On-demand fast authorization and multi-domain a priori equalization method for multi-terminal concurrent communication network

Abstract

The application discloses a multi-terminal concurrent communication network on-demand fast authorization and multi-domain priori balancing method, belongs to the technical field of communication networks, and particularly relates to a concurrent communication method for realizing single handshake, i.e., completing on-demand fast authorization of resources, by using multi-domain priori probability information in a multi-terminal concurrent communication scene, which can be applied to optical communication bus networks, industrial internet wireless multiple access, wired / wireless time fusion sensitive networks and the like. The method analyzes the structures of different domains in a communication mechanism, fully utilizes the priori sparse characteristics of different domains, and accumulates multi-channel observation and transmission resource observation through a frame structure of multi-channel observation-transmission resource joint expansion. Therefore, the activity detection of concurrent terminals, channel estimation and data demodulation can be efficiently realized under a limited number of hardware array type acquisition units, i.e., an efficient on-demand fast authorization concurrent access is realized in a multi-terminal concurrent communication network.

Description

Technical Field

[0001] This invention relates to the field of communication network technology, specifically to on-demand rapid authorization and multi-domain prior balancing methods for multi-terminal concurrent communication networks. Background Technology

[0002] Based on the concept of "Internet of Everything," future communication technologies are expected to integrate the physical and digital worlds, gradually realizing the vision of "Intelligent Internet of Everything," thus ushering in a ubiquitous and intelligent era of broadband communication. The deep integration of wide area networks and local access networks has become an ideal platform to support parallel communication across multiple terminals. At the same time, communication centered on massive machine-type communications is experiencing explosive growth. Its typical characteristics of parallel and bursty access across multiple terminals pose severe challenges to the access capacity, resource scheduling flexibility, and concurrent processing capabilities of communication networks.

[0003] The high density of concurrent terminals and their bursty characteristics make traditional scheduling-dependent sequential authorization handshake protocols unsuitable for meeting the practical requirements of efficient, ultra-low latency communication due to frequent handshakes and complex control signaling. Therefore, single-handshake access technologies and non-orthogonal resource sharing technologies are now widely used in multi-terminal access scenarios to address the problems of excessive access latency and excessive multi-dimensional resource consumption by devices due to transmission resource allocation. This single-handshake method combines authorization requests with uplink transmission, and the central node completes on-demand authorization in real time through efficient conflict inference and processing of concurrent signals. Furthermore, large-scale multidimensional multiplexing technology can increase the number of concurrently accessing terminals through the multidimensional multiplexing capability of multiple transceiver units, and the high-dimensional resource resolution capability allows the link to exhibit sparse prior characteristics in a specific transform domain, thereby promoting the flexible application of algorithms based on prior probability models in access technologies.

[0004] Based on channel state, single-handshake multi-terminal access schemes are mainly divided into two types. When the channel is known at the central node, devices are allowed to carry a small amount of data when sending access requests. The central node needs to simultaneously identify active devices and detect their transmitted data based on the aliased access signals, thus completing joint activity and data detection (JADD). When the channel is unknown, the signal sent by the device consists of two parts: a pilot sequence and data. The central node needs to first identify active devices and estimate their channels based on the received pilot sequence, thus completing joint activity detection and channel estimation (JADCE). Then, the central node uses the estimated channel state information to perform data detection on active access devices. Since future communication aims to increase the number of devices that can access simultaneously, for traditional multiple access, if the condition of a processing dimension greater than or equal to the number of devices still needs to be met (well-determined or over-determined), the required computing and hardware resources will become a huge burden. In addition, when the number of access devices is large, if the link characteristics of two terminals are highly correlated in a certain critical domain, it will lead to ill-conditioned equivalent channel matrix, causing a decrease in detection performance.

[0005] Therefore, how to achieve a fast, on-demand concurrent access solution in scenarios with a large number of connected devices is an unsolved problem. Summary of the Invention

[0006] In view of this, the present invention provides an on-demand fast authorization and multi-domain prior equalization method for multi-terminal concurrent communication networks. This method analyzes the structure of different domains in the communication mechanism, fully utilizes the prior sparsity characteristics of different domains, and accumulates multi-channel observations and transmission resource observations through a frame structure jointly extended by multi-channel observations and transmission resources. Thus, with a limited number of hardware array acquisition units, it can efficiently realize the activity detection, channel estimation, and data demodulation of concurrent terminals, that is, achieve an efficient on-demand fast authorization concurrent access in multi-terminal concurrent communication networks.

[0007] To achieve the above objectives, the technical solution of this invention is: an on-demand fast authorization and multi-domain prior equalization method for multi-terminal concurrent communication networks. This method is applied to a single-handshake uplink data transmission scenario where a central node terminal serves multiple concurrent devices. The central node is equipped with a multi-array acquisition unit and includes the following steps:

[0008] Step 1: A certain number of active terminals use a single handshake protocol to send pilot signals concurrently to the central node at the same time and using the same transmission resources.

[0009] Step 2: The central node, equipped with multiple array-type acquisition units, after receiving the pilot signals of all active terminals in Step 1, realizes active terminal identification and detection and link status estimation through alternating iteration, obtains the index set of active terminals and the link status information of active terminals, and completes the authorization of active terminals.

[0010] Step 3: The active terminals authorized in Step 2 use a single handshake protocol and a multi-channel observation-transmission resource domain joint extended frame structure to send their respective data symbols concurrently to the central node.

[0011] Step 4: The central node receives the mixed data signals from each authorized active terminal. Based on the active terminal index set finally determined in Step 2 and the estimated link state information, it separates and demodulates the data symbols sent by each active terminal from the mixed signal, thereby completing the data detection.

[0012] Furthermore, the method is applied to a single-handshake uplink data transmission scenario where the central node terminal serves multiple concurrent devices. Specifically, the number of array-type acquisition units in the central node is... , For row direction dimension, As a column-oriented dimension, the total number of services is M The terminal has an array of data acquisition units of [number]. , For row direction dimension, It is a column-oriented dimension; OFDM modulation is used for uplink transmission; the spreading sequence of each terminal contains K A transmission resource unit, represented as , where row vector For the serial number m The spreading sequence corresponding to the terminal, For dimension is M×K The set of complex numbers, This is the set of spreading sequences corresponding to all terminals.

[0013] Furthermore, in step 1, the specific implementation process is as follows: the quantity is The active terminals use a single handshake protocol to send pilot symbols to the central node, which are then processed by baseband for uplink pilot transmission. OFDM communication is used when sending pilot symbols, with sequence numbers as follows: terminal The same symbol is transmitted on each transmission resource unit (CRU), and the signal transmitted in each CRU is a constellation symbol and a spreading sequence. The product of; the length of the spreading sequence is Each transmission resource unit has a spread spectrum sequence known at the central node for all terminals; active terminals simultaneously and continuously transmit data to the central node during each uplink pilot transmission phase. The transmission of constellation symbols; the central node receives Pilot signals in one time slot.

[0014] Furthermore, in step 1, the quantity is The active terminals send several frames of data to the central node using a single handshake protocol. After baseband processing, these frames are transmitted via uplink pilot signals. The implementation method is as follows: Each terminal is assigned a unique but non-orthogonal spreading sequence with a length of [missing information]. One transmission resource unit, i.e., the sequence number is m The spreading sequence corresponding to the terminal is First, the terminal converts the data bits into constellation symbols; the terminal transmits in a certain frame... T In the OFDM pilot symbols, each OFDM corresponds to a constellation symbol; the symbol is... m The terminal sends the first frame in a certain frame. t Each constellation symbol is represented as Then, the active terminal sends the first... t The frequency domain sequence corresponding to each OFDM symbol is .

[0015] Furthermore, in step 2, active terminal identification and link state estimation are achieved through alternating iterations. The noisy signals received by the array-type acquisition units of all central nodes in the multi-channel observation-transmission resource domain are used as the initial input signals for the iterations. Specifically, the following iterative steps are executed:

[0016] Step 2.1: Use compressed sensing algorithm for the input signal to obtain the posterior active probability or confidence level of each terminal. Based on the preset threshold, the terminals are initially divided into reliable set and rough set.

[0017] Step 2.2: Utilize the inverse DFT to transform the multi-channel observation-transmission resource domain channel and received signal into the virtual space-time lag domain channel and received signal. Using the noisy signals received by the transmission resource units of all central nodes in the virtual space-time lag domain and employing a compressed sensing algorithm, obtain the virtual space-time lag domain channel. The estimated values ​​are obtained by using DFT transform to obtain the multi-channel observation-transmission resource domain channel. The estimated value.

[0018] Step 2.3: Multiply the channel estimate obtained in Step 2.2 with the pilot signal to obtain the estimated noisy observation. Subtract the estimated noisy observation from the original signal in Step 2.1 to obtain the residual signal, which is used as the input signal for Step 2.1 in the next iteration.

[0019] The above three steps 2.1-2.3 are executed iteratively until the maximum number of iterations is met or the residual signal energy is lower than a set threshold; finally, the rough set obtained from the last iteration is... As the final set of active terminal indexes The equivalent multi-channel observation-transmission resource domain channel estimate is obtained from step 2.2 of the last execution. It simultaneously detects active terminals and estimates channels, obtains the index set of active terminals and the link state information of active terminals, and completes the authorization of active terminals.

[0020] Furthermore, in step 2.1, the MMV-AMP algorithm is used to obtain the posterior active probability of the terminal. Set a high threshold and low threshold When the posterior active probability of a terminal Greater than the low threshold Furthermore, the average value across all interfaces and transmission resource units is greater than the proportional threshold. If so, then this terminal is considered to belong to the rough set. When the posterior active probability of a terminal Greater than the high threshold Furthermore, the average value across all interfaces and transmission resource units is greater than the proportional threshold. If so, then the terminal belongs to the reliable set. .

[0021] Further, step 4 specifically involves the central node using the active terminal index set obtained in step 2. and the estimated multi-channel observation-transmission resource domain equivalent channel By using a linear model, the data sent by the terminal is estimated by combining the signals received by all array-type acquisition units.

[0022] Furthermore, the linear model used in step 4 is:

[0023]

[0024] in, Indicates the central node number n Each array-type acquisition unit receives continuous T Data signals in each time slot; Indicates active terminal set Data symbols transmitted uplink from the terminal Indicates active terminal set From the middle terminal to the central node n Estimated value of the equivalent channel of the multi-channel observation-transmission resource domain for each array-type acquisition unit; Indicates superposition on then A matrix composed of Gaussian white noise on the signal received by an array of acquisition units;

[0025] The matrices formed by the three are as follows:

[0026] ,

[0027] as well as

[0028] ;

[0029] Then we have:

[0030]

[0031] right Perform LMMSE estimation to obtain The expression is as follows:

[0032]

[0033] in It is the identity matrix. This represents the estimated number of elements in the active terminal set; ultimately, the data estimate is... The signal is then converted into constellation symbols and then into data bits; this completes the processing of one frame of signal.

[0034] Beneficial effects:

[0035] 1. Existing on-demand rapid authorization concurrent access solutions face challenges in high-density connection scenarios. Especially when the number of active devices exceeds the number of central node array acquisition units, the detection problem becomes underdetermined and difficult to solve. Furthermore, the high correlation between channels between terminals in the uplink and the high dimensionality of the channel matrix severely degrade detection performance and introduce high computational latency. This invention proposes a frame structure and a novel resource partitioning strategy based on multi-channel observation-transmission resource joint expansion. This design is specifically designed to solve the large-scale concurrent access problem in uplink transmission, enabling effective data detection. Therefore, the frame structure and resource partitioning method of this invention can effectively address the underdetermined problem in high-density scenarios and reduce processing latency.

[0036] 2. Existing concurrent access schemes do not fully utilize the unique channel characteristics that may exist in the communication link, such as sparsity in specific transform domains, resulting in poor performance of active terminal detection and channel estimation. This invention models uplink access as a compressed sensing problem and fully utilizes multi-domain sparsity. To this end, this invention designs an efficient multi-domain alternating active terminal detection and channel estimation scheme, implemented through the proposed Iterative Residual Feedback Multiple Measurement Vectors (MMV)-AMP algorithm. This algorithm specifically utilizes the sparse structures of these two different domains by alternating execution between the detection and estimation phases. Therefore, this invention achieves more efficient and accurate terminal activity detection and channel estimation by fully utilizing the unique multi-domain sparsity of the communication uplink.

[0037] 3. In traditional joint active terminal detection and channel estimation, errors accumulate in the initial detection and estimation stages, making it difficult to recover terminals with weak signals or those previously undetected in subsequent iterations. The iterative residual feedback mechanism proposed in this invention first reconstructs the signal components of reliably detected terminals based on high-precision channel estimation. Then, this reconstructed signal is subtracted from the total received signal to generate a sparser received signal for the next iteration. The detection stage processes this received signal in subsequent iterations, enabling more accurate identification of previously missed terminals. The detection and estimation stages are executed alternately until convergence. Therefore, this residual feedback mechanism effectively suppresses error propagation, iteratively refines the detection and estimation results, significantly improves the accuracy of terminal detection and channel estimation, and thus delivers superior data detection performance. Attached Figure Description

[0038] Figure 1 This is a flowchart of the on-demand fast authorization and prior probability multi-domain balancing method for multi-terminal concurrent communication networks disclosed in this example.

[0039] Figure 2 The performance curves of the mean square error (MSE) of channel estimation under different effective pilot lengths in the embodiments of the present invention are shown, and compared with the OAMP-MMV and SOMP algorithms.

[0040] Figure 3 This is a comparison chart of the channel estimation performance of this invention with two other schemes, using MSE as the evaluation metric under different signal-to-noise ratios.

[0041] Figure 4 The bit error rate (BER) is used as the evaluation metric for different numbers of active terminals. This example compares the data detection performance with two other comparison schemes. Detailed Implementation

[0042] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0043] This invention provides an on-demand fast authorization and multi-domain prior equalization method for multi-terminal concurrent communication networks. This method is applied to a single-handshake uplink data transmission scenario where a central node terminal equipped with multi-array acquisition units serves multiple concurrent devices. The method includes the following steps:

[0044] Step 1: A certain number of active terminals use a single handshake protocol to send pilot signals concurrently to the central node at the same time and using the same time-frequency resources;

[0045] Step 2: The central node, equipped with multiple array-type acquisition units, after receiving the pilot signals from all terminals in Step 1, realizes active terminal identification and detection and link status estimation through alternating iteration, obtains the index set of active terminals and the link status information of active terminals, and completes the rapid authorization of active terminals.

[0046] Step 3: Active terminals identified and authorized in Step 2 use a single handshake protocol and interface-frequency joint extended frame structure similar to those in Step 1 to concurrently send their respective data symbols to the central node.

[0047] Step 4: The central node receives the mixed data signals from each active terminal. Based on the active terminal index set finally determined in Step 2 and the estimated link state information, it separates and demodulates the data symbols sent by each active terminal from the mixed signals, thereby completing the data detection.

[0048] This embodiment considers a scenario where a central node terminal equipped with a multi-array acquisition unit serves multiple concurrent devices in a single-handshake uplink data transmission scenario, such as... Figure 1 As shown. The number of array-type acquisition units in the central node is... , For row direction dimension, As a column-oriented dimension, the total number of services is M The terminal has an array of data acquisition units of [number]. , For row direction dimension, This is a column-oriented dimension. OFDM modulation is used for uplink transmission. The spreading sequence for each terminal contains... K A transmission resource unit, represented as , where row vector For the serial number m The spreading sequence corresponding to the terminal. For dimension is M×K The set of complex numbers, This is the set of spreading sequences corresponding to all terminals.

[0049] This embodiment considers the following channel model:

[0050]

[0051] in, Indicates the sequence number is m Between the terminal and the central node, corresponding to the first A channel matrix with time lag components; the channel of each terminal contains Q Path; The serial number is m The terminal corresponding to the first Complex gain of the path, The serial number is m The terminal corresponding to the first The relative channel delay of the path use Pulse shaping filter with sampling interval, It is the guide vector of the central node. The serial number is m The terminal corresponding to the first The guide vector of the path, This represents the guide vector in the row or column direction, where , , and For intermediate referential variables, For wavelength, It is the spacing between array-type acquisition units. and They are respectively the serial numbers m Given the azimuth and pitch angles of arrival from the terminal to the center node, the steering vector of the center node is calculated as follows: Similarly, we can obtain the sequence number as m The guiding vector of the terminal. For the serial number m The terminal corresponding to the first The azimuth starting angle of the path to the central node. For the serial number m The terminal corresponding to the first The pitch departure angle of the path to the center node.

[0052] Performing a Fourier transform on the above channel model yields the sequence number as follows: m The terminal in the Multi-channel observation on a transmission resource unit - transmission resource domain channel:

[0053]

[0054] in, Let L be the total number of transmission resource units and L be the total number of time-lag components of the channel. To simplify the channel coefficients without loss of generality, in the line-of-sight path, let L = L. In non-line-of-sight paths, let , It is the power distribution factor. Define beamformer as , For the serial number m The azimuth departure angle of the line-of-sight path from the terminal to the central node. For the serial number m The pitch departure angle of the line-of-sight path from the terminal to the central node; For the serial number m The steering vector on the terminal line-of-sight path, and the channel vector of the multi-channel observation-transmission resource domain are... The channel tensor of the multi-channel observation-transmission resource domain is And satisfy ,as well as This is the channel matrix of the multi-channel observation-transmission resource domain from the terminal to the central node with sequence number m.

[0055] In existing two-stage transmission schemes, when the number of active devices exceeds the number of array-type acquisition units at the central node, the data detection problem becomes underdetermined, making it difficult to solve. Furthermore, the increased channel correlation between high-density terminals severely degrades detection performance. A frame structure based on the joint extension of multi-channel observation and transmission resource domains effectively performs data detection and supports access from more terminals by expanding single-dimensional observation into a product of multi-channel observation and transmission resource observation. The specific scheme is as follows: In uplink transmission, all... F Division of transmission resource units P Each transport resource group contains [number] transport resource groups. One transmission resource unit, and The integer is used; a transport resource block is formed by extracting one transport resource with the same index from each transport resource group. Transport resource units within the same transport resource block are evenly distributed, and the same constellation symbol is transmitted on each transport resource unit within the same transport resource block; for the first... A central node array-type acquisition unit, from the full transmission resource channel matrix Extract the channel matrix corresponding to the selected transmission resource block. The method is z is the index of each extracted transmission resource unit, where It is a transmission resource unit selection matrix, used to extract K specific transmission resource units from all F transmission resource units to form a transmission resource block; Each column contains a non-zero element located at the index. At that point, that is, for the channel matrix ,when When the condition is met, it is 1; otherwise, it is 0. Finally, the resource block-level multi-channel observation-transmission resource domain channel tensor is defined as... In its first n The slice of each array acquisition unit is .

[0056] This embodiment discloses a multi-device access method in a central node communication network based on a multi-channel observation-transmission resource domain joint extended frame structure and multi-domain sparsity. The specific implementation steps are as follows:

[0057] Step 1: Quantity is The active terminals use a single handshake protocol to send pilot symbols to the central node, which are then processed by baseband for uplink pilot transmission. OFDM communication is used when sending pilot symbols, with sequence numbers as follows: terminal The same symbol is transmitted on each transmission resource unit (CRU), and the signal transmitted in each CRU is a constellation symbol and a spreading sequence. The product of; the length of the spreading sequence is Each transmission resource unit has a spread spectrum sequence known at the central node for all terminals; active terminals simultaneously and continuously transmit data to the central node during each uplink pilot transmission phase. The transmission of constellation symbols; the central node receives Pilot signals in one time slot.

[0058] Furthermore, in step 1, the active terminal sends several frames of data to the central node using a single handshake protocol. The uplink data transmission method using baseband processing is as follows: each terminal is assigned a unique but non-orthogonal spreading sequence with a length of [missing information]. One transmission resource unit, i.e., the sequence number is m The spreading sequence corresponding to the terminal is First, the terminal converts the data bits into constellation symbols; the terminal transmits in a certain frame... T In the OFDM pilot symbols, each OFDM corresponds to a constellation symbol; therefore, the symbol is... m The terminal sends the first frame in a certain frame. t Each constellation symbol is represented as superscript p This represents the pilot signal; then, the first signal sent by the active terminal... t The frequency domain sequence corresponding to each OFDM symbol is .

[0059] Step 2: After receiving the pilot signals sent by the active terminals in Step 1, the central node derives the active terminal set and channel state information based on the received signals. Based on the active terminal set and channel state information, it alternately performs active terminal identification and detection and channel estimation. The result of superimposing the signals sent by the terminals in Step 2 at the central node is shown in the following formula:

[0060]

[0061] in, The first central node n The array-type acquisition unit in the first t Pilot signals received from all active terminals in each time slot, superscript b∈{p,d} As a transmission stage identifier, when b=p The time indicates the pilot transmission phase, when b=d "Time" indicates the data transmission stage; For the first m The number of active terminals and the central node n Multi-channel observation-transmission resource domain sub-channels between array-type acquisition units For the channel, for transpose, To extract the m-th column; ;definition As the central node n The equivalent channel matrix of the multi-channel observation-transmission resource domain on each array-type acquisition unit, and satisfying the following conditions: As the equivalent channel tensor of the multi-channel observation-transmission resource domain No. n A slice of an array-type acquisition unit; ; For the first t The first time slot center node n Additive white Gaussian noise (AWGN) at each array acquisition unit, with each element following an independent complex Gaussian distribution. .symbol This indicates that the matrix is ​​multiplied element by element.

[0062] The vector form above can be transformed into the matrix form below:

[0063]

[0064] in, The central node is represented by the first... n Continuous data received by an array-type acquisition unit G Pilot signals in one time slot; Indicates continuous active terminals G Pilot signals transmitted in each time slot; Indicates the central node number n Continuous at each array-type acquisition unit G This invention utilizes a time-slot AWGN to estimate the active terminal set and multi-channel observation-transmission resource domain channel state information based on noisy observations and known pilot sequences. These correspond to the three main steps of the central node algorithm. The derivation process of the model for each of the three steps is described below.

[0065] Since the transmitted pilot sequence is known, the problem becomes a classic linear model. By using the noisy signals received by the array-type acquisition units of all central nodes in the multi-channel observation-transmission resource domain and employing a compressed sensing algorithm, the probability of terminal activity can be obtained. This step is called active terminal detection, denoted as step 2.1.

[0066] After step 2.1, preliminary results of activity detection have been obtained. Considering the sparsity of the virtual space domain and the time lag domain, the multi-channel observation-transmission resource domain channel and received signal are transformed into the more sparse virtual space-time lag domain channel and received signal using the inverse DFT transform. By utilizing the noisy signals received by the transmission resource units of all central nodes in the virtual space-time lag domain and employing a compressed sensing algorithm, the virtual space-time lag domain channel can be obtained. The estimated value is used to obtain the virtual space-time lag domain channel using DFT transformation. The estimated value. This step is called channel estimation, denoted as step 2.2.

[0067] Furthermore, multiplying the channel estimate obtained in step 2.2 with the pilot signal yields a result that is closer to the original noisy observation. Subtracting the estimated noisy observation from the original signal in step 2.1 makes the received signal sparser, resulting in a more accurate activity detection result. This sparser received signal can lead to a more accurate channel estimation result in the channel estimation stage. This process can be iterated between the activity detection stage and the channel estimation stage, simultaneously improving the performance of both. This iterative process is considered a step and is denoted as step 2.3.

[0068] The following sections introduce the classical linear models and solution methods corresponding to steps 2.1 and 2.2, and describe the detailed execution process of step 2.3.

[0069] Step 2.1: Activity Detection; for sequence number... The central node array acquisition unit receives noisy observations as shown in the following formula.

[0070]

[0071] The uplink in the central node communication network has a unique propagation environment, and the multi-channel observation-transmission resource domain channel to be estimated... Essentially, it exhibits multi-level sparsity. First, analyzing terminal activity, only a few terminals are active at any given time, which is the sparsity of terminal traffic; second, for the central node... n The array-type acquisition unit, in the... k The channel matrix observed on each transmission resource unit is sparse, and all transmission resource units share the same sparse pattern, i.e. This phenomenon is the sparse structure of the transmission resource domain in large-scale uplink access; finally, since the active probability of the sub-channels corresponding to different array acquisition units is the same, the channel matrix also shares the same sparse pattern in the multi-channel observation domain, that is... In summary, the multi-channel observation-transmission resource domain channel exhibits two-dimensional structured sparsity due to the active sparsity of the terminals. In a single-handshake random access scheme, the data must be determined based on noisy measurements. and known pilot Estimated active terminal set Due to the structured sparsity of the multi-channel observation-transmission resource domain, the active terminal detection problem can be formulated as an MMV compressed sensing problem.

[0072] This invention relates to the elements in the channel matrix. (matrix The ( m, k The minimum mean square error estimation is used for each element, which means calculating the mean of the marginal posterior probabilities. And calculate the posterior distribution based on a Bayesian framework. Channel elements The prior distribution is expressed by the following formula:

[0073]

[0074] in, Sparsity represents The prior non-zero probability, Represents element It follows a complex Gaussian distribution, where It is the mean. It is variance, express The probability density component with a value of zero; marginal probability distribution calculated based on factor graph and sum-product algorithm. High-dimensional integration is required, resulting in enormous computational costs and difficulty in solving the problem. Therefore, this invention employs the AMP algorithm to calculate the marginal posterior probability distribution. The AMP algorithm approximates the calculation using a low-complexity heuristic. This paper employs the existing MMV-AMP algorithm to solve the activity detection problem. (For more information on the MMV-AMP algorithm, please refer to the reference "Translated title: Adaptive Active User Detection and Channel Estimation Method Based on Compressed Sensing: Massive Access Meets Massive MIMO", cited as "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."). Using the MMV-AMP algorithm, terminal representations can be obtained. The probability of a non-zero posterior time. When the posterior active probability of a certain terminal Greater than the low threshold (like =0.3) and the average value across all array-type acquisition units and transmission resource units is greater than the proportional threshold. If so, then this terminal is considered to belong to the rough set. When the posterior active probability of a certain terminal Greater than the high threshold (like =0.9) and the average value across all array-type acquisition units and transmission resource units is greater than the proportional threshold. (like If the value is 0.9, then the terminal is considered to belong to the reliable set. The basis for the above judgment can be expressed by the following formula:

[0075]

[0076] in, i Let P represent the reliable set obtained in the i-th iteration, P represent the total number of transmission resource units, M represent the total number of terminals, and the threshold function represent the threshold function. It means that when hour, ,otherwise, .

[0077] Step 2.2: Channel estimation; by superimposing the sequence number from step 2.1. Multi-channel observation-transmission resource domain signal reception of the central node array acquisition unit The multi-channel observation-transmission resource domain received signal tensor of the central node is obtained. The formula is:

[0078]

[0079] in, This represents additive white Gaussian noise (AWGN) at the center node. Let be the equivalent channel tensor. Besides the structured sparsity of the multi-channel observation-transmission resource domain, the large distance between the central node and the terminal results in very small spread, with significant energy existing only in a few directions, causing the virtual space domain to exhibit obvious cluster sparsity. Simultaneously, the channel in the time-lag domain consists of line-of-sight components and a few non-line-of-sight components, exhibiting significant cluster sparsity. The equivalent virtual space-time-lag domain channel tensor is defined as: The formula is as follows:

[0080]

[0081] in, , , and It is a normalized DFT matrix; Indicates conjugate; and These represent the 3-modulus product and 2-modulus product of the tensor, respectively; similarly, the received signal tensor in the virtual space-time lag domain... It can be calculated using the formula below:

[0082]

[0083] in, The virtual space-time lag domain received signal tensor is sliced ​​to obtain the first... n The virtual space-time lag received signal matrix corresponding to each interval The formula is as follows:

[0084]

[0085] in, Indicates the first n The equivalent virtual space-time lag domain channel matrix corresponding to each interval; Indicates the first n The noise slices correspond to each interval. This is the linear model for channel estimation in step 2.2. In a single-handshake random access scheme, the noise slices must be determined based on the noisy measurements. and known pilot Estimating the virtual space-time lag domain equivalent channel and via channel tensor The virtual space-time lag domain equivalent channel tensor is obtained from the DFT transform. ,Right now Due to the aforementioned enhanced cluster sparsity, the channel estimation problem can be formulated as an MMV compressed sensing problem. The subsequent process for channel estimation using compressed sensing is similar to step 2.1, also utilizing the MMV-AMP algorithm; simply all the... Change to That's all. For matrix The ( m, k ( ) elements. The MMV-AMP algorithm can obtain an estimate of the virtual space-time lag domain equivalent channel. Finally, through the aforementioned domain transformation, the estimated value of the equivalent channel in the multi-channel observation-transmission resource domain can be obtained. .

[0086] Step 2.3: Iteration; Initialization definition: Reliable set It is an empty set, that is The first iteration ( The received signal in step 2.1 of the process is: .

[0087] Iteration begins ( Both the rough set and the cancellation set are empty sets, i.e. In the activity detection phase, a rough set is obtained according to step 2.1. and reliable set The rough set obtained in step 2.1 and reliable set Incoming channel estimation stage; according to step 2.2, the rough set is... Channel estimation using the MMV-AMP algorithm yields the virtual space-time lag domain equivalent channel. That is, the formula below:

[0088]

[0089] in, ; ; The multi-channel observation-transmission resource channel is obtained after the domain transformation in step 2.2. ;according to The proportion of the reliable set obtained in step 2.1 is randomly selected. Part as offset set ,Right now , It can be set to 0.8; cancels the pilot signal of the set terminal. Multiply with the estimated offset set terminal's multi-channel observation-transmission resource domain equivalent channel The received signal is canceled; the original received signal is obtained. The residual obtained by subtracting the canceled received signal is the received signal for activity detection in step 2.1 of the next iteration. This makes the received signal residuals more sparse; the next iteration ( i +1) In, The activity detection phase of the next iteration is then passed in, and activity detection is performed according to step 2.1, that is, the MMV-AMP algorithm is applied to the following formula to obtain the estimated active terminals.

[0090]

[0091] in, It is the first i The received signal residual of the next iteration; ,in , The process involves using the judgment in step 2.1 to determine whether the detected terminal belongs to the rough set or the reliable set, thereby updating the rough set and the reliable set. Then, the rough set and the reliable set are passed to step 2.2 for channel estimation. This method iterates repeatedly between steps 2.1 and 2.2; the iteration terminates when the number of iterations exceeds the maximum number of iterations or the energy of the received signal falls below a threshold. The rough set obtained from the last execution of step 2.1 is then used for estimation. As the final set of active terminal indexes The equivalent multi-channel observation-transmission resource domain channel estimate is obtained from step 2.2 of the last execution. It simultaneously achieves active terminal detection and channel estimation.

[0092] Step 3: Quantity is Active terminals send data symbols to the central node using a one-way handshake protocol, which are then processed by baseband for uplink data transmission; similar to step 1, data symbols are used... Replace the pilot symbol in step 1 Furthermore, the received signal is also changed accordingly. The remaining processing is the same as in step 1, and will not be repeated here.

[0093] Step 4: The central node uses the active terminal index set obtained in Step 2. and the estimated multi-channel observation-transmission resource domain equivalent channel By using a linear model, the data sent by the terminal is estimated by combining the signals received by all array-type acquisition units.

[0094] The linear model used in step 4 is:

[0095]

[0096] in, Indicates the central node number n Each array-type acquisition unit receives continuous T Data signals in each time slot; Indicates active terminal set Data symbols transmitted uplink from the terminal Indicates active terminal set From the middle terminal to the central node n Estimated value of the equivalent channel of the multi-channel observation-transmission resource domain for each array-type acquisition unit; Indicates superposition on the n The matrix formed by the Gaussian white noise on the signal received by the array-type acquisition units is as follows: , as well as Then we have:

[0097]

[0098] right Perform LMMSE estimation to obtain The expression is as follows:

[0099]

[0100] in It is the identity matrix. This represents the estimated number of elements in the active terminal set. Finally, the data estimates are... The signal is then converted into constellation symbols, and then into data bits. This completes the processing of one frame of signal.

[0101] In this embodiment of the invention, for a total of 500 terminals, 50 active terminals, and 5×5=25 central node array acquisition units, the OAMP-MMV algorithm and SOMP algorithm from the following two articles are selected as comparison algorithms.

[0102] The OAMP-MMV algorithm is designed for edge nodes under the JADCE access mode and has excellent performance. Please refer to: “K. Ying et al., 'Quasi-Synchronous Random Access for Massive MIMO-Based LEO Satellite Constellations,' in IEEE Journal on SelectedAreas in Communications, vol. 41, no. 6, pp. 1702-1722, June 2023, doi:10.1109 / JSAC.2023.3273699.”

[0103] SOMP algorithm. Please refer to the literature: "X. Zhouet al., 'Active Terminal Identification, Channel Estimation, and Signal Detection for Grant-Free NOMA-OTFS in LEOSatellite Internet-of-Things,' in IEEE Transactions on WirelessCommunications, vol. 22, no. 4, pp. 2847-2866, April 2023, doi: 10.1109 / TWC.2022.3214862."

[0104] The advantages of this invention by utilizing the AMP algorithm are illustrated by comparing it with the SOMP algorithm.

[0105] To ensure fairness in the algorithm comparison, the pilot overhead (i.e., the number of OFDM symbols) used in one frame of data transmission in this invention is controlled to be equal to the length of the intersymbol interference-free region in the OAMP-MMV and SOMP methods. The number of system transmission resource units used by the three methods is equal, all chosen to be 512. Mean Square Error (MSE) is used to measure the accuracy of channel estimation for each method. MSE is defined as: the sum of the Frobenius norms of the interpolation between the estimated channel matrix and the true channel matrix, divided by the total number of elements. Figure 2 The diagram shows a comparison of the channel estimation performance of this example with two other comparative schemes under different effective pilot lengths. Figure 3The diagram shows a comparison of the channel estimation performance of this example and two other schemes under different signal-to-noise ratios (SNRs). The OAMP-MMV algorithm fails to achieve ideal results due to its limited effective pilot overhead, while SOMP suffers from poor performance due to the lack of prior information. This invention fully utilizes the cooperative gain across different domains and sufficient prior information, thus achieving superior channel estimation performance even with limited pilot overhead or low SNR. Figure 4 The diagram shows a comparison of the data detection performance of this example and two other comparative schemes under different numbers of active terminals. As the number of active terminals increases from 40 to 60, the data detection performance of the two comparative schemes deteriorates and tends to fail. However, this invention fully utilizes multi-domain sparsity and eliminates interference from high-confidence terminals in each iteration through residual feedback, thus exhibiting stronger robustness and a significant advantage in data demodulation.

[0106] In summary, the above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for on-demand rapid authorization and multi-domain prior load balancing in multi-terminal concurrent communication networks, characterized in that, The method is applied to a single-handshake uplink data transmission scenario where a central node terminal serves multiple concurrent devices, wherein the central node is equipped with a multi-array acquisition unit, and includes the following steps: Step 1: A certain number of active terminals use a single handshake protocol to send pilot signals concurrently to the central node at the same time and using the same transmission resources; Step 2: The central node, equipped with multiple array-type acquisition units, after receiving the pilot signals of all active terminals in Step 1, realizes active terminal identification and detection and link status estimation through alternating iteration, obtains the index set of active terminals and the link status information of active terminals, and completes the authorization of active terminals. Step 3: Active terminals authorized in Step 2 concurrently send their data symbols to the central node using a single handshake protocol and a multi-channel observation-transmission resource domain joint extended frame structure. The frame structure based on the multi-channel observation-transmission resource domain joint extension effectively performs data detection and supports the access of more terminals by expanding single-dimensional observation into a product of multi-channel observation and transmission resource observation. All data will be transmitted in the uplink. F Division of transmission resource units P Each transport resource group contains [number] transport resource groups. One transmission resource unit, and It is an integer; a transmission resource with the same index is extracted from each transmission resource group to form a transmission resource block. Transmission resource units within the same transmission resource block are equally spaced. The same constellation symbol is transmitted on each transmission resource unit within the same transmission resource block. Step 4: The central node receives the mixed data signals from each authorized active terminal. Based on the active terminal index set finally determined in Step 2 and the estimated link state information, it separates and demodulates the data symbols sent by each active terminal from the mixed signal, thereby completing the data detection.

2. The on-demand fast authorization and multi-domain prior equalization method for multi-terminal concurrent communication networks as described in claim 1, characterized in that, The method is applied to a single-handshake uplink data transmission scenario where a central node terminal serves multiple concurrent devices. Specifically, the number of array-type acquisition units in the central node is... , For row direction dimension, As a column-oriented dimension, the total number of services is M The terminal has an array of data acquisition units of [number]. , For row direction dimension, For column-oriented dimensions; Uplink transmission is performed using OFDM modulation; the spreading sequence for each terminal contains K A transmission resource unit, represented as , where row vector For the serial number m The spreading sequence corresponding to the terminal, For dimension is M×K The set of complex numbers, This is the set of spreading sequences corresponding to all terminals.

3. The on-demand fast authorization and multi-domain prior equalization method for multi-terminal concurrent communication networks as described in claim 1, characterized in that, The specific implementation process of step 1 is as follows: Quantity is The active terminals use a single handshake protocol to send pilot symbols to the central node, which are then processed by baseband for uplink pilot transmission. OFDM communication is used when sending pilot symbols, with sequence numbers as follows: terminal The same symbol is transmitted on each transmission resource unit (CRU), and the signal transmitted in each CRU is a constellation symbol and a spreading sequence. The product of; the length of the spreading sequence is Each transmission resource unit has a spread spectrum sequence known at the central node for all terminals; active terminals simultaneously and continuously transmit data to the central node during each uplink pilot transmission phase. The transmission of constellation symbols; the central node receives Pilot signals in one time slot.

4. The on-demand fast authorization and multi-domain prior equalization method for multi-terminal concurrent communication networks as described in claim 3, characterized in that, In step 1, the quantity is The active terminal sends several frames of data to the central node using a single handshake protocol, which are then processed by baseband and transmitted via uplink pilot signals. The implementation method is as follows: Each terminal is assigned a unique but non-orthogonal spreading sequence of length [length missing]. One transmission resource unit, i.e., the sequence number is m The spreading sequence corresponding to the terminal is First, the terminal converts the data bits into constellation symbols. The data sent by the terminal in a certain frame T In the OFDM pilot symbols, each OFDM corresponds to a constellation symbol; the symbol is... m The terminal sends the first frame in a certain frame. t Each constellation symbol is represented as ; Then, the active terminal sends the first t The frequency domain sequence corresponding to each OFDM symbol is .

5. The on-demand fast authorization and multi-domain prior equalization method for multi-terminal concurrent communication networks as described in claim 4, characterized in that, In step 2, active terminal identification and link state estimation are achieved through alternating iterations. The noisy signals received by the array-type acquisition units of all central nodes in the multi-channel observation-transmission resource domain are used as the initial input signals for the iterations. The specific iterative steps are as follows: Step 2.1: Use compressed sensing algorithm for the input signal to obtain the posterior active probability or confidence level of each terminal. Based on the preset threshold, initially divide the terminals into reliable set and rough set. Step 2.2: Utilize the inverse DFT to transform the multi-channel observation-transmission resource domain channel and received signal into the virtual space-time lag domain channel and received signal. Using the noisy signals received by the transmission resource units of all central nodes in the virtual space-time lag domain and employing a compressed sensing algorithm, obtain the virtual space-time lag domain channel. The estimated values ​​are obtained by using DFT transform to obtain the multi-channel observation-transmission resource domain channel. The estimated value; Step 2.3: Multiply the channel estimate obtained in Step 2.2 with the pilot signal to obtain the estimated noisy observation. Subtract the estimated noisy observation from the original signal in Step 2.1 to obtain the residual signal, which is used as the input signal for Step 2.1 in the next iteration. The above three steps 2.1-2.3 are executed iteratively until the maximum number of iterations is met or the residual signal energy is lower than a set threshold; finally, the rough set obtained from the last iteration is... As the final set of active terminal indexes The equivalent multi-channel observation-transmission resource domain channel estimate is obtained from step 2.2 of the last execution. It simultaneously detects active terminals and estimates channels, obtains the index set of active terminals and the link state information of active terminals, and completes the authorization of active terminals.

6. The on-demand fast authorization and multi-domain prior equalization method for multi-terminal concurrent communication networks as described in claim 5, characterized in that, In step 2.1, the MMV-AMP algorithm is used to obtain the posterior active probability of the terminal. Set a high threshold and low threshold When the posterior active probability of a terminal Greater than the low threshold Furthermore, the average value across all interfaces and transmission resource units is greater than the proportional threshold. If so, then this terminal is considered to belong to the rough set. When the posterior active probability of a terminal Greater than the high threshold Furthermore, the average value across all interfaces and transmission resource units is greater than the proportional threshold. If so, then the terminal belongs to the reliable set. .

7. The on-demand fast authorization and multi-domain prior equalization method for multi-terminal concurrent communication networks as described in claim 6, characterized in that, Step 4 specifically includes: The central node uses the active terminal index set obtained in step 2. and the estimated multi-channel observation-transmission resource domain equivalent channel By using a linear model, the data sent by the terminal is estimated by combining the signals received by all array-type acquisition units.

8. The on-demand fast authorization and multi-domain prior equalization method for multi-terminal concurrent communication networks as described in claim 7, characterized in that, The linear model used in step 4 is: in, Indicates the central node number n Each array-type acquisition unit receives continuous T Data signals in each time slot; Indicates active terminal set Data symbols transmitted uplink from the terminal Indicates active terminal set From the middle terminal to the central node n Estimated value of the equivalent channel of the multi-channel observation-transmission resource domain for each array-type acquisition unit; Indicates superposition on the n A matrix composed of Gaussian white noise on the signal received by an array of acquisition units; The matrices formed by the three are as follows: ， as well as ； Then we have: right Perform LMMSE estimation to obtain The expression is as follows: in It is the identity matrix. This represents the estimated number of elements in the active terminal set; ultimately, the data estimate is... The signal is then converted into constellation symbols and then into data bits; this completes the processing of one frame of signal.

Images