and the nature of the corresponding inverted differential correlation function (5), we start with the assumptions given in document 1 Is generated. The sampling frequency is 1.92 MHz, the subcarrier separation is 15 kHz, and the maximum number of occupied subcarriers is Nosc = 76 among all N = 128 subcarriers within the 1.92 MHz frequency band ( The transmission bandwidth is 1.25 MHz). The occupied subcarrier is modulated by a component of a pseudorandom sequence from the set of sequences with good cross-correlation properties. Different sequences from the set are labeled with different cell identification numbers (IDs). After DFT demodulation of the received OFDM signal, the transmitted sequence may be identified by inverse mapping from the subcarrier followed by specific signal processing. Low cross-correlation between sequences contributes to more reliable identification of the sequence when multiple signals are received from different cells simultaneously.

  A sample output OFDM synchronization signal s (k) of length N = 128 is

Is obtained by IDFT of spectrum H (n) of N = 128 Fourier coefficients.

If H (n) = H (N−n), n = 0, 1, 2,..., N−1 and if H (N) = H (0) holds due to the periodicity of the DFT, s (K) is also symmetric around its s (N / 2) samples, ie
Only when H (n) = H (N−n), s (k) = s (N−k), k, n = 1,..., N−1 (12)
It can be shown that. The proof of (12) continues.

  s (k)

Starting with the definition

Follow. Here we introduced the variable transformation n = N−1, reordered the sums and used the DFT periodicity (H (n) = H (n + N)). From (A.1) and (A.2), a sufficient condition, if H (n) = H (N−n), then S (k) = s (N−k) follows. As it can be shown by starting from the representation of H (n), it is s (k) = s (N−k) only when H (n) = H (N−n). It is also a necessary condition.

  The spectrum H (n) is acquired by using the components of the pseudo-random sequence c (l), l = 0, 1,..., L−1, L = Nosc as Fourier coefficients at the occupied subcarrier frequency. Also good.

  Where c (l), l = 0,1,..., L-1 is an odd length L centrally symmetric sequence and if we do a mapping between c (l) and H (n)

It is obvious that the condition in (12) is satisfied.

  Thus, the resulting synchronization signal s (k), k = 0, 1, 2,..., N−1 means that only sample s (0) does not have its symmetric pair with respect to s (N / 2). It is a low-pass baseband OFDM signal that is symmetric around its s (N / 2) samples. In other words, the resulting OFDM synchronization signal can be considered as comprising two parts. The first part comprises one sample and the second part is N−1, where s (k) = s (N−k), k = 1, 2,..., N−1. Centrosymmetric samples.

  For blind detection of the above OFDM signal, we use a block of N-1 input signal samples and

Further means that an inverted differential correlation needs to be performed.

  However, once accurate timing has been obtained, a block of N total samples should be used for OFDM demodulation and subsequent identification of information content (cell ID).

  The remaining question is what kind of odd-length L centrosymmetric sequence is selected for subcarrier modulation. L−1 pseudo-noise sequences {ar (l)}, r = 1,..., L−1, used in document 1 to generate a repetitive OFDM synchronization signal with L as a prime number,

Is an odd length L Zadoff-Chu (ZC) sequence defined as here,

It is.

  If L is odd, the ZC sequence (15) is centrosymmetric (around its (L-1) / 2 + 1th component), ie ar (l) = ar (L-1-l) , L = 0, 1,..., L−1. In order to make the sequence length equal to or less than the maximum number of occupied subcarriers, we determine that the first and last specific number of sequence components of the ZC sequence are symmetric about the resulting shortened sequence. You can throw it away to stay in

  Since the maximum possible number of occupied subcarriers is Nosc = 76 and the length of the ZC sequence must be a prime number, in order to generate a prototype ZC sequence, in (15), L = 79, and then discarding the first two and last two components of the prototype ZC sequence so that the resulting shortened ZC sequence remains centrosymmetric, the length L = 75 Shortened to The shortened sequence is then used in (13) to generate an OFDM synchronization signal (11) after the IDFT of H (n).

  By choosing different values of r in (15) we can obtain up to M = L−1 = 74 different OFDM synchronization signals, each carrying different information about the cell ID. The number of cell IDs is almost twice as large as the number of cell IDs (41) in the document 1 with respect to the synchronization signal of the same size. At the same time, the principle of ZC sequence detection by using differential encoding and IDFT from document 1 may also be applied within this embodiment.

In order to ensure demodulation robustness in the case of multipath propagation channels, the OFDM synchronization signal is guided by a cyclic prefix. Inverse differential correlation function of OFDM synchronization signal (11) obtained from a shortened ZC sequence of length L = 75 with a cyclic prefix consisting of cell ID = r = 29 and L _CP = 10 samples Is represented in FIG.

  The cyclic prefix makes the inverted differential correlation function asymmetric with a slightly increased sidelobe level for negative delay. However, since the sidelobe levels are still relatively low compared to the main peak, the probability of timing acquisition errors is not expected to be affected by them.

The Zadoff-Chu sequence is [6]
c (l) = a (l) b (l mod m), l = 0, 1,..., L-1, (16)
Is the basis for the generation of the GCL sequence (c (l)). Where L = sm ² , s and m are positive integers, {b (l)} is an arbitrary sequence of m complex numbers of unit amplitude, and {a (l) } Is a length L Zadoff-Chu sequence. Therefore, in order to obtain a centrally symmetric GCL sequence, L must be an odd number and the modulation sequence {b (l)} must be centrally symmetric. Centro-symmetric GCL sequences, if used within the present invention, have the potential for more information transmission because of their larger number. Besides, they maintain optimal correlation properties apart from the selection of their modulation sequences.

Timing Acquisition Performance Within the cellular system user equipment (UE), the initial frequency error (immediately after power-on) of the RF signal may be on the order of tens of thousands of Hz. Once the receiver is fixed to the signal received from the base base station, this frequency error is reduced in the range of several hundred Hz. The UE is fixed to the base base station after the initial cell search, and the task is performed after it is switched on by the UE. Once the UE has found its “camping” cell, the cell search procedure can be handed over if the UE is in active mode, or if the UE is in idle mode (for better signal reception). For any of the possible cell reselections, it enters a monitoring mode in which it monitors available neighboring cells. In the monitoring mode, the frequency error between the received signal and the UE RF signal is substantially reduced. This is because all cells are strongly frequency synchronized and the UE is already synchronized to one of them.

  In this way, during the initial cell search, it should be possible to detect the arrival time of the synchronization signal transmitted from the base base station under a relatively high frequency error in the receiver.

  The timing acquisition performance of the synchronization signal from embodiment 1 is evaluated by simulation using the probability of accurate timing acquisition as a function of the signal-to-noise ratio (SNR) on the additive white Gaussian noise (AWGN) channel. . The four values of the initial frequency error df between the UE and the base base station are simulated at df = 0, 1, 2 and 3 ppm at a 2.6 GHz carrier frequency. The cyclic prefix is 10 samples long in all cases.

  Timing estimation is considered accurate if the estimated arrival time is within an error tolerance that precedes the correct timing position so that it overlaps the cyclic prefix of the OFDM signal. The size of the error tolerance must not be longer than the length of the cyclic prefix and must be equal to the portion of the cyclic prefix that was not previously covered by the channel response of the OFDM symbol. Since the length of the cyclic prefix should not be much longer than the maximum expected channel response, the actual error tolerance can be longer than a few samples. Absent. However, since the repetitive synchronization signal from document 1 is evaluated as a basis for comparison, we will make the error tolerance equal to the cyclic prefix to obtain the best performance for the signal from document 1. Take to be.

  Since the amplitude of the differential correlation does not depend on the frequency error, it can easily be seen that the signal from document 1 is evaluated without frequency error. The results are shown in FIG.

  If there is no initial frequency error, the centrally symmetric signal detected by the inverted differential correlation is more than 1 dB at a probability 0.5 of accurately acquiring the repetitive signal detected by the differential correlation and the probability of accurately acquiring it. 0.9 is better than 5 dB.

  While the performance of the centrosymmetric signal deteriorates with increasing frequency error, the performance of the repetitive signal remains unchanged for non-zero frequency errors. At a frequency error of 1 ppm (2600 Hz), the relative performance is almost unchanged. The repetitive signal is better at a very low signal-to-noise ratio, but at a frequency error of 2 ppm, the centrosymmetric signal is better with a probability of getting accurately higher than 0.5. However, with a 3 ppm frequency error, the centrally symmetric signal fails to acquire timing synchronization regardless of the signal to noise ratio. This is because some of the side lobes of the inverted differential correlation are larger than the main lobe even in the absence of noise.

Embodiment 2
The timing acquisition performance result for the signal from embodiment 1 shows that if the frequency error is above a certain threshold, the differential correlation produces a better timing acquisition than the inverted differential correlation, while Below the frequency error, the example illustrates that it is the opposite.

  This result suggests that if the frequency error during the initial cell search is above 2 ppm, it is beneficial that the synchronization signal is both centrosymmetric and periodic. Such a signal may differ depending on the UE cell search mode, i.e., depending on the maximum frequency error expected between the carrier frequency of the received signal and the frequency of the reference RF signal in the receiver. It can be detected in the UE by both dynamic correlation and inverted differential correlation.

  In this way, the initial cell search in which a synchronization signal is transmitted from the base station should be performed using differential correlation. Once the cell search enters the monitoring mode, the synchronization signal can be detected by inverting differential correlation, which provides much better timing acquisition performance if the frequency error is low and fast detection of neighboring cells Enable. It should be noted that within the cell search monitoring mode, rapid detection of neighboring cells with better signal quality reduces transmission in the system as it allows transmission with low power of the UE.

  Assuming the same conditions as in Embodiment 1, a set of centrally symmetric and periodic OFDM synchronization signals is obtained by using mapping (13) and IDFT (11), and 36 ZC sequences with a prime length L = 37. Can be generated from a set of Here, N = 64. The signal of 64 samples in length obtained by (11) is then periodically extended, i.e., finally repeated to produce a centrosymmetric and periodic sync signal of length 128 samples. Is done. As in the previous example, in the resulting length N = 128 samples of signal s (k), only sample s (0) does not have its symmetric pair with respect to s (N / 2).

  Identical signal (11) and subsequent general mapping

May be obtained directly (without periodic extension). Here, c (l), l = 0, 1,..., L−1 is an odd length L centrally symmetric sequence, and R = 2 is the number of iterations, ie, a specific basic waveform inside the signal. In addition to the period, N = 128 is the IFFT size. In general, mapping (17) produces a centrosymmetric signal with period R if N mod R = 0.

  The timing acquisition performance of the synchronization signal is evaluated by simulation using the probability of accurate timing acquisition as a function of the signal-to-noise ratio (SNR) on the additive white Gaussian noise (AWGN) channel. The four values of the initial frequency error df between the UE and the base base station are simulated with a carrier frequency of 2.6 GHz and df = 0, 1, 2, and 3 ppm. In all cases, the cyclic prefix is 10 samples long. The result is shown in FIG.

  3 and 4, it can be seen that the inverted differential correlation of the centrally symmetric and periodic OFDM signals is more robust to a 3 ppm frequency error than the inverted differential correlation function of the aperiodic OFDM signal. Starting with the similarity between Eqs. (7) and (8), the description of the different timing acquisition performances of FIGS. It can be derived from the nature of the periodic autocorrelation function. This function is a two-dimensional function of delay and frequency error.

  A chirp-type signal, such as the non-repetitive signal from FIG. 3, is a ridge-type ambiguity that is distinguished by a non-zero delay position whose main lob is shifted with a high frequency error. It is known to have a tee function. This effect is the main reason for the collapse of the inverted differential correlation with a frequency error of 3 ppm. Signals with several other cell IDs are somewhat less sensitive to this effect and can converge to an acquisition probability equal to 1 with a higher signal-to-noise ratio, , Collapse at somewhat higher frequency errors.

  On the other hand, periodic signals, such as one from FIG. 4, are distinguished by so-called bed-of- needles that are distinguished by the placement of fairly high side lobes at regular intervals in the time-frequency plane. Nails) ambiguity function, but the position of the main lobe corresponding to zero delay is invariant with respect to frequency. Basically, these signals behave as having substantially shorter lengths that result in less distortion with high frequency errors. On the other hand, even if there is no frequency error, a signal having two periods of the same basic waveform has a high side of inverted differential correlation so that it has an inverted differential correlation sidelobe equal to at least half of the main lobe. The lobe is derived from the repetitive nature of the signal. This results in a loss of acquisition performance for low frequency errors (below 2 ppm), as can be appreciated by comparing FIGS. 3 and 4.

Embodiment 3
As indicated above, the actual error tolerance cannot be more than a few samples. However, in that case, as can be seen in FIG. 5, even though the differential correlation (used to detect the repetitive synchronization signal from document 1) shows a rather bad performance at high frequency errors. The timing acquisition performance of the signal from FIG. 3 is evaluated with a tolerance of two samples.

  The reason for the poor performance of differential correlation lies in the plateau shown in FIG. 1, which makes it very likely that noise will generate a correlation peak with a delay inside the plateau that is less than (exact) zero delay. . In this way, the curve corresponding to the repetitive signal converges to the value 1 very slowly as the signal to noise ratio increases.

  Previous discussions on different types of ambiguity functions lead to consideration of other types of pseudo-noise sequences with ambiguity functions that are more tolerant of frequency errors. Such a pseudo-noise sequence is, for example, a set of orthogonal Golay (binary) complementary sequences. M.M. J. et al. E. Golay, “Complementary Series”, IRE Transactions on Information Theory, Vol. IT-7, pp. 82-87, Apr. See 1961 (this paper is referred to below as Document 8). A set of complementary Golay sequences exists for even sequence length L and is distinguished by the property that the sum of the non-periodic autocorrelation functions of the sequence is equal to zero for all non-zero delays. The A set of length L orthogonal Golay sequences can be obtained by bitwise multiplication of a single Golay complement sequence of length L and all L Walsh sequences of length L (document 8). Sequences within such a set can be grouped into L / 2 different complement pairs.

  If the bits of the Golay sequence from the set of orthogonal Golay complement pairs are used as Fourier coefficients H (n) in (11), the resulting OFDM synchronization signal s (k) is similar to (9) As well as nature

It comprises.

  Such a signal is represented by a modified inverted differential correlation (10).

Can be detected.

  It can be readily seen that the amplitudes of the inverted differential correlations (10) and (19) remain unchanged under any frequency error in the signal received on the single path propagation channel. This is a general property that is valid for arbitrary signals (9), (9.1) and (18).

  If the component of the Golay sequence c (k) is, for example,

As such, the resulting OFDM signal has a peak-to-average power ratio of less than 3 dB if mapped as Fourier coefficients of equidistant continuous subcarriers. B. M.M. Popovic, “Synthesis of Power Efficient Multisignals with Flat Amplitude Spectrum”, IEEE Transactions on Communications, Vol. 39, no. 7, pp. See 1031-1033, July 1991. That is, all OFDM synchronization signals based on different Golay sequences from a set of orthogonal complement pairs will in that way maximize the average transmission power, i.e. the signal-to-noise ratio received at the cell edge. It further means having a small PAPR value that allows.

  The timing acquisition performance of the OFDM signal acquired from the Golay complement sequence of length L = 64 mapped to the OFDM signal of length N = 128 according to (20) and (11) is shown in FIG. It is understood that the timing acquisition performance of the OFDM signal acquired from the Golay complement sequence does not change with increasing frequency error.

  Information embedding in this scenario can be achieved, for example, by labeling each of the orthogonal Golay sequences in the current set. After receiving the signal and demodulating the data from the OFDM signal, a particular sequence can be identified by correlating with all the sequences in the current set. Such a bank of correlators can be effectively implemented, for example, by using a fast Hadamard transform. Differential coding may be applied to the demodulated sequence before correlation to remove channel distortion. In this case, the reference sequence used for correlation should also be differentially encoded.

  Referring now to FIG. 7, the present invention also constitutes a wireless communication system that may comprise, for example, a base base station 120 of the cellular system 100 and a terminal 130 that communicates with the base base station. The base base station and / or the terminal comprises at least one transmitter unit with means for generating and transmitting a signal with a centrally symmetric part s (k), where the centrally symmetric part s (k) is optional Is the length N. The base base station and / or the terminal also comprises at least one receiver unit comprising means for receiving and processing signals generated by the transmitter unit.

Applications and alternatives The present invention is applicable to all applications where a synchronization signal is transmitted to support and mitigate timing acquisition in the receiver, and also the signal is such as a transmitter identification number, etc. It can be used when carrying some information. One such application is a cell search procedure in a cellular system.

  The proposed centrosymmetric synchronization signal may be of the OFDM type, which provides certain benefits for the demodulation of information from signals that have passed through a multipath (time dispersive) propagation channel.

  However, centrosymmetric synchronization signals such as other types of noise detected by inverted differential correlation, such as direct sequence discrete spectrum signals, can also be deployed with similar timing acquisition performance.

100 wireless communication system 120 transmitter unit 130 receiver unit

Claims (15)

Generating a signal having a time- centric symmetry portion s (k) that is available for synchronization;
Transmitting the signal on a communication channel;
Comprising
The signal is based on a sequence c (l) that is uniquely identifiable from a set of sequences that are available for information transmission;
The centrally symmetric part s (k) is centrally symmetric in the shape of its absolute value, and the centrally symmetric part s (k) has an arbitrary length N;
The signal is generated such that s (k) is obtained as an IDFT of a spectrum H (n) of N subcarrier weights, where the spectrum H (n) is a sequence c (l), l = 0, 1,..., L−1, by using components of L ≦ N, generated as Fourier coefficients of the subcarrier frequency occupied by the signal,
The components of the sequence c (l) are used to modulate the subcarriers occupied by the signal;
The sequence c (l) is

A method of synchronization within a communication system , where c (l) is an odd-length L centrosymmetric sequence, mapped onto the spectrum H (n) as The centrally symmetric part s (k) is

The method of claim 1, wherein the signal is generated to be one of: 3. The signal of claim 1 or 2 , wherein s (k) generates the signal to be guided by a cyclic prefix consisting of L _CP samples that are identical to the last L _CP samples of s (k). The method described. The sequence c (l) is

Zadoff-Chu sequence defined as follows, where

The method according to any one of claims 1 to 3 , wherein: The sequence c (l) is
c (l) = a (l) b (l mod m), l = 0, 1,..., L−1
Is a generalized chirped sequence defined as
Here, L = sm ² is an odd number, s and m are positive integers, {b (l)} is a centrally symmetric sequence having m complex numbers of unit amplitude, and {a ( 4. The method according to any one of claims 1 to 3 , wherein l)} is a length L Zadoff-Chu sequence. The method according to any one of claims 1 to 3 , wherein the sequence c (l) is a Golay complement sequence having an even length L selected from a set of orthogonal Golay complement pairs. Receiving the signal;
Finding a delay of a block of N samples of the received signal that results in a maximum amplitude of correlation;
Selecting such delay as the delay of the first sample in a block of N samples relative to the correct position of the first sample of the signal;
Detecting a unique sequence c (l) from the set of sequences and thereby extracting transmission information;
Further comprising
The correlation of the N samples r (k), k = 0,1 , ..., claim 1 is inverted differential correlation D blocks of N-1 (p) 6 according to any one of Method. The inverted differential correlation D (p) is

here,

The method according to claim 7 , which represents a ceiling function of x, ie the smallest integer greater than or equal to x. The sequence c (l), mapped to the identification information of the transmitter of the signal s (k), and / or the sequence c (l) is according to any one of claims 1 8 is a binary the method of. Different sequences from the set A method according to any one of claims 1 8 to be labeled by the different cell identification numbers. Generating a signal with a time- centric symmetry part s (k) that is available for synchronization;
Comprising a circuit configured to transmit the signal over a communication channel;
The signal is based on a sequence c (l) that is uniquely identifiable from a set of sequences that are available for information transmission;
The centrally symmetric part s (k) is centrally symmetric in the shape of its absolute value, and the centrally symmetric part s (k) has an arbitrary length N;
The signal is generated such that s (k) is obtained as an IDFT of a spectrum H (n) of N subcarrier weights, where the spectrum H (n) is a sequence c (l), l = 0, 1,..., L−1, by using components of L ≦ N, generated as Fourier coefficients of the subcarrier frequency occupied by the signal,
The components of the sequence c (l) are used to modulate the subcarriers occupied by the signal;
The sequence c (l) is

A transmitter unit (120) used in the communication system that maps onto the spectrum H (n) as follows, where c (l) is a centrosymmetric sequence of odd length L. Receiving a signal having a time- centrosymmetric part s (k) that is available for synchronization;
Find a delay of a block of N samples of the signal that yields the maximum amplitude of correlation;
Selecting such a delay as the delay of the first sample in a block of N samples relative to the correct position of the first sample of the signal;
Comprising a circuit configured to detect a unique sequence c (l) from a set of sequences;
The correlation is an inverted differential correlation D (p) of a block of N samples r (k), k = 0, 1,..., N−1, and the signal can be used for information transmission. Based on a sequence c (l) uniquely identifiable from a set of sequences, the centrally symmetric part s (k) is centrally symmetric in the form of its absolute value, and the centrally symmetric part s (k) is Any length N,
The signal is generated such that s (k) is obtained as an IDFT of a spectrum H (n) of N subcarrier weights, where the spectrum H (n) is a sequence c (l), l = 0, 1,..., L−1, by using components of L ≦ N, generated as Fourier coefficients of the subcarrier frequency occupied by the signal,
The components of the sequence c (l) are used to modulate the subcarriers occupied by the signal;
The sequence c (l) is

The receiver unit (130) used in the communication system , where c (l) is an odd-length L centrosymmetric sequence, mapping onto the spectrum H (n) as follows . Receiving and processing by a receiver a signal having a centrosymmetric portion s (k) that is available for synchronization, said signal being unique from a set of sequences available for information transmission; The centrally symmetric part s (k) is centrally symmetric in the form of its absolute value, and the centrally symmetric part s (k) has an arbitrary length N And
The signal is generated such that s (k) is obtained as an IDFT of a spectrum H (n) of N subcarrier weights, where the spectrum H (n) is a sequence c (l), l = 0, 1,..., L−1, by using components of L ≦ N, generated as Fourier coefficients of the subcarrier frequency occupied by the signal,
The components of the sequence c (l) are used to modulate the subcarriers occupied by the signal;
The sequence c (l) is

A method of synchronization within a communication system , where c (l) is an odd-length L centrosymmetric sequence, mapped onto the spectrum H (n) as Comprising a circuit configured to receive and process a signal having a centrosymmetric portion s (k) that is available for synchronization, said signal from a set of sequences that are available for information transmission Based on a uniquely identifiable sequence c (l), the centrally symmetric part s (k) is centrally symmetric in the form of its absolute value, and the centrally symmetric part s (k) has an arbitrary length. N,
The signal is generated such that s (k) is obtained as an IDFT of a spectrum H (n) of N subcarrier weights, where the spectrum H (n) is a sequence c (l), l = 0, 1,..., L−1, by using components of L ≦ N, generated as Fourier coefficients of the subcarrier frequency occupied by the signal,
The components of the sequence c (l) are used to modulate the subcarriers occupied by the signal;
The sequence c (l) is

A receiver used in a communication system that maps onto the spectrum H (n) as follows, where c (l) is a centrosymmetric sequence of odd length L. At least one transmitter unit according to claim 11 and (120), a wireless communication system including at least one receiver as claimed in claim 14 (100).

Images