Signal estimation method and device and base station
By introducing graph neural network and multi-base station collaboration in the expected propagation detection model, the posterior probability estimation is optimized, and the performance degradation caused by multi-user interference in small intervals in super-large-scale MIMO systems is solved, achieving more efficient signal estimation and reception performance improvement .
Patent Information
- Application Number
- CN202310547355.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-16
- Publication Date
- 2024-11-22
AI Technical Summary
The prior art has degraded performance when dealing with the problem of inter-cell multi-user interference in ultra-large-scale MIMO systems, especially the MIMO reception scheme based on deep learning has failed to effectively overcome the uplink capacity bottleneck caused by inter-cell interference.
Using an improved expected propagation detection model, combined with a graph neural network, the graph neural network is inserted between linear and nonlinear estimation modules, optimize posterior probability estimation, improve processing capabilities for multiple users, and improve reception through multi-base station collaboration interaction performance.
It significantly improves the resistance and reception performance of multi-user interference in MIMO system, reduces computing overhead, and effectively overcomes the compromises in complexity and performance of traditional methods.
Smart Images

Figure CN119010960A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of communication technology, and in particular to a signal estimation method, device and base station. Background Art
[0002] In recent years, Multiple-input Multiple-output (MIMO) technology has gradually evolved towards ultra-large-scale MIMO. The frequency efficiency and energy efficiency gains brought by ultra-large-scale MIMO depend on the data detection strategy at the receiving end, and the related design becomes extremely complex as the number of antennas increases.
[0003] In the prior art, Approximate Message Passing (AMP) and Expectation Propagation (EP) algorithms are often used to detect and estimate user sent data. However, the AMP detector is not applicable to the real channel environment with spatial correlation. Although the EP detector is more robust to the channel environment, when the system has serious multi-user interference, there will be serious approximation inaccuracy problems, resulting in a significant degradation in performance.
[0004] Artificial Intelligence (AI) technology, led by Deep Learning (DL), has developed rapidly in recent years. However, existing DL-based MIMO reception solutions have not yet considered how to deal with inter-cell interference, a performance bottleneck that constrains uplink capacity. It can be seen that existing technologies lack effective solutions to the problem of inter-cell multi-user interference. Summary of the invention
[0005] The embodiments of the present application provide a signal estimation method, device and base station to solve the problem that the prior art is difficult to cope with multi-user interference between cells.
[0006] In a first aspect, an embodiment of the present application provides a signal estimation method, including:
[0007] The first base station determines first uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range;
[0008] The first base station detects, by the first user terminal, a service data signal sent by the first user terminal to obtain a first observation signal, wherein the first base station is a local base station of a cell where the first user terminal is located;
[0009] The first base station inputs the first uplink channel state information and the first observation signal into a pre-trained first signal estimation model to obtain first estimation information of the service data signal output by the first signal estimation model, wherein the first signal estimation model adopts an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model.
[0010] Optionally, the method further comprises:
[0011] The first base station receives second estimation information of the service data signal sent by a second base station, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located;
[0012] The first base station determines target estimation information for the service data signal according to the first estimation information and the second estimation information.
[0013] Optionally, the first signal estimation model is trained by the first base station using local training data, or the first signal estimation model is distributed to the first base station by a central server and is trained by the central server using training data of each base station.
[0014] Optionally, the first base station determines, according to the first estimation information and the second estimation information, target estimation information for the service data signal, including:
[0015] In a case where the first estimation information and the second estimation information are represented by posterior probabilities, the first base station multiplies the first estimation information and the second estimation information to obtain the target estimation information;
[0016] In a case where the first estimation information and the second estimation information are represented using a posterior log-likelihood ratio, the first base station adds the first estimation information and the second estimation information to obtain the target estimation information.
[0017] Optionally, the training process of the first signal estimation model includes:
[0018] Acquire a training sample set, wherein the training sample set includes uplink channel state samples, uplink data samples sent by the user terminal, and observation signal samples of the uplink data sent by the user terminal at the base station end after passing through the uplink channel;
[0019] Inputting uplink channel state samples and observation signal samples in the training sample set into an initially constructed first signal estimation model to obtain an estimated signal output by the initially constructed first signal estimation model;
[0020] Determining a model loss value based on a difference between the estimated signal and the uplink data sample;
[0021] Based on the model loss value, the structural parameters of the initially constructed first signal estimation model are adjusted to obtain a trained first signal estimation model.
[0022] In a second aspect, the embodiment of the present application also provides another signal estimation method, including:
[0023] The second base station determines second uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range;
[0024] The second base station detects, by the first user terminal, a service data signal sent by the first user terminal to obtain a second observation signal, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located;
[0025] The second base station inputs the second uplink channel state information and the second observation signal into a pre-trained second signal estimation model to obtain second estimation information of the service data signal output by the second signal estimation model, wherein the second signal estimation model is an improved desired propagation detection model, and a graph neural network is inserted between a linear estimation module and a nonlinear estimation module in each network layer of the improved desired propagation detection model;
[0026] The second base station sends the second estimation information to the first base station, where the first base station is a local base station of a cell where the first user terminal is located.
[0027] In a third aspect, an embodiment of the present application further provides a signal estimation device, which is arranged in a first base station, and includes:
[0028] A first determination module, configured to determine first uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range;
[0029] A first detection module, configured to detect a service data signal sent by a first user terminal to obtain a first observation signal, wherein the first base station is a local base station of a cell where the first user terminal is located;
[0030] The first signal estimation module is used to input the first uplink channel state information and the first observation signal into a pre-trained first signal estimation model to obtain first estimation information of the service data signal output by the first signal estimation model, wherein the first signal estimation model adopts an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model.
[0031] Optionally, the signal estimation device further includes:
[0032] a receiving module, configured to receive second estimation information of the service data signal sent by a second base station, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located;
[0033] The third determination module is used to determine target estimation information for the service data signal according to the first estimation information and the second estimation information.
[0034] Optionally, the first signal estimation model is trained by the first base station using local training data, or the first signal estimation model is distributed to the first base station by a central server and is trained by the central server using training data of each base station.
[0035] Optionally, the third determination module is used to, when the first estimation information and the second estimation information are represented by posterior probabilities, multiply the first estimation information and the second estimation information to obtain the target estimation information;
[0036] Alternatively, the third determination module is used to add the first estimation information and the second estimation information to obtain the target estimation information when the first estimation information and the second estimation information are represented by a posterior log-likelihood ratio.
[0037] Optionally, the training process of the first signal estimation model includes:
[0038] Acquire a training sample set, wherein the training sample set includes uplink channel state samples, uplink data samples sent by the user terminal, and observation signal samples of the uplink data sent by the user terminal at the base station end after passing through the uplink channel;
[0039] Inputting uplink channel state samples and observation signal samples in the training sample set into an initially constructed first signal estimation model to obtain an estimated signal output by the initially constructed first signal estimation model;
[0040] Determining a model loss value based on a difference between the estimated signal and the uplink data sample;
[0041] Based on the model loss value, the structural parameters of the initially constructed first signal estimation model are adjusted to obtain a trained first signal estimation model.
[0042] In a fourth aspect, an embodiment of the present application further provides another signal estimation device, which is arranged in a second base station and includes:
[0043] A second determination module, configured to determine second uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range;
[0044] A second detection module, configured to detect a service data signal sent by a first user terminal to obtain a second observation signal, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located;
[0045] a second signal estimation module, configured to input the second uplink channel state information and the second observation signal into a pre-trained second signal estimation model, and obtain second estimation information of the service data signal output by the second signal estimation model, wherein the second signal estimation model is an improved desired propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved desired propagation detection model;
[0046] A sending module is used to send the second estimation information to a first base station, wherein the first base station is a local base station of a cell where the first user terminal is located.
[0047] In the fifth aspect, an embodiment of the present application also provides a base station, comprising: a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps in the signal estimation method as described in the first aspect are implemented; or the steps in the signal estimation method as described in the second aspect are implemented.
[0048] In a sixth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps in the signal estimation method as described in the first aspect are implemented; or the steps in the signal estimation method as described in the second aspect are implemented.
[0049] In an embodiment of the present application, the first base station determines the first uplink channel state information according to the uplink reference signal sent by the user terminal within the detection range; the first base station detects the service data signal sent by the first user terminal to obtain the first observation signal, wherein the first base station is the local base station of the cell where the first user terminal is located; the first base station inputs the first uplink channel state information and the first observation signal into a pre-trained first signal estimation model to obtain the first estimation information of the service data signal output by the first signal estimation model, wherein the first signal estimation model adopts an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model. In this way, the service data sent by the user terminal is estimated by adopting the improved expected propagation detection model, and the structured dependency characteristics between the transmitted data streams are learned by means of the graph neural network, and the posterior probability estimation is optimized to provide more accurate information to the nonlinear estimation module, so as to make up for the problem that the expected propagation detection model has a large error when using the Gaussian approximation when performing a posteriori inference on the transmitted data, thereby improving the model's ability to resist severe multi-user interference in the MIMO system. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0051] Figure 1 It is one of the flow charts of the signal estimation method provided in the embodiment of the present application;
[0052] Figure 2 is a schematic diagram of the structure of an improved desired propagation detection model provided in an embodiment of the present application;
[0053] Figure 3 It is a schematic diagram of the wireless cellular network architecture used in the embodiments of the present application;
[0054] Figure 4 is a flowchart of the interaction between a base station and a user terminal for signal estimation provided by an embodiment of the present application;
[0055] Figure 5 This is the second flowchart of the signal estimation method provided in the embodiment of the present application;
[0056] Figure 6 is one of the structural diagrams of the signal estimation device provided in the embodiment of the present application;
[0057] Figure 7 This is the second structural diagram of the signal estimation device provided in the embodiment of the present application;
[0058] Figure 8 It is a structural diagram of a base station provided in an embodiment of the present application. DETAILED DESCRIPTION
[0059] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0060] In order to make the embodiments of the present application clearer, the relevant technical knowledge involved in the embodiments of the present application is first introduced as follows:
[0061] In recent years, MIMO technology has gradually evolved towards ultra-large-scale MIMO. By deploying more antennas in base stations and wireless access points, the spectrum efficiency of the system is greatly improved, and the capacity and coverage of wireless networks are expanded exponentially. However, the frequency efficiency and energy efficiency gains brought by ultra-large-scale MIMO depend on efficient data detection strategies at the receiving end. The related design becomes extremely complex as the number of antennas increases. For example, the maximum likelihood detection scheme, which is the best in terms of performance, becomes impractical due to the exponential growth of the number of decision variables. Researchers have therefore proposed a series of suboptimal alternatives. Among them, the traditional linear receiving scheme can complete the detection with lower complexity, but it has been criticized for its large performance gap with the maximum likelihood scheme. The sphere decoding detector can provide performance comparable to the maximum likelihood, but its complexity still grows exponentially with the system dimension. Even in a medium-sized multi-antenna system, it is difficult to meet the detection delay requirements of the existing system. Therefore, the design of the receiving scheme needs to focus on the compromise between performance and complexity, which is also the focus and difficulty of the current research in the industry.
[0062] As representative schemes of approximate inference in Bayesian theory, the AMP algorithm and the EP algorithm have attracted widespread attention in the design of MIMO reception schemes due to their moderate complexity and excellent performance. The AMP detector can achieve Bayesian optimal performance in large-scale MIMO systems where the channels follow independent sub-Gaussian distributions, but it is not applicable to real channel environments with spatial correlation. The EP detector is more robust to the channel environment, but its significant drawback is that it performs poorly when the system has severe multi-user interference.
[0063] AI technology, led by deep learning (DL), has developed rapidly in the past decade. Intelligent communication with endogenous AI has become one of the recognized core enabling technologies for future wireless networks. Physical layer communication integrated with DL is also expected to break the traditional design bottleneck and obtain a new architecture with better overall performance. In particular, deep neural networks have become a powerful tool for balancing the performance and complexity of MIMO reception methods due to their excellent feature extraction and accurate inference capabilities. However, existing DL-based MIMO reception schemes have not yet considered how to deal with inter-cell interference, a performance bottleneck that constrains uplink capacity. Therefore, how to use DL technology to overcome the remaining challenges of uplink multi-cell and multi-user scenario reception schemes is urgently in need of research.
[0064] According to the analysis of existing technologies, it is a common problem that it is difficult to achieve a good compromise between performance and complexity. On the one hand, due to complexity constraints, traditional maximum likelihood and spherical decoding reception schemes cannot adapt to the growth of the number of antennas and high-order modulation schemes. On the other hand, low-complexity linear reception schemes cannot provide guarantees for the potential gains of MIMO technology due to their poor performance. Although AMP and EP technologies based on approximate inference provide a certain degree of performance and complexity compromise, their lack of robustness to actual environments restricts their development. Taking EP technology as an example, it uses a simple Gaussian distribution to approximate the posterior probability distribution of transmitted symbols. As the number of system users increases, there will be serious approximation inaccuracies, resulting in a significant decrease in performance.
[0065] Among the mainstream AI-based MIMO uplink reception solutions in the industry, there are mainly two categories. The first category is a purely data-driven approach based on classic DL models such as deep neural networks and convolutional neural networks. Usually, there are problems such as large number of network parameters leading to excessive training overhead, poor generalization and interpretability. The second category is a deeply expanded model-driven network, which can be regarded as a primary form of combining classic algorithms in the communication field with DL technology. Although it overcomes the problems of the first category of solutions to a certain extent, the resulting model is strongly constrained by the prototype algorithm and cannot give full play to the advantages of DL technology.
[0066] In addition, the industry's current research on AI-based MIMO uplink reception solutions has not yet taken into account the multi-cell scenarios that are widely present in actual cellular networks. Interference from neighboring cells in multi-cell scenarios is recognized as an important factor restricting uplink capacity. Combining DL technology to deal with inter-cell interference has therefore become a technical gap that needs to be filled.
[0067] In order to address the shortcomings and problems of existing solutions, the present application proposes a data model dual-driven uplink reception method, using a graph neural network-assisted expected propagation detector as an efficient data model dual-driven detection scheme on the base station side, overcoming the performance limitations of traditional expected propagation schemes when multi-user interference is severe, thereby achieving a good compromise between performance and complexity. On the other hand, it can also take advantage of the potential advantages of multi-base station collaboration, combined with the ability of the graph neural network-assisted expected propagation detector to output soft estimation information, in exchange for a significant improvement in reception performance at a lower interaction overhead, effectively combating inter-cell interference.
[0068] See also Figure 1 , Figure 1 is a flow chart of a signal estimation method provided in an embodiment of the present application. Figure 1 As shown, the following steps are included:
[0069] Step 101: A first base station determines first uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range.
[0070] In the embodiment of the present application, before detecting user service data, the base station needs to first obtain the uplink channel state information of each user terminal within the detection range, so as to estimate the user service data based on the uplink channel state information and the observation signal of the user service data. Among them, the channel state information reflects the quality of the wireless channel between the user terminal and the base station. The specific acquisition method is that the base station configures the uplink reference signal of the user terminal within the detection range, and the user terminal sends the uplink reference signal to the base station according to the configuration. The base station then gives an estimate of the channel response based on the observation of the uplink reference signal, that is, determines the uplink channel state information. The uplink channel state information specifically includes the channel matrix H ul and noise variance N0, channel matrix H ul It consists of multiple channel information.
[0071] The first base station is a target base station for the first user terminal to send a service data signal, that is, a local base station of a cell where the first user terminal is located, that is, the first user terminal is located in the cell where the first base station is located. The first base station can determine the first uplink channel state information according to the uplink reference signal sent by the user terminal within the detection range.
[0072] Step 102: The first base station detects a service data signal sent by a first user terminal to obtain a first observation signal, wherein the first base station is a local base station of a cell where the first user terminal is located.
[0073] The first user terminal may send a service data signal to the first base station when performing a service, so that the first base station may detect the service data signal and obtain a corresponding observation signal, that is, a first observation signal.
[0074] Step 103: The first base station inputs the first uplink channel state information and the first observation signal into a pre-trained first signal estimation model to obtain first estimation information of the service data signal output by the first signal estimation model, wherein the first signal estimation model adopts an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model.
[0075] In an embodiment of the present application, in order to address the problem of multi-cell interference, an improved expected propagation detection model is proposed as a signal estimation model. This model optimizes the posterior probability estimation by leveraging the graph neural network's ability to capture structured dependency features between transmitted data streams, thereby improving the detection model's ability to combat severe multi-user interference in MIMO systems, so that the detection model can recover the service data sent by the user terminal with high precision.
[0076] Specifically, a graph neural network-assisted desired propagation detector AI model can be constructed by iteratively expanding the desired propagation detector into a deep neural network and inserting a graph neural network between its linear estimation module and nonlinear estimation module. The input of the model is the observed signal and uplink channel state information, and the output is an estimate of the uplink service data signal sent by the user terminal, that is, the posterior probability information, and the optimal antenna weight vector that adapts to the strength relationship of the user signal and the time-varying fading characteristics of the channel can be inferred.
[0077] The graph neural network is a neural network model built on a fully connected factor graph. It completes message transmission on the factor graph through three steps: message propagation, message aggregation and probability reading, and obtains optimized probability estimation with the help of a multi-layer perceptron embedded in the graph model. Specifically, the signal mean and covariance estimated by the linear module can be input into the graph neural network as prior information, and the graph neural network's ability to capture structured dependency features between transmitted data streams is used to optimize the posterior probability estimate to provide more accurate information to the nonlinear estimation module, thereby compensating for the problem of large errors in the expected propagation detector when using Gaussian approximation to perform posterior inference on the transmitted data, and improving the detector's ability to combat severe multi-user interference in MIMO systems.
[0078] In this step, the first base station may utilize a pre-trained first signal estimation model, take the first observation signal and the first uplink channel state information as model inputs, and output an estimate of the service data signal sent by the first user terminal, namely, first estimation information.
[0079] like Figure 2As shown, the first signal estimation model, that is, the improved expected propagation detection model, includes T network layers, and each network layer adopts the following method: the linear estimation module, that is, the linear minimum mean square error estimation module, calculates the signal mean and covariance of the linear estimation; the external information module is used to eliminate the coupling information in the estimation information of the linear estimation module to obtain the mean and covariance of the edge distribution; the mean and covariance of the edge distribution are input into the graph neural network as prior information, and the optimized posterior probability distribution is output; the nonlinear estimation module calculates the posterior mean and covariance according to the posterior probability distribution optimized by the graph neural network; if it is the last layer of the network, the posterior probability information is output, and the optimal antenna weight vector adapted to the strength relationship of the user signal and the time-varying fading characteristics of the channel is inferred; if it is not the last layer of the network, the estimated parameters are updated through the external information module and output to the next network layer.
[0080] Optionally, the training process of the first signal estimation model includes:
[0081] Acquire a training sample set, wherein the training sample set includes uplink channel state samples, uplink data samples sent by the user terminal, and observation signal samples of the uplink data sent by the user terminal at the base station end after passing through the uplink channel;
[0082] Inputting uplink channel state samples and observation signal samples in the training sample set into an initially constructed first signal estimation model to obtain an estimated signal output by the initially constructed first signal estimation model;
[0083] Determining a model loss value based on a difference between the estimated signal and the uplink data sample;
[0084] Based on the model loss value, the structural parameters of the initially constructed first signal estimation model are adjusted to obtain a trained first signal estimation model.
[0085] In order to train the first signal estimation model, training data may be collected first. The training data of the first signal estimation model includes uplink channel state samples, user uplink data signal samples x ul And the observation signal sample y of the user uplink data signal at the base station after passing through the uplink channel ul , the uplink channel state sample specifically includes the channel matrix H ul and noise variance N0, so each training data in the obtained training sample set is a four-tuple {x ul ,y ul ,H ul ,N0}, the training data can be generated using a specific channel model or measured in advance in a deployed channel environment, and can provide a sufficient set of training data for training the first signal estimation model.
[0086] The training of the first signal estimation model can be achieved through the classic error back propagation algorithm. In a supervised learning manner, the loss function between the estimated signal output by the model and the actual uplink data signal sample is calculated, and then the gradient of the loss function with respect to each weight parameter in the network is back-propagated, and each weight parameter is optimized using the stochastic gradient descent method or adaptive moment estimation (Adaptive Moment Estimation, Adam) and other optimizers; using a large amount of training data and multiple rounds of back propagation, the graph neural network in the first signal estimation model can learn the structured dependency characteristics between the transmitted data streams, thereby giving an accurate inference of the posterior probability distribution of the transmitted data symbols, and then the detection model can recover the transmitted data with high precision.
[0087] In this way, through this implementation, a high-precision signal estimation model can be trained to ensure the accuracy of the model's estimation of user service data signals.
[0088] Optionally, the method further comprises:
[0089] The first base station receives second estimation information of the service data signal sent by a second base station, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located;
[0090] The first base station determines target estimation information for the service data signal according to the first estimation information and the second estimation information.
[0091] In one implementation, multiple base stations can collaborate to more accurately estimate user service data signals. Figure 3 The wireless cellular network architecture shown has the following characteristics:
[0092] 1) A multi-base station cooperative system includes at least two base stations equipped with multiple receiving antennas, which are respectively located in at least two adjacent cellular cells, and a number of co-channel users using the same time-frequency resources are distributed in the cellular cells; the base station of the cell where each user is located is its local base station, and the cooperative base station of each user's local base station is determined according to the detection range of each base station, where the detection range refers to the area where the base station can detect the user signal.
[0093] 2) Each base station is connected to a central server and can exchange information error-free through the interface.
[0094] In this implementation, the second base station, i.e., the cooperative base station of the first base station, can estimate the service data signal sent by the first user terminal, and send the estimated second estimation information to the first base station, and the first base station can use the second estimation information provided by the cooperative base station to optimize its estimation of the user signal of the cell, i.e., optimize the first estimation information, thereby obtaining the optimized target estimation information. The number of the second base station, i.e., the cooperative base station, can be one or more.
[0095] Similarly, the second base station can also use a pre-trained second signal estimation model to estimate the service data signal sent by the first user terminal. The second signal estimation model has the same structure as the first signal estimation model and also adopts an improved expected propagation detection model, that is, a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the second signal estimation model. And the training process of the second signal estimation model is also similar to that of the first signal estimation model. Therefore, the structure and training process of the second signal estimation model can refer to the aforementioned introduction to the first signal estimation model, which will not be repeated here.
[0096] Before estimation, the second base station may also determine the second uplink channel state information according to the uplink reference signal sent by the user terminal within the detection range, and detect the service data signal sent by the first user terminal to obtain the second observation signal, and then input the second uplink channel state information and the second observation signal into the second signal estimation model to obtain the second estimation information output by the model, that is, the soft estimation information of the service data signal sent by the first user terminal. The second base station, i.e., the cooperative base station, may transmit the soft estimation information about the user service data signal to the first base station, i.e., the local base station. The soft estimation information may be the data symbol posterior probability or the bit posterior log-likelihood ratio.
[0097] After receiving the second estimation information, the first base station can combine the first estimation information and the second estimation information to determine the target estimation information for the service data signal. For example, the first estimation information can be multiplied or added with the second estimation information to obtain the target estimation information, or the first estimation information can be appropriately adjusted based on the second estimation information to obtain the target estimation information.
[0098] In this way, through this implementation, cooperation between base stations can be used to obtain a more accurate global estimate of user service data signals, thereby significantly improving reception performance in exchange for lower interaction overhead and effectively combating inter-cell interference.
[0099] Optionally, the first base station determines, according to the first estimation information and the second estimation information, target estimation information for the service data signal, including:
[0100] In a case where the first estimation information and the second estimation information are represented by posterior probabilities, the first base station multiplies the first estimation information and the second estimation information to obtain the target estimation information;
[0101] In a case where the first estimation information and the second estimation information are represented using a posterior log-likelihood ratio, the first base station adds the first estimation information and the second estimation information to obtain the target estimation information.
[0102] In one implementation, based on the representation form of the estimation information, it may be determined how to optimize the local estimation information based on the soft estimation information of the cooperative base station to obtain the optimized target estimation information.
[0103] When the first estimation information and the second estimation information are represented using a posteriori probability, the estimation information determined by the local base station can be multiplied with the estimation information sent by each collaborative base station to serve as the final estimate of the user service data signal, that is, the target estimation information is obtained by multiplying the first estimation information and the second estimation information.
[0104] When the first estimation information and the second estimation information are represented using a posterior log-likelihood ratio, the estimation information determined by the local base station and the estimation information sent by each collaborative base station can be accumulated as the final estimate of the user service data signal, that is, the target estimation information is obtained by adding the first estimation information and the second estimation information.
[0105] Taking the a posteriori probability representation as an example, the last layer network output of the second signal estimation model optimizes the a posteriori probability distribution p(s i |y c ,H c ), where s i is the data symbol sent by the user terminal with sequence number i, y c is the observation signal of base station numbered c, H c is the uplink channel state information from the user within the detection range of the base station numbered c to it; p(s i |y c ,H c ) is transmitted to the local base station of the cell where the user terminal with sequence number i is located; the last layer network of the first signal estimation model outputs the optimized posterior probability distribution p(s i |y i H i ), where y iis the observation signal of the local base station, H i is the uplink channel state information from the user within the detection range of the local base station to it. Then the local base station determines the posterior probability of global cooperation, expressed as:
[0106]
[0107] Among them, y coop is the set of observed signals of the local base station and all cooperative base stations, H coop is the uplink channel state information set of the local base station and all coordinated base stations, and C is the number of coordinated base stations.
[0108] In this way, through this implementation, the final estimation of the user service data signal can be given by using the posterior probability of global collaboration, thereby improving the accuracy of the estimation of the user service data signal.
[0109] Optionally, the first signal estimation model is trained by the first base station using local training data, or the first signal estimation model is distributed to the first base station by a central server and is trained by the central server using training data of each base station.
[0110] In one embodiment, the first base station and the second base station can respectively use their own local training data to train their own local signal estimation models, or a central server can collect training data from each base station to obtain a global training data set to train a shared signal estimation model, which is then distributed to each base station.
[0111] The interaction process among the local base station, the cooperative base station and the user terminal in the embodiment of the present application can be as follows: Figure 4 shown.
[0112] The embodiments of the present application can make full use of the advantage that the desired propagation detection model assisted by the graph neural network can output soft estimation information, give play to the macro-diversity gain of multi-base station collaboration, and effectively reduce the impact of inter-cell interference; the graph neural network can also be used to overcome the problem of Gaussian approximation inaccuracy in traditional desired propagation detection technology, and significantly improve the receiving performance; compared with the maximum likelihood and spherical decoding receiving technology, the complexity of the embodiments of the present application is a polynomial order of the system dimension, which can significantly reduce the computational overhead; in addition, a single user data signal reception only requires the exchange of soft estimation information between base stations once, thus having the advantage of low interaction overhead.
[0113] In the signal estimation method of the embodiment of the present application, the first base station determines the first uplink channel state information according to the uplink reference signal sent by the user terminal within the detection range; the first base station detects the service data signal sent by the first user terminal to obtain the first observation signal, wherein the first base station is the local base station of the cell where the first user terminal is located; the first base station inputs the first uplink channel state information and the first observation signal into a pre-trained first signal estimation model to obtain the first estimation information of the service data signal output by the first signal estimation model, wherein the first signal estimation model adopts an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model. In this way, the service data sent by the user terminal is estimated by adopting the improved expected propagation detection model, and the structured dependency characteristics between the transmitted data streams are learned by means of the graph neural network, and the posterior probability estimation is optimized to provide more accurate information to the nonlinear estimation module, so as to make up for the problem that the expected propagation detection model has a large error when using Gaussian approximation when performing posterior inference on the transmitted data, thereby improving the model's ability to resist severe multi-user interference in the MIMO system.
[0114] See also Figure 5 , Figure 5 is a flow chart of another signal estimation method provided in an embodiment of the present application. Figure 5 As shown, the following steps are included:
[0115] Step 501: A second base station determines second uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range.
[0116] Step 502: The second base station detects a service data signal sent by the first user terminal to obtain a second observation signal, wherein the second base station is a base station that can detect the signal of the first user terminal and is not in the cell where the first user terminal is located.
[0117] Step 503: The second base station inputs the second uplink channel state information and the second observation signal into a pre-trained second signal estimation model to obtain second estimation information of the service data signal output by the second signal estimation model, wherein the second signal estimation model is an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model.
[0118] Step 504: The second base station sends the second estimation information to the first base station, where the first base station is a local base station of the cell where the first user terminal is located.
[0119] It should be noted that this embodiment is Figure 1 The implementation method of the second base station side corresponding to the embodiment shown in the figure can be found in the above Figure 1 To avoid repetition, the relevant introduction in the illustrated embodiment will not be repeated here.
[0120] In the signal estimation method of the embodiment of the present application, the second base station determines the second uplink channel state information according to the uplink reference signal sent by the user terminal within the detection range; the second base station detects the service data signal sent by the first user terminal to obtain the second observation signal, wherein the second base station is a base station that can detect the signal of the first user terminal and is not in the cell where the first user terminal is located; the second base station inputs the second uplink channel state information and the second observation signal into a pre-trained second signal estimation model to obtain the second estimation information of the service data signal output by the second signal estimation model, wherein the second signal estimation model is an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model; the second base station sends the second estimation information to the first base station, wherein the first base station is the local base station of the cell where the first user terminal is located. In this way, the user service data signal is estimated by utilizing base station collaboration, and the collaborative base station also adopts the improved expected propagation detection model, so that the local base station obtains a more accurate global estimation, thereby effectively combating inter-cell interference and improving receiving performance.
[0121] The embodiment of the present application also provides a signal estimation device, which is arranged in the first base station. Figure 6 , Figure 6 1 is a structural diagram of a signal estimation device provided in an embodiment of the present application. Figure 1 The signal estimation method shown is similar, so the implementation of the signal estimation device can refer to the implementation of the method, and the repeated parts will not be repeated.
[0122] like Figure 6 As shown, the signal estimation device 600 includes:
[0123] A first determining module 601 is used to determine first uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range;
[0124] A first detection module 602 is configured to detect a service data signal sent by a first user terminal to obtain a first observation signal, wherein the first base station is a local base station of a cell where the first user terminal is located;
[0125] The first signal estimation module 603 is used to input the first uplink channel state information and the first observation signal into a pre-trained first signal estimation model to obtain first estimation information of the service data signal output by the first signal estimation model, wherein the first signal estimation model adopts an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model.
[0126] Optionally, the signal estimation device 600 further includes:
[0127] a receiving module, configured to receive second estimation information of the service data signal sent by a second base station, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located;
[0128] The third determination module is used to determine target estimation information for the service data signal according to the first estimation information and the second estimation information.
[0129] Optionally, the first signal estimation model is trained by the first base station using local training data, or the first signal estimation model is distributed to the first base station by a central server and is trained by the central server using training data of each base station.
[0130] Optionally, the third determination module is used to, when the first estimation information and the second estimation information are represented by posterior probabilities, multiply the first estimation information and the second estimation information to obtain the target estimation information;
[0131] Alternatively, the third determination module is used to add the first estimation information and the second estimation information to obtain the target estimation information when the first estimation information and the second estimation information are represented by a posterior log-likelihood ratio.
[0132] Optionally, the training process of the first signal estimation model includes:
[0133] Acquire a training sample set, wherein the training sample set includes uplink channel state samples, uplink data samples sent by the user terminal, and observation signal samples of the uplink data sent by the user terminal at the base station end after passing through the uplink channel;
[0134] Inputting uplink channel state samples and observation signal samples in the training sample set into an initially constructed first signal estimation model to obtain an estimated signal output by the initially constructed first signal estimation model;
[0135] Determining a model loss value based on a difference between the estimated signal and the uplink data sample;
[0136] Based on the model loss value, the structural parameters of the initially constructed first signal estimation model are adjusted to obtain a trained first signal estimation model.
[0137] The signal estimation device 600 provided in the embodiment of the present application can execute Figure 1 The implementation principle and technical effect of the method embodiment shown are similar, and this embodiment will not be repeated here.
[0138] The signal estimation device 600 of the embodiment of the present application determines the first uplink channel state information according to the uplink reference signal sent by the user terminal within the detection range; detects the service data signal sent by the first user terminal to obtain the first observation signal, wherein the first base station is the local base station of the cell where the first user terminal is located; inputs the first uplink channel state information and the first observation signal into a pre-trained first signal estimation model to obtain the first estimation information of the service data signal output by the first signal estimation model, wherein the first signal estimation model adopts an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model. In this way, the service data sent by the user terminal is estimated by adopting the improved expected propagation detection model, and the structured dependency characteristics between the transmitted data streams are learned by means of the graph neural network, and the posterior probability estimation is optimized to provide more accurate information to the nonlinear estimation module, so as to make up for the problem that the expected propagation detection model has a large error when using the Gaussian approximation when performing a posteriori inference on the transmitted data, thereby improving the model's ability to combat severe multi-user interference in the MIMO system.
[0139] The embodiment of the present application also provides another signal estimation device, which is arranged in the second base station. Figure 7 , Figure 7 1 is a structural diagram of a signal estimation device provided in an embodiment of the present application. Figure 5 The signal estimation method shown is similar, so the implementation of the signal estimation device can refer to the implementation of the method, and the repeated parts will not be repeated.
[0140] like Figure 7 As shown, the signal estimation device 700 includes:
[0141] A second determination module 701 is used to determine second uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range;
[0142] A second detection module 702 is configured to detect a service data signal sent by a first user terminal to obtain a second observation signal, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located;
[0143] A second signal estimation module 703 is configured to input the second uplink channel state information and the second observation signal into a pre-trained second signal estimation model to obtain second estimation information of the service data signal output by the second signal estimation model, wherein the second signal estimation model is an improved desired propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved desired propagation detection model;
[0144] The sending module 704 is configured to send the second estimation information to a first base station, where the first base station is a local base station of a cell where the first user terminal is located.
[0145] The signal estimation device 700 provided in the embodiment of the present application can execute Figure 5 The implementation principle and technical effect of the method embodiment shown are similar, and this embodiment will not be repeated here.
[0146] The signal estimation device 700 provided in the embodiment of the present application determines the second uplink channel state information according to the uplink reference signal sent by the user terminal within the detection range; detects the service data signal sent by the first user terminal to obtain the second observation signal, wherein the second base station is a base station that can detect the signal of the first user terminal and is not in the cell where the first user terminal is located; inputs the second uplink channel state information and the second observation signal into a pre-trained second signal estimation model to obtain the second estimation information of the service data signal output by the second signal estimation model, wherein the second signal estimation model is an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model; sends the second estimation information to the first base station, wherein the first base station is the local base station of the cell where the first user terminal is located. In this way, the user service data signal is estimated by utilizing base station collaboration, and the collaborative base station also adopts the improved expected propagation detection model, so that the local base station obtains a more accurate global estimation, thereby being able to effectively combat inter-cell interference and improve receiving performance.
[0147] The embodiment of the present application also provides a base station. Since the principle of solving the problem by the base station is similar to the signal estimation method in the embodiment of the present application, the implementation of the base station can refer to the implementation of the method, and the repeated parts will not be repeated. Figure 8 As shown, the base station in the embodiment of the present application includes:
[0148] In one implementation, the base station is a first base station, and the processor 800 is configured to read a program in the memory 820 and execute the following process:
[0149] Determine first uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range;
[0150] Detecting a service data signal sent by a first user terminal to obtain a first observation signal, wherein the first base station is a local base station of a cell where the first user terminal is located;
[0151] The first uplink channel state information and the first observation signal are input into a pre-trained first signal estimation model to obtain first estimation information of the service data signal output by the first signal estimation model, wherein the first signal estimation model adopts an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model.
[0152] The transceiver 810 is configured to receive and send data under the control of the processor 800 .
[0153] Optionally, the processor 800 is further configured to read a program in the memory 820 and execute the following steps:
[0154] Receiving, by means of the transceiver 810, second estimation information of the service data signal sent by a second base station, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located;
[0155] Target estimation information for the service data signal is determined according to the first estimation information and the second estimation information.
[0156] Optionally, the first signal estimation model is trained by the first base station using local training data, or the first signal estimation model is distributed to the first base station by a central server and is trained by the central server using training data of each base station.
[0157] Optionally, the processor 800 is further configured to read a program in the memory 820 and execute the following steps:
[0158] In a case where the first estimation information and the second estimation information are represented by posterior probabilities, multiplying the first estimation information and the second estimation information to obtain the target estimation information;
[0159] When the first estimation information and the second estimation information are represented by a posterior log-likelihood ratio, the first estimation information and the second estimation information are added to obtain the target estimation information.
[0160] Optionally, the training process of the first signal estimation model includes:
[0161] Acquire a training sample set, wherein the training sample set includes uplink channel state samples, uplink data samples sent by the user terminal, and observation signal samples of the uplink data sent by the user terminal at the base station end after passing through the uplink channel;
[0162] Inputting uplink channel state samples and observation signal samples in the training sample set into an initially constructed first signal estimation model to obtain an estimated signal output by the initially constructed first signal estimation model;
[0163] Determining a model loss value based on a difference between the estimated signal and the uplink data sample;
[0164] Based on the model loss value, the structural parameters of the initially constructed first signal estimation model are adjusted to obtain a trained first signal estimation model.
[0165] In another implementation manner, the base station is a second base station, and the processor 800 is configured to read the program in the memory 820 and execute the following process:
[0166] Determine second uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range;
[0167] Detecting a service data signal sent by a first user terminal to obtain a second observation signal, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located;
[0168] Inputting the second uplink channel state information and the second observation signal into a pre-trained second signal estimation model to obtain second estimation information of the service data signal output by the second signal estimation model, wherein the second signal estimation model is an improved desired propagation detection model, and a graph neural network is inserted between a linear estimation module and a nonlinear estimation module in each network layer of the improved desired propagation detection model;
[0169] The second estimation information is sent to a first base station through the transceiver 810, wherein the first base station is a local base station of the cell where the first user terminal is located.
[0170] The transceiver 810 is configured to receive and send data under the control of the processor 800 .
[0171] Among them, Figure 8 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically linking together various circuits of one or more processors represented by processor 800 and memory represented by memory 820. The bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 810 may be a plurality of components, namely, a transmitter and a transceiver, providing a unit for communicating with various other devices on a transmission medium. The processor 800 is responsible for managing the bus architecture and general processing, and the memory 820 may store data used by the processor 800 when performing operations.
[0172] The base station provided in the embodiment of the present application can execute the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.
[0173] In addition, the computer-readable storage medium of the embodiment of the present application is used to store a computer program, and the computer program can be executed by the processor to implement Figure 1 or Figure 5 The various steps in the method embodiment are shown.
[0174] In the several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0175] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may be physically included separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.
[0176] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform some steps of the sending and receiving methods described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program codes.
[0177] The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles described in the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.
Claims
1. A signal estimation method, characterized in that: include: The first base station determines first uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range; The first base station detects, by the first user terminal, a service data signal sent by the first user terminal to obtain a first observation signal, wherein the first base station is a local base station of a cell where the first user terminal is located; The first base station inputs the first uplink channel state information and the first observation signal into a pre-trained first signal estimation model to obtain first estimation information of the service data signal output by the first signal estimation model, wherein the first signal estimation model adopts an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model.
2. The method according to claim 1, characterized in that The method further comprises: The first base station receives second estimation information of the service data signal sent by a second base station, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located; The first base station determines target estimation information for the service data signal according to the first estimation information and the second estimation information.
3. The method according to claim 2, characterized in that The first signal estimation model is trained by the first base station using local training data, or the first signal estimation model is distributed to the first base station by a central server and is trained by the central server using training data of each base station.
4. The method according to claim 2, characterized in that: The first base station determines target estimation information for the service data signal according to the first estimation information and the second estimation information, including: In a case where the first estimation information and the second estimation information are represented by posterior probabilities, the first base station multiplies the first estimation information and the second estimation information to obtain the target estimation information; In a case where the first estimation information and the second estimation information are represented using a posterior log-likelihood ratio, the first base station adds the first estimation information and the second estimation information to obtain the target estimation information.
5. The method according to claim 1, characterized in that The training process of the first signal estimation model includes: Acquire a training sample set, wherein the training sample set includes uplink channel state samples, uplink data samples sent by the user terminal, and observation signal samples of the uplink data sent by the user terminal at the base station end after passing through the uplink channel; Inputting uplink channel state samples and observation signal samples in the training sample set into an initially constructed first signal estimation model to obtain an estimated signal output by the initially constructed first signal estimation model; Determining a model loss value based on a difference between the estimated signal and the uplink data sample; Based on the model loss value, the structural parameters of the initially constructed first signal estimation model are adjusted to obtain a trained first signal estimation model.
6. A signal estimation method, characterized in that: include: The second base station determines second uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range; The second base station detects, by the first user terminal, a service data signal sent by the first user terminal to obtain a second observation signal, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located; The second base station inputs the second uplink channel state information and the second observation signal into a pre-trained second signal estimation model to obtain second estimation information of the service data signal output by the second signal estimation model, wherein the second signal estimation model is an improved desired propagation detection model, and a graph neural network is inserted between a linear estimation module and a nonlinear estimation module in each network layer of the improved desired propagation detection model; The second base station sends the second estimation information to the first base station, where the first base station is a local base station of a cell where the first user terminal is located.
7. A signal estimation device, arranged at a first base station, characterized in that: include: A first determination module, configured to determine first uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range; A first detection module, configured to detect a service data signal sent by a first user terminal to obtain a first observation signal, wherein the first base station is a local base station of a cell where the first user terminal is located; The first signal estimation module is used to input the first uplink channel state information and the first observation signal into a pre-trained first signal estimation model to obtain first estimation information of the service data signal output by the first signal estimation model, wherein the first signal estimation model adopts an improved expected propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved expected propagation detection model.
8. A signal estimation device, arranged at a second base station, characterized in that: include: A second determination module, configured to determine second uplink channel state information according to an uplink reference signal sent by a user terminal within a detection range; A second detection module, configured to detect a service data signal sent by a first user terminal to obtain a second observation signal, wherein the second base station is a base station that can detect the signal of the first user terminal and is not a base station in a cell where the first user terminal is located; a second signal estimation module, configured to input the second uplink channel state information and the second observation signal into a pre-trained second signal estimation model, and obtain second estimation information of the service data signal output by the second signal estimation model, wherein the second signal estimation model is an improved desired propagation detection model, and a graph neural network is inserted between the linear estimation module and the nonlinear estimation module in each network layer of the improved desired propagation detection model; A sending module is used to send the second estimation information to a first base station, wherein the first base station is a local base station of a cell where the first user terminal is located.
9. A base station, comprising: A transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor; characterized in that the processor is used to read the program in the memory to implement the steps in the signal estimation method as described in any one of claims 1 to 5; or to implement the steps in the signal estimation method as described in claim 6.
10. A computer-readable storage medium for storing a computer program, characterized in that: When the computer program is executed by a processor, the computer program implements the steps in the signal estimation method according to any one of claims 1 to 5; or implements the steps in the signal estimation method according to claim 6.
