CN107231326B - A Cell Search System in Downlink of NB-IoT System - Google Patents

Abstract

The invention discloses a cell searching method in an NB-IoT system downlink, which uses the following technical means, npss signal detection adopts a method of combining segment correlation and module value correlation, adopts the segment correlation to carry out timing coarse synchronization, and then adopts the module value correlation to carry out timing coarse synchronization correction and timing fine synchronization. When the local npss signal is stored, only the npss time domain down-sampled signal and the frequency domain signal of one symbol are needed to be stored, and the space complexity is reduced. When frequency offset estimation is carried out, the frequency offset of a certain step length multiplied by a local npss signal is subjected to correlation operation with a received master synchronization signal, so that rough frequency offset estimation is obtained and compensated, and the frequency offset of the signal is controlled within a certain step length; when the fine frequency offset estimation is carried out, symbol is crossed, the fine frequency offset estimation range can be reduced to a certain range, and the frequency offset estimation precision is improved. When performing nsss signal detection, before using the received nsss frequency domain signal to correlate with the local nsss frequency domain signal, determining n_fAnd the candidate value of q, and then the received nss frequency domain signal is correlated with the local nss frequency domain signal in the candidate value, so that the time complexity is reduced.

Description

Translated fromChinese

一种NB-IoT系统下行链路中的小区搜索系统A Cell Search System in Downlink of NB-IoT System

技术领域technical field

本发明涉及一种NB-IoT系统下行链路中的小区搜索系统，主要涉及专利分类号H04电通信技术H04B传输H04B1/00不包含在H04B 3/00至H04B 13/00单个组中的传输系统的部件；不以所使用的传输媒介为特征区分的传输系统的部件H04B1/69扩频技术H04B1/707利用直接序列调制的H04B1/7073同步方面H04B1/7083小区搜寻，例如使用三级方法。The present invention relates to a cell search system in the downlink of an NB-IoT system, and mainly relates to a transmission system not included in a single group ofH04B 3/00 to H04B 13/00, patent classification number H04 Electrical communication technology H04B transmission H04B1/00 Component of the transmission system not distinguished by the transmission medium used H04B1/69 Spread spectrum technique H04B1/707 H04B1/707 Synchronization aspect using direct sequence modulation H04B1/7083 Cell search, eg using a three-level approach.

背景技术Background technique

NB-IoT系统中，初始小区搜索过程是用户终端与基站之间建立下行通信链路的前提，其主要目的是实现下行链路在时间、频率上的同步和获取物理层小区标识(cellID)。NB-IoT系统下行采用正交频分复用(OFDM)技术，该技术的基本原理是将高速串行数据流经过串并转换后变成多路并行低速数据流，并调制到多个正交的子载波上传输，因为其要求子载波间满足正交性，因此与单载波系统相比，OFDM系统易受频率偏差的影响。由于发射机的载波频率与接收机本地振荡器间存在频率偏差，或由于多普勒效应，OFDM系统的子载波间的正交性会遭到破坏，从而导致子载波间干扰，所以频率同步要尽可能精确。时间同步是为了确定系统帧的起始位置和OFDM符号的起始位置，在多经信道中，时间同步不准确会造成符号间干扰，对系统性能影响很大，因此要求时间同步尽可能精确。In the NB-IoT system, the initial cell search process is the premise of establishing a downlink communication link between the user terminal and the base station. Its main purpose is to synchronize the downlink in time and frequency and to obtain the physical layer cell ID (cellID). The downlink of the NB-IoT system adopts Orthogonal Frequency Division Multiplexing (OFDM) technology. The basic principle of this technology is to convert high-speed serial data streams into multiple parallel low-speed data streams after serial-to-parallel conversion, and modulate multiple Compared with the single-carrier system, the OFDM system is susceptible to frequency deviation because it requires orthogonality between the sub-carriers. Due to the frequency deviation between the carrier frequency of the transmitter and the local oscillator of the receiver, or due to the Doppler effect, the orthogonality between the sub-carriers of the OFDM system will be destroyed, resulting in inter-sub-carrier interference. Therefore, frequency synchronization requires Be as precise as possible. Time synchronization is to determine the starting position of the system frame and the starting position of the OFDM symbol. In a multi-pass channel, inaccurate time synchronization will cause inter-symbol interference, which has a great impact on system performance. Therefore, the time synchronization is required to be as accurate as possible.

在现有技术中典型的NB-IoT系统采样时间T_s＝1/30720000秒，其时域结构如图1所示，每个系统帧时间为10毫秒(ms)，包含10个子帧，每个子帧时间为1ms，包含2个时隙，每个时隙时间为0.5ms，包含7个正交频分复用符号(symbol)，每个符号长度为2048个采样点并加上CP的长度，其中第0个symbol的CP长度为160个采样点，第1到第6个symbol的CP长度为144个采样点，所以每一个子帧包含14个symbol，长度为30720个采样点。npss信号位于系统帧的第5个子帧，nsss信号位于偶位系统帧的第9个子帧。In the prior art, a typical NB-IoT system sampling time T_s = 1/30720000 second, its time domain structure is shown in Figure 1, each system frame time is 10 milliseconds (ms), including 10 subframes, each subframe The frame time is 1ms, including 2 time slots, each time slot is 0.5ms, including 7 orthogonal frequency division multiplexing symbols (symbol), each symbol length is 2048 sampling points and plus the length of the CP, The CP length of the 0th symbol is 160 samples, and the CP length of the 1st to 6th symbols is 144 samples, so each subframe contains 14 symbols and the length is 30720 samples. The npss signal is located in the 5th subframe of the system frame, and the nsss signal is located in the 9th subframe of the even-bit system frame.

发明内容SUMMARY OF THE INVENTION

针对现有npss信号检测技术在频偏较大情况下，时间同步不准的问题，本发明提出了一种可以抵抗较大频偏的npss信号检测系统，首先采用分段相关进行定时粗同步，然后采用模值相关的方法进行定时粗同步修正和定时细同步，模值相关方法的引入使得定时同步的抗频偏能力大大提高。Aiming at the problem of inaccurate time synchronization in the existing npss signal detection technology when the frequency offset is large, the present invention proposes an npss signal detection system that can resist the large frequency offset. Then, the modulo value correlation method is used to correct the timing coarse synchronization and the timing fine synchronization. The introduction of the modulo value correlation method greatly improves the anti-frequency offset capability of the timing synchronization.

在储存本地npss信号时，本发明只需储存一个symbol的npss时域降采样信号和频域信号即可，而不需要将整个子帧的npss时域降采样信号和频域信号全部储存至本地，空间复杂度降低。When storing the local npss signal, the present invention only needs to store the npss time-domain down-sampling signal and the frequency-domain signal of one symbol, and does not need to store all the npss time-domain down-sampling signal and the frequency domain signal of the whole subframe to the local , the space complexity is reduced.

进行频偏估计时，首先将本地npss信号乘以一定步长的频偏与接收到的主同步信号进行相关运算，求得粗频偏估计，粗频偏补偿后可以保证信号的频偏控制在一定步长以内；进行精频偏估计时，不是将一个symbol内的npss信号与本地信号共轭相乘后分为两段做相关，而是跨symbol进行，这样可以将精频偏估计范围减小至一定范围内，既满足了频偏估计的范围要求又最大限度的提高了精频偏估计的精度。When performing frequency offset estimation, first multiply the local npss signal by the frequency offset of a certain step size and perform a correlation operation with the received primary synchronization signal to obtain a rough frequency offset estimate. After the rough frequency offset compensation, the frequency offset of the signal can be controlled within Within a certain step size; when performing the fine frequency offset estimation, instead of multiplying the npss signal in a symbol with the local signal conjugate and dividing it into two sections for correlation, it is performed across symbols, which can reduce the fine frequency offset estimation range. As small as a certain range, it not only meets the range requirements of frequency offset estimation, but also maximizes the precision of precise frequency offset estimation.

进行nsss信号检测时，根据ZC序列的中心对称性，可在接收的nsss频域信号与本地nsss频域信号相关之前，先确定出n_f和q的候选值，然后使用接收到的nsss频域信号与候选值内的本地nsss频域信号做相关，而无需与所有的本地nsss频域信号做相关，减小了时间复杂度。When performing nsss signal detection, according to the center symmetry of the ZC sequence, the candidate values of n_f and q can be determined before the received nsss frequency domain signal is correlated with the local nsss frequency domain signal, and then the received nsss frequency domain can be used. The signal is correlated with the local nsss frequency domain signal in the candidate value, and does not need to be correlated with all local nsss frequency domain signals, which reduces the time complexity.

附图说明Description of drawings

为了更清楚的说明本发明的实施例或现有技术的技术方案，下面将对实施例或现有技术描述中所需要使用的附图做一简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。In order to illustrate the embodiments of the present invention or the technical solutions of the prior art more clearly, the following briefly introduces the accompanying drawings required in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description These are just some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can also be obtained from these drawings without any creative effort.

图1为本发明NB-IoT时域结构示意图FIG. 1 is a schematic diagram of the NB-IoT time domain structure of the present invention

图2为本发明NB-IoT系统小区搜索功能模块图FIG. 2 is a block diagram of the cell search function of the NB-IoT system of the present invention

图3为本发明npss信号检测功能模块图FIG. 3 is a functional block diagram of npss signal detection according to the present invention

图4为本发明频率同步功能模块图Fig. 4 is the frequency synchronization function module diagram of the present invention

图5为本发明nsss信号检测功能模块图FIG. 5 is a functional block diagram of the nsss signal detection of the present invention

图6为本发明n_f和cellID检测流程图Fig. 6 is the flow chart of n_f and cellID detection of the present invention

图7为本发明时偏估计对比图Fig. 7 is the time offset estimation comparison diagram of the present invention

图8为本发明频偏估计对比图FIG. 8 is a comparison diagram of frequency offset estimation according to the present invention

具体实施方式Detailed ways

为使本发明的实施例的目的、技术方案和优点更加清楚，下面结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚完整的描述：In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention:

本发明使用的技术缩略语如下：The technical abbreviations used in the present invention are as follows:

NB-IoT Narrow Band Internet of Things窄带物联网NB-IoT Narrow Band Internet of Things

npss Narrowband primary synchronization signal主同步信号npss Narrowband primary synchronization signal

nsss Narrowband secondary synchronization signal辅同步信号nsss Narrowband secondary synchronization signal

CP Cyclic Prefix循环前缀CP Cyclic Prefix

IFFT Inverse Fast Fourier Transformation快速傅里叶反变换IFFT Inverse Fast Fourier Transformation

OFDM Orthogonal Frequency Division Multiplexing正交频分复用OFDM Orthogonal Frequency Division Multiplexing

cellID物理层小区标识cellID physical layer cell identifier

n_f系统帧号n_f system frame number

symbol正交频分复用符号symbol OFDM symbol

ε归一化频偏，即频偏与子载波间隔(15kHz)的比值ε normalized frequency offset, that is, the ratio of frequency offset to subcarrier spacing (15kHz)

如图1-6所示，一种NB-IoT系统下行链路中的小区搜索系统主要包括：As shown in Figure 1-6, a cell search system in the downlink of an NB-IoT system mainly includes:

npss信号检测单元npss signal detection unit

npss信号检测功能框图如图3所示，首先生成本地npss频域信号，变换至时域并进行降采样处理，然后对接收的21ms数据进行降采样处理，使用降采样后的数据进行定时粗同步，定时粗同步修正后使用未降采样的数据进行定时细同步，完成npss信号的检测。The functional block diagram of npss signal detection is shown in Figure 3. First, the local npss frequency domain signal is generated, transformed to the time domain and down-sampling processing, and then down-sampling processing is performed on the received 21ms data, and the down-sampled data is used for timing coarse synchronization , after the timing coarse synchronization is corrected, the data that is not downsampled is used for timing fine synchronization, and the detection of the npss signal is completed.

npss信号在频域是一个Zadoff-Chu序列，其生成表达式为：The npss signal is a Zadoff-Chu sequence in the frequency domain, and its generation expression is:

其中，根序列索引u＝5，l为symbol号，S(l)取值如表1所示：Among them, the root sequence index u=5, l is the symbol number, and the value of S(l) is shown in Table 1:

表1S(l)取值表Table 1S(l) value table

可以发现，每个symbol的npss信号只相差一个符号位S(l)，为了减少空间复杂度，只需生成一个symbol的频域npss信号即可，其生成表达式为：It can be found that the npss signal of each symbol differs only by one sign bit S(l). In order to reduce the space complexity, it is only necessary to generate a frequency domain npss signal of a symbol, and its generation expression is:

然后，将npss_loc_fre(n)变换至时域，生成一个symbol的时域npss信号：Then, transform npss_loc_fre(n) to the time domain to generate a symbol's time domain npss signal:

为了减小计算量，对信号npss_loc_ti(n)进行16倍降采样处理，如式(4)所示，得到的信号记为npss_loc_ti_sam(n)，长度为128个采样点。In order to reduce the amount of calculation, the signal npss_loc_ti(n) is subjected to 16-fold down-sampling processing, as shown in equation (4), the obtained signal is recorded as npss_loc_ti_sam(n), and the length is 128 sampling points.

npss_loc_ti_sam(n)＝npss_loc_ti(16n),n＝0,1,...,127 (4)npss_loc_ti_sam(n)=npss_loc_ti(16n),n=0,1,...,127 (4)

npss信号位于系统帧的第5个子帧，则在接收到的21ms数据中，必包含两个子帧的npss信号。The npss signal is located in the 5th subframe of the system frame, and the received 21ms data must contain the npss signal of two subframes.

npss信号检测实际使用前11ms数据，记为re(n)，长度为337920个采样点，对信号re(n)进行16倍降采样处理，得到的信号记为re_sam(n)，长度为21120个采样点。The npss signal detection actually uses the first 11ms data, denoted as re(n), the length is 337920 sampling points, the signal re(n) is down-sampled by 16 times, and the obtained signal is denoted as re_sam(n), the length is 21120 Sampling point.

用npss_loc_ti_sam(n)和re_sam(n)做滑动相关，每滑动一个点，从re_sam(n)中取出一个子帧的数据，记为re_sam_sf(m)，m＝0,1,…,M-1，M＝1920，则re_sam_sf(m)＝re_sam(n+m)，n＝0,1,…,19200，将re_sam_sf(m)分为14个symbol，第l个symbol的数据(不包含CP)记为re_sam_l(n)，n＝0,1,…,127，则re_sam_l(n)＝re_sam_sf(137l+10+(l≥7)+n)，l＝0,1,…,13，其中(l≥7)表示一个值，当l<7时，(l≥7)＝0，当l≥7时，(l≥7)＝1。对于npss信号，前3个symbol信号为0，使用后11个symbol信号re_sam_l(n)与本地信号npss_loc_ti_sam(n)做分段相关(段数取2，每段长度64)，并求和，如式(5)所示：Use npss_loc_ti_sam(n) and re_sam(n) to do sliding correlation, each time a point is slid, the data of a subframe is taken out from re_sam(n), recorded as re_sam_sf(m), m=0,1,...,M-1 , M=1920, then re_sam_sf(m)=re_sam(n+m), n=0,1,...,19200, divide re_sam_sf(m) into 14 symbols, the data of the lth symbol (excluding CP) Denoted as re_sam_l (n), n=0,1,...,127, then re_sam_l (n)=re_sam_sf(137l+10+(l≥7)+n), l=0,1,...,13, Where (l≥7) represents a value, when l<7, (l≥7)=0, and when l≥7, (l≥7)=1. For the npss signal, the first 3 symbol signals are 0, and the last 11 symbol signals re_sam_l (n) are used for segmental correlation with the local signal npss_loc_ti_sam (n) (the number of segments is 2, and the length of each segment is 64), and summed, such as Formula (5) shows:

其中，n＝0,1,…,19200，即共滑动一帧的长度。求取corr(n)的最大值，其最大值对应的标号n即为定时粗同步位置，即

Among them, n=0,1,...,19200, that is, the length of one frame of sliding. Find the maximum value of corr(n), the label n corresponding to the maximum value is the timing coarse synchronization position, that is

因为频偏的影响，上述过程得到的定时粗同步有一定偏差，需要对其进行修正，在得到的粗同步位置coarse_start_time两侧各取20个点，用npss_loc_ti_sam(n)和re_sam(n)做滑动相关。Due to the influence of frequency offset, the timing coarse synchronization obtained in the above process has a certain deviation, which needs to be corrected. Take 20 points on both sides of the coarse synchronization position coarse_start_time obtained, and use npss_loc_ti_sam(n) and re_sam(n) for sliding related.

在使用信号npss_loc_ti_sam(n)与信号re_sam_l(n)做相关时，不再使用分段相关，而使用npss_loc_ti_sam(n)和re_sam_l(n)的模进行相关，以抵抗频偏对时偏的影响，如式(6)所示。When using the signal npss_loc_ti_sam(n) to correlate with the signal re_sam_l (n), the piecewise correlation is no longer used, but the modulus of npss_loc_ti_sam(n) and re_sam_l (n) is used for correlation to resist the frequency offset to the time offset. influence, as shown in Equation (6).

其中，n＝coarse_start_time-20,coarse_start_time-19,…,coarse_start_time+20。求取corr_rev(n)的最大值，其最大值对应的标号n即为定时粗同步的修订值，即

Wherein, n=coarse_start_time-20, coarse_start_time-19, ..., coarse_start_time+20. Find the maximum value of corr_rev(n), and the label n corresponding to the maximum value is the revised value of timing coarse synchronization, that is

以上求取时偏的过程都是采用的16倍降采样信号，为了得到定时细同步，在得到start_time_rev后，The above process of obtaining the time offset uses 16 times down-sampling signal. In order to obtain timing fine synchronization, after obtaining start_time_rev,

首先将其乘以16变换至未降采样数据的时偏估计，然后在其两侧各取64个点，在未降采样的数据re(n)上进行滑动，每滑动一个点，从re(n)中取出一个子帧的数据，记为re_sf(n)，n＝0,1,…,N-1，N＝30720，然后再进行16倍降采样，得到的信号记为re_sam_sf'(n)，n＝0,1,…,N'-1，N'＝1920，第l个symbol的数据记为re_sam'_l(n)，长度为128，对信号npss_loc_ti_sam(n)和re_sam'_l(n)的模进行相关，如式(7)所示：First, multiply it by 16 to transform it to the time offset estimate of the un-downsampled data, then take 64 points on each side of it, and slide it on the un-downsampled data re(n), each sliding a point, from re( Take out the data of one subframe from n), denoted as re_sf(n), n=0,1,...,N-1, N=30720, and then perform 16-fold downsampling, and the obtained signal is denoted as re_sam_sf'(n ), n=0,₁ ,_. The modulus of n) is correlated, as shown in formula (7):

其中，n＝start_time_rev*16-64,start_time_rev*16-15,…,start_time_rev*16+64。求取corr_acc(n)的最大值，其最大值对应的标号n即为定时细同步，即

该值为npss信号开始的位置。Wherein, n=start_time_rev*16-64, start_time_rev*16-15,...,start_time_rev*16+64. Find the maximum value of corr_acc(n), and the label n corresponding to the maximum value is the timing fine synchronization, that is

The value is where the npss signal starts.

频率同步单元frequency synchronization unit

频率同步功能框图如图4所示，根据start_time_acc可以得到接收到的npss信号，首先对信号进行16倍降采样，然后根据降采样的npss信号计算粗频偏估计，对信号进行粗频偏补偿，然后进行精频偏估计。The frequency synchronization function block diagram is shown in Figure 4. The received npss signal can be obtained according to start_time_acc. First, the signal is down-sampled by 16 times, and then the coarse frequency offset estimation is calculated according to the down-sampled npss signal, and the signal is subjected to coarse frequency offset compensation. Then perform precise frequency offset estimation.

将接收到的npss信号进行16倍降采样处理，记为npss_re_ti_sam(n)，则：The received npss signal is down-sampled by 16 times, recorded as npss_re_ti_sam(n), then:

npss_re_ti_sam(n)＝re(start_time_acc+16n),n＝0,1,...,N-1 (8)npss_re_ti_sam(n)=re(start_time_acc+16n),n=0,1,...,N-1 (8)

其中N＝1920，npss_re_ti_sam(n)为一个子帧的数据，将其分为14个symbol，第l个symbol的数据(不包含CP)记为npss_re_ti_sam_l(n)，n＝0,1,…,128-1，则npss_re_ti_sam_l(n)＝npss_re_ti_sam(137l+10+(l≥7)+n)，l＝0,1,…,13。Among them, N=1920, npss_re_ti_sam(n) is the data of one subframe, which is divided into 14 symbols, and the data of the l-th symbol (excluding CP) is recorded as npss_re_ti_sam_l (n), n=0,1,… , 128-1, then npss_re_ti_sam_l (n)=npss_re_ti_sam(137l+10+(l≥7)+n), l=0,1,...,13.

本地npss降采样信号为npss_loc_ti_sam(n)，发生频偏时若忽略噪声的影响，则接收到第l个symbol的npss信号为：The local npss down-sampling signal is npss_loc_ti_sam(n). If the influence of noise is ignored when the frequency offset occurs, the npss signal of the lth symbol received is:

其中，ε为归一化频偏，△_l＝137l+10+(l≥7)。将接收到的第l个symbol的npss降采样信号与本地npss降采样信号共轭相乘：Among them, ε is the normalized frequency offset, Δ_l =137l+10+(l≥7). Multiply the received npss downsampled signal of the l-th symbol with the local npss downsampled signal conjugate:

y_l(n)＝S(l)npss_loc_ti_sam^*(n)npss_re_ti_sam_l(n),n＝0,1,...,127 (10)y_l (n)=S(l)npss_loc_ti_sam^* (n)npss_re_ti_sam_l (n),n=0,1,...,127 (10)

对y_l(n)补偿频偏并求和，然后将后11个symbol的结果相加如式(11)所示。Compensate the frequency offset for y_l (n) and sum it up, and then add the results of the last 11 symbols as shown in equation (11).

其中ε为归一化频偏估计。求取y_sum(ε)的最大值，其最大值对应的标号ε即为粗频偏估计，即求得的粗频偏估计为

对接收到的npss降采样信号npss_re_ti_sam(n)使用

进行补偿，得到补偿后的npss信号为：where ε is the normalized frequency offset estimate. Find the maximum value of y_sum(ε), and the label ε corresponding to the maximum value is the rough frequency offset estimate, that is, the obtained rough frequency offset estimate is

Use npss_re_ti_sam(n) for the received npss downsampled signal

After compensation, the compensated npss signal is:

其中，N＝1920。Among them, N=1920.

下面对信号进行精频偏估计。npss_cmp(n)为一个子帧的数据，将其分为14个symbol，第l个symbol的数据(不包含CP)记为npss_cmp_l(n)，n＝0,1,…,128-1，则npss_cmp_l(n)＝npss_cmp(137l+10+(l≥7)+n)，l＝0,1,…,13。对信号npss_cmp_l(n)做如式(9)、式(10)相同操作，可得：Next, the precise frequency offset estimation of the signal is performed. npss_cmp(n) is the data of a subframe, which is divided into 14 symbols, and the data of the l-th symbol (excluding CP) is denoted as npss_cmp_l (n), n=0,1,...,128-1, Then npss_cmp_l (n)=npss_cmp(137l+10+(l≥7)+n), l=0,1,...,13. Do the same operation as formula (9) and formula (10) on the signal npss_cmp_l (n), we can get:

其中，

为精频偏估计，△_l＝137l+10+(l≥7)。令y₃'(n)与y₅'(n)共轭相乘并相加：in,

For the precise frequency offset estimation, △_l =137l+10+(l≥7). Conjugate y₃ '(n) and y₅ '(n) to multiply and add:

其中，△＝△₅-△₃＝274。同理，对y₄'(n)与y₆'(n)、y₇'(n)与y₉'(n)、y₈'(n)与y₁₀'(n)、y₁₁'(n)与y₁₃'(n)做相同的操作，求出p值并求和，得到p_sum，求得的精频偏估计为：Among them, Δ=Δ₅ -Δ₃ =274. Similarly, for y₄ '(n) and y₆ '(n), y₇ '(n) and y₉ '(n), y₈ '(n) and y₁₀ '(n), y₁₁ '( n) Do the same operation as y₁₃ '(n), find the p value and sum it up to get p_sum, and the obtained precision frequency offset estimate is:

精频偏的估计范围为(-0.234,0.234)，粗频偏估计误差为(-0.2,0.2)，所以精频偏估计可以补偿粗频偏估计误差并最大限度的提高了频偏的估计精度。最终估计出的信号的频偏估计为

The estimation range of the fine frequency offset is (-0.234, 0.234), and the estimation error of the coarse frequency offset is (-0.2, 0.2), so the fine frequency offset estimation can compensate the error of the coarse frequency offset estimation and maximize the estimation accuracy of the frequency offset . The frequency offset of the final estimated signal is estimated as

nsss信号检测单元nsss signal detection unit

nsss信号检测功能框图如图5所示，首先将降采样后的npss信号补偿频偏后变换至频域，根据本地生成的npss频域信号得到信道估计和nsss信号阈值，然后根据start_time_acc得到接收到的nsss信号，将信号降采样并补偿频偏后变换至频域，对nsss频域信号进行信道补偿，利用nsss信号自身特征确定帧号与cellID候选值，然后生成本地的nsss频域信号，与nsss频域接收信号做互相关，最终获取n_f的低3bit信息与cellID。The functional block diagram of nsss signal detection is shown in Figure 5. First, the down-sampled npss signal is compensated for frequency offset and then transformed into the frequency domain. The channel estimation and nsss signal threshold are obtained according to the locally generated npss frequency domain signal, and then the received signal is obtained according to start_time_acc. The nsss signal is down-sampled and the frequency offset is compensated, and then converted to the frequency domain, channel compensation is performed on the nsss frequency domain signal, and the frame number and cellID candidate values are determined by using the characteristics of the nsss signal itself, and then the local nsss frequency domain signal is generated. The received signal in the nsss frequency domain is cross-correlated, and finally the low 3-bit information and cellID of n_f are obtained.

接收到的npss降采样信号记为npss_re_ti_sam(n)，频偏估计为freoffset，按照式(12)将

替换为freoffset计算补偿频偏后的npss信号记为npss_cmp(n)，将其分为14个symbol，第l个symbol的数据(不包含CP)记为npss_cmp_l(n)，n＝0,1,…,127，则npss_cmp_l(n)＝npss_cmp(137l+10+(l≥7)+n)，l＝0,1,…,13。将信号npss_cmp_l(n)按照式(17)变换至频域，得到第l个symbol的npss的频域接收号npss_re_fre_l(k)。The received npss down-sampling signal is recorded as npss_re_ti_sam(n), and the frequency offset is estimated as freoffset. According to equation (12), the

Replace the npss signal with freoffset to calculate the compensated frequency offset and denote it as npss_cmp(n), divide it into 14 symbols, and denote the data of the lth symbol (excluding CP) as npss_cmp_l (n), n=0,1 ,...,127, then npss_cmp_l (n)=npss_cmp(137l+10+(l≥7)+n), l=0,1,...,13. Transform the signal npss_cmp_l (n) into the frequency domain according to equation (17) to obtain the frequency domain reception number npss_re_fre_l (k) of the npss of the lth symbol.

其中l＝3,4,…,13。设npss信号频域的资源网格为npss_re_fre_vec(k,l)，其中k＝0,1,…,11，l＝0,1,…,13。资源网格表示信号的时-频分布，将npss_re_fre_l(k)按照l值和k值对应填入资源网格中即得到了接收到的npss信号的频域资源网格。where l=3,4,...,13. Let the resource grid of the frequency domain of the npss signal be npss_re_fre_vec(k,l), where k=0,1,...,11, l=0,1,...,13. The resource grid represents the time-frequency distribution of the signal, and npss_re_fre_l (k) is filled into the resource grid according to the value of l and the value of k to obtain the frequency domain resource grid of the received npss signal.

由式(2)可得，一个symbol的本地npss频域信号为npss_loc_fre(n)，则第l个symbol的本地npss频域信号为S(l)npss_loc_fre(n)，将其填入资源网格中即得到了本地的npss频域资源网格npss_loc_fre_vec(k,l)，则得到的信道估计为如式(18)所示：From equation (2), the local npss frequency domain signal of a symbol is npss_loc_fre(n), then the local npss frequency domain signal of the l-th symbol is S(l)npss_loc_fre(n), which is filled into the resource grid The local npss frequency domain resource grid npss_loc_fre_vec(k,l) is obtained in

ce′(k,l)＝npss_re_fre_vec(k,l)/npss_loc_fre_vec(k,l) (18)ce'(k,l)=npss_re_fre_vec(k,l)/npss_loc_fre_vec(k,l) (18)

其中k＝0,1,…,10，l＝3,4,…,13。ce′(k,l)第l个symbol上的幅度均差为

abs表示求模值运算，角度均差为

angle表示求角度运算，采用典型内插算法将信道估计扩展，得到ce(k,l)，其中k＝0,1,…,10,11，l＝3,4,…,13，则ce(11,l)的幅度值为ce_mag(11,l)＝abs(ce′(10,l))+ce_mag′(l)，角度值为ce_ang(11,l)＝angle(ce′(10,l))+ce_ang′(l)，则ce(k,l)取值如式(19)所示。where k=0,1,...,10 and l=3,4,...,13. The mean amplitude difference on the l-th symbol of ce'(k,l) is

abs represents the modulo value operation, and the average angle difference is

angle represents the angle operation, and the channel estimation is extended by a typical interpolation algorithm to obtain ce(k,l), where k=0,1,...,10,11, l=3,4,...,13, then ce( The amplitude value of 11,l) is ce_mag(11,l)=abs(ce'(10,l))+ce_mag'(l), and the angle value is ce_ang(11,l)=angle(ce'(10,l ))+ce_ang'(l), then the value of ce(k,l) is shown in formula (19).

将npss_re_fre_vec(k,l)和npss_loc_fre_vec(k,l)反资源网格映射，即按照资源网格映射的位置，依次取出每一个symbol中的数据，然后将数据串接，即可得到npss频域接收信号npss_re_fre(n)和频域本地信号npss_loc_fre(n)，长度为121。nsss信号阈值由频域npss信号得到，如式(20)所示：Inverse resource grid mapping of npss_re_fre_vec(k,l) and npss_loc_fre_vec(k,l), that is, according to the location of the resource grid mapping, take out the data in each symbol in turn, and then concatenate the data to get the npss frequency domain The received signal npss_re_fre(n) and the frequency domain local signal npss_loc_fre(n) have a length of 121. The nsss signal threshold is obtained from the frequency domain npss signal, as shown in equation (20):

nsss信号位于偶数位系统帧的第9个子帧，则在21ms数据中，必存在一个子帧的nsss信号。The nsss signal is located in the 9th subframe of the even-numbered system frame, so in the 21ms data, there must be an nsss signal of one subframe.

根据start_time_acc可以得到两个接收到的nsss信号，只有一个是实际存在的nsss信号，将其记为nsss_re_ti(n)，对信号nsss_re_ti(n)进行16倍降采样得到信号nsss_re_ti_sam(n)，对其按照上述操作进行频偏补偿并变换至频域，得到nsss信号资源网格nsss_re_fre_vec(k,l)，按照式(21)对其进行信道补偿，得到补偿后的nsss频域信号nsss_re_fre_vec_cmp(k,l)。According to start_time_acc, two received nsss signals can be obtained, only one is an actual nsss signal, which is recorded as nsss_re_ti(n), and the signal nsss_re_ti(n) is down-sampled by 16 times to obtain the signal nsss_re_ti_sam(n). Perform frequency offset compensation according to the above operation and transform to the frequency domain to obtain the nsss signal resource grid nsss_re_fre_vec(k,l), perform channel compensation on it according to equation (21), and obtain the compensated nsss frequency domain signal nsss_re_fre_vec_cmp(k,l ).

nsss_re_fre_vec_cmp(k,l)＝nsss_re_fre_vec(k,l)/ce(k,l) (21)nsss_re_fre_vec_cmp(k,l)=nsss_re_fre_vec(k,l)/ce(k,l) (21)

将nsss_re_fre_vec_cmp(k,l)反资源网格映射，即可得到nsss频域信号nsss_re_fre(n)，长度为131。The nsss frequency domain signal nsss_re_fre(n) can be obtained by mapping the nsss_re_fre_vec_cmp(k,l) inverse resource grid, and the length is 131.

nsss本地信号在频域是一个Zadoff-Chu序列，其生成表达式为：The nsss local signal is a Zadoff-Chu sequence in the frequency domain, and its generating expression is:

其中in

n＝0,1,...,131n=0,1,...,131

n′＝nmod131n′=nmod131

m＝nmod128m=nmod128

mod表示取余操作。b_q(m)如表2所示：mod represents the remainder operation. b_q (m) is shown in Table 2:

表2b_q(m)取值表Table 2b_q (m) value table

对于接收信号，若忽略噪声影响，则：For the received signal, ignoring the effect of noise, then:

对信号nsss_re_fre(n)补偿b_q(m)得：Compensating b_q (m) for the signal nsss_re_fre(n) gives:

由式(24)可得：From formula (24), we can get:

当n_f＝0或4时，zc(n)具有中心对称性即zc(n)＝zc(130-n)。令q＝0,1,2,3，n_f＝0,2,4,6，对nsss_re_fre(n)做补偿得：When n_f =0 or 4, zc(n) has centrosymmetric ie zc(n)=zc(130-n). Let q = 0, 1, 2, 3, n_f = 0, 2, 4, 6, and make compensation for nsss_re_fre(n):

对其前后两部分共轭求和得：The conjugate sum of its two parts before and after it gets:

定义corr(q,n_f)最大值对应的n_f为n_f_max，q为q_max，则n_f的候选值为(n_f_max,(n_f_max+4)mod8)，q的候选值为q_max。Define the n_f corresponding to the maximum value of corr(q,n_f ) as n_f _max and q as q_max, then the candidate value of n_f is (n_f _max,(n_f _max+4)mod8), and the candidate value of q is q_max.

令n_f＝0,2,4,6，cellID＝0,1,…,503，如式(22)所示生成本地nsss频域信号nsss_loc_fre(n,n_f,cellID)保存至本地。根据start_time_acc按照前文所述方法从接收到的21ms数据中得到两个nsss频域补偿信号分别记为nsss_re_fre₀(n)、nsss_re_fre₁(n)，计算出的n_f的候选值设为n_f0、n_f1，使用nsss_re_fre_i(n)与nsss_loc_fre(n,n_f,cellID)进行相关运算如式(28)所示，检测流程如图6所示，当corr_nsss>nsss_thresh时，本地信号对应的n_f和cellID值即为计算得到的n_f的低3bit信息与cellID。Let n_f = 0, 2, 4, 6, cellID = 0, 1, . . . , 503, generate a local nsss frequency domain signal nsss_loc_fre(n, n_f , cellID) as shown in equation (22) and save it locally. According to start_time_acc, two nsss frequency domain compensation signals are obtained from the received 21ms data according to the method described above, which are respectively recorded as nsss_re_fre₀ (n) and nsss_re_fre₁ (n), and the calculated candidate values of n_f are set as n_f0 , n_f1 , use nsss_re_fre_i (n) and nsss_loc_fre (n, n_f , cellID) to perform the correlation operation as shown in equation (28), the detection process is shown in Figure 6, when corr_nsss >nsss_thresh, the corresponding n of the local signal The values of_f and cellID are the calculated low 3-bit information and cellID of n_f .

实施例1Example 1

为了验证本发明的有效性，进行了若干测试。实验输入为从接收天线接收到的21ms数据。To verify the effectiveness of the present invention, several tests were performed. The experimental input is the 21ms data received from the receiving antenna.

图7为时偏估计对比图，使用分段相关法与本发明方法进行对比，对输入数据加入-1.8～1.8的归一化频偏，每个频偏对应的每种算法仿真100次，检测到的时偏的均方根误差(RMSE)如图所示，由图像可知本发明的时偏估计方法抗频偏能力较强，在大频偏情况下估计的时偏依然准确。图8为频偏估计对比图，使用将一个symbol内的npss信号与本地信号共轭相乘后分为两段做相关的方法(记为方法一)与本发明方法进行对比，在不同的信噪比(SNR)下，对数据随机加入(-1,1)范围的归一化频偏，每种信噪比下每种算法仿真100次，使用频率同步跟踪算法计算补偿过频偏后的数据的归一化频偏，再将其乘以15kHz作为估计频偏与实际频偏的差值，检测到的频偏的RMSE如图所示，由图像可知本发明的频偏估计结果更接近于实际频偏值，精度更高。表3为nsss信号检测时间对比表，使用直接相关的方法与本发明方法进行对比，数据表明本发明的nsss信号检测所用时间更少，时间复杂度降低。在储存本地npss信号时，本发明只需储存一个symbol的npss时域降采样信号和频域信号即可，空间复杂度降低。Fig. 7 is a comparison diagram of time offset estimation. The segmented correlation method is used to compare with the method of the present invention. A normalized frequency offset of -1.8 to 1.8 is added to the input data, and each algorithm corresponding to each frequency offset is simulated 100 times to detect The root mean square error (RMSE) of the arrived time offset is shown in the figure. It can be seen from the image that the time offset estimation method of the present invention has a strong anti-frequency offset capability, and the estimated time offset is still accurate in the case of a large frequency offset. Fig. 8 is a comparison diagram of frequency offset estimation. The method of multiplying the npss signal in a symbol with the local signal conjugate and then dividing it into two sections for correlation (referred to as method 1) is compared with the method of the present invention. Under the noise ratio (SNR), the normalized frequency offset in the range of (-1, 1) is randomly added to the data, and each algorithm is simulated 100 times under each signal-to-noise ratio, and the frequency synchronization tracking algorithm is used to calculate the frequency offset after compensation. The normalized frequency offset of the data is multiplied by 15kHz as the difference between the estimated frequency offset and the actual frequency offset. The RMSE of the detected frequency offset is shown in the figure. It can be seen from the image that the frequency offset estimation result of the present invention is closer to Compared with the actual frequency offset value, the accuracy is higher. Table 3 is a comparison table of nsss signal detection time, using the method of direct correlation to compare with the method of the present invention, the data shows that the nsss signal detection of the present invention takes less time and reduces the time complexity. When storing the local npss signal, the present invention only needs to store the npss time-domain down-sampling signal and the frequency-domain signal of one symbol, and the space complexity is reduced.

表3nsss信号检测时间对比表Table 3 nsss signal detection time comparison table

以上所述，仅为本实施例较佳的具体实施方式，但本实施例的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本实施例揭露的技术范围内，根据本实施例的技术方案及其发明构思加以等同替换或改变，都应涵盖在本实施例的保护范围之内。The above descriptions are only preferred specific implementations of this embodiment, but the protection scope of this embodiment is not limited to this. The technical solutions and the inventive concept of the examples are equivalently replaced or changed, which should all be covered within the protection scope of this embodiment.

Claims (7)

Translated fromChinese

1.一种NB-IoT系统下行链路中的小区搜索系统，其特征在于包括：1. a cell search system in the downlink of an NB-IoT system, is characterized in that comprising:npss信号检测单元，将生成的本地主同步信号npss的频域信号变换为时域信号，并进行降采样处理，然后对从接收天线接收到的数据进行降采样处理，使用降采样后的数据进行定时粗同步；对定时粗同步的结果进行修正得到定时粗同步的修正值，然后使用未降采样的数据进行定时细同步，完成npss信号的检测，给出定时偏差start_time_acc；The npss signal detection unit converts the generated frequency domain signal of the local primary synchronization signal npss into a time domain signal, and performs downsampling processing, and then performs downsampling processing on the data received from the receiving antenna, and uses the downsampled data to perform Timing coarse synchronization; correct the result of timing coarse synchronization to obtain the correction value of timing coarse synchronization, and then use the undownsampled data to perform timing fine synchronization, complete the detection of npss signal, and give the timing deviation start_time_acc;频率同步单元，该单元根据start_time_acc得到接收到的npss信号，对信号进行降采样，根据降采样的npss信号，计算得出粗频偏估计并进行粗频偏补偿；a frequency synchronization unit, which obtains the received npss signal according to start_time_acc, down-samples the signal, calculates a coarse frequency offset estimate and performs coarse frequency offset compensation according to the down-sampled npss signal;对粗频偏补偿后的npss信号进行精频偏估计，根据npss信号计算的频偏估计可近似为接收信号的频偏估计，即最终得出接收信号的频偏估计；Perform precise frequency offset estimation on the npss signal after coarse frequency offset compensation, and the frequency offset estimation calculated according to the npss signal can be approximated as the frequency offset estimation of the received signal, that is, the frequency offset estimation of the received signal is finally obtained;nsss信号检测单元，该单元将降采样后的npss信号补偿频偏后变换至频域；an nsss signal detection unit, which converts the down-sampled npss signal to the frequency domain after compensating for frequency offset;根据本地生成的npss频域信号得到信道估计和辅同步信号nsss信号阈值，然后根据start_time_acc得到接收到的nsss信号，将信号降采样并补偿频偏后变换至频域，对nsss频域信号进行信道补偿，利用nsss信号自身特征确定帧号与cellID候选值，然后生成本地的nsss频域信号，与nsss频域接收信号做互相关，最终获取n_f的低3bit信息与cellID；其中，cellID表示物理层小区标识，n_f表示系统帧号。Obtain the channel estimation and the nsss signal threshold of the secondary synchronization signal according to the locally generated npss frequency domain signal, then obtain the received nsss signal according to start_time_acc, downsample the signal and compensate the frequency offset, and then transform it into the frequency domain, and perform channel analysis on the nsss frequency domain signal. Compensation, use the characteristics of the nsss signal to determine the frame number and the candidate value of cellID, then generate the local nsss frequency domain signal, cross-correlate with the nsss frequency domain received signal, and finally obtain the low 3bit information and cellID of n_f ; among them, cellID represents the physical layer cell identifier, n_f represents the system frame number.2.根据权利要求1所述的一种NB-IoT系统下行链路中的小区搜索系统，其特征还在于所述的npss信号检测单元中只生成一个symbol的本地npss频域信号变换为时域信号，并进行降采样处理的过程如下：2. the cell search system in the downlink of a kind of NB-IoT system according to claim 1, it is characterized in that the local npss frequency domain signal that only generates a symbol in the described npss signal detection unit is transformed into time domain The process of downsampling the signal is as follows:npss信号在频域是一个Zadoff-Chu序列，其生成表达式为：The npss signal is a Zadoff-Chu sequence in the frequency domain, and its generation expression is:

其中，根序列索引u＝5，l为symbol号，S(0)＝S(1)＝S(2)＝0只生成一个symbol的频域npss信号即可，其生成表达式为：Among them, the root sequence index u=5, l is the symbol number, S(0)=S(1)=S(2)=0, only one symbol frequency domain npss signal can be generated, and its generation expression is:

将npss_loc_fre(n)变换至时域，生成一个symbol的时域npss信号：Transform npss_loc_fre(n) to the time domain to generate a symbol's time domain npss signal:

对信号npss_loc_ti(n)进行16倍降采样处理，如式(4)所示，得到的信号记为npss_loc_ti_sam(n)，长度为128个采样点；The signal npss_loc_ti(n) is subjected to 16 times down-sampling processing, as shown in formula (4), the obtained signal is recorded as npss_loc_ti_sam(n), and the length is 128 sampling points;npss_loc_ti_sam(n)＝npss_loc_ti(16n),n＝0,1,...,127 (4)npss_loc_ti_sam(n)=npss_loc_ti(16n),n=0,1,...,127 (4)npss信号位于系统帧的第5个子帧，则在接收到的21ms数据中，必包含两个子帧的npss信号。The npss signal is located in the 5th subframe of the system frame, and the received 21ms data must contain the npss signal of two subframes.3.根据权利要求2所述的一种NB-IoT系统下行链路中的小区搜索系统，其特征还在于所述的定时粗同步过程如下：3. the cell search system in a kind of NB-IoT system downlink according to claim 2, it is characterized in that described timing coarse synchronization process is as follows:对从接收天线接收到的21ms数据re(n)进行降采样处理，得到的信号记为re_sam(n)，用npss_loc_ti_sam(n)和re_sam(n)做滑动相关，每滑动一个点，从re_sam(n)中取出一个子帧的数据，记为re_sam_sf(m)，m＝0,1,…,M-1，M＝1920，则re_sam_sf(m)＝re_sam(n+m)，n＝0,1,…,19200，将re_sam_sf(m)分为14个symbol，第l个symbol的数据，其中不包含CP，记为re_sam_l(n)，n＝0,1,…,127，则re_sam_l(n)＝re_sam_sf(137l+10+(l≥7)+n)，l＝0,1,…,13，其中(l≥7)表示一个值，当l<7时，(l≥7)＝0，当l≥7时，(l≥7)＝1；对于npss信号，前3个symbol信号为0，使用后11个symbol信号re_sam_l(n)与本地信号npss_loc_ti_sam(n)做分段相关，其中，段数取2，每段长度64，并求和，如式(5)所示：Perform down-sampling processing on the 21ms data re(n) received from the receiving antenna, and the obtained signal is recorded as re_sam(n). Use npss_loc_ti_sam(n) and re_sam(n) for sliding correlation. Take out the data of one subframe from n), denoted as re_sam_sf(m), m=0,1,...,M-1, M=1920, then re_sam_sf(m)=re_sam(n+m), n=0, 1,...,19200, divide re_sam_sf(m) into 14 symbols, the data of the l-th symbol, which does not contain CP, is recorded as re_sam_l (n), n=0,1,...,127, then re_sam_l (n)=re_sam_sf(137l+10+(l≥7)+n), l=0,1,...,13, where (l≥7) represents a value, when l<7, (l≥7) =0, when l≥7, (l≥7)=1; for the npss signal, the first 3 symbol signals are 0, and the last 11 symbol signals re_sam_l (n) and the local signal npss_loc_ti_sam(n) are used for segmentation Relevance, where the number of segments is 2, the length of each segment is 64, and the sum is shown, as shown in formula (5):

其中，n＝0,1,…,19200，即共滑动一帧的长度；求取corr(n)的最大值，其最大值对应的标号n即为定时粗同步位置，即

Among them, n=0,1,...,19200, that is, the length of one frame of sliding; to find the maximum value of corr(n), the label n corresponding to the maximum value is the timing coarse synchronization position, that is

在得到的粗同步位置coarse_start_time两侧各取20个点，用npss_loc_ti_sam(n)和re_sam(n)做滑动相关；Take 20 points on both sides of the coarse synchronization position coarse_start_time obtained, and use npss_loc_ti_sam(n) and re_sam(n) for sliding correlation;使用npss_loc_ti_sam(n)和re_sam_l(n)的模进行相关，以抵抗频偏对时偏的影响，如式(6)所示Use the modulus of npss_loc_ti_sam(n) and re_sam_l (n) for correlation to resist the effect of frequency offset on time offset, as shown in equation (6)

其中，n＝coarse_start_time-20,coarse_start_time-19,…,coarse_start_time+20；求取corr_rev(n)的最大值，其最大值对应的标号n即为定时粗同步的修订值，即

Among them, n=coarse_start_time-20, coarse_start_time-19,..., coarse_start_time+20; find the maximum value of corr_rev(n), and the label n corresponding to the maximum value is the revised value of timing coarse synchronization, that is

4.根据权利要求3所述的一种NB-IoT系统下行链路中的小区搜索系统，其特征还在于所述的定时细同步过程如下：4. the cell search system in a kind of NB-IoT system downlink according to claim 3, it is characterized in that described timing fine synchronization process is as follows:在得到定时粗同步的修订值start_time_rev后，将其乘以16变换至未降采样数据的时偏估计；After obtaining the revised value start_time_rev of the timing coarse synchronization, multiply it by 16 and transform it to the time offset estimation of the undownsampled data;在其两侧各取64个点，在未降采样的数据re(n)上进行滑动，每滑动一个点，从re(n)中取出一个子帧的数据，记为re_sf(n)，n＝0,1,…,N-1，N＝30720，然后再进行16倍降采样，得到的信号记为re_sam_sf'(n)，n＝0,1,…,N'-1，N'＝1920，第l个symbol的数据记为re_sam'_l(n)，长度为128，对信号npss_loc_ti_sam(n)和re_sam'_l(n)的模进行相关，如式(7)所示：Take 64 points on each side of it, slide on the unsampled data re(n), and take out a subframe of data from re(n) for each point you slide, denoted as re_sf(n), n =0,1,...,N-1, N=30720, and then down-sampling by 16 times, the obtained signal is recorded as re_sam_sf'(n), n=0,1,...,N'-1, N'= In 1920, the data of the l-th symbol is recorded as re_sam'_l (n), the length is 128, and the modulo of the signals npss_loc_ti_sam(n) and re_sam'_l (n) are correlated, as shown in formula (7):

其中，n＝start_time_rev*16-64,start_time_rev*16-15,…,start_time_rev*16+64；求取corr_acc(n)的最大值，其最大值对应的标号n即为定时细同步，即

该值为npss信号开始的位置。Among them, n=start_time_rev*16-64, start_time_rev*16-15,...,start_time_rev*16+64; find the maximum value of corr_acc(n), and the label n corresponding to the maximum value is the timing fine synchronization, that is

The value is where the npss signal starts.5.根据权利要求1所述的一种NB-IoT系统下行链路中的小区搜索系统，其特征还在于所述的频率同步单元得出粗频偏估计并给出粗频偏补偿的过程如下：5. the cell search system in a kind of NB-IoT system downlink according to claim 1, it is characterized in that described frequency synchronization unit obtains rough frequency offset estimation and provides the process of rough frequency offset compensation as follows :根据计算出的定时偏差start_time_acc从接收数据中确定出接收到的npss信号，对该确定出的npss信号进行16倍降采样处理，记为npss_re_ti_sam(n)，则：According to the calculated timing deviation start_time_acc, the received npss signal is determined from the received data, and the determined npss signal is subjected to 16-fold down-sampling processing, which is recorded as npss_re_ti_sam(n), then:npss_re_ti_sam(n)＝re(start_time_acc+16n),n＝0,1,...,N-1 (8)npss_re_ti_sam(n)=re(start_time_acc+16n),n=0,1,...,N-1 (8)其中N＝1920，npss_re_ti_sam(n)为一个子帧的数据，将其分为14个symbol，第l个symbol的数据，其中不包含CP，记为npss_re_ti_sam_l(n)，n＝0,1,…,128-1，则npss_re_ti_sam_l(n)＝npss_re_ti_sam(137l+10+(l≥7)+n)，l＝0,1,…,13；Where N=1920, npss_re_ti_sam(n) is the data of a subframe, which is divided into 14 symbols, the data of the l-th symbol, which does not contain CP, is denoted as npss_re_ti_sam_l (n), n=0,1, ...,128-1, then npss_re_ti_sam_l (n)=npss_re_ti_sam(137l+10+(l≥7)+n), l=0,1,...,13;本地npss降采样信号为npss_loc_ti_sam(n)，忽略噪声的影响，则接收到第l个symbol的npss信号为：The local npss downsampling signal is npss_loc_ti_sam(n), ignoring the influence of noise, the npss signal of the lth symbol received is:

其中，ε为归一化频偏，△_l＝137l+10+(l≥7)；将接收到的第l个symbol的npss降采样信号与本地npss降采样信号共轭相乘：Among them, ε is the normalized frequency offset, △_l =137l+10+(l≥7); multiply the received npss down-sampling signal of the lth symbol with the local npss down-sampling signal conjugate:y_l(n)＝S(l)npss_loc_ti_sam^*(n)npss_re_ti_sam_l(n),n＝0,1,...,127 (10)y_l (n)=S(l)npss_loc_ti_sam^* (n)npss_re_ti_sam_l (n),n=0,1,...,127 (10)对y_l(n)补偿频偏并求和，然后将后11个symbol的结果相加如式(11)所示：Compensate the frequency offset for y_l (n) and sum it up, and then add the results of the last 11 symbols as shown in equation (11):

其中ε为归一化频偏估计；where ε is the normalized frequency offset estimate;求取y_sum(ε)的最大值，其最大值对应的标号ε即为粗频偏估计，即求得的粗频偏估计为

对接收到的npss降采样信号npss_re_ti_sam(n)使用

进行补偿，得到补偿后的npss信号为：Find the maximum value of y_sum(ε), and the label ε corresponding to the maximum value is the rough frequency offset estimate, that is, the obtained rough frequency offset estimate is

Use npss_re_ti_sam(n) for the received npss downsampled signal

After compensation, the compensated npss signal is:

其中，N＝1920。Among them, N=1920.6.根据权利要求1所述的一种NB-IoT系统下行链路中的小区搜索系统，其特征还在于所述精频偏估计过程如下：6. the cell search system in a kind of NB-IoT system downlink according to claim 1, it is characterized in that described fine frequency offset estimation process is as follows:npss_cmp(n)为一个子帧的数据，将其分为14个symbol，第l个symbol的数据记为npss_cmp_l(n)，n＝0,1,…,128-1，则npss_cmp_l(n)＝npss_cmp(137l+10+(l≥7)+n)，l＝0,1,…,13；npss_cmp(n) is the data of a subframe, which is divided into 14 symbols, and the data of the l-th symbol is recorded as npss_cmp_l (n), n=0,1,...,128-1, then npss_cmp_l (n )=npss_cmp(137l+10+(l≥7)+n), l=0,1,...,13;对信号npss_cmp_l(n)做如式(9)、式(10)相同操作，可得：Do the same operation as formula (9) and formula (10) on the signal npss_cmp_l (n), we can get:

其中，

为精频偏估计，△_l＝137l+10+(l≥7)；令y₃'(n)与y₅'(n)共轭相乘并相加：in,

For precise frequency offset estimation, △_l =137l+10+(l≥7); let y₃ '(n) and y₅ '(n) conjugate multiply and add:

其中，△＝△₅-△₃＝274；同理，对y₄'(n)与y₆'(n)、y₇'(n)与y₉'(n)、y₈'(n)与y₁₀'(n)、y₁₁'(n)与y₁₃'(n)做相同的操作，求出p值并求和，得到p_sum，求得的精频偏估计为：Among them, △=△₅ -△₃ =274; in the same way, for y₄ '(n) and y₆ '(n), y₇ '(n) and y₉ '(n), y₈ '(n) Do the same operation as y₁₀ '(n), y₁₁ '(n) and y₁₃ '(n), find the p value and sum it up to get p_sum, and the obtained precision frequency offset estimate is:

精频偏的估计范围为(-0.234,0.234)，粗频偏估计误差为(-0.2,0.2)，精频偏估计可以补偿粗频偏估计误差并最大限度的提高了频偏的估计精度，最终估计出的信号的频偏估计为

The estimation range of the fine frequency offset is (-0.234, 0.234), and the error of the coarse frequency offset estimation is (-0.2, 0.2). The fine frequency offset estimation can compensate the coarse frequency offset estimation error and maximize the estimation accuracy of the frequency offset. The frequency offset of the final estimated signal is estimated as

7.根据权利要求6所述的一种NB-IoT系统下行链路中的小区搜索系统，其特征还在于所述nsss信号检测单元具体工作过程如下：7. The cell search system in the downlink of a kind of NB-IoT system according to claim 6, is characterized in that the specific working process of described nsss signal detection unit is as follows:接收到的npss降采样信号记为npss_re_ti_sam(n)，频偏估计为freoffset，按照式(12)将

替换为freoffset计算补偿频偏后的npss信号记为npss_cmp(n)，将其分为14个symbol，第l个symbol的数据(不包含CP)记为npss_cmp_l(n)，n＝0,1,…,127，则npss_cmp_l(n)＝npss_cmp(137l+10+(l≥7)+n)，l＝0,1,…,13；将信号npss_cmp_l(n)按照式(17)变换至频域，得到第l个symbol的npss的频域接收号npss_re_fre_l(k)；The received npss down-sampling signal is recorded as npss_re_ti_sam(n), and the frequency offset is estimated as freoffset. According to equation (12), the

Replace the npss signal with freoffset to calculate the compensated frequency offset and denote it as npss_cmp(n), divide it into 14 symbols, and denote the data of the lth symbol (excluding CP) as npss_cmp_l (n), n=0,1_,. To the frequency domain, the frequency domain receiving number_{npss_re_frel} (k) of the npss of the l-th symbol is obtained;

其中l＝3,4,…,13；设npss信号频域的资源网格为npss_re_fre_vec(k,l)，其中k＝0,1,…,11，l＝0,1,…,13；资源网格表示信号的时-频分布，将npss_re_fre_l(k)按照l值和k值对应填入资源网格中即得到了接收到的npss信号的频域资源网格；where l=3,4,...,13; let the resource grid of npss signal frequency domain be npss_re_fre_vec(k,l), where k=0,1,...,11, l=0,1,...,13; resource The grid represents the time-frequency distribution of the signal, and npss_re_fre_l (k) is filled into the resource grid according to the value of l and the value of k to obtain the frequency domain resource grid of the received npss signal;由式(2)得，一个symbol的本地npss频域信号为npss_loc_fre(n)，则第l个symbol的本地npss频域信号为S(l)npss_loc_fre(n)，将其填入资源网格中即得到了本地的npss频域资源网格npss_loc_fre_vec(k,l)，则得到的信道估计如式(18)所示：From equation (2), the local npss frequency domain signal of a symbol is npss_loc_fre(n), then the local npss frequency domain signal of the l-th symbol is S(l)npss_loc_fre(n), which is filled in the resource grid That is, the local npss frequency domain resource grid npss_loc_fre_vec(k,l) is obtained, and the obtained channel estimation is shown in equation (18):ce′(k,l)＝npss_re_fre_vec(k,l)/npss_loc_fre_vec(k,l) (18)ce'(k,l)=npss_re_fre_vec(k,l)/npss_loc_fre_vec(k,l) (18)其中k＝0,1,…,10，l＝3,4,…,13；ce′(k,l)第l个symbol上的幅度均差为

abs表示求模值运算，角度均差为

angle表示求角度运算，采用典型内插算法将信道估计扩展，得到ce(k,l)，其中k＝0,1,…,10,11，l＝3,4,…,13，则ce(11,l)的幅度值为ce_mag(11,l)＝abs(ce′(10,l))+ce_mag′(l)，角度值为ce_ang(11,l)＝angle(ce′(10,l))+ce_ang′(l)，则ce(k,l)取值如式(19)所示：where k=0,1,...,10, l=3,4,...,13; the mean amplitude difference on the lth symbol of ce'(k,l) is

abs represents the modulo value operation, and the average angle difference is

angle represents the angle operation, and the channel estimation is extended by a typical interpolation algorithm to obtain ce(k,l), where k=0,1,...,10,11, l=3,4,...,13, then ce( The amplitude value of 11,l) is ce_mag(11,l)=abs(ce'(10,l))+ce_mag'(l), and the angle value is ce_ang(11,l)=angle(ce'(10,l ))+ce_ang'(l), then the value of ce(k,l) is shown in formula (19):

将npss_re_fre_vec(k,l)和npss_loc_fre_vec(k,l)反资源网格映射，即按照资源网格映射的位置，依次取出每一个symbol中的数据，然后将数据串接，即可得到npss频域接收信号npss_re_fre(n)和频域本地信号npss_loc_fre(n)，长度为121；nsss信号阈值由频域npss信号得到，如式(20)所示：Inverse resource grid mapping of npss_re_fre_vec(k,l) and npss_loc_fre_vec(k,l), that is, according to the location of the resource grid mapping, take out the data in each symbol in turn, and then concatenate the data to get the npss frequency domain The received signal npss_re_fre(n) and the frequency domain local signal npss_loc_fre(n) have a length of 121; the nsss signal threshold is obtained from the frequency domain npss signal, as shown in equation (20):

nsss信号位于偶数位系统帧的第9个子帧，则在21ms数据中，必存在一个子帧的nsss信号；If the nsss signal is located in the 9th subframe of the even-numbered system frame, in the 21ms data, there must be an nsss signal of one subframe;根据start_time_acc得到两个接收到的nsss信号，只有一个是实际存在的nsss信号，将其记为nsss_re_ti(n)，对信号nsss_re_ti(n)进行16倍降采样得到信号nsss_re_ti_sam(n)，对其按照上述操作进行频偏补偿并变换至频域，得到nsss信号资源网格nsss_re_fre_vec(k,l)，按照式(21)对其进行信道补偿，得到补偿后的nsss频域信号nsss_re_fre_vec_cmp(k,l)；According to start_time_acc, two received nsss signals are obtained, only one is an actual nsss signal, which is recorded as nsss_re_ti(n), and the signal nsss_re_ti(n) is down-sampled by 16 times to obtain the signal nsss_re_ti_sam(n). The above operations perform frequency offset compensation and transform to the frequency domain to obtain the nsss signal resource grid nsss_re_fre_vec(k,l), perform channel compensation on it according to equation (21), and obtain the compensated nsss frequency domain signal nsss_re_fre_vec_cmp(k,l) ;nsss_re_fre_vec_cmp(k,l)＝nsss_re_fre_vec(k,l)/ce(k,l) (21)nsss_re_fre_vec_cmp(k,l)=nsss_re_fre_vec(k,l)/ce(k,l) (21)将nsss_re_fre_vec_cmp(k,l)反资源网格映射，即可得到nsss频域信号nsss_re_fre(n)，长度为131；Mapping the nsss_re_fre_vec_cmp(k,l) inverse resource grid to get the nsss frequency domain signal nsss_re_fre(n) with a length of 131;nsss本地信号在频域是一个Zadoff-Chu序列，其生成表达式为：The nsss local signal is a Zadoff-Chu sequence in the frequency domain, and its generating expression is:

其中inn＝0,1,...,131n=0,1,...,131n′＝n mod 131n′=n mod 131m＝n mod 128m=n mod 128

对于接收信号，若忽略噪声影响，则：For the received signal, ignoring the effect of noise, then:

对信号nsss_re_fre(n)补偿b_q(m)得：Compensating b_q (m) for the signal nsss_re_fre(n) gives:

由式(24)可得：From formula (24), we can get:

当n_f＝0或4时，zc(n)具有中心对称性即zc(n)＝zc(130-n)，令q＝0,1,2,3，n_f＝0,2,4,6，对nsss_re_fre(n)做补偿得：When n_f =0 or 4, zc(n) has centrosymmetric ie zc(n)=zc(130-n), let q=0,1,2,3,_nf =0,2,4, 6. Compensate for nsss_re_fre(n):

对其前后两部分共轭求和得：Conjugate sum of its two parts before and after to get:

定义corr(q,n_f)最大值对应的n_f为n_f_max，q为q_max，则n_f的候选值为(n_f_max,(n_f_max+4)mod8)，q的候选值为q_max；Define the n_f corresponding to the maximum value of corr(q,n_f ) as n_f _max and q as q_max, then the candidate value of n_f is (n_f _max,(n_f _max+4)mod8), and the candidate value of q is q_max;令n_f＝0,2,4,6，cellID＝0,1,…,503，Let n_f = 0, 2, 4, 6, cellID = 0, 1, ..., 503,如式(22)所示生成本地nsss频域信号nsss_loc_fre(n,n_f,cellID)保存至本地；根据start_time_acc按照上述操作从接收到的21ms数据中得到两个nsss频域补偿信号分别记为nsss_re_fre₀(n)、nsss_re_fre₁(n)，计算出的n_f的候选值设为n_f0、n_f1，Generate the local nsss frequency domain signal nsss_loc_fre(n,n_f , cellID) as shown in formula (22) and save it locally; according to start_time_acc, two nsss frequency domain compensation signals are obtained from the received 21ms data according to the above operation and are recorded as nsss_re_fre₀ (n), nsss_re_fre₁ (n), the calculated candidate values of n_f are set to n_f0 and n_f1 ,使用nsss_re_fre_i(n)与nsss_loc_fre(n,n_f,cellID)进行相关运算如式(28)所示：Use nsss_re_fre_i (n) and nsss_loc_fre (n, n_f , cellID) to perform the correlation operation as shown in equation (28):

当corr_nsss>nsss_thresh时，本地信号对应的n_f和cellID值即为计算得到的n_f的低3bit信息与cellID。When corr_nsss >nsss_thresh, the values of n_f and cellID corresponding to the local signal are the calculated low 3-bit information and cellID of n_f .

Images