Method and Apparatus for Decomposing Multi-path Signals

Abstract

A multipath signal decomposition method is provided that enables the estimation of paths and parameters for each path by allowing the received signal to be represented in the sparsest representation while reducing the signal parameter estimation error. The multipath signal decomposition method comprises: (a) acquiring observation data containing multiple multipath signal components as a received reference signal received by a communication receiver; (b) determining one or more of a plurality of basis vectors corresponding to each of the multipath signal components and adding them to a basis vector dictionary; (c) calculating residuals using basis vectors whose distance from the newly added basis vector to the basis vector dictionary is greater than a predetermined threshold, and partially optimizing by updating basis vectors whose distance is less than the predetermined threshold so that the projection of the residual is maximized; (d) if a predetermined termination condition is not satisfied, repeating steps (b) and (c) to determine additional basis vectors, while determining a minimum number of basis vectors while avoiding duplicate basis vectors. and (e) when a predetermined termination condition is satisfied, the step of determining the coefficients and parameters of multipath signal components corresponding to the basis vectors included in the basis vector dictionary; is included.

Description

Technology Field

[0001] Exemplary embodiments relate to a wireless channel estimation method and apparatus, and more specifically, to a method and apparatus for decomposing each multipath signal component and estimating parameters associated with each path in a channel environment where a plurality of multipath signal components exist. Background Technology

[0002] In wireless communication systems, radio waves travel through space along various paths, arriving with varying time delays, angles, and received power; channels in such environments are referred to as multipath channels. Signals transmitted through multipath channels may overlap, leading to distortion and degraded signal reliability. Modeling of multipath channels and multipath signal component decomposition techniques are essential for the development and performance evaluation of wireless transmission technologies. Furthermore, multipath signal component decomposition techniques can improve signal quality in areas such as equalization, frequency offset estimation, and synchronization by performing channel estimation and signal distortion correction. Moreover, multipath signal component decomposition techniques can serve as a foundational technology for various applications, such as increasing transmission speeds through MIMO spatial multiplexing, or reducing interference between wireless systems and expanding communication coverage through beamforming.

[0003] Beyond communications, the radio wave delay and angle of arrival estimation information provided by multipath parameter estimation technologies can be applied in various fields, such as radar systems and environmental monitoring. Particularly in 6G, technologies such as Integrated Sensing and Communications (ISAC), which combines sensing and communication, and Orthogonal Time Frequency Space (OTFS), which enhances signal transmission performance in time-varying channels, are receiving significant attention. Furthermore, if multipath channels between two links are distinguished within an interference channel between the reference signal of a received signal and the reference signal of an interference signal, multi-signal decomposition technology can also be applied as an interference reduction technique. Distinguishing multipath channels and distinguishing multiple signals ultimately involve separating superimposed signals, and thus share the same underlying technology.

[0004] Multipath channel parameter estimation is the process of formulating the channel response into a sparse representation using parameters such as delay time, angle of arrival, and received strength. Sparse representation techniques have been widely applied in the field of compressed sensing. In compressed sensing theory, the Restricted Isometry Property (RIP) refers to the attribute that, in sparse representation, the radius of error is limited to a smaller size as the orthogonality of the signal's basis vectors increases. An environment where signal parameter estimation is difficult is one where multipath channels are clustered. In a Clustered Delay Line (CDL) channel environment, multiple multipaths converge to form a single cluster, resulting in significantly degraded orthogonality between the multipaths; consequently, sensitivity to noise increases, leading to a very large radius of error.

[0005] Matching Pursuit (MP) is a widely used technique in the field of compressed sensing. MP is a method that finds multiple basis vectors by repeatedly identifying and removing signal components (basis vectors) from the measured signal; various variations of this technique are derived from the process of optimizing the magnitudes of the found basis vectors. While MP is advantageous for real-time implementation, its performance in decomposing clustered multipath components is poor. On the other hand, Space-Alternating Generating Expectation Maximization (SAGE) is frequently used in the field of wireless channel research. SAGE estimates multiple parameters through stochastic optimization; although this technique offers much higher accuracy than MP, it has the disadvantage of requiring a very large amount of computation. The problem to be solved

[0006] Exemplary embodiments provide a multipath signal decomposition method that enables the estimation of paths and parameters for each path by minimizing the number of multipaths within a multipath cluster in a clustered delay line (CDL) channel environment in which multiple multipaths are combined to form a single channel, thereby reducing signal parameter estimation error and enabling the received signal to be represented in the sparsest representation.

[0007] Exemplary embodiments provide a multipath signal decomposition device suitable for implementing a multipath signal decomposition method. means of solving the problem

[0008] According to one aspect of an exemplary embodiment, a multipath signal decomposition method comprises: (a) acquiring observation data containing a plurality of multipath signal components as a reception reference signal received by a communication receiver; (b) determining one or more of a plurality of basis vectors corresponding to each of the multipath signal components and adding them to a basis vector dictionary; (c) calculating residuals using basis vectors whose distance from the newly added basis vector to the basis vector dictionary is greater than a predetermined threshold, and partially optimizing by updating basis vectors whose distance is less than the predetermined threshold so as to maximize the projection of the residuals; (d) if a predetermined termination condition is not satisfied, repeating steps (b) and (c) to determine additional basis vectors, while avoiding duplicate basis vectors, and determining a minimum number of basis vectors; and (e) if a predetermined termination condition is satisfied, determining the coefficients of multipath signal components corresponding to the basis vectors included in the basis vector dictionary and their parameters.

[0009] The above step (c) may include: calculating residuals using basis vectors whose distance from the new basis vector is greater than a predetermined threshold, and updating basis vectors whose distance is less than the predetermined threshold so as to maximize the projection of the residuals; determining a first subset to which basis vectors whose distance is greater than the threshold belong among the basis vectors included in the basis vector dictionary, and a second subset to which basis vectors whose distance is less than the threshold belong; calculating residuals between the observed data and combinations of basis vectors belonging to the first subset; and determining basis vectors to which the projection of the residuals is maximized for basis vectors belonging to the second subset.

[0010] The above step (c) may further include the step of updating all basis vectors to new optimal values ​​if the new basis vector is correlated with the existing basis vector.

[0011] The above termination condition may include a predetermined objective function value falling below a certain allowable level.

[0012] The above termination condition may include the number of the basis vectors reaching a predetermined maximum value.

[0013] The above termination condition may include the amount of computation reaching a predetermined upper limit of computation.

[0014] According to another aspect of an exemplary embodiment, a multipath signal decomposition device comprises: a memory for storing program instructions; and a processor connected to the memory and executing the program instructions stored in the memory. When the program instructions are executed by the processor, the processor causes: (a) to acquire observation data containing a plurality of multipath signal components as a reception reference signal received by a communication receiver; (b) to determine one or more of a plurality of basis vectors corresponding to each of the multipath signal components and add them to a basis vector dictionary; (c) to calculate residuals using basis vectors whose distance from the newly added basis vector to the basis vector dictionary is greater than a predetermined threshold, and to partially optimize by updating basis vectors whose distance is less than the predetermined threshold so that the projection of the residual is maximized; (d) if a predetermined termination condition is not satisfied, to repeat the operation (b) and the operation (c) to determine additional basis vectors, while determining a minimum number of basis vectors while avoiding duplicate basis vectors. and (e) when a predetermined termination condition is satisfied, the operation of determining the coefficients and parameters of multipath signal components corresponding to the basis vectors included in the basis vector dictionary is performed.

[0015] The instructions for performing the above operation (c) may cause the processor to perform the following operations: calculating residuals using basis vectors whose distance from the new basis vector is greater than a predetermined threshold, and updating basis vectors whose distance is less than the predetermined threshold so as to maximize the projection of the residuals; determining a first subset to which basis vectors whose distance is greater than the threshold belong among the basis vectors included in the basis vector dictionary, and a second subset to which basis vectors whose distance is less than the threshold belong; calculating residuals between the observed data and the combination of basis vectors belonging to the first subset; and determining basis vectors to which the projection of the residuals is maximized for the basis vectors belonging to the second subset.

[0016] The instructions for performing the above (c) operation may additionally cause the processor to perform the operation of updating all basis vectors to new optimal values ​​if the new basis vector is correlated with the existing basis vector. Effects of the invention

[0017] According to an exemplary embodiment, superposition signals can be decomposed based on projection maximization. The method according to the exemplary embodiment utilizes a stochastic optimization technique similar to SAGE, but by improving the optimization method, it can provide more accurate channel parameter estimation results with less computational load than SAGE. In other words, the method according to the exemplary embodiment demonstrates faster and higher parameter estimation performance than SAGE.

[0018] The slow convergence of SAGE stems from the inaccuracy of basis vector estimation. In SAGE, when finding a new basis vector, all previously found basis vectors are assumed to be true, and the multipath component most similar to the residual—the remainder obtained by subtracting the reconstructed signal from the measured signal—is searched. Due to this structure, the inaccuracy of previously found basis vector estimations negatively affects the performance of subsequent basis vector searches. This estimation error accumulates as the number of basis vectors increases, leading to continuous incorrect estimations. An exemplary embodiment of the present invention ensures fast convergence of the algorithm by reducing this accumulated error.

[0019] The method according to the exemplary embodiment can be applied to various uses, such as evaluation of channel environments and communication quality of communication systems, as well as equalization, frequency offset estimation, synchronization, MIMO spatial multiplexing, beamforming, sensing, object detection, and interference mitigation. Brief explanation of the drawing

[0020] FIG. 1 is a block diagram of a communication system according to an exemplary embodiment. Figure 2 is a block diagram showing the physical configuration of a multipath signal decomposition device. FIG. 3 is a flowchart showing a multipath signal decomposition method according to an exemplary embodiment. Figure 4 shows pseudocode for a projection maximization (ProMax) algorithm according to an exemplary embodiment. Figure 5 shows the parameter estimation process by the multipath signal decomposition method according to an exemplary embodiment in comparison with the conventional estimation process according to the conventional SAGE algorithm. Figure 6 shows a comparison of simulation results of the method according to an exemplary embodiment and conventional methods. Figure 7 shows the results of a performance comparison between a method according to an exemplary embodiment and conventional methods. Specific details for implementing the invention

[0021] The present invention is capable of various modifications and may have various embodiments, and specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and it should be understood that it includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the invention.

[0022] While ordinal numbers such as first, second, etc. may be used to describe various components, said components shall not be limited by said terms. The said terms are used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component. The term "and / or" includes a combination of a plurality of related described items or any of a plurality of related described items.

[0023] In the embodiments of the present application, "at least one of A and B" may mean "at least one of A or B" or "at least one of one or more combinations of A and B". Additionally, in the embodiments of the present application, "at least one of A and B" may mean "at least one of A or B" or "at least one of one or more combinations of A and B".

[0024] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. On the other hand, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.

[0025] The terms used in this application are used merely to describe specific embodiments and are not intended to limit the invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, terms such as "comprising" or "having" are intended to specify the presence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0026] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0027] Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the accompanying drawings. In describing the present invention, to facilitate overall understanding, the same reference numerals are used for identical components in the drawings, and redundant descriptions of identical components are omitted.

[0028] FIG. 1 is a block diagram of a communication system according to an exemplary embodiment. The communication system may include a transmitter (100) that transmits a signal within a communication network and a receiver (200) that receives a signal. Additionally, the communication system may include a multipath signal decomposition device (300) that decomposes a multipath signal from a signal received by the receiver (200).

[0029] A transmitter (100) transmits a signal that has been coded and modulated according to a predetermined wireless communication protocol. A receiver (200) can demodulate and decode the signal received from the transmitter (100) to restore the original data or original signal prior to the transmitter (100) coding. The signal received by the receiver (200) may be a superposition of signal components that have arrived along various paths with various time delays, origin angles, arrival angles, Doppler frequencies, dual polarizations, and received power. A multipath signal decomposition device (300) can receive and store the signal received by the receiver (200), and can decompose the real-time signal or the stored signal into multiple multipath signal components.

[0030] In one embodiment, the transmitter (100) may be a base station and the receiver (200) may be a terminal. However, the transmitter (100) may be a terminal and the receiver (200) may be a base station. Meanwhile, in one embodiment, the multipath signal decomposition device (300) may be provided separately from the receiver (200). However, in a modified embodiment, the multipath signal decomposition device (300) may be implemented as one unit with the receiver (200). For example, the multipath signal decomposition device (300) may be implemented within the receiver (200).

[0031] In this specification, a base station may refer to an access point, a radio access station, a node B, an evolved node B, a base transceiver station, a mobile multihop relay (MMR)-BS, etc., and may include all or part of the functions of a base station, an access point, a radio access station, a node B, an eNodeB, a base transceiver station, an MMR-BS, etc.

[0032] In this specification, the term "terminal" may refer to a mobile station, mobile terminal, subscriber station, portable subscriber station, user equipment, access terminal, etc., and may include all or part of the functions of a terminal, mobile station, mobile terminal, subscriber station, portable subscriber station, user equipment, access terminal, etc.

[0033] Here, a desktop computer, laptop computer, tablet PC, wireless phone, mobile phone, smartphone, smart watch, smart glass, e-book reader, PMP (portable multimedia player), portable game console, navigation device, digital camera, DMB (digital multimedia broadcasting) player, digital audio recorder, digital audio player, digital picture recorder, digital picture player, digital video recorder, digital video player, etc., capable of communicating with a terminal can be used.

[0034] The above base station and terminal may support 4G communication (e.g., LTE (long term evolution), LTE-A (advanced)), 5G communication (e.g., NR (new radio)), etc., as defined in the 3GPP (3rd generation partnership project) standard. The communication system to which the embodiments according to the present invention are applied is not limited thereto, and the present invention may be applied to various communication systems. Here, "communication system" may be used with the same meaning as "communication network," "LTE" may refer to a "4G communication system," an "LTE communication system," or an "LTE-A communication system," and "NR" may refer to a "5G communication system" or an "NR communication system."

[0035] For example, for 4G communication and 5G communication, base stations and terminals can support communication protocols based on CDMA (code division multiple access), WCDMA (wideband CDMA), TDMA (time division multiple access), FDMA (frequency division multiple access), OFDM (orthogonal frequency division multiplexing), Filtered OFDM, CP (cyclic prefix)-OFDM, DFT-s-OFDM (discrete Fourier transform-spread-OFDM), OFDMA (orthogonal frequency division multiple access), SC (single carrier)-FDMA, NOMA (Non-orthogonal Multiple Access), GFDM (generalized frequency division multiplexing), FBMC (filter bank multi-carrier) based communication protocol, UFMC (universal filtered multi-carrier) based communication protocol, SDMA (Space Division Multiple Access) based communication protocol, etc.

[0036] Additionally, the communication system illustrated in FIG. 1 may further include a core network. If the communication system supports 4G communication, the core network may include an S-GW (serving-gateway), a P-GW (PDN (packet data network)-gateway), an MME (mobility management entity), etc. If the communication system supports 5G communication, the core network may include a UPF (user plane function), an SMF (session management function), an AMF (access and mobility management function), etc.

[0037] FIG. 2 is a block diagram showing the physical configuration of a multipath signal decomposition device (300). The multipath signal decomposition device (300) may include a processor (400), memory (402), a storage device (404), and a data transmission and reception unit (406). Additionally, the multipath signal decomposition device (300) may further include an input interface device (410) and an output interface device (412). Each component included in the multipath signal decomposition device (300) may be connected by a bus and communicate with each other.

[0038] The processor (400) can execute program instructions stored in memory (402) and / or storage device (404). The processor (400) may be implemented by at least one central processing unit (CPU) and / or general-purpose processing unit (GPU), and may be any other processing device capable of performing the method according to the present invention. The processor (400) can execute program instructions for implementing the multipath signal decomposition method according to the present invention.

[0039] The memory (402) may include volatile memory such as RAM (Random Access Memory) and non-volatile memory such as ROM (Read Only Memory). The memory (402) may load program instructions stored in the storage device (404) and provide them to the processor (400), thereby enabling the processor (400) to execute them.

[0040] The storage device (404) is a recording medium suitable for storing program instructions and data, and may include, for example, magnetic media such as a hard disk, floppy disk, and magnetic tape; optical recording media such as a CD-ROM (Compact Disk Read Only Memory) and DVD (Digital Video Disk); magneto-optical media such as a floptical disk; flash memory or EPROM (Erasable Programmable ROM); or semiconductor memory such as an SSD manufactured based thereon. The storage device (404) may store program instructions for implementing the multipath signal decomposition method according to the present invention.

[0041] Before describing the multipath signal decomposition method performed by the multipath signal decomposition device (300), the theoretical basis of the Projection Maximization (ProMax) algorithm, which forms the basis of the multipath signal decomposition method, is explained.

[0042] If we consider a time-invariant channel with two multipaths, the received signal can be expressed by Equation 1.

[0043]

[0044] Here, ' 'class ' ' represents the reference signal transmitted by the transmitting device and the received signal received by the receiving device, respectively. 'class ' ' represents the complex amplitude and delay of the n-th multipath signal component, respectively. When the reference signal is a Dirac impulse, the multipath signals are orthogonal to each other with respect to different delays. However, due to limited sampling, the observed multipath signals become correlated with one another. For a rectangular signal with a finite bandwidth, the multipath signals take the form of a sinc function determined by the convolution with the Dirac impulse. Generally, if the spacing between multipath components (MPCs) is small, there is a high probability of correlation between the multipath signals, and the correlation increases as the bandwidth decreases. The received signal is represented as a superposition of correlated multipath signals. Correlation between multipath signals occurs due to limitations such as bandwidth, measurement time, and antenna size.

[0045] A sampling kernel can be expressed as a function of parameters such as signal label, time, space, and Doppler, depending on the propagation environment and sampling method. A sampled signal having N signal components can be expressed by Equation 2.

[0046]

[0047] Here, ' ', ' ', ' ' represents the measurement data, the sampling kernel, and the observation error due to signal noise, respectively. ' represents the input parameter vector of the sampling kernel for the n-th multipath signal component (hereinafter, 'n-th signal component'). ' represents the n-th signal component and can also be referred to as the n-th basis vector. For example, assuming stationary uncorrelated scattering in a broad sense, the sampling kernel in channel estimation of a cellular system using a time-variable multiple-input multiple-output (MIMO) channel model can be expressed by Equation 3.

[0048]

[0049] Here, ' 'class ' ' represents the physical cell identifier (PCI) or base station index and the n-th reference signal determined by the PCI, respectively. 'and ' ' represents a steering vector determined by the Angle of Departure (AoD) and Angle of Arrival (AoA) of the multipath, and represents the Kronecker product. ', ' ', ' ', ' ' are the departure angle, arrival angle, delay time, and Doppler of the n-th multipath, respectively. 'and ' ' is a vector of measurement time and frequency.

[0050] Mathematical equations 2 and 3 can be expressed as the determinant of mathematical equation 4.

[0051]

[0052] A multipath signal decomposition method according to an exemplary embodiment uses observation data ' From ' multipath signal components, i.e., the parameter matrix of multipath signal components ( ) and size matrix( ...is obtained. In particular, the Projection Maximization (ProMax) algorithm, which forms the basis of the multipath signal decomposition method according to the exemplary embodiment, focuses on finding a set of MPCs consisting of the minimum number of multipath components (MPCs) while avoiding redundant basis vectors. The algorithm according to the exemplary embodiment has a double loop structure for probabilistic search. If a newly discovered MPC is correlated with previously discovered MPCs, all MPCs are updated with the new optimal values. The algorithm according to the exemplary embodiment increments N in the outer loop and performs probabilistic optimization in the inner loop. According to the exemplary embodiment, the essential basis vectors for the sparsest representation are searched in this manner.

[0053] The problem of the most sparsity representation can be expressed by the following mathematical formula 5.

[0054]

[0055] The Most Spacious Representation problem is a non-deterministic polynomial-time hard problem and is considered a combinatorial optimization problem. Since the number of parameter combinations increases exponentially as the number of parameters increases, exploring all combinations is inefficient. For this reason, greedy algorithms or stochastic optimization techniques are commonly used. Greedy algorithms sacrifice accuracy for fast implementation, while stochastic optimization methods aim for accurate parameter estimation but are difficult to implement in real-time. The method of the present invention can be considered a type of stochastic optimization method for finding the Most Spacious Representation.

[0056] According to an exemplary embodiment, to find the sparsest representation method, the optimization problem is reconfigured from Equation 5 into Equation 6 as follows.

[0057]

[0058] According to an exemplary embodiment, in the optimization problem reconfigured as described above, the algorithm finds the minimum N by incrementing N one by one until a certain termination condition is met. One example of a termination condition is when the objective function, such as residuals, falls below a pre-set tolerance level. Another example of a termination condition is when the value of N reaches a pre-limited maximum value. Yet another example of a termination condition is when the amount of computation reaches a pre-defined upper limit for total computation. In addition, it is possible to implement termination conditions in various other ways.

[0059] Meanwhile, one of the other features of the algorithm according to the exemplary embodiment is the suboptimization method in each iteration. This suboptimization is based on assumptions about optimal coefficients when searching for new basis vectors and consideration of MPCs found in the previous search. Given basis vectors, the complex amplitude of basis vectors that minimizes the residual can be calculated using the least squares method as shown in Equation 7.

[0060]

[0061] If we substitute Equation 7 into Equation 6, the objective function can be expressed as Equation 8.

[0062]

[0063] Here, Z is expressed by mathematical formula 9.

[0064]

[0065] According to the definition of, mathematical equation 10 holds.

[0066]

[0067] Using Cholesky decomposition, a positive definite Hermite matrix can be decomposed into the product of a lower triangular matrix and its conjugate transpose, as shown in Equation 11.

[0068]

[0069] Here, L is a lower triangular matrix. Using Equations 10 and 11, Equation 8 can be simplified into Equation 12.

[0070]

[0071] Here It is expressed by mathematical formula 13.

[0072]

[0073] Equation 12 shows that the residual is the observation data ( Projection of observed data from ) to basis vectors ( It shows that it is equivalent to subtracting ). Minimizing the residuals is the same problem as maximizing the projection. Therefore, the new objective function can be expressed by Equation 14.

[0074]

[0075] In SAGE, the magnitude of the residual in the objective function of the probabilistic search process ( To minimize ) the magnitude of the basis vectors ( ) is used as the input value of the objective function. On the other hand, in the method according to the exemplary embodiment, the objective function for maximizing projection in the probabilistic search process ( The magnitude of the basis vector as the input value of ) ...is not used. Since SAGE performs stochastic optimization using inaccurate basis vector magnitude information, the convergence and accuracy of the algorithm are significantly reduced. In contrast, according to an exemplary embodiment, the convergence and accuracy are superior to SAGE because the basis vector magnitude estimation results are not used in the stochastic optimization process and the optimal value is always assumed.

[0076] According to an exemplary embodiment, the amount of computation required for a probabilistic search to maximize projection increases significantly as the number of basis vectors increases. Therefore, computational efficiency may decrease when the number of basis vectors is large. For this reason, computational efficiency can be enhanced by limiting the basis vectors used to maximize projection to a subset of the entire set of basis vectors. The performance of basis vector magnitude estimation depends on the orthogonality between the basis vectors. Therefore, by creating a subset of basis vectors to be optimized by excluding basis vectors that are highly orthogonal to the most recently added basis vector during the probabilistic optimization process, a significant amount of computation can be reduced with almost no performance degradation compared to performing the algorithm using the entire set of basis vectors. In this case, if the most recently added basis vector has low correlation with other basis vectors, the subset selected during the probabilistic optimization process may be an empty set, and if all basis vectors have high correlation with each other, the subset may become the entire set. In addition, it is desirable to perform residual calculation after removing basis vector components that do not belong to the aforementioned subset used in the stochastic optimization process from the measurement signal.

[0077] FIG. 3 is a flowchart showing a multipath signal decomposition method according to an exemplary embodiment.

[0078] First, observation data containing multipath signal components is obtained from a signal received by a receiver (200) (step 500). The signal from which the observation data is extracted may be a reception reference signal received by the receiver (100), but the present invention is not limited thereto. Meanwhile, the observation data may be the reception reference signal itself.

[0079] Next, a new basis vector is searched for and added to the basis vector dictionary (Step 510).

[0080] In step 520, the indices for the basis vectors included in the basis vector dictionary are divided into two subsets N1 and N2. Here, N1 is the set of indices for basis vectors for which the distance d(n) from the basis vector added in step 510 is greater than a predetermined threshold. N2 is the set of indices for basis vectors for which the distance d(n) from the added basis vector is less than the threshold.

[0081] Next, residuals can be calculated using N1 (Step 530). Then, an index N is selected from N2, and the N-th basis vector is updated to maximize the projection onto the basis vector (Step 540). Here, projection refers to the projection of the observed data onto the basis vector in the first iteration, and may refer to the projection of the residual onto the basis vector in subsequent iterations. The suboptimization of Steps 510 through 540 is repeated until the termination condition, i.e., the convergence condition, is satisfied in Step 550, and the addition and updating of basis vectors can be performed iteratively.

[0082] If it is determined in step 550 that the termination condition is satisfied, the coefficients for the basis vectors are estimated (step 560). One example of a termination condition may be that the value of the objective function falls below a certain tolerance level. Another example of a termination condition may be that the value of index N reaches a predetermined maximum value. Yet another example of a termination condition may be that the amount of computation reaches a predetermined upper limit for the total amount of computation.

[0083] As described above, the multipath signal decomposition method according to the exemplary embodiment can find the most sparse solution by finding the solution to the objective function while starting the index N from 1 and increasing it by 1 until the residual falls below an acceptable level. As previously mentioned, the algorithm according to the exemplary embodiment has a double loop structure for probabilistic search. If a newly discovered MPC is correlated with previously discovered MPCs, all MPCs are updated with the new optimal values. The algorithm according to the exemplary embodiment incrementally increases the index N in the outer loop and performs probabilistic optimization in the inner loop.

[0084] Accordingly, assuming that the current index N is the maximum value of N, a new basis vector is searched and a partial optimization process is performed. Then, after confirming that it is possible to derive the optimal coefficients for the current index N, the index N is increased and the process is repeated. The multipath signal decomposition method according to the exemplary embodiment is computationally intensive because it involves matrix inverse operations during the partial optimization process. Fortunately, for matrix inverse operations, definite Hermitian matrices It includes, which can be efficiently implemented using Cholesky decomposition. The impact of partial optimization on the computation increases significantly as the index N increases. When N is small, the effect of partial optimization on the computation is minimized. Due to these characteristics, the multipath signal decomposition method according to the exemplary embodiment is suitable for achieving the sparseest representation with a minimum N value.

[0085] FIG. 4 shows pseudocode for the Projection Maximization (ProMax) algorithm according to an exemplary embodiment. As can be seen in the pseudocode, the coefficient matrix ( The estimation is performed only once at the end. Therefore, the performance of partial optimization within the iterative loop is not degraded by errors in coefficient estimation from the previous step. This feature is valuable even though each sub-optimization step involves a significant amount of computation, as it significantly accelerates the convergence of the algorithm.

[0086] FIG. 5 illustrates the parameter estimation process by the multipath signal decomposition method according to an exemplary embodiment in comparison with the conventional estimation process according to the conventional SAGE algorithm. The observed data is (3,1), and the basis vectors that can result from the parameter estimation result are (1,0), ( , ), ( , It is assumed that it is limited to three types of ).

[0087] In the parameter estimation process according to the exemplary embodiment, (1,0), which is closest to the observed data, is first selected as the first basis vector. Then, as the second basis vector, ( , If you find ), the residual( Since the calculation result becomes a zero matrix, the algorithm terminates immediately ((a) state --> (b) operation --> (c) operation).

[0088] In contrast, in the case of SAGE, the algorithm does not terminate due to basis vector magnitude estimation errors. If no limit is placed on the number of basis vectors, unnecessary basis vectors are additionally found, and the range of parameter estimation errors increases due to basis vectors with low orthogonality. For example, after finding the second basis vector, the algorithm can calculate residuals orthogonal only to the second basis vector (Operation (d)). Subsequently, the algorithm may perform a greedy search using redundant basis vectors (Operation (f)) (State (a) --> Operation (b) --> Operation (d) --> Operation (f)). In this case, the parameter estimation results can be sensitive to noise and unstable. In other words, while the algorithm provides fast convergence, stability is low due to the singular basis matrix.

[0089] If the number of basis vectors is appropriately restricted, after calculating residuals orthogonal only to the second basis vector via operation (d), the algorithm can perform statistical optimization to calculate residuals orthogonal only to the first basis vector, for example (operation (e)). Then, through numerous iterations, the final parameter estimation result can be derived (operation (c)). ((a) state --> (b) operation --> (d) operation --> (e) operation --> (c) operation). In other words, even if the number of basis vectors is appropriately restricted, the algorithm converges slowly due to basis vector magnitude estimation errors, resulting in a large amount of computation.

[0090] A simulation was performed to compare the performance of the method according to the exemplary embodiment with that of conventional methods. In the simulation, two multipaths were generated, and channel parameter estimation algorithms were applied while varying the delay intervals of the two multipaths. As performance indicators, the absolute error (AE) of the estimation result of the delay difference of the multipaths, the mean squared error (MSE) of the reconstructed signal, and the empirical cumulative distribution function (eCDF) of the computational amount (computation time and number of algorithm iterations) were examined.

[0091] Figure 6 shows a comparison of simulation results between the method according to the exemplary embodiment and conventional methods. Conventional methods used included the Matched Estimation (MP) technique, which is considered the most basic method, the Orthogonal MP (OMP) technique, which is one of the derivative techniques of MP, and SAGE. MP-based techniques had very low computational loads due to the absence of algorithm iteration, but their parameter estimation performance was low. It was confirmed that the method according to the exemplary embodiment demonstrated the highest parameter estimation performance while having a lower computational load than SAGE.

[0092] FIG. 7 shows the results of a performance comparison between a method according to an exemplary embodiment and conventional methods. FIG. 7(a) shows a 28 GHz channel measured on a train traveling at a speed of 116 km / h in a tunnel, with an SNR of approximately 14 dB. A sequence with a maximum length of 4095 was used, and the measurement bandwidth was 500 MHz. The number of frames was 305, and the frame interval was 32.768 μs. Among the 4095 × 305 samples, 585 × M samples (where i represents an integer from (-M-1) / 2 to (M-1) / 2) consisting of the {4, 11, 18, …, 4092} frequency samples of the (153 + 4i)th frame were used for MPC estimation. FIG. 7(b) to (f) show the results of MPC estimation. The solid and dotted lines represent the Mean Squared Error (MSE) between the total measurements and the reconstructed signal using the observational residuals and estimated MPCs, respectively, and were normalized by the received signal power. The results for OMP, SAGE, and ProMax are shown in green, red, and blue, respectively. While the residuals of all algorithms decrease as the number of MPCs increases, the MSE may increase due to overfitting. Initially, as the number of MPCs increased, both SAGE and ProMax showed faster decreases in residuals and MSE compared to OMP. However, SAGE exhibited unstable MSE due to overfitting, whereas ProMax consistently demonstrated stable performance. For the algorithm implementation, a thorough search was performed to maximize the projection onto the residuals using measurement resolution only when initializing the MPCs, followed by partial optimization using coordinate descent and line search algorithms, where the step size update factor was set to 0.5. The maximum number of iterations was set to 300 for SAGE and 20 for ProMax. These values ​​are the point where the RMSE is approximately 0.07 / B.

[0093] As mentioned above, the operation according to the embodiment of the present invention can be implemented as a computer-readable program or code on a computer-readable recording medium. A computer-readable recording medium includes all types of recording devices in which data that can be read by a computer system is stored. Additionally, the computer-readable recording medium may be distributed across networked computer systems, allowing computer-readable programs or code to be stored and executed in a distributed manner.

[0094] The above computer-readable recording medium may include hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Program instructions may include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

[0095] Some aspects of the invention have been described in the context of a device, but may also be described according to a corresponding method, wherein a block or device corresponds to a method step or a feature of a method step. Similarly, aspects described in the context of a method may also be described according to a corresponding block or item or a feature of a corresponding device. Some or all of the method steps may be performed by (or using) a hardware device, such as, for example, a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, one or more of the most important method steps may be performed by such a device.

[0096] In the embodiments, a programmable logic device (e.g., a field-programmable gate array) may be used to perform some or all of the functions of the methods described herein. In the embodiments, the field-programmable gate array may operate with a microprocessor to perform one of the methods described herein. Generally, it is preferable that the methods be performed by some hardware device.

[0097] Although the present invention has been described above with reference to preferred embodiments, those skilled in the art will understand that various modifications and changes can be made to the invention without departing from the spirit and scope of the invention as described in the following claims.

Claims

Claim 1 (a) acquiring observation data containing multiple multipath signal components from a signal received by a communication receiver; (b) for each of the multipath signal components, determining one or more basis vectors that maximize projection and adding them to a basis vector dictionary; (c) calculating residuals using basis vectors whose orthogonality with the newly added basis vectors in the basis vector dictionary is greater than a predetermined reference level, and partially optimizing the basis vector dictionary by updating basis vectors among which the orthogonality with the new basis vectors is less than the reference level that maximize the projection of the residuals; (d) if a predetermined termination condition is not satisfied, repeating steps (b) and (c) to determine additional basis vectors, while avoiding duplicate basis vectors, determining a minimum number of basis vectors capable of representing the corresponding multipath signal components; and (e) when the above termination condition is satisfied, determining the coefficients of the minimum number of basis vectors and path parameters for the corresponding multipath signal component; a multipath signal decomposition method comprising. Claim 2 A multipath signal decomposition method according to claim 1, wherein in step (a), the observation data is obtained from a reception reference signal received by the communication receiver. Claim 3 A multipath signal decomposition method according to claim 1, wherein in step (b), the projection is an orthogonal projection of the observed data during the first execution, and an orthogonal projection of the residual during the second and subsequent executions. Claim 4 A multipath signal decomposition method according to claim 1, wherein step (c) comprises: determining a first subset of basis vectors among the basis vectors included in the basis vector dictionary in which the distance from the new basis vector is greater than a predetermined threshold, and a second subset of basis vectors among which the distance is less than the threshold; calculating the residual using the observation data and the basis vectors belonging to the first subset; and determining a basis vector among the basis vectors belonging to the second subset that maximizes the basis vector in which the projection of the residual is maximized. Claim 5 A multipath signal decomposition method according to claim 3, further comprising the step of updating all basis vectors to new optimal values ​​if the new basis vector is correlated with the existing basis vector. Claim 6 A multipath signal decomposition method according to claim 1, wherein the termination condition comprises a predetermined objective function value falling below a certain allowable level. Claim 7 A multipath signal decomposition method according to claim 1, wherein the termination condition comprises the number of basis vectors reaching a predetermined maximum value. Claim 8 A multipath signal decomposition method according to claim 1, wherein the termination condition comprises the amount of computation reaching a predetermined upper limit of computation. Claim 9 Memory for storing program instructions; A processor connected to the memory and executing the program instructions stored in the memory; wherein, when the program instructions are executed by the processor, the processor: (a) acquires observation data containing a plurality of multipath signal components from a signal received by a communication receiver; (b) for each of the multipath signal components, determines one or more basis vectors that maximize projection and adds them to a basis vector dictionary; (c) calculates residuals using basis vectors whose orthogonality with the new basis vectors newly added to the basis vector dictionary is greater than a predetermined reference level, and partially optimizes the basis vector dictionary by updating basis vectors among which the orthogonality with the new basis vectors is less than the reference level that maximize the projection of the residuals; (d) if a predetermined termination condition is not satisfied, repeats the operation (b) and the operation (c) to determine additional basis vectors, while avoiding duplicate basis vectors, and determines a minimum number of basis vectors capable of representing the corresponding multipath signal components. and (e) when the above termination condition is satisfied, the multipath signal decomposition device performs the operation of determining the coefficients of the minimum number of basis vectors and path parameters for the corresponding multipath signal component. Claim 10 A multipath signal decomposition device according to claim 9, wherein, when performing the operation (a), the observation data is obtained from a reception reference signal received by the communication receiver. Claim 11 A multipath signal decomposition device according to claim 9, wherein, when performing the operation (b), the projection is an orthogonal projection of the observed data at the first execution and an orthogonal projection of the residual at the second and subsequent executions. Claim 12 A multipath signal decomposition apparatus according to claim 9, wherein the instructions for performing the operation (c) cause the processor to perform the operation of: determining a first subset of basis vectors among the basis vectors included in the basis vector dictionary in which the distance from the new basis vector is greater than a predetermined threshold, and a second subset of basis vectors in which the distance is less than the threshold; calculating the residual using the observation data and the basis vectors belonging to the first subset; and determining a basis vector among the basis vectors belonging to the second subset in which the projection of the residual is maximized. Claim 13 A multipath signal decomposition device according to claim 12, wherein the instructions for performing the operation (c) additionally cause the processor to perform the operation of updating all basis vectors to new optimal values ​​if the new basis vector is correlated with the existing basis vector. Claim 14 A multipath signal decomposition device according to claim 9, wherein the termination condition comprises a predetermined objective function value falling below a certain allowable level. Claim 15 A multipath signal decomposition device according to claim 9, wherein the termination condition comprises the number of basis vectors reaching a predetermined maximum value. Claim 16 A multipath signal decomposition device according to claim 9, wherein the termination condition includes the amount of computation reaching a predetermined upper limit of computation.

Images