CN101355383B - Signal detection method for stochastic access channel - Google Patents

Abstract

The invention provides a method for detecting the signals randomly accessed to a signal channel. The method comprises the steps: A. after a base station acquires N time domain correlation values R(m) of the u root sequence, and all m values corresponding to the N time domain correlation values are used as detection points; B. among all detection points, the detection point m_max corresponding to the biggest time domain correlation value is found out, when the biggest time domain correlation value R(m_max) is bigger than the peak detection threshold, whether lead codes randomly accessed to the signal channel is detected by a circular shift search window corresponding to the m_max is judged, the m_max and the point m1 and/or point m2 of the present detection points are used as non-detection points, wherein m1=(N+m_max-du)mod N, and m2=(m_max+du)mod N; C. if detection points are still left, the biggest time domain correlation value of the remained detection points is bigger than the peak detection threshold, the step B is jumped to, N is more than and equal to Nzc.

Description

A kind of signal detecting method of RACH

Technical field

The present invention relates to the communications field, relate in particular to a kind of signal detecting method of RACH.

Background technology

At LTE (Long Term Evolution; Long Term Evolution) in the system, at first passes through SCH (Synchronization Channel after UE (User Equipment, the user terminal) start; Synchronizing channel) carries out down-going synchronous, find the reception starting point and the cell id (CellID) of radio frames and subframe; Obtain system information through BCH (Broadcast Channel) then, comprise the configuration information of RACH (RandomAccess Channel, RACH) in the system information; Carry out uplink synchronous through RACH at last, accomplish the work of connecting system.

In the process of uplink synchronous; The radio frames that UE confirms during at first with down-going synchronous and the reception starting point of subframe are the transmitting site that the RACH channel is found on the basis; Confirm that from the system information that obtains this sub-district is used for the available sequences of RACH transmission, at random one of selection is sent as lead code (preamble) from available sequence then.The base station is detected with definite up timing adjustment amount lead code, and sends it to UE.UE realizes up time synchronized according to the delivery time of this timing adjustment amount adjustment upward signal.

What the uplink random access lead sign indicating number of LTE used is ZC (Zadoff-Chu) sequence, being defined as of u root ZC sequence:

$x_u\left(n\right)=e^{-j\frac{\mathrm{\πun}(n+1)}{N_{\mathrm{ZC}}}},0\≤n\≤N_{\mathrm{ZC}}-1;$

Wherein, j is an imaginary unit; U is a root sequence index number; N _ZCBe the length and the N of ZC sequence _ZCBe prime number, in LTE, be defined as 839.

In LTE, 64 sequences that are used for lead code all can be distributed in each sub-district, and these 64 sequences can be the different cyclically shifted sequences from same sequence, also can be the different cyclically shifted sequences from different root sequences.

The ZC sequence is the permanent width of cloth zero auto-correlation sequence of (Constant Amplitude Zero Auto-correlation Code is called for short CAZAC), and the correlation of ZC sequence has following characteristics: the correlation between the different cyclically shifted sequences of same root sequence is 0; Correlation between different root sequences and the cyclically shifted sequences thereof is

; Be that correlation between the different root sequences is very little, be approximately 0.Therefore the base station can utilize the relevant nature of ZC sequence that accidental access signal is carried out the time domain coherent detection and obtain up timing adjustment amount.

Be defined as on the time domain related detecting method is directly perceived the complex conjugate of each cyclic shift of the signal received and local sequence is carried out the time domain correlation that dot product and summation obtain each cyclic shift sampling point, after frequency-region signal that can equivalence on the mathematics becomes to receive and local frequency domain sequence complex conjugate dot product, be transformed into time domain through inversefouriertransform.The mathematical form of time domain coherent detection is represented as follows:

The time domain form of supposing the reception signal is y (m), and frequency domain form is Y (k); Local sequence time domain form is x (m), and frequency domain form is X (k), and the complex conjugate time domain form of local sequence is x ^*(m), frequency domain form X ^*(k); Both correlation function R (m) are formulated as:

$R\left(m\right)=\underset{n=0}{\overset{N-1}{\mathrm{\Σ}}}{x}^{*}\left(n\right)\·y(m+n)=\frac{1}{N}\underset{k=0}{\overset{N-1}{\mathrm{\Σ}}}Y\left(k\right)\·X^{*}\left(k\right)\·e^{j\frac{2\mathrm{\π}}{N}\mathrm{km}}---\left(1\right)$

Wherein, m is the cyclic shift sampling point, and N is the number of samples of ZC sequence.Therefore; To the different cyclic shifts of using same sequence RACH user as lead code; The conversion of signals that receives frequency domain sequence complex conjugate dot product with itself and this root sequence behind the frequency domain; Result after the dot product is carried out inversefouriertransform be transformed into time domain, can obtain the corresponding time domain correlation of each cyclic shift sampling point.

Carry out peak value through the time domain correlation to the corresponding cyclic shift search window of each cyclically shifted sequences of local root sequence and detect, which bar lead code what can know that UE uses is, and obtains the Timing Advance of UE according to the lead code that detected UE uses.

Fig. 1 is the method sketch map of RACH time domain coherent detection; Be example with two reception antennas among the figure, described the implementation method of carrying out the time domain coherent detection to received signal, briefly be described below:

The base station is at first carried out frequency correction and is fallen sampling the reception signal of each reception antenna, carries out the FFT (Fast Fourier Transform, FFT) of M point (such as M=1024) then, extracts 839 RACH frequency-region signal; Then with 839 RACH frequency-region signal and each local frequency domain root sequence complex conjugate dot product; (N is greater than 839 to the N point in zero padding; Can equal 2048 such as N), carry out the IFFT (Inverse Fast Fourier Transform, invert fast fourier transformation) that N orders and obtain this local root sequence of N time domain correlation; This N time domain correlation asked mould square, and the N point time domain correlation size that obtains has reflected the size of signal and noise power; Time domain correlation with two reception antennas merges at last, obtains N time domain correlation after antenna merges, and carries out peak value then and detects, and obtains the position and the time corresponding lead of lead code.

The performance of RACH time domain coherent detection (promptly lead code is detected, be also referred to as the RACH preamble detection) can characterize with loss, false alarm rate and the detection threshold of lead code.When RACH detects, at first when not sending signal, confirm under the given false alarm rate target corresponding detection threshold, test the loss that accidental access signal detects when transmitting according to this thresholding then.

The definition of false alarm rate is the probability that when not having lead code to send, detects lead code; But the definition of loss is the lead code that detects a mistake, or do not detect a lead code of having sent, or detect correct lead code estimate the probability that wrong situation such as timing adjustment amount take place.

Generally requiring the target false alarm rate is 10 ^-3Or be slightly less than 10 ^-3When not sending signal, reaching false alarm rate is 10 ^-3Or be slightly less than 10 ^-3Detection threshold be the thresholding that peak value detects.

The thresholding that peak value detects is divided into two kinds of absolute thresholding and relative thresholds.In general; Absolute thresholding is relevant with the size of noise; And with the ratio of the correlation of accidental access signal and noise power during as thresholding, then this thresholding just have nothing to do with noise power, so the ratio of signal correlation values and noise power is defined as the relative threshold of peak value detection.When the RACH signal is detected; It generally is relative threshold of systemic presupposition; The size of estimating noise in detection then; Noise size multiplies each other with relative threshold can obtain absolute thresholding, thus above saidly confirm under given false alarm rate target that when not sending signal the detection threshold of correspondence is meant relative threshold.When the ratio of correlation that receives signal and noise power during greater than relative threshold, thinking has the RACH signal.Under normal circumstances, it is high more that relative threshold is established, the more difficult access of user, and that establishes is low more, and false-alarm is just many more.In general, this thresholding value is: the signal when false alarm rate is 10-3 and the ratio of noise (therefore also can be described as the false-alarm thresholding).

Under the false-alarm thresholding, when UE emission RACH signal, if do not have frequency shift (FS) (abbreviation frequency deviation) and other interference, false alarm rate generally can not be higher than 10 ^-3But when frequency shift (FS) was arranged, the energy of RACH signal can disperse, because the lead code use is the ZC sequence, the definition of this sequence makes RACH signal energy when frequency deviation is arranged can be distributed on the other cyclically shifted sequences, thereby faults occurs.

Introduce this frequency bias properties of ZC sequence below in detail.

After descending frequency offset correction, the frequency deviation of up RACH can be not big especially.Because the movement velocity of UE is generally less than 375km/h, so it has been generally acknowledged that, only there is the frequency deviation smaller or equal to 1 times in RACH.For the ease of analyzing, our supposition has earlier produced 1 frequency multiplication (can be that 1 inclined to one side or negative frequency multiplication of positive 1 frequency multiplication is inclined to one side) partially.If the RACH sequence of the frequency domain that receives is Z (k), its time domain form is designated as z (n), and according to the character of IDFT (Inverse Discrete Fourier Transform, discrete inverse-Fourier transform), to the IDFT that N is ordered, we have:

Expression formula according to the ZC sequence: $z\left(n\right)=\mathrm{Exp}(-j\frac{\mathrm{\π}}{N_{\mathrm{Zc}}}\mathrm{Un}(n+1)),$ To sequence length is N _ZcThe ZC sequence at cyclic shift integer d _uAfterwards, we obtain:

$z(n+d_u)=\exp (-j\frac{\mathrm{\π}}{N_{\mathrm{zc}}}\mathrm{un}(n+1))\·\exp (-j\frac{2\mathrm{\π}}{N_{\mathrm{zc}}}u\·d_u\·n)\·\exp (-j\frac{\mathrm{\π}}{N_{\mathrm{zc}}}{\mathrm{ud}}_u(d_u+1));---\left(3\right)$

Order $\mathrm{\β}=\mathrm{Exp}(-j\frac{\mathrm{\π}}{N_{\mathrm{Zc}}}{\mathrm{Ud}}_u(d_u+1)),$ β is and the irrelevant amount of n, as u and d _uWhen confirming, β is a constant.

Then $z(n+d_u)=z\left(n\right)\mathrm{Exp}(-j\frac{2\mathrm{\π}}{N_{\mathrm{Zc}}}u\·d_u\·n)\·\mathrm{\β}---\left(4\right)$

Can find out to have only the ud of working as by top analysis _u=mN _ZcIn the time of ± 1, z (n+d is arranged _u)=z (n) β, promptly when generation 1 frequency multiplication was inclined to one side, the relevant peaks of RACH sequence can produce skew, i.e. and the signal in the whole circulation displacement search window all can have skew.Here m is an integer, and the value of m guarantees ud _u=mN _Zc± 1 sets up, at this moment d _uCan be expressed as: $d_u=\frac{m{N}_{\mathrm{Zc}}\±1}{u}.$

A frequency multiplication that aligns is inclined to one side, has $u\·d_u^{+}=m^{+}\·N_{\mathrm{Zc}}+1$ , m ⁺With It all is integer; $u\·d_u^{-}=m^{-}\·N_{\mathrm{Zc}}-1,$ M-with

It all is integer.Because $u\&CenterDot;(N_{\mathrm{Zc}}-d_u^{-})=(u-m^{-})\&CenterDot;N_{\mathrm{Zc}}+1$ , and u-m ^-Be integer.When $0<d_u^{+}<N_{\mathrm{Zc}}/2$ , make N _Zc-d ^- _uBe positive integer, then $N_{\mathrm{Zc}}/2<{d^{-}}_u=N_{\mathrm{Zc}}-d_u^{+}<N_{\mathrm{Zc}};$ When $N_{\mathrm{Zc}}/2<d_u^{+}<N_{\mathrm{Zc}}$ , make N _Zc-d ^- _uBe positive integer, then have $0<{d^{-}}_u=N_{\mathrm{Zc}}-d_u^{+}<N_{\mathrm{Zc}}/2.$ Note N _ZcBe prime number, so N _Zc/ 2 is not integer, and

With

It all is positive integer.

In addition, observe below formula (formula (5)) d can find also that a positive and negative frequency multiplication is inclined to one side the time _uThe value rule.

$\exp (-j\frac{2\mathrm{\&pi;}}{N_{\mathrm{zc}}}u\&CenterDot;(\mathrm{Nzc}-d_u)\&CenterDot;n)=\exp (j\frac{2\mathrm{\&pi;}}{N_{\mathrm{zc}}}u\&CenterDot;d_u\&CenterDot;n)=\exp (j\frac{2\mathrm{\&pi;}}{N_{\mathrm{zc}}}(m\&CenterDot;N_{\mathrm{zc}}\&PlusMinus;1)\&CenterDot;n)=\exp (\&PlusMinus;j\frac{2\mathrm{\&pi;}}{N_{\mathrm{zc}}}\&CenterDot;n)$

(5)。

Can know from the top derivation of equation.When if frequency deviation is the decimal times of subcarrier spacing, because ud _uDecimal be can not be,, but former sequence when frequency deviation (no sequence) and former sequence are in through cyclic shift d so the time domain sequences that receives can not produce cyclic shift _uAfter sequence (former sequence sequence that 1 frequency multiplication obtains after partially) between state, the signal that at this moment receives and do not have between the former sequence of frequency deviation and and the inclined to one side sequence that produces of 1 frequency multiplication between (be former sequence cyclic shift d _uSequence afterwards) all can produce relevant peaks, at this time will occur two copies (be called the copy search window that causes because of frequency deviation, or abbreviate the copy window as) of cyclic shift search window simultaneously.

In the physical layer standard TS36.211 of 3GPP, to d to LTE _uValue clear and definite regulation is arranged:

$d_u=\left\{\begin{array}{cc}{u}^{-1}\mathrm{<figure-callout\; id="0"\; label="mod\; N\; ZC"\; filenames="H2008102115131E00041.png"\; state="\{\{state\}\}">mod</figure-callout>}{N}_{\mathrm{ZC}}{}_{}& 0\&le;u^{-1}\mathrm{mod}{N}_{\mathrm{ZC}}<N_{\mathrm{ZC}}/2\begin{array}{}\end{array}\\ N_{\mathrm{ZC}}-u^{-1}\mathrm{mod}{N}_{\mathrm{ZC}}& \mathrm{otherwise}\end{array}\right.,$

Wherein, u ^-1ModN _ZCExpression ud _u=mN _ZC± 1, m makes d _uBe the smallest positive integral of positive integer, d _uAlways less than N _Zc/ 2.In Fig. 2, provided cyclic shift and be example of 0, therefrom be appreciated that

And d _uRelation and the relation of cyclic shift window and copy window, wherein dotted portion is that the negative frequency deviation copy window of equivalence and negative frequency deviation copy window that the time domain correlation method uses are the relations of loop cycle displacement.It is not corresponding cyclic shift window of 0 cyclic shift and an example of copy window that Fig. 3 has provided certain.Can find out from Fig. 2 and Fig. 3, only with the d that defines among the TS36.211 _u(promptly less than N _ZC/ 2 d _uValue), just can derive the distance of two copy windows and its cyclic shift window.In fact, d _uIt is the distance of cyclic shift search window starting point and copy window starting point.

In LTE PHY TS36.211; In LTE, be divided into two kinds of unrestricted collection unrestricted sets (also being the conventional normal of collection set) and restriction set restricted set (also making high speed collect high speed set) to producing cyclic shift that targeting sequencing uses.Restriction set (collection at a high speed) has considered that the frequency deviation of bringing at a high speed influences, for the search window that guarantees each targeting sequencing does not overlap each other, and the cyclic shift amount C that the generation of targeting sequencing is used _vCan be restricted.But the unrestricted collection (conventional collection) to LTE is used for the very little scene of frequency deviation (such as the low speed sub-district), and the cyclic shift amount that its targeting sequencing produces can be not restricted, produces the value C of leading cyclic shift _v=vN _CS, v=0 wherein, 1,2 ...,

The maximum integer that is less than or equal to x is got in expression, promptly x is rounded N downwards _CSLength (promptly producing the step-length of cyclically shifted sequences) for search window.

According to the frequency bias properties of ZC sequence,, in the RACH preamble detection, except cyclic shift corresponding search window (being called for short the cyclic shift search window), also have peak value at the copy window because the RACH signal that receive the base station has frequency deviation.To the high speed sub-district; Doppler frequency deviation is bigger; Therefore the correlation of the RACH signal in the copy window can be bigger; When detecting, think that cyclic shift window and two copy windows all signal may occur, be to cyclic shift window and two copy window comprehensive detection, because restriction set is used in the high speed sub-district; The corresponding cyclic shift window of each lead code and its copy window all do not overlap with its copy window with the cyclic shift window of other lead code, the phenomenon of two different lead codes a search window therefore can not occur.Centering low speed sub-district; Doppler frequency deviation is smaller, and often the signal correlation values in the cyclic shift window is far above the signal correlation values in the copy window, therefore when detecting, only thinks to have signal in the cyclic shift window; Promptly only carry out preamble detection, do not detect the copy window at the cyclic shift window.

But always there is certain frequency deviation in middle low speed sub-district, such as Doppler frequency deviation; Like this; The relevant peaks that in the copy window, always has signal occurs; What used middle low speed sub-district is non-restriction set; Promptly do not have the cyclic shift restriction, though the cyclic shift window of the cyclic shift window of each lead code correspondence and other lead codes does not overlap, the corresponding cyclic shift window of each lead code possibly overlap with the copy window of other lead codes.In middle low speed sub-district, the signal correlation values in the cyclic shift window will be in the copy window because the signal correlation values that frequency deviation is brought, but when signal to noise ratio was higher, the correlation that this frequency deviation is brought in the copy window also can be big.For example, only send lead code C1, do not send lead code C2.When the cyclic shift search window of the copy window of lead code C1 and lead code C2 overlapped, the frequency deviation copy signal of C1 possibly fallen the cyclic shift search window of C2.When the cyclic shift window to C2 detects, be easy to judge into the frequency deviation copy signal of lead code C1 the signal of C2, and C2 is actually not transmission, this has just produced false-alarm.

When certain RACH signal had frequency deviation and signal to noise ratio very high, this false-alarm phenomenon can be clearly.When a plurality of high s/n ratio users inserted simultaneously, false alarm rate can be very high.A large amount of false-alarms can increase the signaling consumption in the random access procedure, increases the processing burden of system, therefore need suppress false-alarm.

Summary of the invention

Technical problem to be solved by this invention is, overcomes the deficiency of prior art, and a kind of signal detecting method of RACH is provided, to the sub-district of using unrestricted collection (middle low speed sub-district) since the false-alarm that frequency deviation causes suppress.

In order to address the above problem, the present invention provides a kind of signal detecting method of RACH, comprising:

A: after the base station obtains a u root sequence of N time domain correlation R (m), with N pairing all m values of time domain correlation as test point, m=1 wherein, 2 ..., N;

B: in current all test points; Find the maximum pairing test point m_max of time domain correlation R (m_max); As the time domain correlation R (m_max) of said maximum during greater than the peak value detection threshold; Judgement has detected the lead code of RACH in the pairing cyclic shift search window of m_max, and with the some m1 in m_max and the current detection point and/or some m2 as non-test point; Wherein, m1=(N+m_max-d _u) mod N, m2=(m_max+d _u) mod N, m1 and m2 are positive integers;

C: if remain test point in addition, and the time domain correlation of the maximum in the residue test point then jumps to step B greater than the peak value detection threshold; Otherwise finish the detection of current root sequence;

Wherein, N>=N _Zc, N _ZcLength for the root sequence; d _uBe the starting point of cyclic shift search window and the distance between the copy window starting point.

In addition, in step B, find said m_max after, also operate as follows:

If m3>m4, then will be more than or equal to m3 and smaller or equal to the point of N, and/or, more than or equal to 1 and smaller or equal to the point of m4 as non-test point; Otherwise, will more than or equal to m3 and smaller or equal to the point of m4 as non-test point; And/or

If m5>m6, then will be more than or equal to m5 and smaller or equal to the point of N, and/or, more than or equal to 1 and smaller or equal to the point of m6 as non-test point; Otherwise, will more than or equal to m5 and smaller or equal to the point of m6 as non-test point;

Wherein, m3=(N+m_max-d _u-X) mod N, m4=(N+m_max-d _u+ X) mod N, m3 and m4 are positive integer; M5=(N+m_max+d _u-X) mod N, m6=(m_max+d _u+ X) modN, m5 and m6 are positive integer; X=int (N/ (N _ZC* 2), int representes to round up or down.

In addition, in step B, find said m_max after, in current detection point, find other test point that surpasses the peak value detection threshold that belongs to same cyclic shift search window with said m_max: m [1] ..., m [K]; With the m in the current detection point [k]-d _uAnd/or m [k]+d _uBe designated non-test point; 1≤k≤K.

In addition, in step B, find said test point: m [1] ..., behind the m [K], also operate as follows:

If m7 [k]>m8 [k], then will be more than or equal to m7 [k] and smaller or equal to the point of N, and/or, more than or equal to 1 and smaller or equal to the point of m8 [k] as non-test point; Otherwise, will more than or equal to m7 [k] and smaller or equal to the point of m8 [k] as non-test point; And/or

If m9 [k]>m10 [k], then will be more than or equal to m9 [k] and smaller or equal to the point of N, and/or, more than or equal to 1 and smaller or equal to the point of m10 [k] as non-test point; Otherwise, will more than or equal to m9 [k] and smaller or equal to the point of m10 [k] as non-test point;

Wherein, m7 [k]=(N+m [k]-d _u-X) mod N, m8 [k]=(N+m [k]-d _u+ X) mod N, m7 [k] and m8 [k] they are positive integer; M9 [k]=(N+m [k]+d _u-X) mod N, m10 [k]=(m [k]+d _u+ X) mod N, m9 [k] and m10 [k] they are positive integer; X=int (N/ (N _ZC* 2), int representes to round up or down.

In addition, in step B, find said m_max after, also all test points that belong to same cyclic shift search window with said m_max in the current detection point are designated non-test point.

In addition, said d _uMore than or equal to 2 and less than N _Zc/ 2, and work as N=N _ZCThe time, said d _uSatisfy following equality: (ud _u) modN _Zc=± 1.

In addition, as N>N _ZCThe time, du=int (d ' _u* N/N _ZC), and d ' _uSatisfy following equality: (ud ' _u) modN _Zc=± 1, d ' _uMore than or equal to 2 and less than N _Zc/ 2; Wherein, int () expression rounds up or rounds downwards.

In addition, adopt following steps to confirm said peak value detection threshold:

Carry out Noise Estimation to received signal, obtain the noise average; With the product of above-mentioned noise average and predefined relative threshold value as the peak value detection threshold;

Said relative threshold value reaches the false-alarm signal of target false alarm rate and the ratio of noise for when not having signal to send.

In addition, adopt following steps to confirm said noise average:

With the maximum time domain correlation that receives signal and Noise Estimation ratio as thresholding A; With in said N the time domain correlation less than the average of the time domain correlation of thresholding A as said noise average; Said Noise Estimation ratio is less than 1 and greater than 0.

In addition, said sequence is the ZC sequence.

In sum; Adopt method of the present invention, can the false-alarm that cause because of frequency deviation in the sub-district of using unrestricted collection (middle low speed sub-district) effectively be suppressed, reduced the signaling consumption that false-alarm causes; Avoided omission simultaneously, guaranteed the successful access of more users accidental access signal.

Description of drawings

Fig. 1 is the method sketch map of RACH time domain coherent detection;

Fig. 2 is that cyclic shift is 0 o'clock positive negative frequency deviation copy window and cyclic shift window distance relation sketch map in the coherent detection of RACH time domain;

Fig. 3 is that certain is not 0 corresponding cyclic shift search window and the frequency deviation copy window sketch map of cyclic shift in the coherent detection of RACH time domain;

Fig. 4 is an embodiment of the invention RACH time domain related detecting method flow chart;

The false-alarm sketch map of Fig. 5 for causing by frequency deviation in the RACH coherent detection.

Embodiment

Basic ideas of the present invention are; Because the frequency shift (FS) meeting causes a large amount of false-alarms; Therefore can be when the peak value in carrying out RACH time domain coherent detection process detects; Find the position of these false-alarms according to the frequency bias properties of ZC sequence, remove the position that these possibly be false-alarms, reach the purpose that false-alarm suppresses.

To combine accompanying drawing and embodiment to describe the present invention below.

Fig. 4 is an embodiment of the invention RACH time domain related detecting method flow chart.For outstanding emphasis of the present invention, Fig. 4 has carried out partly merging and simplifying to Fig. 1, and is that example is described with single reception antenna only.As shown in Figure 4, this method comprises the steps:

401: from receive signal, extract N _ZcThe RACH frequency-region signal value of individual point;

As stated, N _ZcBe the length of ZC sequence, in LTE, be defined as 839.

402: with above-mentioned N _ZcThe RACH frequency-region signal value of individual point and the complex conjugate of each local root sequence are carried out dot product and are obtained N _ZcIndividual dot product value, and after zero padding becomes N point (promptly carrying out time domain oversampling), carry out the IFFT that N is ordered, obtain each local root sequence of N time domain correlation: R (m), 1≤m≤N;

Wherein, but above-mentioned zero padding is treated to selection operation, and its effect is to use inversefouriertransform fast, reduces amount of calculation, does not handle if do not carry out zero padding, then N=N _Zc

403: above-mentioned N time domain correlation carried out mould square;

This step is an optional step, and in fact the time domain correlation of process mould square processing has reflected the signal power value.

Each root sequence of N time domain correlation being carried out peak value detect, below is that example is described with u root sequence, specifically comprises following steps:

404: in N time domain correlation, with all pairing m values of time domain correlation (cyclic shift sampling point, or be called test point) greater than peak value detection threshold (being designated as thresholding B) as initial some set to be detected;

Thresholding B can adopt following method to calculate:

404A: carry out Noise Estimation to received signal, obtain noise gate (being designated as thresholding A);

Thresholding A can equal: maximum time domain correlation * Noise Estimation ratio; Wherein, the Noise Estimation ratio is greater than 0 and less than 1, and for example, the Noise Estimation ratio value can be taken as 0.6.

404B: calculate all averages (being called the noise average) less than the time domain correlation of thresholding A;

In fact the noise average has reflected the size of noise power.

Certainly, thresholding A and noise average also can adopt other method of prior art to calculate.

404C: with the product of above-mentioned noise average and predefined relative threshold value as thresholding B;

Relative threshold can be got when not having signal to send, and reaches the false-alarm signal of target false alarm rate (for example 0.1%) and the ratio of noise.

It should be noted that; The effect of peak value detection threshold (thresholding B) is to distinguish noise and signal; Only play the effect (promptly the time domain correlation less than this thresholding does not detect) that stops to detect in the present invention, also have other multiple setting and computational methods in the prior art.

405: the maximum R (m_max) that finds the time domain correlation of each measuring point to be checked in the current some set to be detected; Judgement has detected lead code in the pairing cyclic shift search window of m_max, and confirms the cyclic shift amount and the Timing Advance of pairing this root sequence of lead code based on m_max;

406: with m_max deletion (to avoid duplicate detection) from current some set to be detected, if m_max-d _uAnd/or m_max+d _uBe positioned at current set to be detected, with m_max-d _uAnd/or m_max+d _uDeletion from current some set to be detected;

Consider m_max-d _uAnd m_max+d _uMight be not in the scope of interval [1, N], the point of actual deletion from some set to be detected is: m1=(N+m_max-d _u) mod N, m2=(m_max+d _u) mod N, m1 and m2 are positive integers.

Except m_max-d _uAnd m_max+d _uIn addition, can also be with interval [m_max-d _u-X, m_max-d _u+ X] and [m_max+d _u-X, m_max+d _u+ X] in the deletion from point to be detected set of other value.Wherein, [] expression closed interval; X=int (N/ (N _Zc* 2), int representes to round up or down.

Equally, consider m_max-d _u± X and m_max+d _uIn fact ± X might should delete the point that meets the following conditions not in the scope of interval [1, N] from some set to be detected:

If m3>m4, then delete more than or equal to m3 and smaller or equal to the point of N, and/or deletion is more than or equal to 1 and smaller or equal to the point of m4; Otherwise deletion is more than or equal to m3 and smaller or equal to the point of m4;

If m5>m6, then delete more than or equal to m5 and smaller or equal to the point of N, and/or deletion is more than or equal to 1 and smaller or equal to the point of m6; Otherwise deletion is more than or equal to m5 and smaller or equal to the point of m6;

Wherein, m3=(N+m_max-d _u-X) mod N, m4=(N+m_max-d _u+ X) mod N, m3 and m4 are positive integer;

M5=(N+m_max+d _u-X) mod N, m6=(m_max+d _u+ X) mod N, m5 and m6 are positive integer.

Above-mentioned d _uBe the starting point of cyclic shift search window and the distance of copy window starting point.Cyclic shift window and copy window almost overlap when du=1, can delete the point of copy window.Therefore only can limit at d _uMore than or equal to the operation of carrying out step 406 at 2 o'clock.

Work as N=N _ZCThe time, satisfy: (ud _u) mod N _Zc=± 1 and d _uLess than N _Zc/ 2;

As N>N _ZCThe time, d _u=int (d ' _u* N/N _ZC), and (ud ' _u) modN _Zc=± 1 and d ' _uLess than N _Zc/ 2, wherein int () expression rounds up or rounds downwards.

U is a root sequence sequence number; Operator mod representes to get remainder; And when 0 < x≤y; (x) value of mod (y) is less than or equal to , and expression rounds up, for example: 838mod839=-1.

Through this step; The lead code that maximum related value place cyclic shift window in the current point to be detected set is corresponding is as detected lead code, and in two copy windows that position m_max, this maximum related value place cyclic shift window of this maximum related value is corresponding with the position m_max correspondence position (m_max-d of this maximum related value _uAnd m_max+d _u) and position m_max-d _u, position m_max+d _u) the plurality of adjacent position removes from point to be detected set as the false-alarm position, to reduce false alarm rate.

407: if also comprise measuring point to be checked in the point to be detected set, then jump to step 405, otherwise finish the peak value detection of u root sequence.

The false-alarm sketch map of Fig. 5 for causing by frequency deviation in the RACH coherent detection; To combine Fig. 5 that an application example of the present invention is introduced below.

After obtaining a certain pairing N of a sequence time domain correlation, the time domain correlation less than thresholding A to be averaged, the average of these data has reflected the noise power average of this root sequence.With noise average and false alarm rate target is 10 ^-3The time the product of relative threshold value as thresholding B.

The length of the search window of supposing the system configuration is Ncs, has a RACH user to insert, and that this user uses is the cyclic shift Cv1 of certain root sequence, and then this user's cyclic shift search window is from Cv1 to Cv1+Ncs, and a copy window is Cv1-d _uTo Cv1-d _u+ Ncs; Another copy window is Cv1+d _uTo Cv1+d _u+ Ncs.To the low speed sub-district, there is not the cyclic shift restriction, so the copy window of cyclic shift Cv1 possibly be other RACH users' a cyclic shift search window.

As shown in Figure 5, after all detection positions are detected, find that some x1 (m1, R (m1)) and some x2 (m2, R (m2)) they are the points that surpasses thresholding B; Corresponding peak power R (m1) maximum of some x1, and drop in the cyclic shift search window of Cv1, therefore judge to detect the RACH user that cyclic shift is Cv1.Point x2 also is the point that surpasses thresholding B; In fact putting x2 is to be caused by the frequency deviation of the RACH of x1 place; But do not have in middle low speed sub-district under the situation of cyclic shift restriction,, do not suppress if do not carry out false-alarm because x2 drops in the cyclic shift window of Cv2; Can judge into another RACH user that cyclic shift is Cv2 to an x2, some x2 will become false-alarm.If signal to noise ratio is very high, each has the corresponding copy window of cyclic shift window of RACH signal all false-alarm can occur, and this will bring very big signaling consumption.

Point x2 just in time is d with the distance of some x1 _uThis explanation point x2 might be that the frequency deviation of an x1 causes, some x2 with and about 2 all may be the false-alarm point, in detection need an x2 with and about 2 from some set to be detected, remove; So just can not detect the signal of these positions, just can not produce false-alarm yet.Because the corresponding cyclic shift search window of some x1 has detected targeting signal; At this moment non-detection position also comprise in the corresponding cyclic shift search window of a some x1 have a few; For fear of duplicate detection, therefore also need all indicate the right and wrong detection position to the position from Cv1 to Cv1+Ncs.Next once more but all detection positions are detected; Find not surpass the time domain correlation of thresholding B; The peak detection process of then judging this root sequence finishes; The end product of this root sequence coherent detection is only detect an x1 place cyclic shift search window correspondence leading, leading not the reporting as the result who detects of some x2 place cyclic shift search window correspondence.

The peak point and the same situation that another cyclic shift of sequence cv2 peak point overlaps that certain cyclic shift cv1 frequency deviation copy under some few situation, can occur; At this moment use our above-mentioned false-alarm inhibition method may cause the omission of the corresponding signal of cyclic shift cv2; But this situation is few the generation; Because each user distance base station location is at random in the real system, its time delay is also different.And when search window length was bigger, the displacement of the available cycles of same sequence can be seldom, and the probability of this coincidence is just littler.

In addition, we are limited in d _uSituation less than 2 is not carried out false-alarm of the present invention and is suppressed to handle (position of promptly in peak value detects, confirming as lead code in a certain cyclic shift search window at the correspondence position of its copy window as non-detection position), can not occur the phenomenon of omission in the time of can guaranteeing to have peak value to overlap between certain cyclic shift window and the copy of himself.

According to basic principle of the present invention, the foregoing description also has multiple mapping mode, for example:

(1) in step 405; After confirming m_max; Can also handle as follows: at first in point to be detected set, find all and m_max to be positioned at the point of same cyclic shift search window, will be arranged in the point of same cyclic shift search window with m_max then and/or each corresponding points of the pairing copy window of this cyclic shift search window are deleted from some set to be detected.

That is to say that if in some set to be detected, have the m_maxl that is positioned at same cyclic shift search window with m_max, the point in the pairing copy window of m_maxl is m_maxl-d _uAnd m_maxl+d _uIf (m_maxl, m_maxl-d _uAnd m_maxl+d _uIn point to be detected set, explain that then its corresponding time domain correlation is greater than thresholding B), then can be with m_maxl deletion from some set to be detected, if m_maxl, m_maxl-d _uAnd/or m_maxl+d _uIn some set to be detected, also can be with m_max-d _uAnd/or m_max+d _uDeletion from some set to be detected.

Certainly, can also be with interval [m_maxl-d _u-X, m_maxl-d _u+ X] and [m_maxl+d _u-X, m_maxl+d _u+ X] in the deletion from point to be detected set of other value.

Generally speaking, in current detection point, find other test point that surpasses the peak value detection threshold that belongs to same cyclic shift search window with m_max: m [1] ..., behind the m [K], can also operate as follows:

If m7 [k]>m8 [k], then will be more than or equal to m7 [k] and smaller or equal to the point of N, and/or, more than or equal to 1 and smaller or equal to the point of m8 [k] as non-test point; Otherwise, will more than or equal to m7 [k] and smaller or equal to the point of m8 [k] as non-test point; And/or

If m9 [k]>m10 [k], then will be more than or equal to m9 [k] and smaller or equal to the point of N, and/or, more than or equal to 1 and smaller or equal to the point of m10 [k] as non-test point; Otherwise, will more than or equal to m9 [k] and smaller or equal to the point of m10 [k] as non-test point;

Wherein, m7 [k]=(N+m [k]-d _u-X) mod N, m8 [k]=(N+m [k]-d _u+ X) mod N, m7 [k] and m8 [k] they are positive integer;

M9 [k]=(N+m [k]+d _u-X) mod N, m10 [k]=(m [k]+d _u+ X) mod N, m9 [k] and m10 [k] are positive integer.

(2) the foregoing description with single reception antenna be example invention has been described, the present invention is equally applicable to the situation of multiple receive antenna, only needs that the pairing N of each a reception antenna time domain correlation is carried out the peak value detection after the addition respectively and gets final product.

(3) used point to be detected set among the embodiment, treated each point in the test point set and carry out peak value and detect, each non-test point or some deletion from some set to be detected of having detected; Can adopt alternate manner when specifically realizing obviously, for example, point that detected or non-test point identified, to avoid duplicate detection or the false-alarm point is detected.

Claims (10)

1. the signal detecting method of a RACH is characterized in that, this method comprises:

A: after the base station obtains a u root sequence of N time domain correlation R (m), with N pairing all m values of time domain correlation as test point, m=1 wherein, 2 ..., N;

B: in current all test points; Find the maximum pairing test point m_max of time domain correlation R (m_max); As the time domain correlation R (m_max) of said maximum during greater than the peak value detection threshold; Judgement has detected the lead code of RACH in the pairing cyclic shift search window of m_max, and with the some m1 in m_max and the current detection point and/or some m2 as non-test point; Wherein, m1=(N+m_max-d _u) mod N, m2=(m_max+d _u) mod N, m1 and m2 are positive integers;

C: if remain test point in addition, and the time domain correlation of the maximum in the residue test point then jumps to step B greater than the peak value detection threshold; Otherwise finish the detection of current root sequence;

Wherein, N>=N _ZC, N _ZCLength for the root sequence; d _uBe the starting point of cyclic shift search window and the distance between the copy window starting point.

2. the method for claim 1 is characterized in that,

In step B, find said m_max after, also operate as follows:

If m3＞m4 then will be more than or equal to m3 and smaller or equal to the point of N, and/or, more than or equal to 1 and smaller or equal to the point of m4 as non-test point; Otherwise, will more than or equal to m3 and smaller or equal to the point of m4 as non-test point; And/or

If m5＞m6 then will be more than or equal to m5 and smaller or equal to the point of N, and/or, more than or equal to 1 and smaller or equal to the point of m6 as non-test point; Otherwise, will more than or equal to m5 and smaller or equal to the point of m6 as non-test point;

Wherein, m3=(N+m_max-d _u-X) mod N, m4=(N+m_max-d _u+ X) modN, m3 and m4 are positive integer;

M5=(N+m_max+d _u-X) mod N, m6=(m_max+d _u+ X) mod N, m5 and m6 are positive integer;

X=int (N/ (N _ZC* 2)), int representes to round up or down.

3. the method for claim 1 is characterized in that,

In step B, find said m_max after, in current detection point, find other test point that surpasses the peak value detection threshold that belongs to same cyclic shift search window with said m_max: m [1] ..., m [K];

With the m in the current detection point [k]-d _uAnd/or m [k]+d _uBe designated non-test point; 1≤k≤K.

4. method as claimed in claim 3 is characterized in that,

In step B, find said test point: m [1] ..., behind the m [K], also operate as follows:

If m7 [k]＞m8 [k] then will be more than or equal to m7 [k] and smaller or equal to the point of N, and/or, more than or equal to 1 and smaller or equal to the point of m8 [k] as non-test point; Otherwise, will more than or equal to m7 [k] and smaller or equal to the point of m8 [k] as non-test point; And/or

If m9 [k]＞m10 [k] then will be more than or equal to m9 [k] and smaller or equal to the point of N, and/or, more than or equal to 1 and smaller or equal to the point of m10 [k] as non-test point; Otherwise, will more than or equal to m9 [k] and smaller or equal to the point of m10 [k] as non-test point;

Wherein, m7 [k]=(N+m [k]-d _u-X) mod N, m8 [k]=(N+m [k]-d _u+ X) modN, m7 [k] and m8 [k] they are positive integer;

M9 [k]=(N+m [k]+d _u-X) mod N, m10 [k]=(m [k]+d _u+ X) mod N, m9 [k] and m10 [k] they are positive integer;

X=int (N/ (N _ZC* 2)), int representes to round up or down.

5. like claim 1 or 3 or 4 described methods, it is characterized in that,

In step B, find said m_max after, also all test points that belong to same cyclic shift search window with said m_max in the current detection point are designated non-test point.

6. the method for claim 1 is characterized in that,

Said d _uMore than or equal to 2 and less than N _ZC/ 2, and work as N=N _ZCThe time, said d _uSatisfy following equality: (ud _u) mod N _ZC=± 1.

7. the method for claim 1 is characterized in that,

As N＞N _ZCThe time, d _u=int (d ' _u* N/N _ZC), and d ' _uSatisfy following equality: (ud ' _u) mod N _ZC=± 1, d ' _uMore than or equal to 2 and less than N _ZC/ 2;

Wherein, int () expression rounds up or rounds downwards.

8. the method for claim 1 is characterized in that,

Adopt following steps to confirm said peak value detection threshold:

Carry out Noise Estimation to received signal, obtain the noise average; With the product of above-mentioned noise average and predefined relative threshold value as the peak value detection threshold;

Said relative threshold value reaches the false-alarm signal of target false alarm rate and the ratio of noise for when not having signal to send.

9. method as claimed in claim 8 is characterized in that,

Adopt following steps to confirm said noise average:

With the maximum time domain correlation that receives signal and Noise Estimation ratio as thresholding A; With in said N the time domain correlation less than the average of the time domain correlation of thresholding A as said noise average;

Said Noise Estimation ratio is less than 1 and greater than 0.

10. the method for claim 1 is characterized in that, said sequence is the ZC sequence.

Images