Analysis method for multi-user random access signals

Abstract

An analysis method for multi-user random access signals is disclosed, which solves the connection problem between a base station and a user equipment when the narrow-band IoT uplink signal transmission is implemented. The analysis method of the present invention utilizes the detection threshold value to effectively determine multiple user equipment to be signaled; then, for each detected user equipment, an effective method based on the phase difference is used to estimate their synchronization parameters, i.e. time-of-arrival (ToA) and residual carrier frequency offset (RCFO). Therefore, according to the present invention, random access signals can be received correctly and efficiently and the user equipment related information can be obtained at the same time to facilitate subsequent communications.

Claims

1. An analysis method for multi-user random access signals, which applies to an uplink system of a narrowband Internet of thing (NB-IoT), and which comprises steps: receiving preamble signals of random access signals from a plurality of user equipment, detecting a plurality of symbol groups of each said preamble signal and acquiring corresponding average power, and comparing said average power with a detection threshold value to determine said user equipment intending to access signals; and acquiring a phase trace of said preamble signal corresponding to each of said user equipment intending to access signals, and calculating parameters of Time of Arrival (ToA) and Residual Carrier Frequency Offset (RCFO) according to phase differences of adjacent said symbol groups of said phase trace, wherein said detection threshold value is a Neyman Pearson threshold value, at least one base station collects bits generated by post-FFT (Fast Fourier Transform) of said symbol groups and detects sufficient statistics of said user equipment to determine said average power, and while said average power is greater than said detection threshold value, said user equipment corresponding to said average power intends to access said base station; while said average power is lower than said detection threshold value, said user equipment corresponding to said average power does not intend to access said base station.

2. The analysis method for multi-user random access signals according to claim 1, wherein said threshold value is determined using a false alarm level and a decision delay of said random access signals, in cooperation with a detected noise power; said decision delay is a count of all said symbol groups of said random access signal.

3. The analysis method for multi-user random access signals according to claim 1, wherein said preamble signals includes 4 symbol groups.

4. The analysis method for multi-user random access signals according to claim 3, wherein said step of calculating parameters of Time of Arrival (ToA) and Residual Carrier Frequency Offset (RCFO) according to said phase differences of adjacent said symbol groups of said phase trace further includes steps: calculating phase differences caused by channel hopping of said symbol groups; averaging all said phase differences corresponding to all said preamble signals to obtain said parameter of Residual Carrier Frequency Offset (RCFO); calculating average phase of each said symbol group to obtain an average phase difference corresponding to each said symbol group; and summing up said average phase differences to obtain said parameter of Time of Arrival (ToA).

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a diagram schematically showing 48 possible channel hopping groups generated according to the 3GPP R13 NPRACH standard;

(2) FIG. 2 is a diagram schematically showing a multi-user system applied to NB-IoT according to one embodiment of the present invention;

(3) FIG. 3 is a flowchart of an analysis method according to one embodiment of the present invention;

(4) FIG. 4 is a diagram showing the relationship between the scale factor A(N; =0.1%) and the decision delay N;

(5) FIG. 5 is a diagram showing theoretical curves of the relationship of P.sub.D,AWGN to SNR.sub.i for different decision delays at =0.1% according to one embodiment of the present invention; and

(6) FIG. 6 is a diagram showing theoretical curves of the relationship of P.sub.D,Fading to SNR.sub.i for different decision delays at =0.1% according to one embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

(7) Refer to FIG. 2, in the random access process of the NB-IoT uplink system, the first piece of uplink signal of the user equipment 12 is called NPRACH, which uses a single-tone frequency hopping preamble sequence. The time unit of frequency hopping is a symbol group. According to the cellular coverage and the distance between the base station and the user equipment, a base station (eNB) 10 specifies the length of the preamble sequence, which is to be sent out by the user equipment 12, in the information block of the downlink system thereof. A detection threshold value is set in the NPRACH receiver algorithm to achieve the false alarm probability and the detection probability, which are regulated by the NB-IoT standard, and to estimate the parameters of ToA and RCFO.

(8) The present invention proposes an analysis method for multi-user random access signals, which is extensively applicable to the NB-IoT uplink system. Refer to FIG. 2 and FIG. 3, the analysis method of the present invention comprises Steps 10-30. In Step S10, the base station 10 receives the preamble signals of the random access signals from a plurality of user equipment 12, detects a plurality of symbol groups of each preamble signal, and acquires the corresponding average energies. In other words, the base station 10 collects the bits generated by the post-FFT of all the symbol groups and detects the sufficient statistics of the user equipment to determine the average power.

(9) In Step S20, determine whether there is any user equipment 12 intending to access the base station 10 according to the corresponding average power and a detection threshold value. In Step S22, the average power is lower than the threshold value, and the corresponding user equipment 12 does not intend to access the base station 10; then, the process would not proceed to the next step but ends herein. In Step S24, the average power is higher than the threshold value, and the corresponding user equipment 12 intends to access the base station 10; then the process proceeds to Step S30. The abovementioned detection threshold value is a Neyman Pearson threshold value. The Neyman Pearson threshold value is determined using a false alarm level and a decision delay, which are obtained beforehand from the random access signal, in cooperation with a detected noise power. The decision delay is the number of all the symbol groups of the random access signal.

(10) In Step S30, while having detected the user equipment 12 intending to access the base station 10, the base station 10 acquires the phase trace of the preamble signal of the corresponding user equipment 12 and calculates the ToA parameter and the RCFO parameter of the user equipment 12 according to the phase difference of the adjacent symbol groups of the phase trace. In Step S30, it is according to the phase trace that the ToA parameters and the RCFO parameters are sequentially estimated. In other words, the phase difference induced by channel hopping of the symbol groups is calculated firstly, and all phase differences corresponding to all preamble signals are averaged, whereby to acquire the RCFO parameter corresponding to the user equipment; next, the average phase of each symbol group is calculated to acquire the average phase difference corresponding to each symbol group; the average phase differences are summed up to obtain the related ToA parameter.

(11) After the technical characteristics of the present invention have been described above, the principles of the present invention will be described thereinafter to prove that the analysis method for multi-user random access signals is practicable and easy to practice.

(12) As there are 48 distinct and possible hopping groups, the test data Q.sub.k of each user equipment is used to compensate for the bits of the corresponding post-FFT. The test data Q.sub.k of UE.sub.k is defined as
Q.sub.k={R(n,i,f.sub.k(n)),n=0, . . . ,N1,i=0,1, . . . 4}
wherein N=4P and means the number of all the symbol groups; P is the number of the preamble signals for detection; N is also regarded as the decision delay in the user equipment detection process. For different coverage class, the base station eNB may designate P to be 1, 2, 4, 8, 16, 32, 64, and 128, which are respectively corresponding to the decision delays N of 4, 8, 16, 32, 64, 128, 256, and 512. Therefore, if the user equipment UE.sub.k has a lower receiving SNR, it is necessary to collect more data, i.e. obtain larger N, leading to a longer decision delay for the farther or weaker user equipment.

(13) As the hopping groups are orthogonal to each other, the superimposed NPRACH detection problems can be decoupled into a parallel of single-UE detection problem. In order to test the presence of UE.sub.k, the energy of the bits of the post-FFT of N symbol groups is collected; then, the sufficient statistics of detecting the user equipment is used to determine the average power:

(14) $\begin{array}{cc}{P}_k\left(N\right)=\frac{1}{5N}\underset{n=0}{\overset{N-1}{.Math.}}\underset{i=0}{\overset{4}{.Math.}}{.Math.R\left(n,i,f_k\left(n\right)\right).Math.}^2& \left(7\right)\end{array}$

(15) Next, the decision rule used to determine the presence of UE.sub.k can be simplified comparing P.sub.k(N) with a threshold value , which may be expressed by

(16) $\begin{array}{cc}{P}_k\left(N\right)\stackrel{{\mathrm{UE}}_k\mathrm{present}}{\underset{{\mathrm{UE}}_k\mathrm{abscent}}{}}& \left(8\right)\end{array}$

(17) The receiver operation characteristic (ROC) or the performance test can be expressed by the false alarm probability P.sub.F
P.sub.D=Pr(P.sub.k(N)>/UE.sub.k is present)(9)

(18) The detection probability P.sub.D is expressed by
P.sub.D=Pr(P.sub.k(N)>/UE.sub.k is present)(10)

(19) Therefore, the main problem of NPRACH detection is to specify the threshold value . The NB-IoT standard demands that the detection probability P.sub.D must exceed 99% and the false alarm probability P.sub.F should not exceed 0.1%. In order to achieve the two conditions, the present invention uses the Neyman Pearson rule to resolve the detection problem, which is expressed by
max{P.sub.D}, such that P.sub.F(11)

(20) It means that the decision rule is most powerful at the significant level for the threshold value . In the case of NB-IoT, it may be selected that =0.1%. According to the Neyman Pearson rule, the threshold value is the function of the decision delay N and has two parameters of a specified false alarm level and a noise power P.sub.n=L.sub.n.sup.2. Suppose that the receiver can detect the noise power P.sub.n. Thus, the threshold value must satisfy the equation:
.sub..sup.g.sub.n(x)dx=(12)
wherein g.sub.n(x) is a probability density function (PDF) of test statistics P.sub.k(N) under the noise-only case. Since P.sub.k(N) is the average power of 10N independent real Gaussian random variables (RVs) with identical distribution N(0, P.sub.n/2). Therefore, PDF g.sub.n(x) is determined by a scaled central Chi-square distribution with 10N degrees of freedom as follows:
g.sub.n(x)=f.sub.c(x;10N)(13)
wherein

(21) $\begin{array}{cc}{f}_c\left(x;m\right)=\frac{1}{2^{k/2}\left(k/2\right)}{x}^{\frac{k}{2}-1}{e}^{-\frac{x}{2}}u\left(x\right)& \left(14\right)\end{array}$
which is a standard .sub.m.sup.2 PDF with in degrees of freedom, wherein =(10N)/(L.sub..sup.2)=10N/P.sub.n is a scale factor. The cumulative distribution function (CDF) of the standard .sub.m.sup.2 PDF is determined by

(22) $\begin{array}{cc}{F}_c\left(x;m\right)={}_0^x{f}_c\left(;m\right)d=\frac{1}{\left(\frac{k}{2}\right)}\left(\frac{k}{2},\frac{x}{2}\right)u\left(x\right)& \left(15\right)\end{array}$
wherein (s,t)=.sub.0.sup.xt.sup.s1e.sup.=tdt is a lower incomplete Gamma function.

(23) It is learned from Equation (11): the false alarm level and the decision delay N can be used to definitely determine the optimized Neyman Pearson threshold value .sub.o with the equation:

(24) $\begin{array}{cc}\begin{array}{c}{}_o=\frac{1}{}{F}_c^{-1}\left(1-;10N\right)\\ =P_n\left(\frac{1}{10N}{F}_c^{-1}\left(1-;10N\right)\right)\\ =P_{n}A\left(N;\right)\end{array}& \left(16\right)\end{array}$
wherein F.sub.x.sup.1(x; m) is the inverse function of .sub.m.sup.2 CDF, i.e. F.sub.c.sup.1(F.sub.c(x; m); m)=x.

(25) From Equation (16), it is noted to determine the threshold for a given and N. The average noise power P.sub.n of post-FFT can be precisely measured using those noise-only resource grids. The threshold value is obtained via multiplying P.sub.n by the scale factor A(N; ). FIG. 4 shows the relationship of the scale factor A(N; ) and the decision delay N. The scale factor is always greater than one, but it decreases with the increase of N.

(26) The 48 average energies of the channel hopping mode along different user equipment can be worked out using only 48 values of 512-point FFT. The scale factor A(N; =0.1%) can be calculated beforehand and picked up from the memory for application. Therefore, the present invention can realize NPRACH detection in very low complexity and very low power consumption.

(27) Next is deduced the detection performance in the AWGN channel and the Rayleigh fading channel. For the AWGN channel, suppose that the composite fading coefficient h.sub.k=1 and that R(n,i,.sub.k (n)) has a non-zero average value Le.sup.jk,n,i. In the AWGN channel, the signal power of post-FFT is P.sub.0=L.sup.2, which is nonrandom. With the signal and noise, a scaled non-central Chi-square distribution PDF may be used to express the statistical data P.sub.k(N) as follows:
g.sub.s(x)=f.sub.nc(x;10N,.sub.0)(17) wherein k is the same as the scale factor in Equation (13).

(28) $\begin{array}{cc}{f}_{\mathrm{nc}}\left(x;m,{}_0\right)=\frac{1}{2}{e^{-\left(x+\right)/2}\left(\frac{x}{{}_0}\right)}^{k/4-1/2}{I}_{k/2-1}\left(\sqrt{{}_{0}x}\right)u\left(x\right)& \left(18\right)\end{array}$
Equation (18) is a standard non-central Chi-square distribution PDF with m degrees of freedom, wherein the non-central parameter cis given by the following equation:
.sub.0=10NP.sub.0/P.sub.n=10NL/.sub.n.sup.2(19) l.sub.v(x) is the modified Bessel function of the first kind with degree . CDF of f.sub.nc(x; m, .sub.0) is expressed by

(29) $\begin{array}{cc}{F}_{\mathrm{nc}}\left(x;m\right)={}_0^x{f}_{\mathrm{nc}}\left(;m,{}_0\right)d=1-Q_{\frac{k}{2}}\left(\sqrt{{}_0},\sqrt{x}\right)& \left(20\right)\end{array}$
wherein

(30) 0$\begin{array}{cc}{Q}_M\left(a,b\right)={}_b^{}{x\left(\frac{x}{a}\right)}^{M-1}\exp \left(-\frac{x^2+a^2}{2}\right)I_{M-1}\left(\mathrm{ax}\right)\mathrm{dx}& \left(21\right)\end{array}$
Equation (21) is a generalized Marcum Q-function; a and b are the parameters substituted into the equation for computation.

(31) After the Neyman Pearson threshold value .sub.0 is acquired, the theoretical detection probability P.sub.D under the AWGN channel may be derived as
P.sub.D,AWGN(.sub.0)=Q.sub.5N({square root over (.sub.0)},{square root over (10N.sub.o/P.sub.n)})(22)

(32) P.sub.D,AWGN(.sub.0) may be used to calculate the detection probability under the Rayleigh fading channel Letting .sub.k.sup.2=1 without loss of generality, then =|h.sub.k|.sup.2 is an exponential redundancy version (RV) with unity mean. The average signal power of post-FFT P.sub.s=E[P.sub.0]=P.sub.0 is unchanged and the non-central parameter becomes =.sub.0. Therefore, the detection probability under the Rayleigh fading channel may be obtained by averaging P.sub.D,AWGN(.sub.0) over the PDF of and may be expressed by the following implicit integral expression:
P.sub.D,Fading(.sub.0)=.sub.0.sup.e.sup.Q.sub.5N({square root over (.sub.0)},{square root over (10N.sub.o/P.sub.n)})d(23)
Equation (23) may have a complicated closed-form formula. However, the present invention can use numerical integration to calculate the theorectical detection performance under the Rayleigh fading channel.

(33) Refer to FIG. 5 and FIG. 6 respectively showing the theorectical curves of the relationships of P.sub.D,AWGN and P.sub.D,Fading to SNR.sub.i for different decision delays at =0.1%. In order to verify the analyses, the performance curves is simulated with the Monte-Carlo method. The results show that the simulation curves almost coincide with the theorectical curves. It is observed in these curves: a longer preamble signal, i.e. a larger N, can significantly improve the UE detection performance of eNB, especially while SNR is low. In other words, eNB can use a longer NPRACH configuration to extend the coverage thereof.

(34) After NPRACH detection has been discussed above, the present invention proposes an algorithm that can effectively estimate the synchronization parameters of ToA and RCFO. The algorithm integrates the abovementioned detection methods, addressing only the detected user equipment and using the decoupling method described below to estimate the parameters of ToA and RCFO.

(35) For the detected user equipment UE.sub.k, the algorithm starts from its unwrapped phase trace of R(n,i, f.sub.k(n)) expressed as:
q.sub.k,n,i=unwrap{arg{R(n,i,f.sub.k(n))}}(24)
In the noise-free case, the phase trace can be obtained using q.sub.k,n,i=.sub.k,n,i=2D.sub.kf.sub.k(n)2.sub.k(n,i)+C in Equation (4), wherein C is a constant phase. As the first two phase terms are directly related to RCFO and ToA parameters, the phase trace, together with suitable phase differences, can be used to estimate the two synchronization parameters. Further, average operation can be used to decrease the estimation variance caused by AWGN.

(36) For each symbol group of 5 symbols, there are four adjacent phase differences that can be calculated:
.sub.k,n,i=q.sub.k,n,i+1q.sub.k,n,i for i=1,2,3,4,(25)
Next, by averaging .sub.k,n,i, over all n and i in the preamble, it can obtain the RCFO estimate for UE.sub.k, as shown by the equation:

(37) $\begin{array}{cc}{\hat{}}_k=\frac{1}{2}\frac{1}{4N}\underset{n=0}{\overset{N-1}{.Math.}}\underset{i=1}{\overset{4}{.Math.}}{\mathrm{.Math.}}_{k,n,i}& \left(26\right)\end{array}$
Next, in order to estimate ToA parameter, the derivative terms of RCFO is removed from q.sub.k,n,i to obtain Equation (27):
q.sub.k,n,i=q.sub.k,n,i2{circumflex over ()}.sub.kft.sub.n,i(27)
wherein t.sub.n,i=[(5n+i)L+(n+1)L.sub.cp]T.sub.s is the starting time instant of the ith symbol of the nth symbol group. The average phase of each symbol group q.sub.k,n,i may be expressed by

(38) $\begin{array}{cc}{z}_{k,n}=\frac{1}{5}\underset{i=0}{\overset{4}{.Math.}}{\stackrel{\_}{q}}_{k,n,i}& \left(28\right)\end{array}$
Then, the sum of the differences of z.sub.k,n is divided by the sum of the differences of channel hopping to obtain ToA, as shown by the equation:

(39) $\begin{array}{cc}{\hat{D}}_k=\frac{{}_k}{2T_s}\frac{\underset{n=1}{\overset{N-1}{.Math.}}.Math.z_{k,n+1}-z_{k,n}.Math.}{f\underset{n=1}{\overset{N-1}{.Math.}}.Math.f_k\left(n+1\right)-f_k\left(n\right).Math.}& \left(29\right)\end{array}$
wherein the sign of ToA may be expressed by

(40) $\begin{array}{cc}{}_k=\mathrm{sign}\left(\frac{-\underset{n=1}{\overset{N-1}{.Math.}}\left(z_{k,n+1}-z_{k,n}\right)}{\underset{n=1}{\overset{N-1}{.Math.}}\left(f_k\left(n+1\right)-f_k\left(n\right)\right)}\right)=+1\mathrm{or}-1& \left(30\right)\end{array}$

(41) Hence, for each detected UE, the above joint RCFO/ToA synchronization algorithm is straight forward, easily implemented, and computationally efficient.

(42) In conclusion, the present invention proposes an analysis method for multi-user random access signals, which can solve the linking problem between the base stations and the user equipment, and which can make the overall detection performance achieve a false alarm probability P.sub.F0.1% and a detection probability P.sub.D>99%, and which can accurately and efficiently detect each user equipment and estimate the synchronization parameters of ToA and RCFO to facilitate subsequent communications. Therefore, the present invention is a high-precision and low-computation burden analysis method with a definite threshold value.

(43) The embodiments are described above to demonstrate the technical contents and characteristics of the present invention to enable the persons skilled in the art to understand, make, and use the present invention. However, these embodiments are only to exemplify the present invention but not to limit the scope of the present invention. Any equivalent modification or variation according to the spirit of the present invention is to be also included by the scope of the present invention.