A method for adaptive demodulation of a spatial modulation system

Abstract

The application provides a kind of adaptive demodulation method of enhanced spatial modulation system, comprising: S1: receiving end obtains real-time channel matrix H and receiving signal y; using orthogonal matrix Q to the receiving signal y is carried out orthogonal transformation and obtains transformed receiving signal;S2: preset maximum reserved node number of each layer, initialize search layer;Generate the candidate symbol set of initial layer, and calculate the initial metric value corresponding to each candidate symbol;All initial metric values are screened, and the optimal initial metric value is reserved;S3: from the 2nd layer to layer, carry out iterative tree search layer by layer;S4: judge the layer number of current layer, and according to the judgment result, the candidate symbol and the corresponding cumulative metric value are reserved, and the reserved candidate symbol is stored in symbol vector;When the current layer is layer, symbol vector is used as the estimated value of transmitting vector x in enhanced spatial modulation system.

Description

Technical Field

[0001] This application belongs to the field of wireless communication technology and relates to signal detection technology in multiple input multiple output (MIMO) systems. Specifically, it relates to an adaptive demodulation method for enhanced spatial modulation (ESM) systems for robust data transmission. Background Technology

[0002] Spatial Modulation (SM) technology, as a novel MIMO transmission scheme, uses antenna indexes and modulation symbols to carry information, improving spectral efficiency while effectively avoiding inter-channel interference. ESM is a significant improvement over traditional spatial modulation; it enhances antenna utilization by introducing a hybrid activation mode of single and dual antennas, thereby further increasing the system's transmission rate.

[0003] In ESM systems, the receiver typically employs the Maximum Likelihood (ML) detection algorithm to achieve optimal demodulation performance. However, the ML detection algorithm requires traversing all possible combinations of transmit vectors, and its computational complexity increases exponentially with the number of transmit antennas and the modulation order. Consequently, this extremely high complexity makes the ML algorithm difficult to apply in practical communication systems, especially in scenarios with high real-time requirements.

[0004] Recent studies have proposed low-complexity demodulation methods for ESM systems, including the Transmit Vector Tree Search Detector (TV-TSD) and the Low-Complexity Detector based on ordered DeduplicatedActivation Antenna Combinations (DAAC-LCD). However, TV-TSD requires initialization based on all possible transmit vectors, which still has high complexity in large-scale ESM systems. On the other hand, the demodulation performance of DAAC-LCD is somewhat inferior to that of ML algorithms, making it difficult to meet the requirements of high-performance communication scenarios.

[0005] Meanwhile, existing methods fail to fully utilize channel state information to dynamically optimize the demodulation process, resulting in an imbalance between complexity and performance. Therefore, there is an urgent need to design an ESM adaptive low-complexity demodulation method for robust data transmission, which can significantly reduce the computational complexity at the receiver and improve the robustness and effectiveness of data transmission while ensuring demodulation performance is close to that of the ML algorithm.

[0006] Application content

[0007] To address the above problems, this application provides an adaptive demodulation method for enhancing spatial modulation systems, the demodulation method comprising:

[0008] S1: The receiving end acquires the real-time channel matrix H and the received signal y, performs QL orthogonal decomposition on the channel matrix H, and decomposes it into an orthogonal matrix Q and a lower triangular matrix L;

[0009] The received signal y is subjected to an orthogonal transformation using an orthogonal matrix Q, and the transformed received signal is obtained. ;

[0010] S2: Preset maximum number of nodes to retain per layer Initialize the search layer;

[0011] Generate the candidate symbol set for the initial layer And calculate each candidate symbol Corresponding initial metric ,in ;

[0012] For all initial metrics Filter and retain those. The optimal initial metric ,in ; and the corresponding candidate symbols Store the symbol vector ;

[0013] S3: Traverse from level 2 to... Layer by layer, an iterative tree search is performed. The iterative tree search at each layer includes:

[0014] Based on the symbol vector of the previous layer By combining antenna activation rules and constellation diagram rules, the candidate symbol set for the current layer is dynamically generated sequentially. , ,in, This refers to the number of antennas at the transmitting end;

[0015] Get the candidate set of the current layer Each candidate symbol And calculate each candidate symbol Cumulative metric ;

[0016] S4: Determine the current layer number and retain candidate symbols based on the determination result. and its corresponding cumulative metric value and retain candidate symbols Store the symbol vector ;

[0017] The current layer is When layering, the symbol vector As an estimate of the transmit vector x in the enhanced spatial modulation system.

[0018] As an optional technical solution, step S1, "performing QL orthogonal decomposition on the channel matrix H and decomposing it into an orthogonal matrix Q and a lower triangular matrix L", includes:

[0019] The receiver performs QL orthogonal decomposition on the channel matrix H.

[0020] ,

[0021] Where Q is The orthogonal matrix, L is The lower triangular matrix, where The number of antennas at the receiving end. .

[0022] As an optional technical solution, in step S1, "using an orthogonal matrix Q to perform an orthogonal transformation on the received signal y and obtaining the transformed received signal" "include:

[0023] Extracting the first orthogonal matrix Q Listed to number Columns form a submatrix ;

[0024] Perform an orthogonal transform on the received signal y to obtain the transformed received signal. ,

[0025] ,

[0026] in, for The conjugate transpose of .

[0027] As an optional technical solution, step S2 "generates a candidate symbol set for the initial layer" In the context of candidate symbol set for:

[0028] ,

[0029] Where 0 represents the antenna being turned off and silent. The main constellation symbol set, This is the first set of constellation symbols. This is the second set of constellation symbols.

[0030] As an optional technical solution, in step S2, "calculating each candidate symbol" Corresponding initial metric "include:

[0031] ,

[0032] in, The first element of the received signal after orthogonal transformation. The first-order main diagonal channel coefficients of the lower triangular matrix L are... The current candidate symbol for traversal;

[0033] Step "For all initial metrics" Filter and retain those. The optimal initial metric "Includes: all initial metrics" Sort in ascending order and keep the top few. Initial metric .

[0034] As an optional technical solution, in step S3, "based on the symbol vector of the previous layer" By combining antenna activation rules and constellation diagram rules, the candidate symbol set for the current layer is dynamically generated sequentially. "include:

[0035] when If the number of non-zero elements is 2, then ;

[0036] when If the number of non-zero elements is 0, and the current layer is the last layer, then... ,otherwise ;

[0037] when If the number of non-zero elements in the array is 1, then let the non-zero element be z. If ,but ;

[0038] like And if the current layer is the last layer, then ,otherwise ;

[0039] like And if the current layer is the last layer, then ,otherwise ;

[0040] Where 0 represents the antenna being turned off and silent. The main constellation symbol set, This is the first set of constellation symbols. This is the second set of constellation symbols.

[0041] As an optional technical solution, step S3, "obtaining the candidate set of the current layer" Each candidate symbol And calculate each candidate symbol Cumulative metric "include:

[0042] ,

[0043] in, , Accumulate metrics for the path in the previous layer. The received signal of the nth layer after orthogonal transformation. The main diagonal channel coefficients of the lower triangular matrix L are... Let be the non-diagonal element in the nth row and kth column of the lower triangular matrix L.

[0044] As an optional technical solution, step S4 involves "determining the current layer number and retaining candidate symbols based on the determination result". and its corresponding cumulative metric value and retain candidate symbols Store the symbol vector "include:

[0045] Determine if the current layer is 2 or higher. Layer, for all candidate symbols in the current layer. Cumulative metric Perform filtering and retain node r and its corresponding... The optimal cumulative metric and the corresponding candidate symbols Symbol vector added to the previous layer In this process, the symbol vector of the current layer is formed. Simultaneously record the corresponding optimal cumulative metric value. As the cumulative path metric for the current layer .

[0046] As an optional technical solution, the step "for all candidate symbols in the current layer" Cumulative metric Perform filtering and retain node r and its corresponding... The optimal cumulative metric "Includes: all candidate symbols for the current layer" Cumulative metric Sort in ascending order and keep the first few. Cumulative metric And the corresponding node r.

[0047] As an optional technical solution, step S4 involves "determining the current layer number and retaining candidate symbols based on the determination result". and its corresponding cumulative metric value and symbol vector "include:

[0048] Determine the current layer is Layer, retain Minimum cumulative metric in the layer and the corresponding candidate symbols Symbol vector added to the previous layer In this process, the symbol vector of the current layer is formed. .

[0049] Compared with the prior art, this application uses QL decomposition to accumulate metric values. The computation is simplified to a hierarchical iterative form, and combined with the node retention mechanism at each level in the tree search, the exponential search complexity is reduced to linear. Furthermore, this application does not change the transmitter structure and transmission protocol of the existing ESM system, but only improves the demodulation method at the receiver, which can be directly applied to the existing ESM system communication architecture, resulting in low application cost. Attached Figure Description

[0050] Figure 1 A flowchart illustrating an adaptive demodulation method to enhance spatial modulation systems. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0052] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0053] It should be noted that the demodulation method in this application is applied to the receiver of an existing ESM (Enhanced Spatial Modulation) system. The transmitter of this ESM system is configured with... One transmitting antenna, the receiving end is equipped with The root receiving antenna. Its antenna activation modes include single-antenna activation and dual-antenna activation.

[0054] When a single antenna is activated, the transmitted symbols change from order to... Selected from the main constellation chart, possessing There are several activation modes. M is the order of the main constellation in the ESM modulation, and the symbol set of the main constellation diagram is... , It contains M modulation symbols.

[0055] When dual antennas are activated, the first step is to activate two antennas of order 1. A constellation is selected from the sub-constellation diagram, and then the symbols to be transmitted are chosen from the selected sub-constellation diagram. The symbols for both antennas come from the same selected sub-constellation. The symbol set in the sub-constellation diagram... This is the first set of constellation symbols. This is the second set of constellation symbols. and All include One modulation symbol. Because the transmitter has... When selecting dual-antenna activation, the first antenna from the root transmit antenna... Choose any one of the transmitting antennas, and the second antenna is also selected from... One of the transmit antennas can be selected at random, so when both antennas are activated, there are a total of There are two activation modes. Therefore, the two antenna activation methods of this application have a total of [number missing]. kind.

[0056] Furthermore, in this application, the received signal model of the receiving end is as follows: ,in for The channel matrix has elements that follow a complex Gaussian distribution CN(0~1). For ESM emission vector, It is Gaussian white noise, and its elements follow a complex Gaussian distribution. , This represents the noise variance.

[0057] like Figure 1 As shown, this application provides an adaptive demodulation method for enhancing spatial modulation systems, the demodulation method comprising:

[0058] S1: The receiving end acquires the real-time channel matrix H and the received signal y, performs QL orthogonal decomposition on the channel matrix H, and decomposes it into an orthogonal matrix Q and a lower triangular matrix L;

[0059] The received signal y is subjected to an orthogonal transformation using an orthogonal matrix Q, and the transformed received signal is obtained. ;

[0060] S2: Preset maximum number of nodes to retain per layer Initialize the search layer;

[0061] Generate the candidate symbol set for the initial layer And calculate each candidate symbol Corresponding initial metric ,in ;

[0062] For all initial metrics Filter and retain those. Optimal initial metric values ,in ; and the corresponding candidate symbols Store the symbol vector ;

[0063] S3: Traverse from level 2 to... Layer by layer, an iterative tree search is performed. The iterative tree search at each layer includes:

[0064] Based on the symbol vector of the previous layer By combining antenna activation rules and constellation diagram rules, the candidate symbol set for the current layer is dynamically generated sequentially. , ,in, This refers to the number of antennas at the transmitting end;

[0065] Get the candidate set of the current layer Each candidate symbol And calculate each candidate symbol Cumulative metric ;

[0066] S4: Determine the current layer number and retain candidate symbols based on the determination result. and its corresponding cumulative metric value and retain candidate symbols Store the symbol vector ;

[0067] The current layer is When layering, the symbol vector As an estimate of the transmit vector x in the enhanced spatial modulation system.

[0068] In this application, accumulated metrics are decomposed using QL. The computation is simplified to a hierarchical iterative form, and combined with the node retention mechanism at each level in the tree search, the exponential search complexity is reduced. Furthermore, this application does not change the transmitter structure and transmission protocol of the existing ESM system, but only improves the demodulation method at the receiver, which can be directly applied to the existing ESM system communication architecture, resulting in low application cost.

[0069] In this application, the search layer is a collective term for all levels in the hierarchical tree search that require path expansion, metric calculation, and pruning / selection. It includes the initial layer (layer 1), intermediate layers (layer 2 to layer 3), and intermediate layers (layer 4 to 5). -1), Tail layer (first) (Layer). In this application, "initializing the search layer" means initializing the search layer and executing "generating the candidate symbol set of the initial layer". "etc."

[0070] Step S1, "performing QL orthogonal decomposition on the channel matrix H and decomposing it into an orthogonal matrix Q and a lower triangular matrix L," includes:

[0071] The receiver performs QL orthogonal decomposition on the channel matrix H.

[0072] ,

[0073] Where Q is The orthogonal matrix, L is The lower triangular matrix, where The number of antennas at the receiving end. .

[0074] Therefore, the channel matrix H is orthogonally decomposed into a lower triangular matrix L, transforming the influence of the transmitted signal into a hierarchical structure, thus enabling subsequent low-complexity tree-based demodulation. Furthermore, in this application, the number of antennas at the receiving end... Greater than the number of antennas at the transmitting end This is to satisfy the structural conditions of the channel matrix QL decomposition, ensure the normal decomposition of the orthogonal matrix Q and the lower triangular matrix L, and provide the mathematical premise for the dimensionality reduction transformation of the received signal and the hierarchical iterative tree demodulation.

[0075] Furthermore, in step S1, "an orthogonal transformation is performed on the received signal y using an orthogonal matrix Q to obtain the transformed received signal". "include:

[0076] Extracting the first orthogonal matrix Q Listed to number Columns form a submatrix ;

[0077] Perform an orthogonal transform on the received signal y to obtain the transformed received signal. ,

[0078] ,

[0079] in, for The conjugate transpose of .

[0080] In this scheme, the latter part of the orthogonal matrix Q is extracted. Columns form a submatrix Since the channel matrix, after QL decomposition, only the lower triangular matrix L carries the effective channel information, and the effective components corresponding to it in the orthogonal matrix Q are its subsequent components. Therefore, this part of the columns is extracted to form a submatrix. It can realize the reception of signals by Wei Xiang Dimensionality reduction can be achieved, and the effective subspace of the signal can be screened out while discarding invalid null space components. Under the premise of maintaining the norm-preserving properties of orthogonal transformation, the equivalent distance transformation can be completed, providing a foundation for hierarchical iterative tree demodulation.

[0081] As described above, the received signal model at the receiving end is: Where y is the actual received noisy signal. Theoretically, the signal that should be received is an ideal, noise-free signal. Therefore, the demodulation process is to find the most suitable transmission vector x such that y and The difference has the smallest L2 norm square, which is to obtain The value of is the smallest. The formula is the square of the Euclidean distance, which is used to determine the magnitude of signal differences: the smaller the distance, the higher the matching degree.

[0082] Since the orthogonal matrix Q satisfies ,in, Let Q be the conjugate transpose of Q. Submatrix It is obtained by truncating a portion of the orthogonal matrix Q. It is a column orthogonal matrix and also a unitary matrix, which satisfies the following conditions: Therefore, the unitary matrix has local normality preservation, ensuring that the subsequent Euclidean distance equivalent transformation holds true.

[0083] Specifically, the equivalent transformation of Euclidean distance is as follows:

[0084] = = = ;

[0085] in .

[0086] As mentioned above, the front of the orthogonal matrix Q Listed as a redundant noise subspace, the following Columns as submatrices , It can be equivalent to a block matrix , for An identity matrix of order 1, therefore, .

[0087] therefore, It can be equivalent to This laid the foundation for subsequent hierarchical iterative calculations.

[0088] Furthermore, step S2 "generates a candidate symbol set for the initial layer" In the context of candidate symbol set for:

[0089] ,

[0090] Where 0 represents the antenna being turned off and silent. The main constellation symbol set, This is the first set of constellation symbols. This is the second set of constellation symbols.

[0091] Step S2 is used to complete the initial configuration of the hierarchical tree search. First, the search layer is initialized, setting n=1, and the search iteratively begins from the first layer. The maximum number of nodes to retain at each layer is defined as follows. This is used to limit the number of search paths at each level, thus implementing pruning control. In this application, the maximum number of nodes to be retained at each level is the same, which is... This results in a lower difficulty level and controllable complexity. Of course, users can flexibly adjust the settings according to the needs of different communication scenarios. parameter.

[0092] Furthermore, a candidate symbol set for the initial layer is defined, which includes the symbol "0" and all available modulation symbols of the system, including the main constellation symbol set. The first set of constellation symbols Second set of constellation symbols The set of main constellation symbols This is the set of modulation symbols used to carry valid information data in a single-antenna active mode. Typically, a standard, high-order constellation diagram is used, such as 16QAM or 64QAM. Symbols are emitted directly from the active single antenna, without involving symbol combinations or interference between antennas. This is the first constellation symbol set. Second set of constellation symbols This is used for dual-antenna activation mode. When the system selects to activate two antennas simultaneously, each antenna selects one symbol from the sub-constellation, thus transmitting two symbols in one time slot and improving spatial multiplexing gain. The sub-constellation is usually obtained from the primary constellation through rotation, scaling, or remapping, and its modulation order is lower than that of the primary constellation. In this application, its order is [not specified]. .

[0093] Thus, the candidate symbol set Including 0 and primary and secondary symbols, it provides a complete candidate space for modulation symbols for the first layer, enabling multi-stream transmission of enhanced spatial modulation. Furthermore, it provides additional redundant paths for hierarchical path selection, reducing the probability of premature pruning of the optimal path and improving detection robustness in low signal-to-noise ratio scenarios.

[0094] Furthermore, in step S2, "calculating each candidate symbol" Corresponding initial metric "include:

[0095] ,

[0096] in, The first element of the received signal after orthogonal transformation. The channel coefficients are the first-order main diagonal coefficients of the lower triangular matrix L. The current candidate symbol for traversal;

[0097] Step "For all initial metrics" Filter and retain those. The optimal initial metric "Includes: all initial metrics" Sort in ascending order and keep the top few. Initial metric .

[0098] The candidate symbols of the initial layer Therefore, in calculating the initial metric value At that time, it is necessary to calculate each candidate symbol separately. initial metric And filter to select the one with the smallest value. Initial metric 'r' represents a node, as mentioned above. .

[0099] The initial metric The calculation method is the same as the Euclidean distance mentioned above. The first layer of error terms after splitting is used to determine whether the initial metric value is affected. smallest The current traversal candidate symbols Therefore, it is equivalent to filtering out Find the optimal path to continue expanding to the next layer.

[0100] In this specific embodiment, for all initial metric values Sort in ascending order and keep the first few Initial metric The retained initial metric value It is the smallest. One. Of course, if the initial metric value Sort in descending order and retain Initial metric Or arrange them in other ways and keep the smallest one. Initial metric Both can achieve the purpose of this application.

[0101] After completing the calculation of the first layer, subsequent iterative calculations are performed. Specifically, in step S3, "based on the symbol vector of the previous layer..." By combining antenna activation rules and constellation diagram rules, the candidate symbol set for the current layer is dynamically generated sequentially. "include:

[0102] when If the number of non-zero elements is 2, then ;

[0103] when If the number of non-zero elements is 0, and the current layer is the last layer, then... ,otherwise ;

[0104] when If the number of non-zero elements in the array is 1, then let the non-zero element be z. If ,but ;

[0105] like And if the current layer is the last layer, then ,otherwise ;

[0106] like And if the current layer is the last layer, then ,otherwise ;

[0107] Where 0 represents the antenna being turned off and silent. The main constellation symbol set, This is the first set of constellation symbols. This is the second set of constellation symbols.

[0108] The antenna activation rules include single-antenna activation rules and dual-antenna activation rules. As mentioned above, single-antenna activation starts from... One of the transmit antennas is selected to be active, while the remaining NT-1 antennas are muted. Dual antenna activation is achieved by... Two transmit antennas are selected for activation, while the remaining NT-2 antennas are muted. The constellation diagram rules include primary and secondary constellation diagram rules, encompassing the selection and combination of modulation symbols and the coding rules bound to the activation modes. The antenna activation rules and constellation diagram rules are reflected in the steps described above. Specifically:

[0109] "when If the number of non-zero elements is 2, then "middle, This is the symbol vector of the previous layer, specifically the complete path vector from layer 1 to layer (n-1). Then in... If there are two non-zero elements, it means that the dual-antenna activation mode has been completed, and the remaining antennas must remain silent, thus affecting the current layer. .

[0110] "when When the number of non-zero elements is 0, it means that no antennas are activated in the path vector from layer 1 to layer (n-1), and the activation rules that can be entered are more diverse. Specifically, if the current layer is the last layer, then to avoid invalid transmission, the single-antenna activation mode must be forcibly enabled, and only the main constellation symbol is selected as a candidate. If the current layer is not the last layer, the system retains full selection rights, and all ESM transport modes are enabled. .

[0111] "when When the number of non-zero elements is 1, it means that there is only one active antenna in the path vector from layer 1 to layer (n-1). Let this non-zero element be z. The candidate symbol set for layer n is determined based on the constellation to which this active antenna belongs. .

[0112] like This indicates that the main constellation symbol has already appeared in the path, meaning the system has selected single-antenna activation mode. Therefore, other antennas must remain silent; a second antenna cannot be activated, nor can any other non-zero symbols be transmitted. Furthermore, in the subsequent layers, only 0 can be selected.

[0113] like This corresponds to the dual-antenna activation mode. Currently, only the first antenna is activated, and the symbol of the second antenna is needed to complete pairing. If the current layer is the last layer, pairing must be completed. Furthermore, as mentioned above, in dual-antenna activation mode, both antennas must use the same sub-constellation symbol. Otherwise, if the current layer is not the last layer, it can either complete dual-antenna pairing or choose to remain silent. .

[0114] like Similarly, if the current layer is the last layer, then pairing must be completed. Otherwise, if the current layer is not the last layer, it can either complete dual-antenna pairing or choose to remain silent. .

[0115] Therefore, the technical solution of this application strictly follows the constellation and activation rules of the ESM system to ensure transmission effectiveness. Furthermore, it directly embeds single-antenna / dual-antenna modes and primary / secondary constellation mapping into the hierarchical tree-structured search path expansion, ensuring that the demodulation results match the ESM coding rules of the transmitter, thereby improving the reliability and accuracy of demodulation. In addition, the number of candidate paths at each layer is strictly limited, while avoiding premature pruning of the optimal path, significantly reducing computational complexity and improving demodulation robustness.

[0116] Furthermore, obtain the candidate symbol set according to the rules. Then, based on each candidate symbol Calculate each candidate symbol accordingly Cumulative metric Specifically, "Get the candidate set for the current layer". Each candidate symbol And calculate each candidate symbol Cumulative metric "include:

[0117] ,

[0118] in, , Accumulate metrics for the path in the previous layer. The received signal of the nth layer after orthogonal transformation. The main diagonal channel coefficients of the lower triangular matrix L are... Let be the non-diagonal element in the nth row and kth column of the lower triangular matrix L.

[0119] The The formula for the above Euclidean distance is... The recursive format. Specifically, the formula can be broken down into two parts, the first part... This is the cumulative metric for the path in the previous layer, equivalent to the sum of the total errors for paths in layers 1 to n-1. (The latter part...) The newly added cumulative metric for the nth layer, i.e., the newly added error component, is the Euclidean distance. The nth term. Specifically, Candidate symbols for the current layer Through channel gain The generated received signal components, This is the sum of the responses generated by all symbols in layer n-1 at layer n, thus pre-calculating the impact of the preceding path on the current layer. Therefore, the latter part... When calculating the cumulative metric value of the current layer, the transformed received signal needs to be... Subtract candidate symbols from the current layer The received signal components, as well as the signal components that affect the current layer from the preceding path.

[0120] For example, if Then the iterative Euclidean distance for each layer is calculated as follows:

[0121] First layer, n=1, ;

[0122] The second layer, n=2, ;

[0123] The third layer, n=3, .

[0124] Additionally, it should be noted that the current layer's total cumulative metric value... Not equal to every candidate symbol Cumulative metric For example, in the above In a specific embodiment, Not equal to , Not equal to This is because, Indicates the first The r-th path of layer r, and the candidate symbols of layer n. The combined temporary metric; and It is the final metric for the r-th retained path after the n-th layer of filtering. Includes candidate symbolic variables , is a combined metric with variables; These are fixed values ​​after filtering; only the top values ​​are retained. The metric results for the optimal path.

[0125] Furthermore, after calculating each candidate symbol in the current layer... Cumulative metric Then, the corresponding candidate symbols need to be retained according to certain rules. And form the symbol vector of the current layer. .

[0126] Specifically, in step S4, "determine the current layer number and retain candidate symbols based on the determination result". and its corresponding cumulative metric value and retain candidate symbols Store the symbol vector "include:

[0127] Determine if the current layer is 2 or higher. Layer, for all candidate symbols in the current layer. Cumulative metric Perform filtering and retain node r and its corresponding... The optimal cumulative metric and the corresponding candidate symbols Symbol vector added to the previous layer In this process, the symbol vector of the current layer is formed. Simultaneously record the corresponding optimal cumulative metric value. As the cumulative path metric for the current layer .

[0128] Therefore, if the current layer is not the last layer, then the current layer still needs to retain the information contained therein. Each node r and its corresponding candidate symbol and cumulative metric This provides a stable foundation for subsequent calculations. Therefore, in this step, we chose to retain [the relevant data / data]. Cumulative metric for each path and the corresponding candidate symbols Symbol vector added to the previous layer In this process, a path is formed to serve as the input for the next layer. Furthermore, in this application, the path can be adjusted under different circumstances. The size is adjusted to meet the transmission conditions of different signals.

[0129] Furthermore, the accumulated metric value at the next layer. In the calculation,

[0130] ,

[0131] The cumulative path metric obtained from the calculation of the previous layer. Substitute into the formula, and use the candidate symbol set in step S3. The calculation yields different candidate symbols. And the path node r obtained after the previous layer's filtering, so in the current layer's calculation, for each preceding path node r, it is necessary to traverse all candidate symbols corresponding to that path node r. This allows for permutation calculations to obtain cumulative metric values ​​of different sizes for different combinations. In step S4, the cumulative metric is... Perform the screening.

[0132] Furthermore, the step "for all candidate symbols in the current layer" Cumulative metric Perform filtering and retain node r and its corresponding... The optimal cumulative metric "Includes: all candidate symbols for the current layer" Cumulative metric Sort in ascending order and keep the first few. Cumulative metric And the corresponding node r.

[0133] For cumulative measures The filtering process is similar to step S2 above, both involving first filtering the cumulative metrics. Sort in ascending order, then keep the first few. Cumulative metric The retained cumulative metric value That is, the smallest value Cumulative metric This keeps the number of candidate paths at each layer within a fixed range. Even if the local metric at a certain layer is not globally optimal, as long as it remains within the global optimum... Within the specified range, there is an opportunity to correct errors in subsequent layers, preventing the optimal path from being pruned prematurely.

[0134] If the current layer is the last layer, the method for retaining symbols is different. Specifically, in step S4, "determine the layer number of the current layer and retain candidate symbols based on the determination result." and its corresponding cumulative metric value and symbol vector "include:

[0135] Determine the current layer is Layer, retain Minimum cumulative metric in the layer and the corresponding candidate symbols Symbol vector added to the previous layer In this process, the symbol vector of the current layer is formed. .

[0136] If the current layer is the last layer In the next layer, only the optimal solution is selected, and only the minimum cumulative metric value is chosen. Corresponding candidate symbols And add it to the symbol vector of the previous layer. In this process, the symbol vector of the current layer is formed. Therefore, the Layer symbol vector This serves as an estimate of the transmit vector x in the enhanced spatial modulation system, at which point demodulation ends.

[0137] In summary, the enhanced spatial modulation (ESM) hierarchical tree demodulation method proposed in this invention achieves an efficient balance between demodulation performance and computational complexity through path-state-based candidate set constraints, recursive metric calculation, and hierarchical pruning screening. Its beneficial effects are specifically reflected in the following aspects:

[0138] (1) Significantly reduced complexity: Traditional maximum likelihood detection (ML detection) requires traversing all possible combinations of transmit vectors, and its complexity increases exponentially with the number of antennas and modulation order, making it difficult to meet the real-time processing requirements of multi-antenna ESM systems. This invention first simplifies the core Euclidean distance metric of ESM demodulation into a hierarchical iterative form through QL decomposition, reducing the complexity of global matrix operations; at the same time, combined with the node retention mechanism of each layer in tree search, only retains The invention finds an optimal path, reducing the overall search complexity from exponential to polynomial, significantly decreasing computational load. Furthermore, it introduces a dynamic candidate set pruning strategy based on inherent ESM rules, making the average candidate set size at each layer much smaller than the full traversal space of traditional algorithms. This further reduces the computation and expansion of invalid symbols, significantly lowering the algorithm's computational complexity and hardware implementation difficulty, and enabling low-latency demodulation in high-speed communication scenarios.

[0139] (2) Demodulation performance close to optimal ML detection: Traditional low-complexity ESM demodulation algorithms often suffer significant loss in bit error rate performance due to premature pruning or insufficient path constraints. This invention incorporates the impact of the preceding path on the current layer into the error calculation in advance through the interference compensation term in the cumulative metric recursive formula, ensuring the accuracy of the layered search; at the same time, the candidate set generation method based on the inherent rules of the ESM system can accurately screen out effective paths that conform to the transmitter coding specifications, and will not mistakenly delete the most likely transmission vector during the pruning process. By retaining multiple paths instead of a single optimal path, performance redundancy is reserved for channel fluctuations and noise interference, and even in low signal-to-noise ratio or deep fading channels, global demodulation errors caused by local optima can be avoided, and robust demodulation performance can still be maintained in dynamic channel environments.

[0140] (3) Strong system adaptability: This invention only improves the demodulation algorithm at the receiving end, without modifying the transmitter structure, modulation method, or transmission protocol of the existing ESM system. It is directly compatible with the existing ESM communication architecture, without the need for additional hardware modifications or protocol upgrades, resulting in low application costs and strong compatibility. At the same time, the hierarchical iterative structure and pruning logic of this invention can be implemented modularly, and the recursive metric formula and candidate set rules can be mapped to general hardware condition branches, which facilitates rapid integration and deployment on existing communication baseband platforms and has good engineering application prospects.

[0141] (4) Parameters are adjustable, flexible and controllable: when As the value increases, the number of paths retained increases, and the demodulation performance becomes closer to ML detection; when When the size is reduced, the computational complexity is significantly lowered, making it more suitable for low-power, low-latency applications. Users can flexibly adjust the size according to the needs of different communication scenarios. The parameters are configured to achieve the optimal balance between performance and complexity, and have broad adaptability and scalability across various scenarios.

[0142] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0143] The detailed descriptions listed above are merely specific descriptions of feasible implementations of the present invention and are not intended to limit the scope of protection of the present invention. All equivalent implementations or modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention.

Claims

1. An adaptive demodulation method for enhancing spatial modulation systems, characterized in that, The demodulation method includes: S1: The receiving end acquires the real-time channel matrix H and the received signal y, performs QL orthogonal decomposition on the channel matrix H, and decomposes it into an orthogonal matrix Q and a lower triangular matrix L; The received signal y is subjected to an orthogonal transformation using an orthogonal matrix Q, and the transformed received signal is obtained. ; S2: Preset maximum number of nodes to retain per layer Initialize the search layer; Generate the candidate symbol set for the initial layer And calculate each candidate symbol Corresponding initial metric ,in ; For all initial metrics Filter and retain those. The optimal initial metric ,in ; and the corresponding candidate symbols Store the symbol vector ; S3: Traverse from level 2 to... Layer by layer, an iterative tree search is performed. The iterative tree search at each layer includes: Based on the symbol vector of the previous layer By combining antenna activation rules and constellation diagram rules, the candidate symbol set for the current layer is dynamically generated sequentially. , ,in, This refers to the number of antennas at the transmitting end; Get the candidate set of the current layer Each candidate symbol And calculate each candidate symbol Cumulative metric ; S4: Determine the current layer number and retain candidate symbols based on the determination result. and its corresponding cumulative metric value and retain candidate symbols Store the symbol vector ; The current layer is When layering, the symbol vector As an estimate of the transmit vector x in the enhanced spatial modulation system.

2. The adaptive demodulation method according to claim 1, characterized in that, Step S1, "performing QL orthogonal decomposition on the channel matrix H and decomposing it into an orthogonal matrix Q and a lower triangular matrix L", includes: The receiver performs QL orthogonal decomposition on the channel matrix H. ， Where Q is The orthogonal matrix, L is The lower triangular matrix, where The number of antennas at the receiving end. .

3. The adaptive demodulation method according to claim 2, characterized in that, In step S1, "an orthogonal transformation is performed on the received signal y using an orthogonal matrix Q to obtain the transformed received signal". "include: Extracting the first orthogonal matrix Q Listed to number Columns form a submatrix ; Perform an orthogonal transform on the received signal y to obtain the transformed received signal. , ， in, for The conjugate transpose of .

4. The adaptive demodulation method according to claim 1, characterized in that, Step S2 generates a candidate symbol set for the initial layer. In the context of candidate symbol set for: ， Where 0 represents the antenna being turned off and silent. The main constellation symbol set, This is the first set of constellation symbols. This is the second set of constellation symbols.

5. The adaptive demodulation method according to claim 1, characterized in that, In step S2, "calculate each candidate symbol" Corresponding initial metric "include: ， in, The first element of the received signal after orthogonal transformation. The first-order main diagonal channel coefficients of the lower triangular matrix L are... The current candidate symbol for traversal; Step "For all initial metrics" Filter and retain those. The optimal initial metric "Includes: all initial metrics" Sort in ascending order and keep the top few. Initial metric .

6. The adaptive demodulation method according to claim 1, characterized in that, In step S3, "based on the symbol vector of the previous layer" By combining antenna activation rules and constellation diagram rules, the candidate symbol set for the current layer is dynamically generated sequentially. "include: when If the number of non-zero elements is 2, then ; when If the number of non-zero elements is 0, and the current layer is the last layer, then... ,otherwise ; when If the number of non-zero elements in the array is 1, then let the non-zero element be z. If ,but ; like And if the current layer is the last layer, then ,otherwise ; like And if the current layer is the last layer, then ,otherwise ; Where 0 represents the antenna being turned off and silent. The main constellation symbol set, This is the first set of constellation symbols. This is the second set of constellation symbols.

7. The adaptive demodulation method according to claim 6, characterized in that, In step S3, "obtain the candidate set of the current layer" Each candidate symbol And calculate each candidate symbol Cumulative metric "include: ， in, , Accumulate metrics for the path in the previous layer. The received signal of the nth layer after orthogonal transformation. The main diagonal channel coefficients of the lower triangular matrix L are... Let be the non-diagonal element in the nth row and kth column of the lower triangular matrix L.

8. The adaptive demodulation method according to claim 7, characterized in that, In step S4, "determine the current layer number and retain candidate symbols based on the determination result". and its corresponding cumulative metric value and retain candidate symbols Store the symbol vector "include: Determine if the current layer is 2 or higher. Layer, for all candidate symbols in the current layer. Cumulative metric Perform filtering and retain node r and its corresponding... The optimal cumulative metric and the corresponding candidate symbols Symbol vector added to the previous layer In this process, the symbol vector of the current layer is formed. Simultaneously record the corresponding optimal cumulative metric value. As the cumulative path metric for the current layer .

9. The adaptive demodulation method according to claim 8, characterized in that, Step "For all candidate symbols in the current layer" Cumulative metric Perform filtering and retain node r and its corresponding... The optimal cumulative metric "Includes: all candidate symbols for the current layer" Cumulative metric Sort in ascending order and keep the first few. Cumulative metric And the corresponding node r.

10. The adaptive demodulation method according to claim 1, characterized in that, In step S4, "determine the current layer number and retain candidate symbols based on the determination result". and its corresponding cumulative metric value and symbol vector "include: Determine the current layer is Layer, retain Minimum cumulative metric in the layer and the corresponding candidate symbols Symbol vector added to the previous layer In this process, the symbol vector of the current layer is formed. .

Images