CN104202287A - Hardware low-complexity carrier frequency offset estimation method for OFDM-WLAN (orthogonal frequency division multiplexing-wireless local area network) system - Google Patents

Abstract

The invention discloses a hardware low-complexity carrier frequency offset estimation method for an OFDM-WLAN (orthogonal frequency division multiplexing-wireless local area network) system. The hardware low-complexity carrier frequency offset estimation method includes the steps of 1), calculating normalized fine frequency offset by the aid of a long training sequence (LTF), and defining the normalized fine frequency offset as epsilon F; 2), calculating delay-related cumulated sum by the aid of a short training sequence (STF), and defining the delay-related cumulated sum as AC; 3), determining normalized coarse frequency offset by the aid of the epsilon F range and the relation between a real part and an imaginary part of the AC, and defining the normalized coarse frequency offset as epsilon C; 4), acquiring integral frequency offset by the aid of the epsilon F and the epsilon C and performing frequency offset compensation. The relation between frequency offset estimation ranges of the long and short training sequences is ingeniously utilized, calculation of one-time inverse trigonometric function and one-time frequency offset compensation is reduced; and meanwhile, by the aid of diversity combining of multiple antennas, use of a complex multiplier is reduced.

Description

A kind of hardware low complex degree carrier frequency bias estimation for OFDM-WLAN system

Technical field:

The invention belongs to wireless communication technology field, be specifically related to a kind of carrier frequency bias estimation for OFDM-WLAN system.

Background technology:

OFDM (OFDM) is a kind of multi-carrier modulation technology, because of its effectively contrary frequency selectivity decline be widely used in (WLAN) in WLAN (wireless local area network), for example IEEE802.11a/g/n/ac agreement.But ofdm system is responsive especially for frequency deviation, the orthogonality of very little frequency deviation between will heavy damage subcarrier, causes and between subcarrier, disturbs (ICI).Cannot utilize follow-up channel estimating to compensate and disturb between subcarrier, therefore frequency deviation is estimated to play vital effect in whole OFDM receiver.

For this burst data transmission mode of OFDM-WLAN system, the general cross-correlation of calculating between continuous two same-sign that adopts realizes Frequency Synchronization.For example, in IEEE802.11ac, utilize that short training sequence (STF) and long training sequence (LTF) do respectively slightly, thin frequency deviation is estimated and compensation, this method need relate to the calculating of twice antitrigonometric function.And in hardware is realized, the calculating of antitrigonometric function normally adopts look-up table (LUT) or CORDIC iterative algorithm, very huge for the consumption of hardware resource.In addition, this method is carried out respectively frequency deviation estimation and compensation for each antenna, and complex multiplier and delayer number required when hardware designs are a lot, the consumption that has further strengthened hardware resource.

FPGA (Filed-Programmable Gate Array), it is field programmable gate array, as a kind of semi-custom circuit in application-specific integrated circuit (ASIC) field, solve the deficiency of custom circuit, and overcome programming device gate circuit number shortcoming very little.Utilize FPGA to realize OFDM-WLAN system, there is the characteristic of quick low delay and low-power consumption.But need to spend a large amount of logical resources based on said frequencies synchronization scenario, and the resource of FPGA is limited, the algorithm of therefore seeking low complex degree and efficiently and accurately is necessary.

Summary of the invention:

For the defect described in background technology and deficiency, the present invention proposes a kind of carrier frequency bias estimation for OFDM-WLAN system, hardware resource consumption significantly reduces and can accurately estimate frequency deviation.

The invention process method example:

1) signal that OFDM-WLAN system receives in real time, calculates the thin frequency deviation of normalization by time delay auto-correlation algorithm and estimates.Because WLAN mostly is centralized MIMO-OFDM system, each transmitting antenna uses same crystal oscillator.The terminal reception antenna distance that WLAN supports is very near, substantially can think that the carrier wave frequency deviation on every reception antenna is identical.Utilize multi-antenna receive diversity gain, on every antenna, get the long training subsequence of equal number and do the cumulative summation of auto-correlation, hardware multiplier just can stagger in time like this, and the utilance that can ensure multiplier in whole computational process is 100%, greatly reduced hardware multiplier quantity, improved the efficiency that hardware uses.What Fig. 1 showed is LTF sequence, and wherein LTF1 and LTF2 are that length is 64 identical sequence.Concrete step is as follows:

(1) definition A _ffor the signal and the front L that receive _fthe relevant cumulative sum of moment signal, r _j(n) be the signal that the label that receives on j root antenna is n, N _rfor reception antenna number, L _ffor long training sequence length, represent to round downwards, what Fig. 2 showed is the relevant schematic diagram of computation delay during for 4 reception antennas.

(2) definition ε _festimate for the thin frequency deviation of normalization, N is ofdm system subcarrier number, and Im, Re represent respectively to get real part, the imaginary part of imaginary number.

${\mathrm{\ϵ}}_F=-\frac{N}{2\mathrm{\π}{L}_F}{\tan }^{-1}\left[\frac{\mathrm{Im}\left\{A_F\left(n\right)\right\}}{\mathrm{Re}\left\{A_F\left(n\right)\right\}}\right]$

2) definition A _c(n) the delay-correlated cumulative sum for being calculated by short training sequence (STF), I _cand Q _cbe respectively its real part and imaginary part.Wherein STF sequence is with reference to Fig. 3, and computation delay is relevant can analogy Fig. 2.

I _C＝Re{A _C(n)}

Q _C＝Im{A _C(n)}

3) utilize ε _fand I _cand Q _cbetween relation determine the thick frequency deviation ε of normalization _c.When delay-correlated calculates frequency deviation, there are close relationship in estimated frequency deviation region and the cycle of sequence, and for the sequence that the cycle is D, estimated frequency deviation region is be 64 and 16 length training sequence for the cycle, estimated normalization frequency deviation region is respectively [0.5,0.5], [2,2].For antenna reception to signal respectively with long training sequence and short training sequence do frequency deviation estimate after obtain the results are shown in Figure 4.By Tu Ke get, because the frequency offset estimation range of long training sequence is less, in [0.5,0.5] interval, thickness frequency deviation is estimated to calculate accurately initial frequency deviation.In [1.5 ,-0.5] interval, thick frequency deviation is estimated to estimate accurately initial frequency deviation, and result and initial frequency deviation that thin frequency deviation calculates differ Δ ε=-1.In like manner, in [0.5,1.5] interval, result and initial frequency deviation that thin frequency deviation calculates differ Δ ε=+ 1.Can obtain thus, in the situation that calculating thin frequency deviation and estimating, only need judge that now the difference DELTA ε of result and initial frequency deviation can obtain correct frequency offset estimation result, and the set of difference DELTA ε be 1,0 ,+1}.The delay-correlated value A that can utilize thick frequency deviation to calculate _c(n) judge the value of Δ ε, in the time that initial frequency deviation is set as 0.5, the delay-correlated value A being calculated by thick frequency deviation _c(n) corresponding phase angle is π/4, is now a critical condition, A _c(n) real part I _cwith imaginary part Q _cnumerical value equates.And in the time that initial frequency deviation is set as [1.5 ,-0.5], now A _c(n) phase angle should be in interval [pi/2 ,-π/4], and pass corresponding to real part imaginary part is :-Q _c> I _c,

Δε＝-1。In like manner in the time that initial frequency deviation is set as [0.5,1.5], pass corresponding to real part imaginary part is: Q _c> I _c,

Δε＝+1。And OFDM-WLAN system specifies that the maximum frequency deviation error of transmitter and receiver is 20e-6, extreme case is that transmitter and receiver reaches maximum frequency deviation simultaneously, and in the time that carrier frequency is 5.3GHz, normalized frequency deviation region is: [0.6784,0.6784].Above-mentioned normalization interval range [1.5,1.5] can meet the demands.Concrete step is as follows:

(1) if | ε _f|≤0.25, ε _c=0.Do not meet and jump to step (2) or (3).

(2) if ε _f> 0, and Q _c> I _c, ε _c=-1.Do not meet and go to step (4).

(3) if ε _f< 0, and-Q _c> I _c, ε _c=1.Do not meet and go to step (4).

(4)ε _C＝0。

4) utilize the ε trying to achieve _fand ε _cthe frequency deviation that the obtains entire system line frequency offset compensation of going forward side by side.The overall frequency deviation that definition ε is system, R _j(n) be the reception signal after compensate of frequency deviation.

ε＝ε _C+ε _F

$R_j\left(n\right)=r_j\left(n\right)\&CenterDot;e^{-j2\mathrm{\&pi;\&epsiv;n}/N_{\mathrm{fft}}}$

Beneficial effect:

The present invention is a kind of hardware low complex degree carrier frequency bias estimation for OFDM-WLAN system, can effectively reduce FPGA implementation complexity and consumed resource and can obviously not reduce frequency deviation estimated performance, is suitable for the application of Practical Project.

Brief description of the drawings:

Fig. 1 is LTF sequential structure;

Fig. 2 is 4 antenna LTF delay-correlated schematic diagrames;

Fig. 3 is STF sequential structure;

Fig. 4 is that thickness frequency deviation is estimated and initial frequency deviation curve.

Specific implementation method:

Below in conjunction with the drawings and specific embodiments, further illustrate the present invention, should understand these embodiment is only not used in and limits the scope of the invention for the present invention is described, after having read the present invention, those skilled in the art all fall within the application's claims limited range to the amendment of the various equivalent form of values of the present invention.

The inventive method author realizes on NI-PXI platform, and synchronization scenario is to complete at the FPGA of one piece Virtex-5 series, and based on IEEE802.11ac agreement, bandwidth is 20MHz, and transmitting antenna and reception antenna number are all 4.Further illustrate implementation method below in conjunction with actual hardware.

1) the synchronous signal of elapsed time of antenna being received calculates A by time delay autocorrelator _f(n) angle ∠ R, concrete steps are as follows:

(1) by the signal the LTF synchronous elapsed time on every antenna part by 64 delayers, doing length of window with the signal now receiving is 16 delay-correlated cumulative sum.Again window is received to time delay backward 16 in data on next road and continue to do delay-correlated cumulative sum, until 4 circuit-switched data are all finished.Finally the delay accumulation of 4 circuit-switched data and addition are obtained to A _f(n).

$A_F\left(n\right)=\underset{j=1}{\overset{4}{\mathrm{\&Sigma;}}}\underset{k=0}{\overset{15}{\mathrm{\&Sigma;}}}{r}_j[n+(j-1)\&CenterDot;16+k]\&CenterDot;r_j^{*}[n+(j-1)\&CenterDot;16+k+64]$

(2) in order to avoid division when the frequency equilibrium, in the time that realizing, hardware first utilize LUT algorithm to calculate A _f(n) angle

$\&angle;R_F={\tan }^{-1}\left[\frac{\mathrm{Im}\left\{A_F\left(n\right)\right\}}{\mathrm{Re}\left\{A_F\left(n\right)\right\}}\right]$

(3) calculate the thin frequency deviation ε of normalization _f

${\mathrm{\&epsiv;}}_F=-\&angle;R_F\&CenterDot;\frac{64}{2\mathrm{\&pi;}\&CenterDot;64}=-\frac{\&angle;R_F}{2\mathrm{\&pi;}}=-\&angle;R_F\&CenterDot;0.1592$

2) the signal the STF synchronous elapsed time on every antenna part is crossed to 16 delayers, doing length of window with the signal now receiving is 4 delay-correlated cumulative sum.Again window is received to time delay backward 16 in data on next road and continue to do delay-correlated cumulative sum, until 4 circuit-switched data are all finished.Finally the delay accumulation of 4 circuit-switched data and addition are obtained to A _c(n).

$A_C\left(n\right)=\underset{j=1}{\overset{4}{\mathrm{\&Sigma;}}}\underset{k=0}{\overset{3}{\mathrm{\&Sigma;}}}{r}_j[n+(j-1)\&CenterDot;4+k]\&CenterDot;r_j^{*}[n+(j-1)\&CenterDot;4+k+16]$

I _C＝Re{A _C(n)}

Q _C＝Im{A _C(n)}

3) determine the thick frequency deviation of normalization, concrete steps are as follows:

(1) if | ε _f|≤0.25, ε _c=0.Do not meet and go to step (2) or (3).

(2) if ε _f> 0, and Q _c> I _c, ε _c=-1.Do not meet and go to step (4).

(3) if ε _f< 0, and-Q _c> I _c, ε _c=1.Do not meet and go to step (4).

(4)ε _C＝0。

4) the computing system entirety angle line frequency offset compensation of going forward side by side.Concrete steps are as follows:

(1) calculate thick frequency deviation angle

$\&angle;R_C=-{\mathrm{\&epsiv;}}_C\&CenterDot;\frac{2\mathrm{\&pi;}\&CenterDot;16}{64}=-{\mathrm{\&epsiv;}}_C\&CenterDot;\frac{\mathrm{\&pi;}}{2}=-{\mathrm{\&epsiv;}}_C\&CenterDot;1.5708$

(2) computing system angle

$\&angle;R=\frac{\&angle;R_C}{16}+\frac{\&angle;R_F}{64}$

In the time that hardware is realized, utilize displacement to realize division, move right respectively 4 and 6

(3) carry out compensate of frequency deviation

R _j(n)＝r _j(n)·e- ^j∠R·n。

Claims (1)

1. for a hardware low complex degree carrier frequency bias estimation for OFDM-WLAN system, comprise following steps:

1) after signal elapsed time of described OFDM-WLAN system being received is in real time synchronous, utilize long training sequence LTF to calculate the thin frequency deviation of normalization, concrete steps are as follows:

(1) definition A _ffor the signal and the front L that receive _fthe relevant cumulative sum of moment signal, r _j(n) be the signal that the label that receives on j root antenna is n, N _rfor reception antenna number, L _ffor long training sequence length, represent to round downwards,

(2) definition ε _festimate for the thin frequency deviation of normalization, N is ofdm system subcarrier number, and Im, Re represent respectively to get real part, the imaginary part of imaginary number,

${\mathrm{\&epsiv;}}_F=-\frac{N}{{2\mathrm{\&pi;L}}_F}{\tan }^{-1}\left[\frac{\mathrm{Im}\left\{A_F\left(n\right)\right\}}{\mathrm{Re}\left\{A_F\left(n\right)\right\}}\right];$

2) definition A _c(n) the delay-correlated cumulative sum for being calculated by short training sequence STF, L _cfor short training sequence length, I _cand Q _cbe respectively its real part and imaginary part,

I _C＝Re{A _C(n)}；

Q _C＝Im{A _C(n)}；

3) utilize ε _fand I _cand Q _cbetween relation determine the thick frequency deviation ε of normalization _c, concrete steps are as follows:

(1) if | ε _f|≤0.25, ε _c=0, do not meet and go to step (2) or (3);

(2) if ε _f> 0, and Q _c> I _c, ε _c=-1, do not meet and go to step (4);

(3) if ε _f< 0, and-Q _c> I _c, ε _c=1, do not meet and go to step (4);

(4)ε _C＝0；

4) utilize the ε trying to achieve _fand ε _cthe frequency deviation that the obtains entire system line frequency offset compensation of going forward side by side, the overall frequency deviation that definition ε is system, R _j(n) be the reception signal after compensate of frequency deviation,

ε＝ε _C+ε _F；

R _j(n)＝r _j(n)·e ^-j2πεn/N。