Wireless signal multi-dimensional array coding method and system based on dynamic rank optimization

Abstract

The invention belongs to the field of wireless communication, and particularly relates to a wireless signal multi-dimensional array coding method and system based on dynamic rank optimization. The method comprises the following steps: detecting state information of a wireless channel, predicting state change of the channel by using a pre-trained deep learning model, and generating a dynamic optimization multiplier; adjusting a preset static optimization multiplier by using the dynamic optimization multiplier, constructing a target function for jointly optimizing the energy efficiency and the spectrum efficiency, and determining an optimal transmission rank by solving the target function; comparing the optimal transmission rank with the current transmission rank to determine a rank-up or rank-down operation; and reconstructing a transmission signal of the wireless channel according to a result of the rank increasing or decreasing operation. According to the method, the optimization target is dynamically adjusted by predicting the channel change, and the asymmetric ascending and descending rank algorithm is adopted, so that the adaptive transmission rank optimization in a complex time-varying channel environment is realized, and the communication performance is remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication, and specifically relates to a method and system for multidimensional array coding of wireless signals based on dynamic rank optimization. Background Technology

[0002] With the commercial deployment of fifth-generation (5G) mobile communication systems and the evolution of next-generation communication technologies (6G), Massive MIMO technology has become one of the core technologies for improving the spectral efficiency and network capacity of wireless communication systems. In MIMO systems, by configuring multiple antennas at both the transmitting and receiving ends, multiple parallel transmission channels can be constructed in space, thereby multiplying the data transmission rate without increasing the spectral bandwidth. The number of parallel transmission channels is usually referred to as the transmission rank.

[0003] The choice of transmission rank is a critical issue, directly impacting the overall system performance. On one hand, increasing the transmission rank can increase the number of parallel data streams, thus significantly improving the system's spectral efficiency (i.e., peak rate). On the other hand, a higher transmission rank means activating more RF links and performing more complex signal processing, which not only drastically increases system energy consumption and reduces energy efficiency but may also introduce more severe inter-stream interference under poor channel conditions, ultimately leading to a deterioration in communication quality. Therefore, there is an inherent contradiction and trade-off between spectral efficiency (SE) and energy efficiency (EE).

[0004] In existing technologies, most adaptive adjustment methods for transmission rank rely on measurements of current or historical channel state information (CSI). These methods are typically reactive: they passively adjust the transmission rank only after a change in channel quality is detected. In scenarios where channel conditions change rapidly (such as high-speed movement or complex scattering environments), this lagging adjustment mechanism struggles to guarantee optimal communication performance and may lead to transmission interruptions or missed opportunities to take advantage of favorable channel conditions.

[0005] Furthermore, some existing methods employ coarse-grained or computationally overly complex update algorithms when deciding on rank adjustments. For example, each rank adjustment requires a complete singular value decomposition (SVD) of the channel matrix or a complete re-decomposition of the channel tensor. This incurs enormous computational overhead for Massive MIMO systems with a large number of antennas, making it difficult to meet real-time requirements. Simultaneously, when reducing transmission rank, these methods often lack fine-grained interference identification, potentially erroneously pruning data streams with better signal quality, thus affecting system stability. Summary of the Invention

[0006] To address the aforementioned problems in the prior art, namely overcoming the reactive lag, high computational complexity, and inflexible trade-off between energy efficiency and spectral efficiency in existing wireless signal transmission rank adjustment strategies, this invention provides, in a first aspect, a multidimensional array coding method for wireless signals based on dynamic rank optimization, comprising: Detect the state information of the wireless channel to extract the channel quality parameters at the current moment; The channel quality parameters are input into a pre-trained deep learning model to predict the channel quality parameters of the wireless channel at a future preset time, and the dynamic optimization multiplier is determined based on the prediction results. The preset static optimization multiplier is adjusted in real time using the dynamic optimization multiplier to construct an objective function for jointly optimizing energy efficiency and spectral efficiency. The optimal transmission rank is determined by solving the objective function. The static optimization multiplier is determined based on the channel quality parameters of the wireless channel under a preset communication scenario. The optimal transmission rank is compared with the current transmission rank of the wireless channel to determine whether to perform a rank increase or decrease operation. When the optimal transmission rank is not less than the current transmission rank, an incremental tensor decomposition algorithm is used to perform a rank increase operation to expand the signal dimension. When the optimal transmission rank is less than the current transmission rank, a sparse constraint tensor decomposition algorithm based on interference identification is used to perform a rank decrease operation to compress the signal dimension. The transmitted signal of the wireless channel is reconstructed based on the result of the rank-up or rank-down operation.

[0007] In some preferred embodiments, detecting the state information of the wireless channel to extract channel quality parameters includes: The channel matrix is ​​obtained by receiving a reference signal characterizing the wireless channel state information through a measurement unit within the communication device. The channel matrix is ​​calculated to extract a set of channel quality parameters to characterize the channel state. The set of channel quality parameters includes at least one or more combinations of channel quality parameters, including parameters to characterize signal received power, parameters to characterize signal received quality, signal-to-noise-interference ratio parameters to characterize channel interference level, Doppler frequency shift parameters to characterize channel time-varying characteristics, and delay spread parameters to characterize multipath effects.

[0008] In some preferred embodiments, the deep learning model is a neural network architecture for processing time series data. The deep learning model takes the channel quality parameters of the wireless channel at the current moment and at multiple consecutive historical moments as input, and predicts and outputs predicted values ​​of the channel quality parameters at one or more future moments.

[0009] In some preferred embodiments, determining the dynamically optimized multiplier based on the prediction results includes: Determine the difference between the channel quality parameters at a future preset time and the channel quality parameters at the current time; the difference is used to indicate the trend of channel quality change. The difference is taken as input, and a dynamic optimization multiplier is output through a preset mapping function. The mapping function is used to map positive trends to adjustment values ​​aimed at improving spectral efficiency weights, and negative trends to adjustment values ​​aimed at ensuring energy efficiency.

[0010] In some preferred embodiments, the objective function is used to evaluate the overall performance of each candidate transmission rank. The evaluation value of the objective function is defined as a weighted combination of spectral efficiency and energy efficiency, wherein the weight of spectral efficiency is determined by the combination of the dynamic optimization multiplier and the static optimization multiplier, and the weight of energy efficiency satisfies a preset relationship with the weight of spectral efficiency. The optimal transmission rank is determined by solving the objective function, and the transmission rank that maximizes the evaluation value of the objective function is the optimal transmission rank.

[0011] In some preferred embodiments, the step of performing the rank-raising operation using an incremental tensor decomposition algorithm includes: While keeping all rank-one tensor components constituting the current transmission rank unchanged, the rank-one decomposition operation is iteratively performed on the residual tensor of the sum of the original channel tensor and all the decomposed rank-one tensor components. The number of iterations is equal to the difference between the optimal transmission rank and the current transmission rank. Each iteration yields a new rank-one tensor component, which is defined by a new set of factor vectors. All factor vectors obtained in the iterative operation are added to the original tensor factor matrix to obtain the pre-encoding matrix after rank increase.

[0012] In some preferred embodiments, the step of performing the rank reduction operation using a sparse constraint tensor decomposition algorithm based on interference identification includes: The interference intensity in each signal dimension is determined by performing eigenvalue decomposition on the covariance matrix of the received signal. An optimization model for performing tensor decomposition is constructed, the objective of which is to impose a penalty proportional to the intensity of the disturbance on factor vectors associated with high disturbance dimensions through a pre-defined sparsity constraint term. By solving the optimization model, multiple rank-tensor components are obtained, and selective elimination is performed on the multiple rank-tensor components according to a preset elimination criterion. The elimination criterion is used to characterize the interference intensity and signal power contribution corresponding to each rank-tensor component until the transmission rank is reduced to the optimal transmission rank.

[0013] In some preferred embodiments, reconstructing the transmitted signal of the wireless channel based on the result of the rank-up or rank-down operation includes: A new transmitter precoding matrix is ​​generated using the updated tensor factor matrix; Based on the optimal transmission rank, the mapping relationship between data and spatial stream is updated synchronously, and the most matching modulation and coding strategy is selected from the preset set of modulation and coding strategies. The reconstructed wireless signal is generated and transmitted using the new transmitter precoding matrix, combined with the updated mapping relationship and the selected modulation and coding strategy.

[0014] A second aspect of this invention proposes a multidimensional array coding system for wireless signals based on dynamic rank optimization, comprising: The data acquisition module is used to detect the status information of the wireless channel in order to extract the channel quality parameters at the current moment; The model prediction module is used to input the channel quality parameters into a pre-trained deep learning model, predict the channel quality parameters of the wireless channel at a preset time in the future, and determine the dynamic optimization multiplier based on the prediction results. The joint solution module is used to adjust the preset static optimization multiplier in real time using the dynamic optimization multiplier, construct an objective function for jointly optimizing energy efficiency and spectral efficiency, and determine the optimal transmission rank by solving the objective function. The static optimization multiplier is determined based on the channel quality parameters of the wireless channel under the preset communication scenario. The signal processing module is used to compare the optimal transmission rank with the current transmission rank of the wireless channel to determine whether to perform a rank increase or decrease operation. When the optimal transmission rank is not less than the current transmission rank, an incremental tensor decomposition algorithm is used to perform a rank increase operation to expand the signal dimension. When the optimal transmission rank is less than the current transmission rank, a sparse constraint tensor decomposition algorithm based on interference identification is used to perform a rank decrease operation to compress the signal dimension. A signal transmission module is used to reconstruct the transmission signal of the wireless channel based on the result of the rank-up or rank-down operation.

[0015] The beneficial effects of this invention are: On one hand, this invention introduces a deep learning model to predict future channel states, thereby achieving forward-looking transmission rank decision-making. It transforms passive reactive adjustment into proactive predictive optimization, which can capture communication opportunities brought about by channel changes or avoid potential performance degradation in a more timely manner. It is especially suitable for fast time-varying channel environments such as high-speed mobile environments.

[0016] On the other hand, this invention dynamically optimizes the multiplier to adjust the optimization target in real time. When the channel is about to improve, it prioritizes improving spectral efficiency to obtain a higher data transmission rate; while when the channel is about to deteriorate, it prioritizes ensuring energy efficiency and communication stability, thus achieving a flexible and intelligent balance of system performance in both energy efficiency and spectral efficiency dimensions.

[0017] Furthermore, this invention employs an asymmetric rank-raising and rank-reducing algorithm. During rank raising, incremental tensor decomposition is used, reusing existing computational results and avoiding the enormous overhead of complete re-decomposition. During rank reducing, sparse constraint tensor decomposition based on interference identification is employed, which can accurately eliminate signal dimensions that contribute little to communication quality or even have a negative impact. Compared to traditional complete decomposition methods, this significantly reduces computational complexity and improves the real-time performance and accuracy of rank adjustment. Attached Figure Description

[0018] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart illustrating a multi-dimensional array coding method for wireless signals based on dynamic rank optimization provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the framework of a wireless signal multidimensional array coding system based on dynamic rank optimization provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a computer system used to implement the methods and system embodiments of this application. Detailed Implementation

[0019] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0020] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] To more clearly illustrate the method of the present invention, the following description is provided in conjunction with... Figure 1 The steps in the embodiments of the present invention will be described in detail below.

[0022] The first embodiment of the wireless signal multidimensional array coding method based on dynamic rank optimization of the present invention includes steps S10-S50, each step of which is described in detail below: Step S10: Detect the state information of the wireless channel in order to extract the channel quality parameters at the current moment; Step S20: Input the channel quality parameters into a pre-trained deep learning model to predict the channel quality parameters of the wireless channel at a future preset time, and determine the dynamic optimization multiplier based on the prediction results; Step S30: The preset static optimization multiplier is adjusted in real time using the dynamic optimization multiplier to construct an objective function for jointly optimizing energy efficiency and spectral efficiency, and the optimal transmission rank is determined by solving the objective function. The static optimization multiplier is determined based on the channel quality parameters of the wireless channel under the preset communication scenario. Step S40: Compare the optimal transmission rank with the current transmission rank of the wireless channel to determine whether to perform a rank increase or rank decrease operation. When the optimal transmission rank is not less than the current transmission rank, an incremental tensor decomposition algorithm is used to perform a rank increase operation to expand the signal dimension. When the optimal transmission rank is less than the current transmission rank, a sparse constraint tensor decomposition algorithm based on interference identification is used to perform a rank decrease operation to compress the signal dimension. Step S50: Reconstruct the transmitted signal of the wireless channel based on the result of the rank-up or rank-down operation.

[0023] This embodiment provides a wireless signal multidimensional array coding method based on dynamic rank optimization. Its core lies in combining predictive analysis of channel state with an efficient signal dimension adjustment algorithm to achieve a dynamic intelligent trade-off between energy efficiency and spectral efficiency.

[0024] Specifically, in this embodiment, the primary task of the communication system (e.g., a base station or terminal device) is to accurately sense the characteristics of the current wireless channel. The measurement unit within the system estimates the instantaneous channel by receiving a predefined reference signal (such as the Channel State Information Reference Signal (CSI-RS) in a 5G NR system). After processing by the channel estimation algorithm, a multi-dimensional channel matrix H is obtained, which characterizes the spatial propagation path characteristics between the transmitting antenna array and the receiving antenna array.

[0025] To quantify the channel state, a set of key channel quality indicators (CQI) needs to be extracted from the channel matrix H.

[0026] In this embodiment, the parameter is constructed as a multi-dimensional feature vector to comprehensively describe the channel state. This multi-dimensional feature vector may include, but is not limited to: Signal received power parameters, such as Reference Received Power (RSRP), are used to characterize the coverage strength of the signal; signal received quality parameters, such as Signal-to-Interference-Ratio (SINR), are core indicators for evaluating the quality of communication links and directly affect the achievable data rate; channel time-varying characteristic parameters, such as Doppler Shift, reflect the fast fading characteristics of the channel caused by user movement or environmental changes and are key to measuring the time-varying speed of the channel; multipath effect parameters, such as Delay Spread, describe the time difference of a signal arriving at the receiver through different paths and reflect the richness of the channel's multipath.

[0027] The above parameters together constitute a channel state snapshot C(t) at time t, providing a data basis for subsequent prediction and decision-making.

[0028] Based on the extracted channel quality parameters, this embodiment utilizes a neural network to make a forward-looking judgment on the future evolution trend of the channel. Specifically, the model in this embodiment is preferably a neural network capable of processing time series data, such as a Long Short-Term Memory (LSTM) network or a Gated Recurrent Unit (GRU).

[0029] The model takes as input a sequence of channel quality parameters for the current time and N consecutive past time points. By learning the temporal correlation in historical data, the model outputs a predicted value C'(t+Δt) of the channel quality parameters for one or more preset time points in the future (e.g., t+Δt).

[0030] After obtaining the prediction results, a dynamic optimization multiplier is generated accordingly. The determination process of the multiplier is based on a preset mapping function f(·). The core logic of the function is to compare the predicted channel quality parameter C'(t+Δt) with the channel quality parameter C(t) at the current time.

[0031] If the prediction results indicate that the channel quality will be significantly improved (e.g., the predicted SINR value is much higher than the current value), the mapping function will output an adjustment amount that encourages the improvement of spectral efficiency (e.g., a large positive value or a factor greater than 1), indicating that the current moment is a good time to increase the transmission rate. Conversely, if channel degradation is predicted (e.g., a predicted decrease in SINR), the mapping function will output an adjustment that tends to ensure energy efficiency and communication stability (e.g., a negative value or a factor less than 1), in which case a more conservative transmission strategy should be adopted.

[0032] For example, the mapping function can be a linear mapping function, and its mapping rule can be that the dynamic optimization multiplier is proportional to the performance deviation; The mapping function can also be a piecewise / step mapping function. In this implementation, the mapping function satisfies the threshold-based discrete control method, which can output a preset fixed value according to the interval where the performance deviation is located. In addition, the mapping function can also be a Sigmoid function, which can use an S-shaped curve to smoothly map potentially infinite input deviations to a finite output range.

[0033] Furthermore, this embodiment constructs an objective function U(R) for jointly optimizing energy efficiency (EE) and spectral efficiency (SE), which is defined as a weighted sum of energy efficiency and spectral efficiency: U(R) =ω_SE SE(R)+ω_EE EE(R), where SE(R) and EE(R) are estimates of the spectral efficiency and energy efficiency achievable by the system when the transmission rank is R, respectively.

[0034] The weight ω_SE for spectral efficiency is determined by two parts: the static optimization multiplier α_static and the dynamic optimization multiplier α_dyn generated in the above process.

[0035] Specifically, in this embodiment, the static optimization multiplier is a pre-set calibration value that represents the optimization preference of the communication system in the default state. Its value can be determined according to the channel quality parameters of the wireless channel in a preset communication scenario, and it usually remains unchanged during a complete communication task, or is only manually updated when there are significant changes in the network configuration (e.g., the operator adjusts the communication strategy).

[0036] In this embodiment, the dynamic optimization multiplier is a real-time calibration value generated dynamically based on the real-time network performance. Its value represents the adjustment amount of the static optimization multiplier. Based on the set basic strategy (determined by the mapping function of the dynamic optimization multiplier mentioned above), the static optimization multiplier is corrected in real time to adapt to the ever-changing channel conditions.

[0037] Specifically, α_static is a preset benchmark value based on long-term statistics or specific business scenarios (such as eMBB scenarios which emphasize spectral efficiency, and mMTC scenarios which emphasize energy efficiency); the dynamic multiplier α_dyn is used to fine-tune the benchmark value in real time, i.e., ω_SE = g(α_static,α_dyn), where g(·) is a combination function.

[0038] Specifically, the combination function receives two input signals and calculates the weights of the spectral efficiency accordingly.

[0039] For example, the combination function can be an addition truncation function, which directly adds the dynamic adjustment indicated by the dynamic optimization multiplier to the static baseline, and then ensures that the result does not exceed a certain range (such as [0,1]) through a truncation operation.

[0040] In addition, the combination function can also be a combination based on the Softmax / Sigmoid function. Other combinations use nonlinear mapping to constrain the weights to [0, 1]. Taking the logarithmic probability space as an example, addition can be performed in the logarithmic probability space, and then mapped back to the probability space of [0, 1] through the Sigmoid or Softmax function.

[0041] Preferably, the energy efficiency weight ω_EE is usually complementary to ω_SE, for example, satisfying the normalization relationship ω_SE+ω_EE=1.

[0042] Based on this, this embodiment determines the optimal transmission rank R by solving an optimization problem. : R =argmax_R U(R). This process typically involves iterating through all candidate transmission ranks (from 1 to the maximum rank R_max supported by the system), calculating the objective function evaluation value for each candidate rank, and selecting the rank with the largest evaluation value as the optimal transmission rank R at the current moment. .

[0043] After determining the optimal transmission rank R Then, it is compared with the current transmission rank R_current, and the corresponding dimension adjustment operation is performed. This embodiment uses an asymmetric ascending-descending rank algorithm.

[0044] Where R When R_current is greater than or equal to R_current, perform a rank increase operation: At this point, an incremental tensor decomposition algorithm is used, which reuses existing calculation results. Specifically, the R_current rank tensor components representing the current channel structure and their corresponding factor matrices remain unchanged. On the residual tensor of the reconstruction results of the original channel tensor and the R_current components, R is decomposed. -R_current iterations of rank-1 decomposition. Each iteration finds and decomposes the most dominant signal component in the residual. Finally, the newly obtained R... The factor vectors corresponding to the -R_current rank tensor components are directly added to the original factor matrix. In this way, the complete re-decomposition of the entire channel tensor is avoided, which greatly reduces the computational complexity.

[0045] When R When R_current is less than R_current, perform a rank reduction operation: At this point, a sparse constraint tensor decomposition algorithm based on interference identification is adopted. The key to rank reduction lies in accurately eliminating signal dimensions that contribute the least to system performance or are even harmful.

[0046] First, by performing eigenvalue decomposition on the received signal covariance matrix, the distribution and intensity of interference in each spatial dimension are identified. Then, a sparsity constraint term (such as weighted L1 norm regularization) is introduced into the objective function of the tensor decomposition. This constraint term imposes a penalty proportional to the interference intensity on factor vectors associated with high-interference dimensions, thereby guiding the decomposition results to automatically suppress or ignore these heavily interfered dimensions during the solution process. Finally, based on a rejection criterion that integrates signal power contribution and interference level, R_current - R_interference is selectively removed from the R_current rank tensor components. The component with the worst overall performance, until the transmission rank drops to R. .

[0047] After adjusting the transmission rank, the signal is reconstructed and transmitted according to the new rank, specifically including: The precoding matrix is ​​updated by using the tensor factor matrix updated in the previous steps to generate a new transmitter precoding matrix P, which is used to map the data stream onto the physical antenna. Data stream mapping and scheduling updates: Generally speaking, the scheduler of a communication system updates the mapping relationship between data and spatial streams to R. A parallel stream; Adaptive selection of modulation and coding strategy (MCS) based on the new transmission rank R In addition to the channel quality of each data stream, the communication system selects the modulation order and coding rate that best match the current channel conditions from the preset MCS set in order to maximize the reliability and efficiency of data transmission.

[0048] Finally, the data blocks to be sent are encoded and modulated using the new MCS, and then allocated to R through the updated data stream mapping relationship. On a spatial stream, after precoding using a new precoding matrix P, it is transmitted through an antenna array.

[0049] This completes a full cycle of multidimensional array coding for wireless signals based on dynamic rank optimization. In practical applications, the communication system will continuously repeat the above steps to achieve continuous adaptation to the dynamic wireless channel.

[0050] As a preferred implementation, detecting the state information of the wireless channel to extract channel quality parameters includes: The channel matrix is ​​obtained by receiving a reference signal characterizing the wireless channel state information through a measurement unit within the communication device. The channel matrix is ​​calculated to extract a set of channel quality parameters to characterize the channel state. The set of channel quality parameters includes at least one or more combinations of channel quality parameters, including parameters to characterize signal received power, parameters to characterize signal received quality, signal-to-noise-interference ratio parameters to characterize channel interference level, Doppler frequency shift parameters to characterize channel time-varying characteristics, and delay spread parameters to characterize multipath effects.

[0051] In this embodiment, a measurement unit (e.g., an RF receive link and a baseband processor) within a communication device (such as a base station or user equipment) periodically receives a reference signal with a known structure transmitted by a peer device. In a 5G NR (NewRadio) system, this reference signal may be a Channel State Information Reference Signal (CSI-RS). By comparing the received distorted reference signal with the locally stored original reference signal, the baseband processor can execute a channel estimation algorithm (such as least squares (LS) or minimum mean square error (MMSE)) to obtain a channel matrix H that accurately describes the channel response between multiple antenna transmit / receive links.

[0052] After obtaining the channel matrix H, a multi-dimensional set of channel quality parameters is calculated and extracted from it. This set constitutes an eigenvector, which comprehensively characterizes the instantaneous state of the current channel. Specifically, this parameter set may contain a combination of one or more of the following parameters: Parameters characterizing signal received power include, for example, the reference signal received power (RSRP). It is obtained by measuring the power of a specific reference signal and directly reflects the degree of signal attenuation after propagation through space. Parameters characterizing signal reception quality: For example, Reference Signal Received Quality (RSRQ). It is the ratio of RSRP to Total Received Signal Power (RSSI), reflecting the level of interference and noise in the channel; The signal-to-interference-plus-noise ratio (SINR) parameter, which characterizes the level of channel interference, is calculated using the channel matrix and is one of the most direct indicators for evaluating data demodulation performance. Doppler frequency shift parameters characterize the time-varying properties of the channel: These parameters are obtained by analyzing the channel's time-dependent correlation function or by estimating the received signal using a dedicated algorithm. They are used to quantify the fading rate of the channel caused by user movement or environmental changes. Delay spread parameter characterizing multipath effect: obtained by analyzing the power delay profile of the channel, used to quantify the time-domain dispersion effect of the signal caused by multipath propagation, reflecting the frequency selectivity of the channel.

[0053] In a preferred embodiment, the deep learning model is a neural network architecture for processing time series data. The deep learning model takes the channel quality parameters of the wireless channel at the current moment and at multiple consecutive historical moments as input, and predicts and outputs the predicted values ​​of the channel quality parameters at one or more future moments.

[0054] In this embodiment, the deep learning model used for channel state prediction is specifically a recurrent neural network (RNN) architecture capable of processing time series data, such as a long short-term memory network (LSTM) or a gated recurrent unit (GRU). This architecture can effectively learn and memorize long-term dependencies in time series data.

[0055] Before actual deployment, the model needs to be pre-trained offline. The training dataset can consist of a large number of historical channel quality parameter sequences collected in real or simulated environments and their corresponding future true values.

[0056] Specifically, during runtime, the model receives input using a sliding time window. Specifically, the system concatenates the channel quality parameter vectors [C(tN), ..., C(t)] from the current time t and the past N sampling times into an input matrix, which is then fed into an LSTM or GRU model. The model processes the input vectors at each time step through its loop structure and utilizes its learned temporal evolution rules to ultimately output the predicted value C'(t+Δt) of the channel quality parameter vector for one or more future preset times (e.g., t+Δt).

[0057] As a preferred embodiment, determining the dynamically optimized multiplier based on the prediction results includes: The prediction results are read, and based on the difference between the prediction results and the current time, a dynamic optimization multiplier is determined through a preset mapping function. The mapping function maps the predicted channel quality improvement trend to an adjustment amount that is biased towards improving spectral efficiency, and maps the predicted channel quality deterioration trend to an adjustment amount that is biased towards ensuring energy efficiency.

[0058] In this embodiment, the predicted channel quality vector C'(t+Δt) output by the deep learning model is first read and compared with the actual measurement value C(t) at the current time. The expected change of key indicators, such as SINR, is calculated, which satisfies ΔSINR = SINR'(t+Δt) - SINR(t).

[0059] Subsequently, this change is input to a preset nonlinear mapping function f(·) to generate a dynamic optimization multiplier α_dyn = f(ΔSINR); For example, the nonlinear mapping function can be a sigmoid or Tanh function that is differentiable everywhere in its domain, using a smooth S-shaped curve to map the input (which can be ΔSINR of any size) to a bounded output range (e.g., -0.2 to +0.2).

[0060] When channel quality is predicted to improve (ΔSINR>0): the mapping function outputs a positive adjustment amount, the magnitude of which is positively correlated with ΔSINR, to increase the weight of spectral efficiency (SE) in the objective function, so as to prompt the excitation system of the communication equipment to adopt a more aggressive transmission strategy (such as increasing the transmission rank) to maximize data throughput.

[0061] When channel quality is predicted to deteriorate (ΔSINR<0): the mapping function outputs a negative adjustment amount whose absolute value is positively correlated with |ΔSINR|, in order to reduce the weight of spectral efficiency and correspondingly increase the weight of energy efficiency (EE) or stability, prompting the excitation system of the communication equipment to adopt a more conservative strategy (such as reducing transmission rank) and prioritize ensuring the reliability and energy efficiency of the connection.

[0062] For example, the mapping function described above can be a piecewise function, a sigmoid function, or a pre-defined look-up table.

[0063] In a preferred embodiment, the objective function is used to evaluate the overall performance of each candidate transmission rank. The evaluation value of the objective function is defined as a weighted combination of spectral efficiency and energy efficiency, wherein the weight of spectral efficiency is determined by the combination of the dynamic optimization multiplier and the static optimization multiplier, and the weight of energy efficiency satisfies a preset relationship with the weight of spectral efficiency. The optimal transmission rank is determined by solving the objective function, and the transmission rank that maximizes the evaluation value of the objective function is the optimal transmission rank.

[0064] In this embodiment, the weight ω_SE of the spectral efficiency is determined by both the static and dynamic optimization multipliers. For example, it can take the form of an additive or multiplicative model, and then be normalized, for example: ω_SE = clip(α_static + α_dyn, ω_min, ω_max); Here, α_static is the static optimization multiplier, α_dyn is the dynamic optimization multiplier, and the clip function is used to ensure that the final weight ω_SE is limited to a preset reasonable range [ω_min, ω_max] (e.g., [0, 1]).

[0065] The process of finding the optimal transmission rank involves performing a traversal search on all candidate transmission ranks supported by the system. For each candidate transmission rank, it is substituted into the objective function to calculate its comprehensive performance evaluation value. Finally, the candidate rank that maximizes the comprehensive performance evaluation value is determined as the optimal transmission rank for the current decision period.

[0066] As a preferred embodiment, the step of performing the rank-raising operation using an incremental tensor decomposition algorithm includes: While keeping all rank-one tensor components constituting the current transmission rank unchanged, an iterative rank-one decomposition operation is performed on the residual tensor of the sum of the original channel tensor and all decomposed rank-one tensor components. The number of iterations is equal to the difference between the optimal transmission rank and the current transmission rank. Each iteration yields a new rank-one tensor component. The factor vectors corresponding to all the new rank-one tensor components obtained in the iterations are added to the original tensor factor matrix to obtain the rank-up precoding matrix.

[0067] Simultaneously, a rank reduction operation is performed using a sparse constraint tensor decomposition algorithm based on interference identification, including: The interference intensity in each signal dimension is determined by eigenvalue decomposition of the covariance matrix of the received signal. A sparsity constraint term is introduced into the objective function of the tensor decomposition, which imposes a penalty proportional to the interference intensity on factor vectors associated with high-interference dimensions during the optimization process. Selective elimination is performed from the multiple rank-tensor components obtained from the solution, according to a preset elimination criterion. This criterion characterizes the interference intensity and signal power contribution corresponding to each rank-tensor component, until the transmission rank is reduced to the optimal transmission rank.

[0068] As a preferred embodiment, reconstructing the transmitted signal of the wireless channel based on the result of the rank-up or rank-down operation includes: A new transmitter precoding matrix is ​​generated using the updated tensor factor matrix; based on the optimal transmission rank, the mapping relationship between data and spatial stream is updated synchronously, and the most matching modulation and coding strategy is selected from the preset modulation and coding strategy set; using the new precoding matrix, combined with the updated mapping relationship and the selected modulation and coding strategy, the reconstructed wireless signal is generated and transmitted.

[0069] In this embodiment, a new precoding matrix for the transmitter is directly constructed using the updated tensor factor matrix (whether it is a supplementary or removed component).

[0070] Taking a common communication system as an example, its Media Access Control (MAC) layer scheduler will receive a physical layer transmission rank that has been updated to R. The signaling is based on which the scheduler will divide the data block to be sent into R. Each data stream is a separate spatial stream, and corresponding time-frequency resources are allocated to each stream. The mapping relationship between data and spatial streams is adjusted synchronously.

[0071] For this R Each independent spatial channel will have its equivalent channel quality estimated separately. Then, the adaptive modulation and coding (AMC) module will independently select the MCS level that best matches its current channel quality for each spatial stream from a preset set of modulation and coding strategies (MCS) (e.g., various combinations from QPSK to 256-QAM).

[0072] Ultimately, this R Each data stream undergoes channel coding and symbol modulation through its selected MCS, and then is linearly combined through a new precoding matrix to form several baseband signals. After digital-to-analog conversion and frequency up-conversion via the RF link, the signals are transmitted by the antenna.

[0073] Please refer to Figure 2 The second embodiment of the present invention provides a wireless signal multidimensional array coding system based on dynamic rank optimization, used to execute the above-described wireless signal multidimensional array coding method based on dynamic rank optimization, comprising: The data acquisition module 210 is used to detect the status information of the wireless channel in order to extract the channel quality parameters at the current moment; The model prediction module 220 is used to input the channel quality parameters into a pre-trained deep learning model, predict the channel quality parameters of the wireless channel at a preset time in the future, and determine the dynamic optimization multiplier based on the prediction results. The joint solution module 230 is used to adjust the preset static optimization multiplier in real time using the dynamic optimization multiplier, construct an objective function for jointly optimizing energy efficiency and spectral efficiency, and determine the optimal transmission rank by solving the objective function. The static optimization multiplier is determined based on the channel quality parameters of the wireless channel under the preset communication scenario. The signal processing module 240 is used to compare the optimal transmission rank with the current transmission rank of the wireless channel to determine whether to perform a rank increase or rank decrease operation. When the optimal transmission rank is not less than the current transmission rank, an incremental tensor decomposition algorithm is used to perform a rank increase operation to expand the signal dimension. When the optimal transmission rank is less than the current transmission rank, a sparse constraint tensor decomposition algorithm based on interference identification is used to perform a rank decrease operation to compress the signal dimension. The signal transmission module 250 is used to reconstruct the transmission signal of the wireless channel based on the result of the rank-up or rank-down operation.

[0074] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the system described above can be found in the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0075] It should be noted that the multidimensional array coding method and system for wireless signals based on dynamic rank optimization provided in the above embodiments are only illustrative examples of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the modules or steps in the embodiments of the present invention can be further decomposed or combined. For example, the modules in the above embodiments can be merged into one module, or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing the various modules or steps and are not considered as an improper limitation of the present invention.

[0076] The following is for reference. Figure 3 It shows a schematic diagram of the structure of a computer system used to implement the system and method embodiments of this application. Figure 3 The server shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0077] like Figure 3 As shown, the computer system includes a Central Processing Unit (CPU) 601, which can perform various appropriate actions and processes based on programs stored in Read Only Memory (ROM) 602 or programs loaded from storage section 608 into Random Access Memory (RAM) 603. The RAM 603 also stores various programs and data required for system operation. The CPU 601, ROM 602, and RAM 603 are interconnected via a bus 604. An Input / Output (I / O) interface 605 is also connected to the bus 604.

[0078] The following components are connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 610 as needed so that computer programs read from it can be installed into storage section 608 as needed.

[0079] Specifically, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 609, and / or installed from removable medium 611. When the computer program is executed by central processing unit (CPU) 601, it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described above in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0080] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0081] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0082] The terms “first”, “second”, etc., are used to distinguish similar objects, not to describe or indicate a specific order or sequence.

[0083] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent in such process, method, article, or apparatus / device.

[0084] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A multidimensional array coding method for wireless signals based on dynamic rank optimization, characterized in that, include: Detect the state information of the wireless channel to extract the channel quality parameters at the current moment; The channel quality parameters are input into a pre-trained deep learning model to predict the channel quality parameters of the wireless channel at a future preset time, and the dynamic optimization multiplier is determined based on the prediction results. The preset static optimization multiplier is adjusted in real time using the dynamic optimization multiplier to construct an objective function for jointly optimizing energy efficiency and spectral efficiency. The optimal transmission rank is determined by solving the objective function. The static optimization multiplier is determined based on the channel quality parameters of the wireless channel under a preset communication scenario. The optimal transmission rank is compared with the current transmission rank of the wireless channel to determine whether to perform a rank increase or decrease operation. When the optimal transmission rank is not less than the current transmission rank, an incremental tensor decomposition algorithm is used to perform a rank increase operation to expand the signal dimension. When the optimal transmission rank is less than the current transmission rank, a sparse constraint tensor decomposition algorithm based on interference identification is used to perform a rank decrease operation to compress the signal dimension. The transmitted signal of the wireless channel is reconstructed based on the result of the rank-up or rank-down operation.

2. The method according to claim 1, characterized in that, The detection of wireless channel state information to extract channel quality parameters at the current moment includes: The channel matrix is ​​obtained by receiving a reference signal characterizing the wireless channel state information through a measurement unit within the communication device. The channel matrix is ​​calculated to extract a set of channel quality parameters to characterize the channel state. The set of channel quality parameters includes at least one or more combinations of channel quality parameters, including parameters to characterize signal received power, parameters to characterize signal received quality, signal-to-noise-interference ratio parameters to characterize channel interference level, Doppler frequency shift parameters to characterize channel time-varying characteristics, and delay spread parameters to characterize multipath effects.

3. The method according to claim 1, characterized in that, The deep learning model is a neural network architecture for processing time series data. The deep learning model takes the channel quality parameters of the wireless channel at the current moment and at multiple consecutive historical moments as input, and predicts and outputs the predicted values ​​of the channel quality parameters at one or more future moments.

4. The method according to claim 1, characterized in that, The step of determining the dynamically optimized multiplier based on the prediction results includes: Determine the difference between the channel quality parameters at a future preset time and the channel quality parameters at the current time; the difference is used to indicate the trend of channel quality change. The difference is taken as input, and a dynamic optimization multiplier is output through a preset mapping function. The mapping function is used to map positive trends to adjustment values ​​aimed at improving spectral efficiency weights, and negative trends to adjustment values ​​aimed at ensuring energy efficiency.

5. The method according to claim 1, characterized in that, The objective function is used to evaluate the overall performance of each candidate transmission rank. The evaluation value of the objective function is defined as a weighted combination of spectral efficiency and energy efficiency. The weight of spectral efficiency is determined by the combination of the dynamic optimization multiplier and the static optimization multiplier, and the weight of energy efficiency satisfies a preset relationship with the weight of spectral efficiency. The optimal transmission rank is determined by solving the objective function, and the transmission rank that maximizes the evaluation value of the objective function is the optimal transmission rank.

6. The method according to claim 1, characterized in that, The step of performing the rank-raising operation using the incremental tensor decomposition algorithm includes: While keeping all rank-one tensor components constituting the current transmission rank unchanged, the rank-one decomposition operation is iteratively performed on the residual tensor of the sum of the original channel tensor and all the decomposed rank-one tensor components. The number of iterations is equal to the difference between the optimal transmission rank and the current transmission rank. Each iteration yields a new rank-one tensor component, which is defined by a new set of factor vectors. All factor vectors obtained in the iterative operation are added to the original tensor factor matrix to obtain the pre-encoding matrix after rank increase.

7. The method according to claim 1, characterized in that, The step of performing the rank reduction operation using a sparse constraint tensor decomposition algorithm based on interference identification includes: The interference intensity in each signal dimension is determined by performing eigenvalue decomposition on the covariance matrix of the received signal. An optimization model for performing tensor decomposition is constructed, the objective of which is to impose a penalty proportional to the intensity of the disturbance on factor vectors associated with high disturbance dimensions through a pre-defined sparsity constraint term. By solving the optimization model, multiple rank-tensor components are obtained, and selective elimination is performed on the multiple rank-tensor components according to a preset elimination criterion. The elimination criterion is used to characterize the interference intensity and signal power contribution corresponding to each rank-tensor component until the transmission rank is reduced to the optimal transmission rank.

8. The method according to claim 1, characterized in that, Reconstructing the transmitted signal of the wireless channel based on the result of the rank-up or rank-down operation includes: A new transmitter precoding matrix is ​​generated using the updated tensor factor matrix; Based on the optimal transmission rank, the mapping relationship between data and spatial stream is updated synchronously, and the most matching modulation and coding strategy is selected from the preset set of modulation and coding strategies. The reconstructed wireless signal is generated and transmitted using the new transmitter precoding matrix, combined with the updated mapping relationship and the selected modulation and coding strategy.

9. A wireless signal multidimensional array coding system based on dynamic rank optimization, characterized in that, include: The data acquisition module is used to detect the status information of the wireless channel in order to extract the channel quality parameters at the current moment; The model prediction module is used to input the channel quality parameters into a pre-trained deep learning model, predict the channel quality parameters of the wireless channel at a preset time in the future, and determine the dynamic optimization multiplier based on the prediction results. The joint solution module is used to adjust the preset static optimization multiplier in real time using the dynamic optimization multiplier, construct an objective function for jointly optimizing energy efficiency and spectral efficiency, and determine the optimal transmission rank by solving the objective function. The static optimization multiplier is determined based on the channel quality parameters of the wireless channel under the preset communication scenario. The signal processing module is used to compare the optimal transmission rank with the current transmission rank of the wireless channel to determine whether to perform a rank increase or decrease operation. When the optimal transmission rank is not less than the current transmission rank, an incremental tensor decomposition algorithm is used to perform a rank increase operation to expand the signal dimension. When the optimal transmission rank is less than the current transmission rank, a sparse constraint tensor decomposition algorithm based on interference identification is used to perform a rank decrease operation to compress the signal dimension. A signal transmission module is used to reconstruct the transmission signal of the wireless channel based on the result of the rank-up or rank-down operation.

Images