CN118019047A - Multi-user access identification detection method for frequency domain pre-equalization in time sensitive network - Google Patents
Multi-user access identification detection method for frequency domain pre-equalization in time sensitive network Download PDFInfo
- Publication number
- CN118019047A CN118019047A CN202410230192.9A CN202410230192A CN118019047A CN 118019047 A CN118019047 A CN 118019047A CN 202410230192 A CN202410230192 A CN 202410230192A CN 118019047 A CN118019047 A CN 118019047A
- Authority
- CN
- China
- Prior art keywords
- user
- data
- estimation
- channel
- base station
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 34
- 238000000034 method Methods 0.000 claims abstract description 60
- 230000005540 biological transmission Effects 0.000 claims abstract description 33
- 230000000694 effects Effects 0.000 claims abstract description 16
- 230000006854 communication Effects 0.000 claims abstract description 11
- 238000004891 communication Methods 0.000 claims abstract description 10
- 238000001228 spectrum Methods 0.000 claims abstract description 8
- 239000011159 matrix material Substances 0.000 claims description 55
- 238000004422 calculation algorithm Methods 0.000 claims description 43
- 230000007480 spreading Effects 0.000 claims description 20
- 238000003892 spreading Methods 0.000 claims description 20
- 239000013598 vector Substances 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 15
- 230000008447 perception Effects 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 238000013507 mapping Methods 0.000 claims description 2
- 230000006835 compression Effects 0.000 claims 1
- 238000007906 compression Methods 0.000 claims 1
- 230000008569 process Effects 0.000 abstract description 8
- 238000009826 distribution Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 7
- 230000009916 joint effect Effects 0.000 description 7
- 230000008901 benefit Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 239000000654 additive Substances 0.000 description 2
- 230000000996 additive effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Landscapes
- Radio Transmission System (AREA)
Abstract
The invention discloses a multi-user access identification detection method for frequency domain pre-equalization in a time sensitive network, belonging to the technical field of data transmission in wireless communication. The implementation method of the invention comprises the following steps: in the downlink stage, a certain antenna of the base station broadcasts a sign signal to all user equipment, and the user performs synchronization, channel estimation and power control. In the uplink transmission stage, the user adopts an unauthorized mode to transmit data. The process of transmitting signals by users comprises constellation symbol modulation, multiplication of a spread spectrum sequence, pre-equalization and OFDM modulation. And the base station performs channel estimation by using the estimated data when combining the user activity with the data rough estimation to identify the active user sequence number and the data rough estimation. More accurate data estimation is performed using the channel estimate. The channel estimation and the data estimation can be iterated. The invention can realize data detection and channel estimation under less time-frequency resources, and realize the data transmission with high real-time and high reliability expected by the time-sensitive network.
Description
Technical Field
The invention relates to an unauthorized mass equipment access identification detection method using pre-equalization assistance and a large-scale multiple-input multiple-output (Multiple input multiple output, MIMO) technology for an Internet of things scene supported by a time-sensitive network, and an unauthorized access method for realizing active user detection, channel estimation and data demodulation under the mechanism, belonging to the technical field of data transmission in wireless communication.
Background
Along with the surge of the digital transformation and upgrading of society, core technologies for supporting wisdom cities, intelligent manufacturing and other prospects related to the Internet of things, edge computing and the like are rapidly developing. Meanwhile, how to realize high-real-time and high-reliability data communication in the internet of things becomes a new challenge. The Time-sensitive network (Time-SENSITIVE NETWORKING, TSN) protocol specification is proposed to construct more efficient and reliable data transmission in the Ethernet, which perfectly meets the development requirements of the Internet of things technology. Therefore, in the future, the internet of things technology must be deeply integrated with the TSN.
With the rapid development of the internet of things technology, the number of internet of things devices can be increased in an explosive manner. The increased number of devices may complicate the handshaking process in the scheduling-based random access protocol, thereby bringing about a longer access delay for accessing. Secondly, the orthogonal resource allocation makes mass devices occupy excessive time-frequency resources. For this purpose, the scholars respectively propose unlicensed random access protocols and non-orthogonal multiple access techniques to solve these two problems. The two technologies are combined to obtain the unlicensed mass access communication scene which is widely focused at present. In this communication scenario, each user is assigned a unique but non-orthogonal spreading sequence. In the uplink transmission process, all devices send data to the base station at the same time. The base station processes the superimposed signals.
Meanwhile, the large-scale MIMO technology can greatly improve the number of base station link users through diversity of an antenna domain and space multiplexing. Meanwhile, the high antenna number of the massive MIMO enables the antenna array to have high angle domain resolution, so that sparsity of channels in the angle domain appears, and the use of a compressed sensing algorithm is promoted. Therefore, massive MIMO is well suited for application in unlicensed massive access scenarios.
At present, signal processing schemes for mass access are mainly divided into two types. Joint ACTIVITY AND DATA detection, JADD for Joint activity and data detection, and Joint activity detection and channel estimation (Joint activity detection AND CHANNEL estimation, JADCE) for Joint activity detection. In JADD-based schemes, it is often assumed that the channel is known at the base station. Then, the user activity and the data are jointly estimated by using a compressed sensing algorithm. However, in a practical application scenario, perfect channel estimation is not possible, resulting in reduced activity and data detection performance. In JADCE-based schemes, users often employ a two-stage transmission scheme. First, active users transmit non-orthogonal pilots, and base stations jointly estimate user activity and channel using compressed sensing algorithms. The active users then transmit data and the base station estimates the data using the estimated channel. However, to ensure reliability of subsequent data demodulation, the JADCE scheme generally has a high pilot number, and it is difficult to achieve low-latency transmission desired by TSN. Meanwhile, the data estimation performance is limited by the channel estimation performance, and under the fast time-varying channel condition, the data estimation performance can be seriously deteriorated, and the reliability of data transmission is greatly reduced.
Disclosure of Invention
In order to solve the problems of serious deterioration of data estimation performance and high frequency guide number in the prior art, the invention aims to provide a multi-user low-delay access method assisted by pre-equalization in a time-sensitive network.
The aim of the invention is achieved by the following technical scheme.
The invention discloses a pre-equalization assisted multi-user low-delay access method in a time sensitive network, which comprises three steps of base station downlink broadcasting, user uplink transmission and data processing; wherein,
Step 1, a base station broadcasts in a downlink mode; the base station is equipped with massive MIMO with N antennas and communicates with the total number K of users. First, the base station randomly selects one antenna, and broadcasts a flag signal to all devices using the single antenna. The serial number of this antenna for downlink broadcasting is denoted by η. And after receiving the marking signal, the user performs power control, synchronization and channel estimation.
Step 2, the active user adopts an unlicensed protocol to send a plurality of frames of data to the base station, and uplink data transmission is carried out through baseband processing; and (2) after receiving the marking signal sent in the step (1), the active user performs uplink data transmission by baseband processing when the active user adopts an unlicensed protocol to send signals. Firstly, an active user obtains a value of a pre-equalization factor according to a downlink channel estimation result; then, when data is transmitted, adopting an orthogonal frequency division multiplexing mode for communication; the data sent on each subcarrier of the user with the sequence number k is the product of the constellation symbol, the pre-equalization factor and the value of the spreading sequence s k on the corresponding subcarrier; the length of the spread spectrum sequence is M subcarriers, and the spread spectrum sequences corresponding to all users are known at the base station end; active users continuously transmit T constellation symbols to a base station at the same time in each uplink transmission stage; the base station receives signals of T slots.
Further, in the step 2, the active user adopts an unlicensed protocol to send a plurality of frames of data to the base station, and the implementation method for uplink data transmission through baseband processing is as follows: each user is allocated a unique but non-orthogonal spreading sequence, M subcarriers in length; the subcarrier corresponding to the user with the sequence number k is denoted as s k; firstly, a user converts data bits into constellation symbols; each OFDM symbol corresponds to one constellation symbol in T OFDM symbols transmitted by a user in a certain frame; therefore, the t-th constellation symbol sent by the user with the sequence number k in a certain frame is denoted as x k,t; then, the user generates a pre-equalization factor lambda k according to the channel estimation result in the step 1; finally, the frequency domain sequence corresponding to the t-th OFDM symbol transmitted by the active user isWherein; symbol/>Representing element-wise multiplication.
Further, in step 2, the specific method for generating the pre-equalization factor is as follows; channel gain on the mth subcarrier estimated by user with sequence number kThe value of the pre-equalization factor lambda k on the m-th subcarrier is denoted lambda m,k; the value of lambda m,k is generated as follows;
Step 3, data processing; after receiving the signals sent by the active users in the step 2, the base station obtains an active user set, channel state information and T constellation symbols sent by each active user according to the received signals; and according to the active user set, the channel state information and T constellation symbols sent by each active user, simultaneously realizing active user identification detection, data detection and channel estimation, namely realizing large-scale MIMO unlicensed access based on pre-equalization assistance in a time-sensitive network.
In step 3, the processing procedure of the base station for one frame of signal comprises the following steps,
Step 3.1: combining user activity with the data rough estimate; for antennas transmitting the signpost signal, the effect of the channel is cancelled by the user's pre-equalization operation, and the received signal is only related to the spreading sequence and the data. Thus, the base station can derive a set of active user sequence numbers and a preliminary estimate of the transmitted constellation symbols from the signal received by the antenna transmitting the beacon signal. Here, the compressed sensing method is used to obtain the active user sequence number setCoarse estimation of transmitted constellation symbols
Step 3.2: channel estimation; the base station obtains the active user sequence number set according to the step 3.1Rough estimation/>, of transmitted constellation symbolsObtaining an estimated value of an equivalent channel and an estimated value of noise power by utilizing signals received by all antennas of a base station; channel estimation is performed using a compressed sensing algorithm. Wherein, the equivalent channel is the product of the real channel of each antenna and the corresponding channel of the antenna transmitting the marking signal;
Step 3.3: estimating data; and (3) the base station performs refined estimation on data sent by the user by utilizing signals received by all antennas according to the equivalent channel estimation value and the noise power estimation value obtained in the step (3.2).
Step 3.4: iteration between channel estimation and data estimation; replacing the data rough estimation obtained in the step 3.1 with the data estimation obtained in the step 3.3, and performing the step 3.2 and the step 3.3; iterating between step 3.2 and step 3.3 according to this method; terminating the iteration after the iteration number exceeds the maximum iteration number; obtaining an estimated value of the equivalent channel from the last executed step 3.3; obtaining an estimated value of the data from the last executed step 3.3, and simultaneously realizing active user identification detection, data detection and channel estimation, and estimating the estimated value of the constellation symbolsConstellation symbol mapping, constellation symbol demodulation, and conversion to a bit sequence is performed.
Further, the specific form of the compressive sensing method in step 3.1 is as follows:
Yη=SX+Wη
Where η is the serial number of the antenna transmitting the tag signal. Y η is the signal received by the antenna from which the base station transmits the flag signal; s= [ S 1,s2,…,sK ] is a matrix formed by spreading sequences of all users; x is a matrix formed by constellation symbols sent by a user, and elements of a kth row and a kth column of the matrix are matrices formed by Gaussian white noise, wherein X k,t;Wη is the element of the kth row and the kth column; because the uplink transmission has sparsity, the column vectors of X are sparse vectors; since the user's activity remains unchanged within a frame, all column vectors of X have the same support set, thus having structured sparsity; therefore, the posterior probability of the constellation symbol x k,t is approximately calculated by adopting GMMV-AMP compressed sensing algorithm, and the average value is taken as the estimated value of the constellation symbol; the GMMV-AMP algorithm is also used to derive a posterior probability gamma k,t that each x k,t is a non-zero element; for users with average posterior probability greater than 0.5, determining that the users are active, and thus estimating a set of active user sequence numbers
Further, in step 3.2, the channel estimation problem is expressed in the spatial domain as:
the compressed sensing problem comprises M equations in total, wherein each equation corresponds to a baseband equivalent model on one subcarrier; wherein the method comprises the steps of A signal received for the base station on the mth subcarrier; the number of elements in the set is represented by the symbol | c; /(I)Each element of the sensing matrix on the m-th subcarrier is the product of the constellation symbol estimated in the step 3.1 and the spread spectrum sequence; /(I)Contains only the collection/>Constellation symbols and spreading sequences of the middle users; /(I)For the equivalent channel on the m-th subcarrier, the element of the equivalent channel is equal to the product of the real channel and the pre-equalization factor and only comprises the setA channel of a user; /(I)Is gaussian white noise superimposed on the reception signal of the m-th subcarrier.
Right multiplying fourier transform matrix on both sides of the channel estimation problem modelTo obtain an angular domain representation of the channel estimation problem:
Wherein the method comprises the steps of An angular domain representation of an equivalent channel on the mth subcarrier; because the large-scale MIMO channel has sparsity in the angle domain, the row vector of A m has sparsity; in addition, the row vector of A m also has clustering sparse characteristics; in addition, the subcarriers in the same OFDM system bandwidth have the same propagation characteristic, so that the A m matrixes corresponding to different subcarriers have the same support set; in order to fully utilize the cluster sparsity and the structured sparsity of A m, the prior GMMV-AMP algorithm is adopted to obtain the estimated value/>, of the angle domain equivalent channelNoise power estimate/>And then reusedObtain the space domain equivalent channel estimation value/> Is gaussian white noise transformed into the angle domain.
Further, in the step 3.3, data estimation is performed according to the following formula:
Wherein the method comprises the steps of Representing signals received by an nth antenna of the base station; /(I)For the collection/>Estimate of equivalent channel from user to base station nth antenna,/>For the collection/>A matrix formed by spreading sequences corresponding to users; /(I)For the collection/>The user sends the estimated value of the constellation symbol; symbol/>The representation matrix is multiplied by elements; perception matrix/>Obtaining; /(I)Is gaussian white noise superimposed on the reception signal of the nth antenna.
Order the/>The following overdetermined matrix equation is obtained:
For a pair of Performing LMMSE estimation to obtain/>The expression is as follows:
Wherein the method comprises the steps of Is a unitary matrix with the dimension of/> The estimated noise variance for step 3.2.
The beneficial effects are that:
1. the existing JADD method detects under the assumption of a perfect channel, but in a practical scenario, estimating the channels of numerous users requires a huge pilot overhead, making this assumption too ideal. The invention discloses a pre-equalization assisted multi-user low-delay access method in a time-sensitive network, which is not realized based on the assumption of a perfect channel, and meanwhile, pre-equalization is adopted in the uplink transmission process of a user, so that a base station can obtain rough estimation of active user sequence numbers and data, and further realize channel estimation. Thus, the present invention enables or is associated with user detection, data detection, and channel estimation.
2. The existing JADCE scheme based on two-stage transmission needs to send pilot frequency first and then send data when the user transmits uplink. Furthermore, the JADCE scheme of two-phase transmission requires simultaneous estimation of active users and large-dimensional channel matrices involved in massive MIMO, requiring a large amount of pilot overhead. The multi-user low-delay access method assisted by pre-equalization in the time sensitive network disclosed by the invention completes channel estimation in the downlink broadcasting stage of the base station, so that users only need to transmit data and do not need to transmit pilot frequency when transmitting uplink. In addition, the invention separates the active user identification detection from the large-dimension channel estimation step, thereby remarkably reducing the time domain resource overhead required under the same accuracy of the active user identification detection. Therefore, the invention has lower access delay and can meet the low-delay transmission expected by TSN.
3. The existing JADCE scheme based on two-stage transmission directly performs data detection by using the estimated channel state information with errors, and the performance of data estimation can be influenced by the performance of channel estimation. The multi-user low-delay access method assisted by pre-equalization in the time sensitive network disclosed by the invention can iterate each other between the channel estimation step and the data estimation step, and the estimation errors of the channel estimation step and the data estimation step can be gradually reduced through iteration, so that the accuracy of data estimation can be improved, and the high-reliability transmission expected by TSN can be satisfied.
Drawings
Fig. 1 is a schematic diagram of a transmitter and a receiver in an uplink transmission stage.
Fig. 2 is a schematic diagram of an uplink and downlink communication process.
Fig. 3 is a schematic diagram of a signal processing flow.
Fig. 4 is a comparison of the data demodulation performance of the present invention with the other two comparison schemes.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and examples.
The embodiment considers an unlicensed uplink data transmission scenario where a base station equipped with a massive MIMO array serves massive internet of things devices, as shown in fig. 1. The base station adopting time division duplex massive MIMO is provided with a uniform linear array with the antenna number of M, and serves user equipment with the total number of K. To combat frequency selective fading of the channel, an orthogonal frequency division multiplexing (Orthogonal frequency division multiplexing, OFDM) modulation scheme is used for uplink transmission. The spreading sequence of each user contains M subcarriers, denoted asWherein the column vector s k is the spreading sequence corresponding to the user with sequence number k. And selecting a partial Fourier transform matrix with S as a semi-unitary body.
The present embodiment considers the following channel model.
The channel gain on the mth subcarrier from the user with sequence number k to all antennas of the base station is represented. Each user's channel contains L paths. For the first path in the user's corresponding channel with sequence number k, its gain, physical angle and time are denoted as β k,l,φk,l and τ k,l, respectively.The array response vector for the first path of user k, where λ is the wavelength. B s denotes the baseband signal bandwidth. Since the distance of each user from the base station is much greater than the distance of the scatterers surrounding the user. Thus, for a single user, the electromagnetic waves it emits can only reach the base station with a small angular spread. Cluster sparsity will be present if the channel matrix is transformed into the angular domain.
As shown in fig. 2, the embodiment discloses a pre-equalization assisted multi-user low-delay access method in a time-sensitive network, which specifically comprises the following implementation steps:
Step 1: the base station broadcasts in a downlink mode; the base station periodically broadcasts a flag signal to all devices. When broadcasting the flag signal, the base station randomly selects one antenna and broadcasts using only the selected antenna. After receiving the flag signal, the user performs downlink channel estimation, synchronization and power control. For a user with sequence number k, the channel gain estimated to be on the mth subcarrier is expressed as
Step 2: uplink data transmission; because of adopting the unlicensed access protocol, the active users simultaneously send a plurality of frames of data to the base station. Within each frame, each active user continuously transmits T OFDM symbols to the base station. Subsequently, the base station performs signal processing, and the processing is performed in units of frames.
For a user with sequence number k, the data that it finally transmits is composed of the product of three parts, data, spreading sequence and pre-equalization factor, respectively. For a user with sequence number k, the data that it transmits in the t-th OFDM symbol is denoted as x k,t. For active users, their data is modulated onto constellation symbols, and the modulation order is L. For an inactive user, it is determined that its transmit data is 0, i.e., x k,t∈Ω={a1,a2,…,aL, 0. And when the user transmits uplink, pre-equalizing the antenna broadcasting the marking signal. For a user with sequence number k, the value of the pre-equalization factor λ m,k on its mth subcarrier is set as follows:
Where H 0 is a constant. When the modulus of the channel estimation value is larger than H 0, taking the reciprocal of the channel estimation value by the pre-equalization factor; otherwise, the equalization factor is set to zero, so that the excessive pre-equalization factor value is avoided, the signal to noise ratio of a receiving end can be obviously improved, and the transmitting power consumption of user equipment can be reduced.
Step 3: base station signal processing; in step 2, after a user sends a frame of data, the base station processes the superimposed signal sent by the user to obtain the active user serial number, the channel and the estimated value of the sent data. In the step 2, the result of the superposition of the signals sent by the user at the base station end is shown in the following formula:
Wherein y m,n,t is a signal of the nth antenna of the base station at the mth subcarrier received by the nth time slot; h n,m,k is the channel gain at the mth subcarrier for the channel between the kth user's antenna to the nth antenna of the base station. Lambda m,k is the pre-equalization factor of the user with sequence number k at the m-th subcarrier; s m,k is the value of the spreading sequence of the user with the sequence number k at the mth subcarrier, that is, s k=[s1,k,s2,k,…,sM,k]T.αk,t is an activity factor for representing the activity of the kth user at the kth time slot; x k,t is a constellation symbol sent by the kth user in the t time slot; w m,n,t is the additive white gaussian noise received by the nth antenna at the mth subcarrier received by the nth slot;
the scalar form of the equation described above is converted into the lower matrix form of the equation:
Wherein the method comprises the steps of Representing the signal received by the nth antenna of the base station. /(I)Representing the frequency domain channel gain between all users to the nth antenna of the base station. /(I)Is a matrix of pre-equalization factors with elements lambda m,k at the mth row and kth column positions. /(I)The element at the kth row and the tth column of the matrix formed by the data sent by the user is x k,t. /(I)For the frequency domain noise received on the nth antenna, each element obeys an independent complex gaussian distributionSymbol/>The representation matrix is multiplied by elements.
The channel state information, the pre-equalization factor and the data are unknown at the base station end, so that the channel estimation and the data estimation become double-blind problems. The invention converts the double-blind problem into three classical linear models for solving by using pre-equalization at the user side, and the three main steps respectively correspond to the receiver algorithm, as shown in figure 3. The process of deriving the linear model corresponding to the three steps is described below.
Part of the effect of the channel is cancelled due to the pre-equalization at the user side. If the channel is pre-equalized with antenna number η, the received signal is equal to the product of the spreading sequence matrix S and the data matrix X without taking into account the additive gaussian noise. The problem becomes a classical linear model as the spreading sequence matrix is known. The data received by the antenna with the sequence number eta is utilized and a compressed sensing algorithm is adopted, so that the rough estimation of the data matrix X can be obtained, and the step is called joint user activity and data rough estimation and is recorded as step 3.1.
Through step 3.1, the data matrix X is known to the base station, making the channel estimation problem a classical linear model. The estimation of all channel state information can be obtained by using the data received by all antennas and adopting a compressed sensing algorithm. This step is referred to as data-aided channel estimation and is denoted step 3.2.
Since step 3.1 uses the observations of only one antenna for estimation, the multi-antenna diversity gain is not used. The channel state information is known throughout, step 3.2, making the data estimation problem a classical linear model. At this time, a Linear Minimum Mean Square Error (LMMSE) estimation algorithm is adopted for the observation of all antennas, so that a more accurate estimation value of the user transmission data can be obtained. This step is referred to as channel estimation. And is designated as step 3.3.
In addition, the more accurate channel estimation can be obtained by substituting the more accurate estimation value obtained in the data estimation stage into the data-assisted channel estimation stage. Similarly, the iteration can be performed in a data-assisted channel estimation stage and a data estimation stage, and the performance of the two stages is improved. The iterative procedure is counted as a step and denoted step 3.4.
The classical linear model and the solving method corresponding to the steps 3.1, 3.2 and 3.3 are described below, and the detailed execution process of the step 3.4 is described below.
Step 3.1: combining activity with data rough estimation; for an antenna with the sequence number eta, the received signal is expressed as the following formula, the symbolThe representation matrix is multiplied by elements. Although the pre-equalization mode adopted by the invention can not perfectly offset the channel, the error can be controlled within a reasonable range by reasonably setting the threshold value H 0, and the data matrix X can still be recovered with higher accuracy in the rough estimation stage.
Since the number of active users is much smaller than the total number of users (K a < K) during the transmission time per frame, the column vectors in the data matrix X are all sparse vectors. In addition, since active users continuously transmit T OFDM symbols in each frame, the column vectors in the data matrix X have the same support set, and thus have structured sparsity. Thus, the recovery problem of the data matrix X is modeled as an MMV compressed sensing problem.
The invention adopts minimum mean square error estimation to the elements in the data matrix, namely, the average value of the edge posterior probability p (x k,t|Yη) is obtained. The posterior distribution is calculated based on bayesian theorem. Modeling the a priori distribution of data x k,t as discrete form by constellation symbol:
Where γ k,t is called sparsity, is the probability that data x k,t is non-zero. Delta (·) is a dirac function. The edge probability distribution p (x k,t|Yη) can be calculated based on the factor graph and the sum-product algorithm, but the integral with huge dimension needs to be calculated, and the calculation is difficult to solve. To this end, the invention employs an approximate message passing algorithm (Approximate MESSAGE PASSING, AMP) based calculation of the edge posterior probability distribution. The AMP algorithm approximates the messages between the factor graph nodes to gaussian distributions, thereby simplifying the computation process.
The invention uses the prior modification of the existing AMP-NNSPL algorithm to the prior probability of the constellation symbols, thereby solving the rough estimation problem. (regarding the AMP-NNSPL algorithm, please refer to the reference "estimation method of sparse large-scale MIMO-OFDM channel with approximately the same support set" for specific steps, the author of the reference is "X.Lin,S.Wu,L.Kuang,Z.Ni,X.Meng and C.Jiang,"Estimation of Sparse Massive MIMO-OFDM Channels With Approximately Common Support,"in IEEE Communications Letters,vol.21,no.5,pp.1179-1182,May 2017,doi:10.1109/LCOMM.2017.2657620.")AMP-NNSPL algorithm, uses the AMP algorithm to solve the edge posterior probability, uses the expectation-maximization EM algorithm to learn parameters such as noise power, sparsity and the like, uses the NNSPL method to correct the sparsity so as to learn the structured sparsity of the data matrix, and finally, the AMP-NNSPL algorithm outputs the posterior probability distribution corresponding to all data
Where pi k,t is the posterior sparsity of data x k,t, ζ k,t,l is the posterior probability of data x k,t as constellation symbol a l.
Calculating average sparsity of each user in the frame, considering the users with average sparsity larger than 0.5 as active users, and arranging the serial numbers of the active users from small to large to form an ordered setI.e.
Then, the mean value of the posterior probability is calculatedAs a result of estimation of data
Step 3.2: data-aided channel estimation; the spatial domain linear model corresponding to the data-aided channel estimation stage is shown in the following formula.
Where |·| c denotes the number of elements in the set,/>The estimated number of active users for step 3.1. Y CE,m is the spatial domain representation of the signal received by the base station on the mth subcarrier. The perception matrix phi CE,m is a matrix formed by the product of the user transmission data estimated value obtained in the rough estimation stage or the data estimation stage and the spread spectrum sequence, and the element of the t th row and the k th column is/>Where [. Cndot ] m,n represents the value of the nth column element of the mth row of the matrix and [ (cndot ] k represents the value of the kth element of the ordered set or vector. /(I)Is the space domain equivalent channel matrix on the m-th subcarrier, is the product of a pre-equalization factor and a real channel, and has the elements of Is gaussian white noise superimposed on the reception signal of the mth subcarrier. Since the pre-equalization factor is unknown to the base station, the base station can only estimate the equivalent channel of the real channel multiplied by the pre-equalization factor. But is independent of the antenna due to the size of the pre-equalization factor. Thus, the relative magnitude of the channel gain between the different antennas is not changed by the pre-equalization factor, and the equivalent channel still contains complete real channel angle domain information.
Because the channel has sparsity in the angle domain, the spatial domain linear model is considered to be transformed into the angle domain, so that the channel is estimated by using a compressed sensing algorithm. The angular domain linear model for channel estimation is shown in the following formula.
Is a discrete fourier transform matrix and is a unitary matrix. R m is an angular domain representation of the signal received by the base station on the mth subcarrier. /(I)The angle domain representation of the equivalent channel matrix is a sparse matrix, and the angle domain channel matrices corresponding to different subcarriers have the same support set. /(I)Is gaussian white noise transformed into the angle domain. Thus, the channel estimation problem is modeled as a multi-vector observation compressed sensing problem.
The present invention solves the channel estimation problem using the existing GMMV-AMP algorithm. (concerning GMMV-AMP algorithm, please refer to "decipher: adaptive active user identification detection and channel estimation method based on compressed sensing: massive access and massive MIMO", its author, english title "M.Ke,Z.Gao,Y.Wu,X.Gao and R.Schober,"Compressive Sensing-Based Adaptive Active User Detection and Channel Estimation:Massive Access Meets Massive MIMO,"in IEEE Transactions on Signal Processing,vol.68,pp.764-779,2020,doi:10.1109/TSP.2020.2967175.")., adopts GMMV-AMP algorithm, can obtain estimation value of angle domain equivalent channelEstimated value of noise power/>Pair/>The transpose of the right-hand matrix U BS can obtain the estimate of the spatial domain equivalent channel/>
Step 3.3: estimating data; the channel state information is known at the base station in its entirety, via step 3.2. At this time, data estimation can be performed using observations of all antennas. The data estimation here is more accurate due to the multi-antenna diversity gain.
The linear model used in step 3.3 is:
Wherein the method comprises the steps of Is the signal received by the nth antenna of the base station. /(I)For sequence numbers belonging to collection/>The data transmitted by the user constitutes a matrix. /(I)And (3) obtaining an estimated value of the active user channel for the channel estimation stage.A matrix of spreading sequences, but wherein only sequence numbers are included as a set/>A spreading sequence corresponding to a user of the mobile station. The calculation formula of the perception matrix is/> A matrix of gaussian white noise superimposed on the signal received by the nth antenna.
Order theThen there is
Since step 3.2 has already estimated the noise, an estimate of the data can be obtained using the LMMSE algorithmThe following expression shows/>Is an identity matrix. Subsequently, and for the data result obtained in step 3.1/>And updating. Specifically, matrix/>The number of intermediate lines belongs to the set/>The sub-matrix of row vectors of (a) is replaced with/>This updating operation is indicated by the left arrow.
Step 3.4: iterating; at step 3.3 pair matrixIt can be carried back to step 3.2 after the update. Due to the/>, obtained at this timeThe actual value is closer than the rough estimation stage, so the/>, at this time is usedThere will be better performance for channel estimation. Further, a more accurate channel may result in a more accurate data estimate. The iteration is repeated between step 3.2 and step 3.3. The execution of step 3.2 once and step 3.3 once is described as an iterative process. Stopping iteration when the iteration number reaches a preset threshold value to obtain an equivalent channel estimation value/>Data estimation value/>Finally, data estimation value/>The data bits are converted into constellation symbols after decision. So far, the processing of one frame signal ends.
As shown in fig. 4, the total number of users is 500 and the number of active users is 50. The number of base station antennas is 128. The pre-equalization factor threshold is set to 0.2. The OAMP-MMV-SSL algorithm and GMMV-AMP algorithm are selected as comparison algorithms, and the advantages of high data demodulation accuracy under low time resource expense are verified.
(1) OAMP-MMV-SSL algorithm. The OAMP-MMV-SSL algorithm is designed for single antenna base stations in JADD access mode, which has excellent performance. Comparing the invention with OAMP-MMV-SSL algorithm shows the advantages brought by the invention using large-scale MIMO base station. (regarding OAMP-MMV-SSL algorithm, please refer to the literature "translation: joint user activity and data detection based on compressed sensing in massive Internet of things access", its author, english title is "Y.Mei et al.,"Compressive Sensing-Based Joint Activity and Data Detection for Grant-Free Massive IoT Access,"in IEEE Transactions on Wireless Communications,vol.21,no.3,pp.1851-1869,March 2022,doi:10.1109/TWC.2021.3107576.")
(2) GMMV-AMP algorithm. GMMV-AMP is a JADCE-based algorithm with excellent performance in a large bandwidth scenario. In addition, after the GMMV-AMP algorithm completes joint activity and channel estimation, the estimated channel is used to coherently detect the data sent by the user. Comparing the present invention with GMMV-AMP algorithm illustrates the advantage of the present invention that less time resources are required. (regarding coherent detection, please refer to the literature "translation: next generation low-delay high-reliability mass equipment access: a unified active passive random access semi-blind detection framework", its English author, titled is :M.Ke,Z.Gao,M.Zhou,D.Zheng,D.W.K.Ng and H.V.Poor,"Next-Generation URLLC With Massive Devices:A Unified Semi-Blind Detection Framework for Sourced and Unsourced Random Access,"in IEEE Journal on Selected Areas in Communications,vol.41,no.7,pp.2223-2244,July 2023,doi:10.1109/JSAC.2023.3280981.").
In order to ensure the fairness of algorithm comparison, the OAMP-MMV-SSL and the time domain resource occupied by one frame of data transmission of the invention are controlled, namely the number of OFDM symbols is equal to the number of pilot frequencies in the GMMV-AMP method. The number of the subcarriers occupied by the three methods is equal, and 60 is selected. In addition, since OAMP-MMV-ASL is also an access method assisted by pre-equalization, the pre-equalization factor setting method is the same as the present invention. The number of iterations in step 3.4 of the present invention is set to 3.
And measuring the data demodulation accuracy of each method by adopting a bit error rate index BER. BER is defined as: (total number of bits of the missing device + number of error bits of the correct detection device)/total number of transmission bits of the active user. Fig. 4 shows the BER performance of the three methods at different OFDM symbol count overheads. Wherein GMMV-AMP algorithm can not converge due to too few observations in case of an OFDM symbol number of 10. The invention has no error code in the simulation process when the OFDM symbol number is more than 40. As shown in fig. 4, the data demodulation performance has a clear advantage when the number of pilots is small. Compared with an OAMP-MMV-SSL algorithm, the method and the device provided by the invention can bring lower error rate by utilizing diversity gain of large-scale MIMO aiming at a large-scale MIMO scene. Compared with the method for estimating the active user and the channel at one time in the GMMV-AMP method, the method and the device separate the two steps, thereby ensuring the accuracy of the active user and the channel estimation and having obvious data demodulation performance advantages.
In summary, the above embodiments are only preferred embodiments of the present invention, and are not intended to limit the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (6)
1. The multi-user access identification detection method for the frequency domain pre-equalization in the time sensitive network comprises three steps of base station downlink broadcasting, user uplink transmission and data processing; the method is characterized in that:
Step 1, a base station is provided with a large-scale MIMO with the antenna number of N and communicates with users with the total number of K; the base station randomly selects an antenna, and broadcasts a marking signal to all devices by utilizing the single antenna; recording the serial number of the antenna used for downlink broadcasting as eta; after receiving the sign signal, the user performs power control, synchronization and channel estimation;
Step 2, after receiving the sign signal sent in the step 1, the active user obtains a value of a pre-equalization factor according to a downlink channel estimation result; then, when data is transmitted, adopting an orthogonal frequency division multiplexing mode for communication; the data sent on each subcarrier of the user with the sequence number k is the product of the constellation symbol, the pre-equalization factor and the value of the spreading sequence s k on the corresponding subcarrier; the length of the spread spectrum sequence is M subcarriers, and the spread spectrum sequences corresponding to all users are known at the base station end; active users continuously transmit T constellation symbols to a base station at the same time in each uplink transmission stage; the base station receives signals of T time slots;
Step 3, after receiving the signals sent by the active users in the step 2, the base station obtains an active user set, channel state information and T constellation symbols sent by each active user according to the received signals; and according to the active user set, the channel state information and T constellation symbols sent by each active user, simultaneously realizing active user identification detection, data detection and channel estimation, namely realizing large-scale MIMO unlicensed access based on pre-equalization assistance in a time-sensitive network.
2. The method for pre-equalization assisted multi-user low-delay access in a time-sensitive network of claim 1, wherein: in the step 2, the specific method for generating the pre-equalization factor is as follows; channel gain on the mth subcarrier estimated by user with sequence number kThe value of the pre-equalization factor lambda k on the m-th subcarrier is denoted lambda m,k; the value of lambda m,k is generated as follows;
3. the method for pre-equalization assisted multi-user low-delay access in a time-sensitive network of claim 2, wherein: in step 3, the signal processing procedure of the base station for one frame of signal comprises the following steps,
Step 3.1: combining user activity with the data rough estimate; for antennas transmitting the signature signal, the effect of the channel is counteracted by a pre-equalization operation by the user, the received signal being related only to the spreading sequence and to the data; therefore, the base station can obtain a set of active user sequence numbers and preliminary estimation of the transmitted constellation symbols according to the signals received by the antenna transmitting the sign signals; here, a compressed sensing method is used to obtain an estimate of the set of active user sequence numbersEstimation of transmitted constellation symbols
Step 3.2: channel estimation; the base station obtains the active user sequence number set according to the step 3.1Coarse data estimationObtaining an estimated value of an equivalent channel and an estimated value of noise power by utilizing signals received by all antennas of a base station; performing channel estimation by using a compressed sensing algorithm; wherein, the equivalent channel is the product of the real channel of each antenna and the corresponding channel of the antenna transmitting the marking signal;
Step 3.3: estimating data; the base station performs refined estimation on data sent by the user by utilizing signals received by all antennas according to the equivalent channel estimation value and the noise power estimation value obtained in the step 3.2;
Step 3.4: iteration between channel estimation and data estimation; replacing the data rough estimation obtained in the step 3.1 with the data estimation obtained in the step 3.3, and performing the step 3.2 and the step 3.3; iterating between step 3.2 and step 3.3 according to this method; terminating the iteration after the iteration number exceeds the maximum iteration number; obtaining an estimated value of the equivalent channel from the last executed step 3.3; obtaining an estimated value of the data from the last executed step 3.3, and simultaneously realizing active user identification detection, data detection and channel estimation, and estimating the estimated value of the constellation symbols And performing constellation symbol mapping, constellation symbol demodulation and conversion into a bit sequence, namely realizing the unlicensed access of the large-scale MIMO based on the pre-equalization assistance in the time-sensitive network.
4. A method of pre-equalization assisted massive MIMO multi-user low-delay access in a time sensitive network as recited in claim 3, wherein:
The specific form of the compression sensing method in the step 3.1 is as follows:
Yη=SX+Wη
Wherein Y η is the signal received by the antenna from which the base station transmits the flag signal; s= [ S 1,s2,…,sK ] is a matrix formed by spreading sequences of all users; x is a matrix formed by constellation symbols sent by a user, and elements of a kth row and a kth column of the matrix are matrices formed by Gaussian white noise, wherein X k,t;Wη is the element of the kth row and the kth column; because the uplink transmission has sparsity, the column vectors of X are sparse vectors; since the user's activity remains unchanged within a frame, all column vectors of X have the same support set, thus having structured sparsity; therefore, adopting GMMV-AMP compressed sensing algorithm to calculate posterior probability of constellation symbol x k,t, and taking the average value as estimated value of constellation symbol; the GMMV-AMP algorithm is also used to derive a posterior probability gamma k,t that each x k,t is a non-zero element; for users with average posterior probability greater than 0.5, determining that the users are active, and estimating to obtain a set of active user serial numbers
5. A method of pre-equalization assisted multi-user low delay access in a time sensitive network as recited in claim 3 wherein:
Channel estimation problem in step 3.2 the channel estimation problem is represented in the spatial domain as:
the compressed sensing problem comprises M equations in total, wherein each equation corresponds to a baseband equivalent model on one subcarrier; wherein the method comprises the steps of A signal received for the base station on the mth subcarrier; the number of elements in the set is represented by the symbol | c; each element of the sensing matrix on the m-th subcarrier is the product of the constellation symbol estimated in the step 3.1 and the spread spectrum sequence; /(I) Contains only the collection/>Constellation symbols and spreading sequences of the middle users; /(I)For the equivalent channel on the m-th subcarrier, the element of the equivalent channel is equal to the product of the real channel and the pre-equalization factor, and only comprises the set/>A channel of a user; /(I)Is gaussian white noise superimposed on the reception signal of the m-th subcarrier;
right multiplying fourier transform matrix on both sides of the channel estimation problem model To obtain an angular domain representation of the channel estimation problem:
Wherein the method comprises the steps of An angular domain representation of an equivalent channel on the mth subcarrier; an estimated value/>, of an angle domain equivalent channel is obtained by adopting the existing GMMV-AMP algorithmNoise power estimate/>And then reusedObtain the space domain equivalent channel estimation value/> Is gaussian white noise transformed into the angle domain.
6. A pre-equalization assisted multi-user low delay access method in a time sensitive network as recited in claim 3 wherein said step 3.3 performs data estimation according to the following equation:
Wherein the method comprises the steps of Representing signals received by an nth antenna of the base station; /(I)For the collection/>Estimate of equivalent channel from user to base station nth antenna,/>For the collection/>A matrix formed by spreading sequences corresponding to users; For the collection/> The user sends the estimated value of the constellation symbol; symbol/>The representation matrix is multiplied by elements; perception matrix/>Obtaining; /(I)Is Gaussian white noise superimposed on the n-th antenna reception signal;
Order the AndThe following overdetermined matrix equation is obtained:
For a pair of Performing LMMSE estimation to obtain/>The expression is as follows:
Wherein the method comprises the steps of Is a unitary matrix with the dimension of/> The estimated noise variance for step 3.2.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410230192.9A CN118019047A (en) | 2024-02-29 | 2024-02-29 | Multi-user access identification detection method for frequency domain pre-equalization in time sensitive network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410230192.9A CN118019047A (en) | 2024-02-29 | 2024-02-29 | Multi-user access identification detection method for frequency domain pre-equalization in time sensitive network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN118019047A true CN118019047A (en) | 2024-05-10 |
Family
ID=90950362
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410230192.9A Pending CN118019047A (en) | 2024-02-29 | 2024-02-29 | Multi-user access identification detection method for frequency domain pre-equalization in time sensitive network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN118019047A (en) |
-
2024
- 2024-02-29 CN CN202410230192.9A patent/CN118019047A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107483088B (en) | Large-scale MIMO robust precoding transmission method | |
CN101378277B (en) | Multi-user pre-coding and dispatching method and realize the base station of the method | |
CN111106859A (en) | Millimeter wave/terahertz network large-scale MIMO wireless transmission method | |
CN111698182A (en) | Time-frequency blocking sparse channel estimation method based on compressed sensing | |
Thompson | Deep learning for signal detection in non-orthogonal multiple access wireless systems | |
CN101427485A (en) | Reduced complexity beam-steered MIMO OFDM system | |
EP2727304B1 (en) | Noise estimation filter | |
WO2017219389A1 (en) | Methods for sending and receiving synchronization signals and signals subjected to perfect omnidirectional pre-coding in large-scale mimo system | |
CN106612135B (en) | Signal sending method, receiving method and device based on multi-carrier spatial modulation | |
CN102223327B (en) | SAGE (Space-alternating Generalized Expectation-maximization) based channel estimation method in CoMP multi-user system | |
CN110881010B (en) | Statistical CSI-assisted multi-user NOMA downlink transmission method | |
CN104737481A (en) | Transmitter and wireless communication method | |
CN104702390A (en) | Pilot frequency distribution method in distributed compressive sensing (DCS) channel estimation | |
CN105429688A (en) | Multi-cell pre-coding method for inhibiting pilot pollution in large-scale distributed antenna system | |
CN105991219B (en) | Method and device for processing transmission signal | |
CN114665930A (en) | Downlink blind channel estimation method of large-scale de-cellular MIMO system | |
CN108600125B (en) | Channel estimation method based on iteration | |
Shin et al. | Blind channel estimation for MIMO-OFDM systems using virtual carriers | |
Ouchikh et al. | Iterative channel estimation and data detection algorithm for MIMO-OTFS systems | |
CN101719816A (en) | Method for realizing low feedback velocity of self-adaptive MIMO-SCFDE system | |
CN118019047A (en) | Multi-user access identification detection method for frequency domain pre-equalization in time sensitive network | |
CN110492956B (en) | Error compensation multi-user detection method and device for MUSA (multiple input multiple output) system | |
Liu et al. | A distributed detection algorithm for uplink massive MIMO systems | |
CN104734756A (en) | MIMO system detection method and device | |
CN105207701B (en) | For the coding/decoding method based on ICA in multiple cell multi-user multi-aerial system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |