TR202016333A1 - METHOD AND APPARATUS BASED ON PHASE CORRECTION IN PILOT PATTERN DIVISION FOR CHANNEL ESTIMATION IN MIMO-OFDM SYSTEMS - Google Patents

Abstract

The present invention relates to a channel estimation method 100 and apparatus for MIMO-OFDM communication systems that have better performance compared to conventional receivers. The method (100) of the invention is based on compensating for phase changes in the source elements in a CDM group during the pilot pattern separation step. The received signal is OFDM demodulated to obtain the received symbols in the source grid and a pilot pilot pattern decomposition is performed on the received symbols. The phase change ratios in the weld grid are determined, the phase correction terms are calculated and applied. The pilot pattern decomposition is re-performed using the phase corrected received signals, and the channel estimates at the pilot positions are updated using the phase change correction terms. The estimated values at the pilot positions are interpolated to obtain the channel estimates for all source elements.

Description

DESCRIPTION METHOD AND APPARATUS BASED ON PHASE CORRECTION IN PILOT DRUNK SEPARATION FOR CHANNEL ESTIMATION IN MIMO-OFDM SYSTEMS Technical Field of the Invention The invention is related to the field of communication and especially to channel estimation methods in multi-input multi-output (MIMO) Orthogonal Frequency Division Multiplexing (OFDM) communication systems. State of the Art In modern communication systems, with the latest developments, achieving high data rates, low latency, high reliability and connectivity have become the main performance goals. Meeting these requirements across a wide range of use cases and deployment scenarios is a difficult task, making the accuracy of channel information at the receiver even more critical. Therefore, channel estimation is one of the most important blocks of a communication system, and this process can be achieved based on previously defined reference signals, also known as pilots. Global standardization organizations such as 3GPP define modern communication standards such as Long Term Evaluation (LTE), Long Term Evaluation (Advanced-LTE-A) and more recently 5G New Radio (NR). carries out the necessary work for the The 5G waveform is based on Cyclic Prefix OFDM (CP-OFDM) (as in LTE and LTE-A) for both 6 GHz and millimeter wave frequencies and supports single- and multi-user MIMO to achieve high data speeds . Both LTE and NR define reference signals for various physical channels to obtain channel estimation at the receiver. Reference signals defined for phase-compatible demodulation of control and user data in both upstream and downstream connections are also known as demodulation reference signals (DMRS). To realize data transfer at multiple layers through a MIMO system, during DMRS design It is necessary to take into account many factors at the same time. Some of the factors taken into consideration in 5G DMRS design include pilot density, power variation in frequency, number of layers supporting orthogonal pilot symbols, configuration flexibility, location of pilots supporting low-latency demodulation, and ensuring the use of a common receiver structure for different configurations. Both 5G and LTE use code division multiplexing (CDM) for orthogonal transmission of pilot signals in different MIMO layers. Orthogonal cover codes (OCC) or cyclic shift (CS) operations are used to support pilot transmission across multiple layers on the same resources and achieve orthogonality between pilots. For example, in 5G, each DM RS configuration (except single-layer transmission) includes an OCC-based pilot allocation in the code space. By combining CDM and frequency domain multiplexing (FDM), it is possible to define up to 12 orthogonal layers. The use of CDM-based pilots has an inherent advantage over other orthogonal methods such as time or frequency domain multiplexing due to processing gain. Channel State Information Reference Signal (CSI-RS) and Sounding Reference Signal (SRS) defined in NR utilize the CDM structure to extend single port operation to multiple ports. Therefore, CDM-based pilot placement is widely used in state-of-the-art communication systems. While CDM-based designs have certain advantages and attractive features, this approach is based on the assumption that the channel does not change over the resource elements for which the CDM is defined. For example, if the CDM group consists of frequency source elements, if the channel is frequency selective or if there is a time synchronization error between the transmitter and the receiver, orthogonality is lost at the receiver in the frequency plane. Similarly, in the case where the CDM group consists of source elements in time, if the channel fades quickly or there is a frequency synchronization error between the transmitter and receiver, orthogonality is lost at the receiver in the time plane. This situation is detrimental to channel estimation performance, especially in high spectral efficiency scenarios. Because with the loss of orthogonality in the receiver, interlayer interference occurs in the pilot symbols, and therefore channel estimation errors can restrict the overall performance. Therefore, it is important to evaluate and deal with such situations to achieve high data rate targets across a wide range of channel scenarios. It is important that the channel estimation algorithm in a MIMO-OFDM system accurately retrieves the CDM pilots at the receiver and interpolates the estimated channel values at the pilot positions to cover the entire resource grid. Obtaining the channel in the pilots of each MIMO layer belonging to the same CDM group from the received signal and obtaining a preliminary estimate at the pilot positions is called pilot pattern parsing. Pilot pattern decomposition and interpolation can optimally be performed jointly in an MMSE estimator. However, such an optimal pattern parsing process is not very practical due to the extremely large complexity. In particular, it requires a large matrix inversion process that must be done in real time for each pilot use and depends on the pilot symbol values. Another problem with optimal pattern separation is that when DMRS pilots are used for multi-user interference measurement and users use different source grid sizes for interpolation, performance degradation occurs due to this incompatibility between users. For this reason, lower complexity channel estimation algorithms are used in realistic receivers. The traditional channel estimation procedure for MIMO systems includes two separate stages: a pilot pattern parsing stage and an MMSE estimation stage that interpolates the estimated channel values at the pilot positions obtained as a result of the pilot pattern parsing process. In this case, the pilot pattern parsing process is carried out according to the least squares (LS) method, assuming that the radio channel remains constant at the pilot pattern parsing points, and this process has linear complexity. Interpolation can be performed independently using pilot values after pattern parsing. This method has much less complexity compared to optimal pattern parsing. If the channel has not changed at all in the CDM sources, the performance of the optimal pattern separation method is the same as the traditional method. However, usually due to multipath and mobility, the wireless channel has a certain delay and Doppler spread. This is time on the channel and! or may cause small or large changes in frequency. Similarly, when there are non-ideal factors such as time synchronization error or carrier frequency offset (CFO) error, the channel cannot remain stable, which causes serious performance loss in channel estimation for traditional receivers. Especially when it is considered that modern communication systems must support high data rate and spectral efficiency through MIMO technology, it is obvious that channel estimation errors can limit the performance of these systems for the reasons mentioned above. In a MIMO-OFDM system, the signal nik received at the kth subcarrier of the nth OFDM symbol at the mth receiver antenna can be expressed as follows: . 95th; . . .. . iwûng ;î 7," ? TGT-Win& I :_ 5311:! 7 ::Fâg 9;. is .5. .. fm ..Mas where, 711 = 1, ...,NR, n = 1, . ..,Ns,o,, k = 1, ..,,.K i' = 1, ...,NL for HÃT) kth of the nth OFDM symbol between the ith layer and the mth receiver antenna x552, the complex pilot symbol carried on the kth subcarrier of the nth OFDM symbol in the ith layer, stands for the noise term. NL is the number of data layers, NR is the number of antennas in the receiver, Nm is the total number of OFDM symbols in the channel estimation window! time slot, and K is the total number of OFDM subcarriers in the transmission band. The source grid for a given layer consists of NslotK source elements. Channel The aim of the estimation problem is to obtain an estimation of the channel coefficients, that is, gg?) values, by using the pilot symbols and the signals received at the locations where the pilot symbols are located. In practical systems, pilots should be placed perpendicularly between the NL MIMO layer to prevent interlayer interaction, otherwise the overall performance can be significantly reduced due to the interference effect. To achieve this goal, different multiplexing options can be used for pilot allocation. In Code Field Multiplexing (CDM), pilot symbols in different layers belonging to the same CDM group use the same source elements, where decomposition is performed through various codes such as orthogonal cover codes (OCC). In Time Domain Multiplexing (TDM), pilot symbols at different layers are transmitted in different OFDM symbols in a time slot. In Frequency Domain Multiplexing (FDM), pilot symbols at different layers are transmitted on different subcarriers within an OFDM symbol. One or more of these schemes can be used to ensure that pilot symbols are orthogonal. In CDM, a given list of MIMO layers belonging to the same CDM group shares the same source elements in time and frequency. Therefore, the layers need to be orthogonalized within the code space. Each CDM group can be characterized by sets of basic position indices determined in terms of subcarrier and OFDM symbol indices. For example, the CDM group d can be specified by sets of basic indexes (Ka,Ld). In other words, all layers in CDM group d ? The lower carrier indices produced according to the CD set and La! It can be said that it has pilot symbols positioned in the OFDM symbol indices produced according to the set. At the same time, each CDM group has at most |Cd| = |JCd||Ld| It can support a number of orthogonal layers, where Cd is the set of layer indices of CDM group d and |. | It shows the cardinality of the set. Different CDM groups should be transmitted in different source grid locations and should not overlap. Therefore, in layers that do not belong to a given CDM group, no symbols should be passed in the resource elements used for that CDM group. For example, for a layer i' that is not included in dz, which is a CDM group that can be expressed with the basis set (Efdzßdz), if E is 13,12, and F: E is 36,12, then X& = 0 since EE is Cdz. Different CDM groups, FDM and! or can be multiplexed using TDM. Let F and T be the numbers of CDM groups separated by FDM and TDM, respectively; In this case, D is 2 FT, where D is the number of different CDM groups within the source grid. In practical systems, the pilot placement pattern specified by (Kala) for CDM base sets, i.e. CDM group 01, needs to be repeated regularly in frequency and time to increase pilot density. This is because a sufficiently dense pilot distribution is required to capture channel effects and changes in the spatiotemporal frequency domain in the given grid. This means that in the entire resource grid, for a CDM group d, the pilots are found not only in the basic clusters (Kdßdyde pilot locations) but also in repeated situations. An example pilot placement is given in Figure 1. In this example, ? Cd : {1,2 } and L,, 2 {3,4} and the source grid contains K: 120 carriers and N51,,, = 14 OFDM symbols. At the same time, for the CDM groupj, the pilots are repeated every 6 subcarriers in the entire grid, frequency .They are also repeated once in the time domain so that the repeated pilot positions include 6 OFDM symbols. Hence, the pilot symbols of CDM group d are found in OFDM symbol indices {1, 2, 7, 8, 109, etc. However, the orthogonal cover When applying the code (length-2 in frequency and length-2 in time), each group of repeated pilots is treated separately, which shows that the pilot pattern parsing process is applied to the repeated pilots separately.In the present description, this is called the pilot pattern parsing group. For example, while the pilots on subcarriers numbered {1, 2} of OFDM symbols numbered {3, 4} form a pilot pattern separation group among themselves, the pilots on subcarriers numbered {7, 8} of OFDM symbols numbered {3, 4} form a pilot pattern separation group among themselves. Pilots form another pilot pattern separation group among themselves. First, traditional channel estimation methods in MlMO-OFDM systems with CDM pilot placement are explained. The common goal in all channel estimation methods is to use the symbols taken in pilot positions and pilot values according to the system model given in equation (1), m : 1, ...,NR, n 2 1, ...,Nslo,, k 2 1 The aim is to obtain the estimates for , ...,K, i' = 1, ...,NL, that is, Hg'g'i) values. For a given layer i' and receiving antenna 111, Interest) can be estimated using similar procedures for each (m, i') pair in a practical communication system. The main operations in a traditional channel estimation procedure are shown in Figure 2. In the first step, the signal is received to a receiving antenna, for example the receiving antenna m, and OFDM demodulation is performed to obtain the received symbols, i.e. Yâ?) values, in the frequency domain in the source grid of the receiving antenna m. In the second step, the traditional pilot pattern parsing process is performed on the pilot pattern parsing groups in the source grids of all layers in a CDM group, for example, CDM group d, using the relevant OCC code and the received symbols. As a result of this step, for a given pilot pattern discrimination group, a calculated channel prediction value is assigned jointly as the prediction of all positions in the group. However, channel estimates in different pilot pattern separation groups will likely be different. Pilot pattern separation should be performed for all CDM groups, i.e. d : 1, ..D. In the third step, the estimated values at the pilot positions are used to obtain the channel estimate for all weld elements in the weld grid using an interpolation method. For example, in the interpolation process; MMSE estimation in both time and frequency (2D- MMSE), MMSE estimation first in time and then in frequency or MMSE estimation first in frequency and then in time (MMSE 1D-1D), linear interpolation, nearest point interpolation, sliding window Different methods such as average can be used. In the last step of the traditional method in the state of the art, all the operations given in the first three steps of the procedure are performed for all remaining receiving antenna signals, that is, m = 1, ...,NR. In an example, the traditional channel estimation procedure given in Figure 2 The traditional pilot pattern parsing process in the second step is performed by Least-Squares (LS) estimation, assuming that the radio channel remains constant in the sources for which the pilot pattern parsing is performed. An example of this method is described below. For this purpose, it is shown how OCC is applied to a pilot pattern parsing group at the transmitter and how the estimate for each layer can be calculated with the pilot pattern parsing process. For example, assuming that a layer belonging to CDM group 1, a pilot pattern parsing group in the source grid, is located in the 1st and 2nd subcarriers of the 1st and 2nd OFDM symbols, there are four pilots in the pilot pattern parsing group. Layer indices in CDM group 1 are given as C1 = {1,2,3,4}. This means that all layers in Cy have the same locations of all pilots in their respective source grids. Applying length-2 000 in frequency and length-2 OCC in time to achieve orthogonalization, the transmitted pilot symbols can be written as: 1.1 X1; where Km is 2, each X792(Kg; Based on , ; and (2), the following equations can be written tü"=w$"-H$"+H$"-H$"Mn+mî) vsr = (Han + Ham - "gr" - asthma.. + msn v$=wy-arwsuarmnwa In order for the pilots using CDM on the receiver side to be orthogonal, for any (mi) pair, the pilot brüntL' of the channel as HE?" :c: HE?" :c: Hgîý'i) z: Hgîý'i) It is assumed that it remains constant in the sources in the parsing group i.Under this assumption, based on equation (2), the following equation is obtained (Hgîii for abbreviation) = hi:) 1 1 1 hi ~ ' ~' -1 1 -1 hz + Wign _ YL? 1 -1 -1 h3 Nm) _ "(m) -1 -1 1 !:4 - 1 where WUnJ_ - Wrfk is ank and ianki- _ 1 for n 2 1,2 and k 2 1,2 and all)- - Y(k) ank. This process can also be expressed in matrix form as Bh + &7 = ýp. The least squares (LS) solution for the estimated vector Ii = is expressed as follows: According to Equation (4), the elements of Ii can be expressed as: Here, EE is the receiving antenna with the ith layer in CDM group 1 m among the resolved pilot ex'L'int'L'i is the common channel estimate value assigned to the pilot positions within the decomposition group, i.e. iii = agri) = HE?" = Hgfý" = agzi". This example method is used for a specific receiving antenna For the signal, it is repeated for all pilot pattern parsing groups in the source grids of all layers in the CDM group.As a result of the operations of the traditional pilot pattern parsing step, the a priori estimates at all pilot positions between the ith layer and the mth receiving antenna are h 53; " It can be stored in a vector like . In the traditional pilot pattern parsing procedure, the main assumption is that the radio channel remains fixed across sources in a pilot pattern parsing group. If this assumption fails, the verticality of the pilots at the receiver is lost. Such an assumption should also be valid for all layers due to the cross terms, as shown in equation (5). For example, based on equations Ãl, (3) and (5), H(m3)+ HU" 3) -Hgm3) -Hgm3) + HSTAL Hfg'Q- HgT'4)+HgZ"4) + in terms of actual channel values i/l/fT)+ Vî'fgi)+ iÃ/gglh ili/g? It can be expressed as . This expression shows that even if the pilot pattern for layer 1 in the CDM group holds the assumption that the channel remains constant in the decomposition group, the estimation will be affected by the changes in the channel values in the {2, 3, 4} indexed layers, and such a situation may bring extra interference to the estimation. In practical systems, the assumption that the channel remains constant rarely holds exactly due to reasons such as large delay spread in the wireless environment, Doppler spread due to mobility, carrier frequency drift (CFO) and possible time synchronization errors. Such a loss in orthogonality can result in severe performance loss, especially in transmission scenarios involving high spectral efficiency (large numbers of MIMO layers, higher modulation order and code rates) required to achieve high data rates. In the document, channel estimation methods for distributed pilot model and MIMO-OFDM systems and an apparatus that enables these methods to be implemented are mentioned. The method and apparatus disclosed in the patent document in question enable reducing the number of pilot symbols in a MIMO-OFDM communication system and improving channel estimation in such a system. However, the patent in question does not describe a channel estimation method for MIMO-OFDM systems based on compensating the phase changes occurring in the source elements in a CDM group during the pilot pattern separation process. Therefore, it is not practical in MIMO-OFDM systems that use CDM in the pilot layout. There is a need for channel estimation methods and devices that provide better and more robust performance with less computational complexity. Brief Description of the Invention The present invention relates to a channel estimation method and apparatus for MIMO-OFDM communication systems with improved performance compared to conventional receivers. According to a particular embodiment of the present invention, there is provided a method and an apparatus for channel estimation in MlMO-OFDM systems using CDM in the pilot pattern, which method is based on compensating phase changes in the source elements in a CDM group during the process of parsing the pilot pattern and compared to conventional receivers. It has better performance. According to an embodiment of the present invention, a channel estimation method is provided for MIMO-OFDM communication systems using CDM in pilot distributions. This method is based on compensating for phase changes in the source elements in a CDM group during the pilot pattern separation phase. The received signal is OFDM demodulated to obtain the received symbols in the source grid, and a preliminary pilot pattern decomposition process is applied to the received symbols. Phase change rates in the source grid are determined, phase correction terms are calculated and applied. The pilot pattern separation process is performed once again using the received and phase-corrected signals, and the channel estimates at the pilot positions are updated using phase change correction terms. The estimated values at the pilot locations are interpolated to obtain the channel estimated values for all source elements. According to an embodiment of the present invention, an apparatus is provided for channel estimation in MIMO-OFDM communication systems using CDM in pilot allocations. The apparatus includes a pre-processing module for receiving the transmitted signal, performing OFDM demodulation to obtain the received symbols in the source grid, and performing a preliminary pilot pattern parsing using the received symbols; a phase correction-based pilot pattern parsing module for determining phase change rates in the source grating, calculating and applying phase correction terms, performing pilot pattern separation using the received phase corrected signals, and updating channel estimates using phase change correction terms at the pilot positions; and a channel estimation module that interpolates the estimated values at the pilot locations to obtain channel estimated values for all source elements. Brief Description of the Drawings Figure 1 is a graph showing an exemplary pilot allocation in the resource grid of a layer of a CDM group. Figure 2 shows the flowchart of the processes for conventional channel estimation in a MlMO-OFDM system with CDM pilot allocation. Figure 3 shows the flow diagram of the processes for the disclosed invention based on the phase correction based pilot pattern separation method for channel estimation in MlMO-OFDM systems. Figure 4 is a chart showing 5G Physical Downlink Data Channel (PDSCH) pilot allocations for two example DMRS configurations. Figure 5 is a graph showing the mean square error (MSE) performances at different signal to noise ratio (SNR) values under the TDL-C channel and DMRS Type 1 configuration with a delay spread of 1 us for eight different implementations of the disclosed invention and traditional channel estimation methods. . Figure 6 is a graph showing the MSE performances at different SNR values under the TDL-C channel and DMRS Type 2 configuration with a delay spread of 1 us for eight different implementations of the described invention and traditional channel estimation methods. Figure 7 is a graph showing the MSE performances at different SNR values under the TDL-C channel and DMRS Type 1 configuration with a delay spread of 2 us for eight different implementations of the disclosed invention and traditional channel estimation methods. Figure 8 is a graph showing eight different implementations of the disclosed invention and It is a graph showing the MSE performances at different SNR values under the TDL-C channel and DMRS Type 2 configuration with a delay spread of 2 us for traditional channel estimation methods. Figure 9 is a graph showing SNR versus MSE performances for various values of 80 used in the disclosed invention, where 50 is an integer representing the single group phase change rate used to calculate phase correction terms. Figure 10 is a graph showing SNR versus MSE performances under two different implementations of the disclosed invention and traditional channel estimation, TDL-C channel with 1 ps delay spread and DMRS Type 1 and Type 2 configurations. . Figure 11 is a graph showing SNR versus MSE performances for two different implementations of the disclosed invention, and for conventional channel estimation, under TDL-C channel with 2 ps delay spread and DMRS Type 1 and Type 2 configurations. Figure 12 shows the block error rate (BLER) performances at different SNR values when using the TDL-C channel with 2 us delay spread and the 16-QAM modulation scheme with 3/4 coding rate under DMRS Type 1 and 2 configurations for the described invention and traditional channel estimation methods. It is a graph showing . Figure 13 is a graph showing the BLER performances at different SNR values for the disclosed invention and conventional channel estimation methods, when using the TDL-C channel with 1 us delay spread and the 64-QAIVi modulation scheme with 3/4 coding rate under DMRS Type 1 and 2 configurations, Figure 13 14 is a graph showing the BLER performances at different SNR values for the described invention and conventional channel estimation methods, when using the TDL-C channel with 1 us delay spread and the 256-QAM modulation scheme with 2/3 coding rate under DMRS Type 1 and 2 configurations. Reference list 100. Computer applied method 101. Receiving the signal and performing OFDM demodulation 102. Primary pilot pattern separation 103. Determination of phase change rates 104. Calculation of phase correction terms 105. Application of phase correction terms 106. Pilot pattern separation process using phase corrected received signals 107. Updating channel estimates using phase change correction terms 108. Interpolating estimated values at pilot positions. Detailed Description Detailed explanations of the embodiments of the present invention will be given together with the figures accompanying the invention. The present invention relates to a channel estimation method and apparatus with better performance compared to conventional receivers for MIMO-OFDM communication systems. The method of the invention is based on eliminating phase changes in the source elements in a CDM group while performing the pilot pattern separation process. Therefore, the focus is on the phase of the channel coefficients in the pilot 'Pattern separation groups'. When there is a time synchronization error, or if the strongest channel plug arrives after a certain delay relative to the beginning of the receive window due to delay spread, there will be a dominant phase rotation in the frequency domain of the effective channel, which will violate the necessary assumptions for orthogonality. When there is a carrier frequency shift or Doppler spread due to mobility, the phase of the channel will change in the time domain. The described method increases the performance of the traditional channel estimation method by estimating the phase change rate in transmission bands in the frequency domain and/or transmission time slots in the time domain and removing this change during the pilot image separation process. The inventive method (100) enables channel estimation in multi-input multiple-output (MIMO) Orthogonal Frequency Division Multiplexing (OFDM) communication systems that use Code Division Multiplexing (CDM) groups, especially in the pilot layout; o receiving each receiving antenna signal and applying OFDM demodulation to them (101) to obtain the symbols received in the source grids of each receiving antenna, Performing a preliminary pilot pattern parsing process in all pilot pattern parsing groups using the received symbols (102), . For each receiving antenna signal, determining the phase change rates within the source grid in all layers in all CDM groups (103), o determining the phase correction terms to be used for each source element in all pilot pattern separation groups in all CDM groups, for each receiving antenna signal. calculation (104), o applying the phase correction term to each source element in all pilot separation groups in all CDM groups, for each receiving antenna signal (105), . Performing the pilot pattern parsing process using phase-corrected received signals in all CDM groups and receiving antenna signals, in all pilot pattern parsing groups, 0 phase change correction of the channel estimates at the pilot positions of all pilot pattern parsing groups in all CDM groups and receiving antenna signals. o updating using the terms (107), o interpolating the estimated values at the pilot positions to obtain the channel estimated values in all source elements in each layer for each receiving antenna signal (108). In Figure 3, MIMO-OFDM systems pilot for channel estimation. A flowchart is given for the disclosed invention, which is based on phase correction during pattern separation. In the method (100) of the invention, firstly, the signal is received at the receiver and OFDM demodulation is done by using the symbols in the frequency domain within the source grid, that is, YTE?, for each receiving antenna m, for m = 1, ...,NR. It is carried out to obtain the values (101). In the second step of the method (100) of the invention, the preliminary pilot pattern parsing process (102) is carried out in all pilot pattern parsing groups, using the relevant OCC code and received signals, for all CDM groups and receiving antennas, that is, CDM group d., d = 1. It is realized for ,...,D, and receiving antenna m, m: 1, ...,NR,. In one embodiment, the preliminary pilot pattern parsing step (102) is performed using the Least Squares (LS) estimation method, assuming that the radio channel remains constant within the pilot pattern parsing group. The first two steps (101), (102) of the present invention can be seen as pre-processing steps before the pilot separation process based on phase correction. In the third step of the method (100) of the invention, the phase change rates (slope) occurring in the frequency domain in the transmission source grid for a specific layer i, receiving antenna m, OFDM pilot symbol n and pilot subcarrier index k can be shown by (slope) and/or phase in the time domain. The rates of change (which can be represented by ErîmÜ) are determined (103). Note that the rate of phase change in frequency is a real, scalar value representing the channel phase changes occurring on the subcarriers in a given OFDM pilot symbol and may vary in different pilot symbols, i.e. different n, so it is a function of n, i.e. slope) It should be noted that it is named. Similarly, the rate of phase change over time is a real, scalar value that represents the channel phase changes occurring in the time symbols for a given pilot subcarrier and can therefore vary across different subcarriers, i.e. different k, so that k The function, that is, Eiîi, is called ±(k). In one embodiment, the slope) can be determined by calculations to be made on the estimated channel values at the pilot positions obtained as a result of the preliminary pilot pattern separation process (102). In another example, &fm-(k) can be determined by calculations to be made on the estimated channel values at the pilot positions obtained as a result of the pilot pattern separation process (102). In another example, the slope) may be determined by taking input from a processing block at the receiver. In yet another example, 8.2141:) may be determined by taking it as input from a processing block at the receiver. At the end of the step (103) of determining the phase change rates, the phase change rate should be obtained in the time domain or frequency domain or in both domains. The process of determining the phase change rates (103) is carried out in each receiving antenna signal, that is, the receiving antenna in all layers in m, m = 1, ...,NR and all CDM groups, for example, CDM group receive and ,01 = 1, ...D must be implemented. In the fourth step of the method (100) of the invention, the phase correction terms to be used in each source element in the pilot pattern separation groups for all CDM groups and received antenna signals are calculated (104). First, after the step (103) of determining the phase change rates in the method (100) of the invention, the phase change rates in the source grids in the frequency domain, that is, the slope) values, if available, are used for each CDM group 01 and receiving antenna m by using the existing values. Group phase change rates in the frequency domain named , êâid (n) are calculated. Then, using êf dm), the phase correction term in the frequency domain can be calculated as follows: In a given CDM group, for example CDM group d, a pilot pattern separation group {k0.k1,..., klmn-Ü is located at the subcarrier positions and here v If k,, < ku for < u, the phase correction term in the frequency domain at the nth pilot symbol and the ith subcarrier position in the relevant pilot pattern decomposition group is jznwu-kmgmcn) AS?" (11) = exp( Nfft ). Here Nfft is the FFT size used for OFDM modulation and j is the virtual unit.If, after the step (103) of determining the phase change rates of the inventive method (100), the phase change rates in the source grids in the time domain, that is, agi-(k) values, are available, the current values group phase change rates in the time domain called EiîiaÜf) are calculated for each CDM group 01 and receiving antenna m. Then, by using êiEnACk), the phase correction term in the time domain can be calculated as follows: In a given CDM group, for example, CDM group at, A pilot pattern parsing group is located at symbol positions {n0,n1,..., niLdH) and where v < u is 11,, < nu, the kth pilot subcarrier and the time at the ith symbol position in the relevant pilot pattern parsing group are The phase correction term in the field is yesterday) = p (W). Here, Ts is the number of samples in an OFDM symbol. Applying the phase correction terms calculation process (104) to all pilot pattern separation groups for all CDM groups, that is, CDM group d = 1, ...,D, and receiving antenna signals, that is, receiving antenna m = 1, ...,NR. is required. In the fifth step of the method (100) of the invention, at least one of the phase correction terms calculated in the frequency and time domain is applied to the received pilot signal in all pilot pattern separation groups (105). For a pilot pattern separation group belonging to CDM group d and located at the nth OFDM symbol position with subcarriers with partial-Ü indexes {k0,k1,..., phase in the frequency domain to the signal received at the receiving antenna m. The process of applying correction terms (105) can be expressed as 177.572 :YISEMSLMÜMY. This process occurs for each u : 0,1,...,|&'Cd| - It is done for 1, where (Aîimlfmnî Aâm) is the complex conjugate of f(n). For a pilot pattern decomposition group belonging to CDM group d and located at the kth subcarrier position with indexed symbols {n0,n1,..., ni5d|_1}, the phase in the time domain is given to the signal received at the receiving antenna m. The process of applying correction terms (105) can be expressed as ýgmzlfgîgmâmuwn*. This process is done for every u = 0,1,...,ii:di - 1, where (Aâm)'t(k))* is the complex conjugate of Aâm)'t(k). The process of applying phase correction terms (105) results in all CDM groups, that is, CDM group 0! = 1, .D and receiving antenna signals, that is, receiving antenna m: 1, ...,NR, it must be done in all pilot pattern separation groups. In the sixth step of the method (100) of the invention, the pilot pattern separation process is carried out in all pilot pattern separation groups using the received signal with phase correction applied, that is, 1775? (106). In one embodiment, the pilot pattern separation process (106) is found by means of the least squares (LS) method. As a result of this step, a specific CDM group d. For , channel estimates 15,20 are obtained at all pilot debris separation positions in all layers belonging to CDM group d. For example, a pilot 'pattern separation group, subcarriers with index {k0,k1,..., klei_1} and OFDM symbols {n0,n1,..., nißdl_1} with index {n0,n1,..., niLdi_1} Considering that it is positioned, 12 < u while k,, < ku ini; < nu is happening. In this case, the pilot says 'As a result of the pattern parsing process (106), the relevant pilot says 'A single channel estimate is produced for all source elements in the pattern parsing group. This value is allocated as the channel estimate for the pilot position (n0,k0), and ? It is called gg. The pilot pattern separation process (106), which uses the phase-corrected received signals, is used for all CDM groups, that is, CDM group d = 1, ...,D, and receiving antenna signals, that is, receiving antenna m = 1, ...,NR, needs to be realised. In the seventh step of the method (100) of the invention, the channel estimates at the pilot positions in all pilot pattern separation groups are updated using a phase term (107). For example, when a pilot pattern parsing group is located on subcarriers with index {k0,k1,..., kd-Ü and OFDM symbols with index {n0,n1,..., nllal-1}, when v < u, k, , < sand,, < nu. In an embodiment, in the process of applying the phase correction terms of the method (100) of the invention (105), if the phase correction is applied in the frequency domain, then for each u = 0, 1, ijcdi - 1, Ãgmkiî) = ggîgkgmlûmy and -12n Nm) and EMO/i) invention is 71 In another application, for each E Cd, -jzmku-koßgmtn) Gâm'ÜCR) = exp ( Nm ) and ââdm) is the ratio of the method of the invention (100) within the frequency domain determined in the step (104) of calculating the phase correction terms. is the group phase change rate. In an embodiment, in the process (105) of applying the phase correction terms of the method (100) of the invention, if the phase correction is applied in the time domain, then for each u = 0, 1, |Ld| - 1, for relevant) = Heat?? Ggmii)(k) and -i27'r(nu-n0)efnii(k) k 5 {k0.k1,.... kw&, agm'ûm) = exp(Ts) and asmak) invention It is the phase change rate in the time domain obtained in the determination step (103) of the phase change rates of the method (100). In another embodiment, for each .i (for 5 Cd -i2rr(nu-n0)êgl_d(k) Gâm'ûü) = exp(TS) and êfmUc) in the phase correction terms calculation step (104) of the method (100) of the invention. ) is the group phase change rate within the specified time domain. The process of updating the channel estimates using a phase term (107) must be performed in all pilot pattern decomposition groups for all CDM groups, that is, CDM group d 2 1, ..D and receiving antenna signals, that is, receiving antenna m : 1, ...,NR. . At the end of step (107) of updating the channel estimates using a phase term, channel estimates are obtained at all pilot positions in all layers for all receiving antenna signals. The channel estimates between the i'inoi layer and the mth receiving (mi) antenna are in a vector. In the eighth step of the method (100) of the invention, the channel estimation values obtained after the process of updating the channel estimates using a phase term (107) are for all receiving antennas. and interpolated to obtain channel estimates for all source elements in the source grid in each layer (108). Interpolation step (MMSE estimation is carried out using at least one of the following methods: first in time and then in frequency, or first in frequency and then in time (MMSE 1D-1D), linear interpolation, nearest point interpolation, sliding window average. The phase change rate in the frequency domain, determined in the step (103) of determining the phase change rates of the method (100) of the invention, represents the average change in the phase of the estimated channels in the consecutive subcarriers of the transmission band in the network layer i, the receiving antenna m and the OFDM pilot symbol n. It should be noted that the rate of phase change does not need to be constant throughout the transmission band, but in order to have low complexity and a robust algorithm, phase correction is done using a single value for the relevant layer, receiving antenna and OFDM symbol. Ideally, 37:11.01 ) can be chosen to minimize the average channel estimation error for a particular channel model. However, since there is no closed-form expression for calculating such an error for statistical channel models, the selection of sgji(n) is achieved using performance simulations. However, this is an impractical approach to implement in a real-time system. In the method (100) of the invention, the phase change rate network-(n) in the frequency domain obtained in the step (103) of determining the phase change rates is calculated using low complexity methods, with calculations based on the estimated channel values at the pilot positions after the preliminary pilot pattern separation process (102). can be determined. In one implementation of this embodiment, the slope) is determined using the IFFT-based approach. This method is based on the intuition that the phase rotation in the conduction band will be determined by the dominant channel taps arriving with a delay. Therefore, first of all, the response in the time domain of the channel effective for OFDM symbol n, layer i' and receiver antenna m is obtained by taking the IFFT of the estimated channel values in the pilot subcarriers located in the relevant OFDM symbol n. Based on this, 551.01) is calculated as a function of the estimated strength of the channel plugs. Some example functions to use for this include the index of the strongest channel plug or the index of the latest plug that is above a predetermined threshold value to avoid missing strong channel plugs that occur after the strongest plug. The selected index needs to be normalized before assigning it to 2.2147:). The normalization factor is equal to the difference between the initial subcarrier indices of the pilot pattern separation groups. An example for this calculation method is given to illustrate the procedure: For a layer 1' within a CDM group d, K = 100, ? Let Cd = {1,3} and Ld = {1} and considering that the pilots (still) given in the basis set are repeated in frequency in every 4 subcarriers in the entire grid of each layer belonging to CDM group d, layer i' ' has a pilot. According to this layout, the first pilot pattern parsing group is located on the subcarriers indexed as {1,3}, while the second pilot pattern parsing group is located on the subcarriers indexed as {5,7}. Therefore, the difference between the initial subcarrier indices of the pilot 'pattern separation groups for the entire source grid is 4. To calculate sm ii(1), first the IFFT of the following vector is calculated: hgnpi)=[H1(mÜ,H1(î"Ü HEM..." HSJÄHEÜJW, where each term of IiSgÜ is the CDM group The primary pilot pattern between 01 and the receiving antenna m is obtained by the decomposition process (102). The index of the strongest channel plug (or alternatively, the latest plug above a predetermined threshold value) is then recorded. Finally, , Em ,(1) :mm/#type In another implementation of this embodiment, &nu-,(11) is determined using the phase response approach as a direct approach to characterize the changes in the phase response of the channel between layer i and the receiving antenna m.This method is The preliminary pilot pattern separation process starts with calculating the phase response of the channel estimation values in the pilot symbols after (102) and recording the phase change between consecutive subcarriers. Based on this, an average value (after normalization with Zn/N",) is 5;_i(n). ) is assigned to. In one embodiment, this average value is obtained using linear regression. In another embodiment, this average value is obtained by averaging the recorded phase changes between consecutive subcarriers. In yet another embodiment, this average value is obtained by taking the median of the recorded phase changes between consecutive subcarriers. In the phase correction terms calculation (104) step of the method (100) of the present invention, the group phase change rate in the frequency domain calculated in the OFDM symbol n = for each CDM group 01 and receiver antenna m is obtained by different methods using sm (n) values. can be done. In one embodiment, êmdûi) is the average of the phase change ratio, SmiÜÜ: values in the OFDM symbol n for the layers belonging to CDM group d. In another embodiment, the phase change ratio in OFDM symbol n for layers belonging to .êgmmy CDM group d is the median of the values 851101). To calculate a single group phase change ratio for use in all CDM groups, receiver antennas and OFDM pilot symbols, âg,_d(n) = so. Here, Np is the total number of pilot symbols in the OFDM time slot. It is also the average of 50 phase change ratio, slope) values calculated for all possible layer, receiving antenna and OFDM pilot symbol combinations. In another example, the 80th is the median of the calculated phase change ratio, 551101) values across all possible layer, receive antenna, and OFDM pilot symbol combinations. Using a single value for the phase change ratio provides both simplicity and the opportunity to use more phase change rate values to calculate the required phase correction term, resulting in a particularly low signal! It is an advantageous choice as it leads to a more reliable estimation in noise rate scenarios. In one embodiment, in the step (108) of interpolating the channel estimates at the pilot positions of the method (100) of the invention, 2D-MMSE interpolation is performed using the following expression: ii (mt) z kahpmgphp + 021)*1 ii (mi) (6) 55.23 , stores the final channel estimation values in the entire source grid between the ith layer and the mth receiving antenna, and is the channel estimation values obtained after the step (107) of updating the ?i estimates using a phase term. Rgihp in Equation (6) is the correlation matrix of wireless channels between all source grid positions and pilot positions for the ith layer. Rgphp in Equation (6) is the wireless channel correlation matrix at the pilot positions for the ith layer, and 0-2 is the noise variance. This process given in Equation (6) is repeated for all layers i = 1,2, ...,NL, using the correlation matrices corresponding to each layer. In one embodiment, the correlation matrices Rghp and Rgphp given in equation (6) can be calculated using the robust channel estimation method in two dimensions. The robust channel estimation method is used in Robust MMSE channel estimation in OFDM systems with practical timing synchronization, by Vineet Srivastava et al., IEEE Wireless Commun. and Networking Conf., Atlanta, GA, USA, 2004. In the explanations below, layer index i' has been removed from the expression in variables for simplicity of explanation. To calculate a correlation matrix, time and frequency correlations between source elements are calculated independently and these are combined with a Kronocker product. The correlation matrices in the frequency domain are expressed as Râ?, and 12,122,, and in the time domain as hphp. First, if the specific correlation matrices are R p and R for the number of channel plugs L, Râ? A matrix named is defined by the following equation, 1 -jZTrÇa - b)l RISÃJÜL b) = Z ECXP (N_) (7) Here, Nfft is the FFT size used in OFDM modulation and Rgg)(a,b), Râ:) is the matrix a' is the value in the th row and bth column, ath and hp matrix in the source grid, &Emin is just the channel correlation value between the bth subcarriers. Râ can be obtained by taking the columns corresponding to the pilot positions in the frequency domain. The R(f) matrix, w› can be obtained by taking only the rows of hphp corresponding to the pilot positions in the frequency domain. Similarly, using the Jakes model, the matrix Râ? is defined by the following equation: where D is the maximum Doppler spread, T is the OFDM symbol duration. Rgmaß) is the element in the ath row and bth column of the Râ:) matrix and is the channel correlation value between the ath and bth OFDM symbols in the source grid. Also, jo(.) is the zeroth order Bessel function of the first kind. The rggp matrix &fal can be obtained by simply taking the columns corresponding to the pilot positions in the time domain. The matrix 3.1221,, &Eg, can be obtained by taking only the rows corresponding to the pilot positions in the time domain. Then, the combined correlation matrices can be calculated as Rhhp = Rgßég RIEQQ and Rhphp = Rhgâp® Rhgg,, where @3 denotes the Kronocker product of the matrices. It should be noted here that the channel values within the source grid or within the pilot positions, that is, the values at 11 and hp respectively, must be sorted first in the frequency domain and then in the time domain. This means that for a given layer, h1 hp›1 h = ' and hp = ', where ! For each hNsloC hpiNp vector h" (n = 1,2, ---iNslot) in 1, it is a vector that carries the channel values on the subcarriers of the nth OFDM symbols in the time slot. Each vector in hp hi" (s = 1 ,2,...,Np) is a vector that carries the channel values on the subcarriers of the sth pilot OFDM symbols in the time slot, and where Np refers to the total number of pilot OFDM symbols in the time slot. The matrix Rhphp is common for each layer in the same CDM group. The method (100) of the invention has a practical calculation complexity, just like in traditional receivers. In the traditional method, traditional pilot pattern parsing can be accomplished in linear time complexity with simple arithmetic operations, and 2D-MMSE estimation-based interpolation can be done by storing the necessary matrices for a set of channels and SNR parameters in hardware memory. In other words, the matrix inversion process can be performed offline beforehand, and during the operation of the system, the matrix to be used for the relevant channel conditions can be retrieved from memory and used to perform MMSE estimation. In this way, only matrix multiplication is required to perform the estimation, which can be implemented very efficiently in hardware. The method (100) of the invention includes a low-cost pre-processing step for the step (103) of determining the channel phase change rate in the subcarriers. For the IFFT-based method, the IFFT block is already used in OFDM modulation and demodulation and can be efficiently implemented in hardware with O(NlogN) complexity. For the phase response-based approach, the phase of the channel estimation values can be obtained efficiently using the CORDIC algorithm without using any multipliers. If the phase change rate is known, the proposed algorithm requires only 2 extra complex multiplications per pilot pattern separation operation compared to the traditional method. Therefore, the complexity of the described method is only slightly higher than the traditional method due to low complexity preprocessing. If the delay spread of the channel does not change very quickly, the phase change rate can be calculated at specific update periods instead of being calculated on each transmission. In this case, the complexity will be almost the same as the traditional method between updates. Another option would be to create a table for the phase change rate for different channel models and select the relevant rate from the table rather than calculating it dynamically. The performance of the method (100) of the invention is evaluated using the pilot structure in the SG Version 15 standards offered by SGPP. In 5G, the pilot symbols used for data demodulation are called Demodulation Reference Signals (DMRS), the downstream link and uplink data channels are called Physical Downlink Data Channel (PDSCH) and Physical Upstream Link Data Channel (PUSCH). DMRS defined for both PDSCH and PUSCH have the same structure. Since there are a wide variety of usage scenarios that need to be supported by 5G, the DMRS structure is very flexible and can be configured with relevant configuration parameters. As given in the standard, the PDSCH mapping type specifies whether the timeslot is a traditional downstream link timeslot (Type A) or a special time slot structure defined for 5G (Type B), also called mini time slots, dmrs- TypeA- Position indicates the starting symbol of the first DMRS in the time slot (3 or 4), dmrs-AdditionalPosition DMRS (0, 1, 2 or 3) indicates whether there is an additional OFDM symbol in the time slot, dmrs-Type (Type 1 or da Type 2) determines the frequency domain pattern of DMRS within a given symbol, and maxLength is 1 (single) or 2 (even) symbols of the CDM group, i.e. |Ld| = 1 or 2 indicates whether it is defined in . Figure 4 gives two examples of DMRS configurations. The first two source grids show the DMRS patterns for the first example. In this case, there is NL = 4 orthogonal layers and 1 source block (1( = 12 subcarriers). The entire transmission band is an aggregate of such source blocks and Nsm = 14 OFDM symbols. For larger source block sizes, the given pattern is in the frequency domain In this case, the DMRS parameters are configured as PDSCH mapping Type = Type A, dmrs-TypeA-Position = 3, dmrs-AdditionalPosition = 3, dmrs-Type = Type 1 and maxLength = 1. Here, the basis set ?( 1 = The layers in CDM group 1 with {1,3} and 21 = {3} are C1 = {1,2}, while the layers in CDM group 2 with basis set ?CZ = {2,4} and .'52 = {3} layers are given as CZ = {3,4} CDM groups are multiplexed with each other in the frequency domain, so F = 2, T = 1 and D = 2. For type 1, each pilot pattern separation group, each source It is used three times in the frequency domain inside the blog.For example, in Figure 4, for the first example pilot layout, the subcarrier starting positions of the pilot pattern parsing groups are {1,5,9} for each symbol. Pilot pattern separation groups are automatically placed in an orderly manner throughout the data transmission band in the frequency domain, using ?61 as the basic reference. However, the frequency of pilots in the time domain can be configured with the dmrs-AdditionaIPosition parameter. For example, in Figure 4, the pilots are located on 4 OFDM symbols (symbols whose positions in the time slot are indexed {3,6,9,12}). For type 1 and maxLength = 1, there can be a maximum of 2 different CDM groups, and since each CDM group can carry a maximum of two layers, the maximum number of orthogonal layers that can be supported in this configuration is 4. In Figure 4, the bottom two source grids show the DMRS patterns for the second example. In this case, the main difference compared to the first example is that dmrs-Type = Type 2 is used instead of Type 1. In this case, the layers in CDM group 1 2 with basis set JC, : {1,3} and 131 = {3} are given as C2 = {3,4}. For Type 2, each pilot pattern decomposition group is used twice in the frequency domain within each source block. For example, in the second example given in Figure 4, the subcarrier starting positions of the pilot pattern separation groups are {1,7} for each symbol. This indicates that Type 2 has lower density in the frequency domain compared to Type 1, although Type 2 can support a greater number of layers overall. The main reason for this is that for Type 2, there can be up to 3 different CDM groups multiplexed within the frequency domain. Additionally, considering that each CDM group can carry a maximum of two layers when maxLength = 1, the maximum number of orthogonal layers that can be supported for this configuration can be found to be 6. In the example given, only NL = 4 of them are used. If maxLength = 2, the maximum number of layers with orthogonal pilots is doubled compared to the single symbol case, becoming 8 and 12 for Type 1 and Type 2 respectively. This is because each CDM group can carry a maximum of four layers instead of two in the case of maxLength = 2, and the maximum number of CDM groups does not change. Although SG PDSCH DMRS pilot signals are used in the examples given in the invention, the invention described can be applied to any pilot allocation scheme containing CDM groups. Examples of these include application scenarios such as multi-port CSI-RS, SRS, channel estimation with PUSCH DMRS in 5G, or channel estimation with UE-specific DMRS in LTE. In the simulations, DMRS parameters are selected as in the examples given in Figure 4 and 5G waveform is used. Additionally, the Tapped Delay Line-C (TDL-C) channel model is used with a delay spread of 1 or 2 us, without user mobility. Modulation type is 16, 64 or 256-QAM. Channel coding is NR LDPC with base graph 1 and code rates are 213 or 3/4. The channel decoder is a 20-iteration minimum sum algorithm. The transmission band consists of K = 240 subcarriers and the subcarrier spacing is 15 kHz. The number of vertical layers is NL = 4 and the number of receiving antennas is NR = 8, MIMO precoding is not used, the carrier frequency is 3.5 GHz. Synchronization is assumed to be perfect and a soft MMSE equalizer is used as the MIMO equalizer. In the channel estimation interpolation stage, both the traditional method and 2D-iVlMSE are used for the described invention. Eight different applications of the invention described in Figures 5, 6, 7 and 8 and the traditional channel estimation method are compared. Conventional channel estimation follows the steps given in Figure 2. In the applications in the figures, the phase change ratio, i.e. 20, used in the phase correction terms calculation (104) step of the method (100) of the method of the invention, is used as a single value in common for all CDM groups, receiving antennas and OFDM pilot symbols. Therefore, for all applicable d,m,n combinations, the first application, called Median phase-based / Set average: median, is based on the method (100) of the invention, in the step of determining the phase change rates (103), in the step of determining the phase change rates in the frequency domain (slope), phase It shows that it is obtained by taking the median of the recorded phase changes between consecutive subcarriers using the response approach. Also, for this application, 20 is the median of the phase change ratio (i.e. slope) values obtained for all possible layer, receiver antenna and OFDM pilot symbol combinations. Median phase-based / Set average: The second application, called arithmetic median, is based on the method (100) of the invention, in the step (103) of determining the phase change rates, the phase change rate in the frequency domain is calculated as the median of the recorded phase changes between consecutive carriers, using the Sifuwl phase response approach. It shows that it was obtained by taking . Also, for this application, 80 is the arithmetic mean of the phase change ratio values, i.e. the network, for all possible layer, receiver antenna and OFDM pilot symbol combinations. Arithmetic middle phase based / Set average: The third application, called median, is the method of the invention (100), in the step (103) of determining the phase change rates, the phase change rate in the frequency domain, 851,101), is recorded between consecutive sub-carriers using the phase response approach. It shows that the phase changes are obtained by taking the arithmetic average. Also, for this application, 50 is the median of the Phase Change Rate (i.e., slope) values for all possible layer, receive antenna, and OFDM pilot symbol combinations. Arithmetic medium phase based 1' Set average: The fourth application, called arithmetic mean, is the method (100) of the invention, in the step (103) of determining the phase change rates, the phase change rate in the frequency domain, 512.101) between consecutive sub-carriers, using the phase response approach, It shows that the recorded phase changes are obtained by taking the arithmetic average. Also, for this application, 80 is the arithmetic mean of the phase change ratio values, i.e. deaf:) values, for all possible layer, receiver antenna and OFDM pilot symbol combinations. The fifth application named shows that the method (100) of the invention is obtained by selecting and normalizing the index of the strongest channel plug in the step (103) of determining the phase change rates, using the IFFT-based approach. Also, for this application, .90 is the arithmetic mean of the phase change ratio values for all possible layer, receiver antenna and OFDM pilot symbol combinations, i.e. 272.01). The sixth application shows that in the step (103) of determining the phase change rates of the method (100) of the invention, the phase change rate in the frequency domain, 55,, (11), is obtained by selecting and normalizing the index of the strongest channel plug using the IFFT-based approach. shows. Also, for this application, 20 is the median of the phase change rate, i.e., slope, values for all possible layer, receive antenna, and OFDM pilot symbol combinations. The application is to obtain and normalize the index of the last plug whose power is above a predetermined threshold value, using the IFFT-based approach (103), in the step (103) of determining the phase change rates of the method of the invention (100). It shows that it was obtained by . The threshold value is taken as 3/4 of the strength of the strongest channel plug. Also, for this application, 5 is the median of the phase change rate values, i.e., slope, for all possible layer, receiver antenna, and OFDM pilot symbol combinations. The application is to obtain and normalize the index of the last plug whose power is above a predetermined threshold value, using the IFFT-based approach (103), in the step (103) of determining the phase change rates of the method of the invention (100). It shows that it was obtained by . The threshold value is taken as 3/4 of the strength of the strongest channel plug. Also, for this application, 50 is the average of the phase change rate, i.e., slope, values for all possible layer, receiver antenna, and OFDM pilot symbol combinations. In Figures 5, 6, T and 8, DMRS Type 1 and channel delay spread is 1 ps, DMRS Type 1 and channel delay spread is 2 ps, DMRS Type 2 and channel delay spread is 1 ps and DiVIRS Type 2 and channel delay spread is shown, respectively. MSE performances were provided at different SNR values for 2 ps propagation. In all of the figures, it is observed that all eight different implementations of the described method improve the performance of the traditional method, especially as the SNR increases. The performances of median phase based, Artimetic mid based and IFFT (threshold value) based applications are very close to each other at large SNR values. The IFFT (maximum peak) based implementation performs slightly worse than the others. At the same time, it can be seen from all four figures that Median phase based implementations are slightly better than Arithmetic median phase based implementations. In terms of the set averaging methods to obtain 80, the performance values for the Set average: arithmetic mean and Set average: median methods are very close to each other and one may be slightly better than the other depending on the phase change rate calculation method and simulation scenario. When we look closer to the figures, we see that the overall performance ranking of the methods changes as the scenario changes. However, it is important to note that these differences will not create an observable performance difference in practical SNR and code rate scenarios in terms of link level performance. Therefore, any of the exemplary embodiments of the disclosed invention has been used to obtain the performance of the present invention. When a single set of phase change ratios is used to calculate phase correction terms at all possible phase correction points, one issue that needs to be addressed is what the best possible value is in terms of channel estimation performance. This issue is important in order to control the performance of the low-complexity phase ratio calculation options described in the invention. Since 80 is a positive scalar real value, its ideal value must normally be sought on real numbers through Monte-Carlo simulations, but this approach is not realistic and practical. Instead, its value can be restricted to integer values, that is, it can be scanned among 80 different positive integer values to achieve good approximation performance. In Figure 8, MSE performances for different SNR values are given when the delay spread is 1 ps, for the DMRS Type 1 configuration, when various integer values are used for 50 in the phase correction terms calculation step (104) of the method (100) of the method of the invention. It can be seen that the performance increases as 50 increases, up to 50 = 5, and then begins to decrease if it is increased more than 20 = 5. The performances of 50 = 5 and 80 : 6 are very close to each other. This shows that although the best value for .50 need not be an integer in general, the performance of the best integer value will approximate the optimal value very closely. For this channel model, similar results are obtained in the Type 2 configuration, and the best integer value is again obtained as 80 = 5, although it is not given in Figure 9. When the delay spread is 2 ps, the best (integer) performances are obtained as .90 : 8 and so = 7 for Type 1 and Type 2, respectively, even though they are not given in Figure 9. In Figure 10, MSE performances are provided at different SNR values for DMRS Type 1 and Type 2 in the TDL-C channel with a delay spread of 1 ps for two different applications of the method (100) of the invention and traditional channel estimation. The disclosed invention (50 = 5) corresponds to a single group phase change ratio, 50, which is used to calculate the phase correction terms at all possible phase correction points in the phase correction terms calculation step (104) of the method (100) of the method subject to the invention. . Using Figure 8, optimal integer values for 80 are used. The invention described (phase response approach) refers to the Median phase-based / Set average: median application used in Figures 5, 6, 7 and 8. When the method of the invention is applied, as the SNR increases, a significant increase in performance is achieved compared to the traditional method. At the same time, in both Type 1 and Type 2 configurations, the performance of the Described invention (phase response approach) method is comparable to the performance obtained with the Described invention (50 = 5), that is, by choosing the optimal integer for the group phase change ratio used in the disclosed invention, 80. It has been observed that it can reach the performance achieved with . This means that the disclosed low-complexity methods for calculating phase change rates achieve the optimal values that can be achieved if a single set of phase change rates is used in the disclosed invention. The reason for this is that, as can be seen from Figure 8, the difference between the performance that will be achieved by choosing the best integer and the performance that will be achieved by using the optimal real number value for 50 is not significant. For all methods, the Type 2 configuration provides better performance compared to Type 1 at medium and high SNR values. The reason for this is that the main factor limiting the performance in this case is the loss of orthogonality at the receiver due to the frequency selective channel seen in the pilot pattern separation groups. The subcarriers in a pilot pattern separation group are consecutive for Type 2, as shown in Figure 4, so they are less affected by loss of orthogonality compared to the Type 1 configuration. On the other hand, Type 1 configuration has better performance than Type 2 at low SNR because Type 1 has a more regular pattern and higher pilot density compared to Type 2. In this way, it can suppress noise more effectively at low SNR. In Figure 11, MSE performances at different SNR values for DMRS Type 1 and Type 2 in the TDL-C channel with a delay spread of 1 ps are given for two different applications of the method (100) of the invention and traditional channel estimation. In this figure, the implementations for the disclosed invention are the same as in Figure 10. Additionally, for the optimal single group phase change ratio, sm, previously determined optimal integer selections for this channel model are used, i.e. 5.3:8 is used for Type 1, while 50:? It is used for type 2. The performance of the disclosed invention (phase response approach) provides the same performance as if optimal integer values for 50 were used in both configurations. For the channel estimation methods given in the figures, it has also been observed that Type 1 provides better performance compared to Type 2 in all SNR regions. The reason for this is that the channel is high frequency selective and in the Type 2 configuration, this configuration does not have sufficient pilot density and cannot adequately capture the change in the channel due to the gaps between the pilot pattern separation groups in the frequency domain. The described method exceeds the performance of the conventional method in both configurations. It is observed that the best performance is achieved when the described method is used with the Type 1 configuration among all the options considered in the figure. In Figure 12, for the method (100) of the invention and traditional channel estimation, in case of using 16-QAM modulation with 3/4 coding rate for DMRS Type 1 and 2 in the TDL-C channel with a delay spread of 2 us, the BLER formed at different SNR values The performance result is presented for comparison purposes in the case where the receiver has perfect (ideal) channel information. For Type 1, the BLER target of 0.01 is obtained with the described invention and the traditional method, when the SNR is 9.7 and 11.8 dB, respectively. Therefore, by applying the described method, the performance of the traditional method is improved by 2.1 dB. The performance difference between the described invention and the case with ideal channel information is 3.8 dB for Type 1. For Type 2, the BLER target is achieved by the described invention and the traditional method, respectively. It is achieved when the SNR is 12.2 and 12.9 dB, indicating a performance improvement of 0.7 dB. It is important to note that the improvement obtained with the described method is greater for Type 1, because the distance between subcarriers in the same pilot pattern separation group is larger in this configuration , therefore the phase change to be compensated is more pronounced than in the Type 2 case. In Figure 13, for the method (100) of the invention and for traditional channel estimation, for DMRS Type 1 and 2 in the TDL-C channel with a delay spread of 1 us, in case of using 64-QAM modulation with 3/4 coding rate, for different SNR values. The resulting BLER performance is shown. In this case, Type 2 gives better performance compared to Type 1 for both methods. With the explained and traditional method for Type 2, the BLER target of 0.01 is achieved when the SNR is 14 and 14.3 dB, respectively, while the target for Type 1 is achieved when the SNR is 15.4 and 18.8 dB, respectively. Accordingly, the present invention increases the performance of the traditional method by 3.4 and 0.3 dB in Type 1 and Type 2 configurations, respectively. The difference between the described method and the ideal channel information situation is 2.8 and 4.2 dB for Type 1 and Type 2, respectively. With the described method, it is observed that the performance improvement is greater in the Type 1 configuration. If you look at Figure 10, when the SNR is 15 dB, 28% and values are also important in terms of MSE for Type 1 and Type 2, respectively, with the explained method. For example, for Type 2, system performance to which the described method is applied is still limited primarily by noise. As a result, in order to achieve a significant improvement in BLER performance, both a good improvement in MSE values and this improvement must be around (or larger) the noise variance values in the SNR region required for a given spectral efficiency. In Figure 14, for the method (100) of the invention and for traditional channel estimation, in the TDL-C channel with a delay spread of 1 us, for DMRS Types 1 and 2, with 2/3 coding rate, under 256-QAM modulation, for different SNR values. BLER results are given. It has been observed that the Type 1 configuration cannot work with this modulation and code rate with the traditional method. The proposed method can achieve 0.1 BLER when the SNR is 30 dB for Type 1. On the other hand, for Type 2, the BLER target of 0.01 can be achieved when the SNR is 19.6 and 21.4 dB with the described invention and the conventional method, respectively. Accordingly, the described method improves the performance of the conventional method in the Type 2 configuration by 1.8 dB. At the same time, the performance difference of the described method with the ideal channel information situation is 4.2 dB. For the method described and the traditional method in the case of Type 1 according to Figure 10, the effective SINR cannot be greater than 20.5 and 18.8 dB due to the MSE estimation error; This explains the large BLER values observed. However, for Type 2, when the SNR is 22 dB, the 21% improvement in MSE is a significant difference in the SNR values required to reach the target BLER, as the MSE values are close to the variance value of 0.0041 when the disclosed invention is used and when it is not used. results in recovery. Therefore, it is possible to achieve significant performance improvements for both Type 1 and Type 2 using the disclosed invention. The results obtained from simulations regarding the performance of the method (100) of the invention can be summarized as follows. First, the disclosed invention can use different low-complexity implementation approaches to calculate phase change rates 103 and phase correction terms 104 in pilot pattern separation groups. It is observed that the performances of these approaches are very close to each other and all of them improve the performance of the traditional method. Their exact performance ranking depends on the channel and DMRS configuration type, but since the performance differences are not significant, the choice of method can be left as an implementation option. Secondly, if the single group phase change rate is used in the process of calculating the phase correction terms (104), the proposed implementation approaches of different low complexity can in all cases achieve the average error performance that occurs when the optimal value for the single phase change rate is used. This is because even if the phase change rate is limited to integer values, a good approximation can be achieved for the real-valued case in the general case, and also because low-complexity implementation methods have almost the same performance as the optimal integer selection. Third, the disclosed invention provides performance improvements for both Type 1 and Type 2 configurations. For example, in some test cases, it is observed that the described method can provide an improvement in the SNR values required to reach the target BLER, up to 3.4 dB in the case of Type 1 and 1.8 dB in the case of Type 2. Since the Type 1 configuration is more sensitive to loss of orthogonality in the case of a frequency selective channel compared to Type 2, the performance increase provided by the invention is generally greater for Type 1. At the same time, greater performance improvements are observed in high spectral efficiency scenarios. This observation is 'Particularly Important' because one of the main reasons to use CDM groups in pilot design is to be able to support a larger number of MIMO layers to achieve high data rates. This invention also relates to an apparatus that uses CDM groups in the pilot layout for channel estimation in MIMO-OFDM wireless communication systems. In a preferred embodiment of the invention, the apparatus in question; o To obtain the received signals in the source grid of each receiving antenna, using the corresponding OCC code and the received symbols in the source grid of all layers in all CDM groups for each receiving antenna signal, receiving each receiving antenna signal and performing OFDM demodulation to them The pre-processing module that performs the pilot 'pattern parsing' process among all pilot pattern parsing groups; o determines the phase change rates within the source grid for all layers in all CDM groups for each receiving antenna signal, calculates the phase correction terms to be used for each source element in all pilot image separation groups in all CDM groups for each receiving antenna signal, For a receiving antenna signal, applying the phase correction term to each source element within all pilot 'Pattern decomposition groups' in all CDM groups and using phase shift correction terms at pilot positions within all pilot pattern decomposition groups in all CDM groups for each receiving antenna signal. a phase correction based pilot pattern separation module that updates channel estimates; it includes a channel estimation module configured to interpolate estimated values at pilot locations to obtain channel estimated values on all source elements at each layer for each receiving antenna signal. Around this basic concept, it is possible to develop various applications of the computer applied method (100) that is the subject of the invention, and the invention cannot be limited to the Examples explained here, it is essentially as stated in the claims.TR TR TR