CN101945078B - Frequency deviation estimation method based on noise subspace - Google Patents

Abstract

The invention discloses a frequency deviation estimation method based on the noise subspace, which comprises the steps of: calculating a phase angle rotational component caused by the frequency deviation according to a received time domain signal; averaging and normalizing the phase angle rotational component to obtain a corresponding fine frequency deviation estimation value; carrying out fine frequency deviation correction on the sequence of the received time domain signal; compensating the corrected time domain signal with an Epsilon phase; carrying out N-point Discrete Fourier Transform on the compensated signal; removing symbols modulated on a subcarrier; carrying out M-point Inverse Discrete Fourier Transform to obtain a time domain signal with the length M; computing the sum of the modulus squares of G values at the head of the time domain signal with the length M and M to G values at the end of the time domain signal with the length M and calculating the minimum ratio of the modulus squares to provide the estimated value of a coarse frequency deviation; and adding the estimated coarse frequency deviation value with the fine frequency deviation value to obtain the total frequency deviation estimation value. The invention is suitable for various block transmitting systems, including OFDM, with cyclic prefixes, and has the advantages of speediness, reliability, light load, wide capture range, high estimation precision, low complexity for implementation and like.

Description

Frequency deviation estimating method based on noise subspace

Technical field

The present invention relates to a kind of frequency synchronization method that is applied to OFDM (OFDM) system that belongs to mobile communication simultaneous techniques field, be specifically related to a kind of fast and reliable, load is smaller, capture range is large, estimated accuracy is high, implementation complexity the is low frequency deviation estimating method based on noise subspace.

Background technology

Frequency Synchronization is the prerequisite of mobile communication system energy proper communication.In order to support high-speed data service, future mobile communication system will be the system of broadband, many (sending and receiving) antenna, and OFDM is the important candidate scheme of future mobile communication system.For following mobile radio telecommunications, the difference of transmitting-receiving two-end crystal oscillator exerts an influence to carrier frequency, and it is offset, thereby destroys the orthogonality between the subcarrier in the ofdm system.Compare with single-carrier system, ofdm system is more responsive to carrier wave frequency deviation, how to reduce between subcarrier and to disturb ICI on the impact of systematic function, is one of the prerequisite that can be used widely of ofdm system.Traditional frequency synchronization method all or based on frequency domain training sequence or based on time-domain training sequence comes carrier wave frequency deviation is estimated, they have such-and-such shortcoming: be not suitable for packet data transmission, load too high, capture range is little, estimated performance is undesirable, computation complexity is high.

Summary of the invention

Goal of the invention: in order to overcome the deficiencies in the prior art, the invention provides a kind of fast and reliable, load is smaller, capture range is large, estimated accuracy is high, implementation complexity the is low frequency deviation estimating method based on noise subspace.

Technical scheme: to achieve these goals, the frequency deviation estimating method based on noise subspace of the present invention may further comprise the steps:

1), the time-domain signal to receiving, according to its P that comprises identical subsequence, calculate the phase angle rotational component that causes owing to frequency deviation;

2), above-mentioned phase angle rotational component is done mean deviation normalization and can obtain corresponding thin frequency deviation estimated value;

3), the time-domain signal sequence that receives is made thin frequency offset correction;

4), the time-domain signal after proofreading and correct being carried out phase place is

Compensation;

5), the signal after the compensation is the discrete Fourier transform (DFT) DFT that N is ordered;

6), remove the symbol of modulating on the subcarrier;

7), be the inverse discrete fourier transform IDFT that M is ordered, obtain the time-domain signal that length is M;

8), with the mould quadratic sum of M-G of this time-domain signal afterbody value divided by stem G mould quadratic sum that is worth, obtain ratio ρ ^ε

9), corresponding minimum ρ ^εThe ε of value has provided the estimated value of thick frequency deviation;

10), the thick frequency deviation value that will estimate and thin frequency deviation value addition, obtain total frequency deviation estimated value;

Wherein, ε is the thick frequency deviation value chosen in advance, for-N/2 between the N/2 and do not comprise-integer of N/2; N is the sub-carrier number of ofdm system; P is subsequence number identical in the training symbol, and M is the sub-carrier number that training sequence takies; G is the length of Cyclic Prefix; Usually, G is greater than the maximum delay expansion of wireless multipath fading channel, and P=N/M.

As preferably, described step 4) in the time-domain signal after proofreading and correct being carried out phase place is that the concrete methods of realizing of the compensation of ε is: i sampling to time-domain signal be multiply by

Beneficial effect: the frequency deviation estimating method based on noise subspace of the present invention, can be for the various block transmission systems that Cyclic Prefix is arranged that comprise OFDM, the method compared with prior art has the following advantages:

1) by introducing the concept of noise subspace, can enlarge the estimation range of thick frequency deviation, improve the precision of estimating, and estimate simultaneously the impulse response of channel; 2) Fourier just/inverse transformation exists efficient implementation structure and algorithm fast, this method by adopt a large amount of Fouriers just/inverse transformation, can the mitigation system implementation complexity.

Description of drawings

Fig. 1 is the block diagram of the frequency deviation estimating method based on noise subspace of the present invention;

Fig. 2 is the implement device structural representation of the frequency deviation estimating method based on noise subspace of the present invention.

Embodiment

Below in conjunction with accompanying drawing the present invention is further described.

As shown in Figure 2, the implement device of the frequency deviation estimating method based on noise subspace of the present invention comprises multiplication adding up device, multiplier, phase calculation device, fast fourier transformation apparatus, phase compensation and demodulating equipment, inverse fast fourier transform device, squaring device, indoor minimum value searcher and adder.

Suppose that the number of sub carrier wave that an ofdm system comprises is N, the length of Cyclic Prefix is G.Training symbol is comprised of two sections identical subsequences, takies the even bit subcarrier at frequency domain.At this moment, M=N/2, P=2.The training sequence that receives is delivered to parallel multiplication, multiplier module, phase calculation module successively, utilize the structure of subsequence, obtain corresponding thin frequency deviation (fractional part of frequency offset) estimated value, and training sequence is carried out phasing.(N/2, N/2] select successively in the scope integer ε as the preset value of thick frequency deviation (integer frequency offset), again compensate proofreading and correct rear signal, and estimate the time domain impulse response of channel.The ratio of calculating noise subspace and signal subspace power is determined the estimated value of integer frequency offset by the ratio of minimum.At last, with thick, thin frequency deviation estimated value addition, export total frequency deviation estimated value.

Specific algorithm is described below:

To receive at first that signal input multiplication is cumulative, multiplication and phase calculation device estimate fractional part of frequency offset, namely

${\widehat{\mathrm{\υ}}}_f=\frac{1}{\mathrm{\π}}\mathrm{angle}\left\{\frac{1}{M}\underset{n=0}{\overset{M-1}{\mathrm{\Σ}}}{r}^{*}\left(n\right)r(n+M)\right\}$ [formula one]

Wherein, signal r (n), and n=0,1 ..., the training sequence behind the removal Cyclic Prefix that N} represents to receive; Angle{} represents to get the phase angle operation.Then, utilize the thin frequency deviation of estimating to proofread and correct to received signal

$\hat{r}\left(n\right)=r\left(n\right)e^{-j\frac{2\mathrm{\π}{\widehat{\mathrm{\υ}}}_f(G+n)}{N}},$ N=0,1 ..., N-1 [formula two]

(N/2, N/2] select in the scope integer ε as the preset value of integer frequency offset, the time-domain signal after proofreading and correct is carried out phase compensation:

[formula three]

Burst after the phase compensation be multiply by the DFT matrixing to frequency domain, then get rid of modulation symbol after, multiply by again the IDFT matrixing to time domain.

When ε was even number, this process can be expressed as:

${\hat{h}}^{\mathrm{\ϵ}}={[{\hat{h}}_0^{\mathrm{\ϵ}},...,{\hat{h}}_{M-1}^{\mathrm{\ϵ}}]}^T=F_1^{-1}{X}^{-1}{F}_2{\stackrel{\‾}{r}}^{\mathrm{\ϵ}}$ [formula four]

Here, F ₂Be the matrix of a M * N, the element of the capable q row of its p is:

${\left[F_2\right]}_{p,q}=e^{-j\frac{4\mathrm{\π}(p-1)(q-1)}{N}}$ [formula five]

F ₁Be the matrix of a M * M, the element of the capable q row of its p is:

${\left[F_1\right]}_{p,q}=e^{-j\frac{2\mathrm{\π\π}(p-1)(q-1)}{M}}$ [formula six]

X is the diagonal matrix of a M * M, and its diagonal element is the known training sequence of receiving terminal.

When ε was odd number, this process can be expressed as:

${\hat{h}}^{\mathrm{\ϵ}}={[{\hat{h}}_0^{\mathrm{\ϵ}},...,{\hat{h}}_{M-1}^{\mathrm{\ϵ}}]}^T=F_1^{-1}{X}^{-1}{F}_3{\stackrel{\‾}{r}}^{\mathrm{\ϵ}}$ [formula seven]

Here, F ₃Be the matrix of a M * N, the element of the capable q row of its p is:

${\left[F_3\right]}_{p,q}=e^{-j\frac{2\mathrm{\π}[2(p-1)-1](q-1)}{N}}$ [formula eight]

With the sequence that obtains

Be sent to quadratic sum adder unit ratio calculated

${\mathrm{\ρ}}^{\mathrm{\ϵ}}=\frac{\underset{i=G}{\overset{M-1}{\mathrm{\Σ}}}{\left|{\hat{h}}_i^{\mathrm{\ϵ}}\right|}^2}{\underset{i=0}{\overset{G-1}{\mathrm{\Σ}}}{\left|{\hat{h}}_i^{\mathrm{\ϵ}}\right|}^2}$ [formula nine]

In the span of ε, by search unit, seek minimum ρ ^ε

${\widehat{\mathrm{\υ}}}_i=\arg \underset{\mathrm{\ϵ}\∈(-N/2,N/2]}{\min }{\mathrm{\ρ}}_{\mathrm{\ϵ}}$ [formula ten]

Minimum ρ _εCorresponding phase place ε is the thick frequency deviation of estimation

Thin, the thick frequency deviation estimated value that estimates is delivered to adder, and can to obtain total frequency deviation estimated value as follows:

$\widehat{\mathrm{\υ}}={\widehat{\mathrm{\υ}}}_i+{\widehat{\mathrm{\υ}}}_f$ [formula 11]

According to above description, can obtain the performing step following (as shown in Figure 1) based on the frequency deviation estimating method of noise subspace:

1), the time-domain signal to receiving, according to its P that comprises identical subsequence, calculate the phase angle rotational component that causes owing to frequency deviation;

2), above-mentioned phase angle rotational component is done mean deviation normalization and can obtain corresponding thin frequency deviation estimated value;

3), the time-domain signal sequence that receives is made thin frequency offset correction;

4), the time-domain signal after proofreading and correct being carried out phase place is Compensation;

5), the signal after the compensation is done discrete Fourier transform (DFT) (DFT) conversion that N is ordered;

6), remove the symbol of modulating on the subcarrier;

7), do inverse discrete fourier transform (IDFT) conversion that M is ordered, obtain the time-domain signal that length is M;

8), with the mould quadratic sum of M-G of this time-domain signal afterbody value divided by stem G mould quadratic sum that is worth, obtain ratio ρ ^ε

9), corresponding minimum ρ ^εThe ε of value has provided the estimated value of thick frequency deviation;

10), the thick frequency deviation value that will estimate and thin frequency deviation value addition, obtain total frequency deviation estimated value.

Wherein, ε is the thick frequency deviation value chosen in advance, for-N/2 between the N/2 and do not comprise-integer of N/2; N is the sub-carrier number of ofdm system; P is subsequence number identical in the training symbol, and M is the sub-carrier number that training sequence takies; G is the length of Cyclic Prefix.Usually, G is greater than the maximum delay expansion of wireless multipath channel, and P=N/M.Step 4) in the time-domain signal after proofreading and correct being carried out phase place is that the concrete methods of realizing of the compensation of ε is: i sampling to time-domain signal be multiply by

Multiplication adding up device, multiplier and phase calculation device are finished the computing that comprises in the formula [], multiplier is finished thin frequency offset correction (formula [two]) and phase compensation (formula [three]) computing, discrete Fourier just/inverse transformation device and demodulating equipment finish the computing that comprises among formula [four], [five], [six], [seven] and [eight], the quadratic sum multiplier is finished formula [nine], and the minimum value searcher is finished the estimation (formula [ten]) of thick frequency deviation in the window.Adder is finished total frequency deviation and is estimated computing (formula [11])

The above only is preferred implementation of the present invention; be noted that for those skilled in the art; under the prerequisite that does not break away from the principle of the invention, can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (2)

1. frequency deviation estimating method based on noise subspace is characterized in that the method may further comprise the steps:

1), the time-domain signal to receiving, according to its P that comprises identical subsequence, calculate the phase angle rotational component that causes owing to frequency deviation;

2), above-mentioned phase angle rotational component is done mean deviation normalization and can obtain corresponding thin frequency deviation estimated value;

3), the time-domain signal sequence that receives is made thin frequency offset correction;

4), the time-domain signal after proofreading and correct being carried out phase place is

Compensation;

5), the signal after the compensation is the discrete Fourier transform (DFT) DFT that N is ordered;

6), remove the symbol of modulating on the subcarrier;

7), be the inverse discrete fourier transform IDFT that M is ordered, obtain the time-domain signal that length is M;

8), with the mould quadratic sum of M-G of this time-domain signal afterbody value divided by stem G mould quadratic sum that is worth, obtain ratio ρ ^ε

9), corresponding minimum ρ ^εThe ε of value has provided the estimated value of thick frequency deviation;

10), the thick frequency deviation value that will estimate and thin frequency deviation value addition, obtain total frequency deviation estimated value;

Wherein, ε is the thick frequency deviation value chosen in advance, for-N/2 between the N/2 and do not comprise-integer of N/2; N is the sub-carrier number of ofdm system; P is subsequence number identical in the training symbol, and M is the sub-carrier number that training sequence takies; G is the length of Cyclic Prefix; Usually, G is greater than the maximum delay expansion of wireless multipath fading channel, and P=N/M.

2. the time synchronization method based on noise subspace according to claim 1 is characterized in that: described step 4) time-domain signal after proofreading and correct being carried out phase place is that the concrete methods of realizing of the compensation of ε is: i sampling to time-domain signal be multiply by

Images