[0108] In order to make the purpose, technical scheme and advantages of the present invention more clear, the present invention will be further explained in detail with examples below. It should be understood that the specific embodiments described here are only for explaining the present invention, but not for limiting the present invention.
[0109] Aiming at the problems existing in the prior art, the present invention provides a CDMA access method based on large-scale MIMO-OTFS, and the present invention will be described in detail with reference to the attached drawings.
[0110] such as Figure 1 It is shown that the CDMA access method provided by the present invention includes the following steps:
[0111] S101: building a large-scale MIMO time-varying uplink channel model based on a large-scale uniform linear array antenna;
[0112] S102, based on OTFS modulation, a large-scale MIMO-OTFS angle-delay-Doppler domain received signal model and an angle-delay-Doppler domain uplink primary and secondary 3D channel model are obtained;
[0113] S103: Based on the channel model, using the dispersion characteristics of large-scale MIMO-OTFS channel in the angle-delay-Doppler domain, designing user angle domain grouping and path scheduling algorithm, reasonably allocating angle domain resources at the user end, and ensuring that the observation areas of different users are orthogonal in the angle-delay-Doppler domain 3D cube area at the receiving end;
[0114] S104: receiving pilots and effective data of different users at the scheduled receiving grid positions, estimating their channels by using the least squares (LS) estimator, and recovering data by using the maximum ratio combining (MRC) method;
[0115] S105: Use the user grouping and path scheduling algorithm in uplink transmission to guide the downlink access process, and design a low-complexity downlink beamforming strategy.
[0116] The CDMA access method based on large-scale MIMO-OTFS provided by the invention can be implemented by other steps by ordinary technicians in the industry, Figure 1 The CDMA access method based on large-scale MIMO-OTFS provided by the present invention is only a specific embodiment.
[0117] such as Figure 2 It is shown that the CDMA access system based on large-scale MIMO-OTFS provided by the present invention includes:
[0118] The first model building module 1 is used for building a large-scale MIMO time-varying uplink channel model based on a large-scale uniform linear array antenna;
[0119] The second model building module 2 is used for obtaining a large-scale MIMO-OTFS angle-delay-Doppler domain received signal model and an angle-delay-Doppler domain uplink primary and secondary 3D channel model based on OTFS modulation;
[0120] User angle domain grouping and path scheduling algorithm design module 3, which is used to design user angle domain grouping and path scheduling algorithm based on the channel model and using the dispersion characteristics of large-scale MIMO-OTFS channel in angle-delay-Doppler domain, allocate angle domain resources reasonably at the user end, and ensure that the observation areas of different users are orthogonal on the 3D cube area of angle-delay-Doppler domain at the receiving end;
[0121] The data recovery module 4 is used to receive the pilots and effective data of different users at the scheduled receiving grid positions, estimate their channels by using the least squares (LS) estimator, and recover the data by using the maximum ratio combining (MRC) method;
[0122] The downlink beamforming strategy module 5 is used for guiding the downlink access process by using the user grouping and path scheduling algorithm in uplink transmission, and designing a low-complexity downlink beamforming strategy.
[0123] The technical scheme of the present invention will be further described with reference to the accompanying drawings.
[0124] The invention specifically relates to a CDMA access method based on large-scale MIMO-OTFS, which can be applied to large-scale MIMO wireless networks, effectively meet the intensive multi-user access requirements of networks in high mobility scenarios, and improve the overall performance of the system. For uplink, angle-delay-Doppler domain resources are allocated through user angle domain grouping and path scheduling. LS estimator is used to estimate the angle-delay-Doppler domain 3D channel in the current OTFS block, and MRC strategy is used to recover data. For the downlink, the same path scheduling strategy as the uplink is used, and the MMSE method is used to design the beamforming vector, so that the downlink equivalent channel estimation and data recovery can be realized.
[0125] The application principle of the present invention will be described in detail with reference to the accompanying drawings.
[0126] such as Figure 3 It is shown that the CDMA access method based on large-scale MIMO-OTFS provided by the example of the present invention includes the following steps:
[0127] OTFS modulation is used in large-scale MIMO system. According to the characteristics of large-scale uniform linear antenna array and OTFS modulation, an angle-delay-Doppler domain channel model is constructed to obtain the uplink channel, and the uplink channel parameters are estimated by existing algorithms.
[0128] According to the sparseness of the large-scale MIMO-OTFS channel in the angle domain and the dispersion characteristics in the delay-Doppler domain, the base station uses the path scheduling algorithm to group the users in space, and allocates the angle-delay-Doppler domain resources to the users to make their receiving areas at the base station orthogonal.
[0129] Each user sends the pilot and data in the delay-Doppler domain resource block allocated by each user in the same OTFS block. The pilot is used to estimate the angle-delay-Doppler domain channel in the current OTFS block, and then it is used for data recovery based on MRC.
[0130] The uplink path scheduling algorithm is used to guide the downlink, and the beamforming vector is designed, so that the downlink equivalent channel estimation and data recovery can be realized.
[0131] such as Figure 4 The uplink access flow of the CDMA access method based on large-scale MIMO-OTFS provided by the embodiment of the present invention is shown. First, the user maps the data on the grid points of the delay-Doppler domain assigned to the user by the scheduling algorithm, and then it is converted to time-frequency domain by OTFS modulation and sent out. At the base station, the received signal is transformed into antenna-delay-Doppler domain by OTFS demodulator, and then the signal is transformed into angle-delay-Doppler domain by DFT along the antenna. The base station demaps and receives each user's signal at the desired angle-delay-Doppler domain grid point, estimates each user's angle-delay-Doppler domain channel using pilot, and then recovers each user's data using MRC. The downlink process is similar to that of uplink, but the design of beamforming vector is added before sending data. Finally, the equivalent channel is estimated and the data is recovered at the user end.
[0132] The application scenario of the CDMA access method based on large-scale MIMO-OTFS provided by the embodiment of the invention is as follows:
[0133] Step 1: Consider a single-cell millimeter-wave large-scale MIMO system in a high-mobility scenario. The base station serves k = 32 randomly distributed single-antenna users. The base station is equipped with n r = a uniform linear array of 128 antennas. There are scatterers in the space, and the user channel consists of multiple propagation paths. The wireless signal can reach the base station along the line-of-sight path, or it can be reflected by multiple scatterers, which means that the channel link between the base station and users will be affected by frequency selective fading. The user's moving speed range is V. s ∈ [120,306] km/h, and the maximum Doppler frequency is 2kHz. Due to the high mobility of users, the channel changes rapidly and experiences time-selective fading. Suppose the channel between a specific user and the base station has P scattering paths, and each scattering path corresponds to a direction of arrival (DOA), a Doppler frequency shift and a time delay.
[0134] Define θ k，p Is the DOA of the path p of the kth user, and the corresponding spatial guidance vector can be expressed as:
[0135]
[0136] Where λ is the carrier wavelength, and the antenna spacing D is set to half wavelength. Therefore, from the geometric channel model, the time-varying channel of user k in time slot n is expressed as:
[0137]
[0138] Where l represents the delay domain index, h k，p 、τ k，p And v k，p And represent that gain, delay and Doppler frequency of the channel of the p-th path of the k-th user respectively. δ () represents Dirac function, Is the system sampling period. Hypothesis τ k，p ＝n τ，p T s ，n τ，p Is an integer. τk，p Distributed in {0, t s ，2T s ，...，15T s } are randomly selected. θ k，p And v k，p Uniformly distributed in [-90, 90] and [-2kHz, 2kHz] respectively.
[0139] In a preferred embodiment of the present invention, the construction of a large-scale MIMO-OTFS channel model includes:
[0140] Step 1: For the k-th user, set the length to L. D N D The data sequence is rearranged into two-dimensional OTFS data blocks in delay-Doppler domain. L among them D = 512 and n D = 128 indicates the number of subcarriers and OFDM symbols, respectively.
[0141] In the preprocessing block, ISFFT is applied to obtain the time-frequency domain data block. among and Is the normalized DFT matrix. Then, yes Do L in each column of D Point inverse DFT to obtain transmission signal block Among them s k，j Represents an OFDM symbol.
[0142] By adding cyclic prefix CP to each OFDM symbol, the invention can obtain one-dimensional transmission signal in time domain. L among them cp = 32 is the CP length. s k Will be in n D Occupy L in T time D Bandwidth of δ, where δ f and t = (L cp +L D )T s They are subcarrier spacing and OFDM symbol period.
[0143] On the base station side, operations symmetrical to the user side, such as rearrangement, CP removal, and L, are continuously applied to the post-processing block. D Point DFT and L D ×N D Operation of dimension SFFT. Related, in the nth r Antenna, two-dimensional received data blocks in delay-Doppler domain can be obtained. (I, j+N) D /2) Elements can be expressed as:
[0144]
[0145] among Is i-i' divided by l D The remainder of, I = 0,1, ..., L. D -1，j＝-N D /2，...，0，...，N D /2-1。 In addition, w i，j The mean is 0 and the variance is. Complex Gaussian noise.
[0146] Step two, using the large-scale MIMO channel model established in step one, transmitting the MIMO channel model (I, j+N) D /2) The element is re-represented as:
[0147]
[0148] among And:
[0149]
[0150]
[0151] The main channel and the secondary channel in the antenna-delay-Doppler domain are respectively represented. In addition, l is the delay domain coordinate of the grid point of the base station receiving signal,
[0152] Step 3, further considering the channel sparseness brought by large-scale antenna array, the invention can perform DFT on the channel according to the antenna dimension to obtain the angle-delay-Doppler domain primary and secondary channel models:
[0153]
[0154]
[0155] among
[0156] definition The received signal model in angle-delay-Doppler domain can be obtained by DFT:
[0157]
[0158] In a preferred embodiment of the present invention, the uplink and downlink CDMA access method in the angle-delay-Doppler domain includes:
[0159] Step 1, channel parameters are estimated on the uplink by the existing algorithm
[0160] Step 2, the 3D channel can be obtained from the large-scale MIMO-OTFS angle-delay-Doppler domain channel model and at There are dominant values, among which:
[0161]
[0162] (i k，p ，j k，p ，q k，p ) is related to the p-th scattering path of the k-th user, and can be considered as the angle-delay-Doppler domain label of this path. Furthermore, the p th scattering path almost contains lattice points (I k，p ，j k，p ，q k，p ) all the energy in the place. Therefore, for the lattice point (I k，p ，j k，p ，q k，p ), which has the following approximation:
[0163]
[0164]
[0165] Accordingly, the following relation can be derived:
[0166]
[0167] Step 3: For the kth user, its angle label set is defined as Users with non-overlapping angles are assigned to the same group. Meet:
[0168]
[0169] among D θ Is the guard interval of the angle domain. right They are allocated the same delay-doppler domain resources, i.e. And different receiving end angle-delay-Doppler domain lattice points, i.e.
[0170] Step 4: For different user groups Allocating them with distinguishable resources in the delay-Doppler domain meets the following restrictions:
[0171]
[0172] among express Guard interval dτ And d ν Used against 3D channels. and Dispersion effect in delay and Doppler domain. Define the maximum dispersion length of delay and Doppler domain:
[0173]
[0174] Typically, d τ And d ν Set as and After scheduling, different users can map their respective data to the scheduled delay-Doppler domain grid points, and at the same time send data to the base station in the same OTFS block, and occupy different 3D resources at the base station. The delay and Doppler domain width of the effective data block of each user are W, respectively. d = 48 and w D ＝50。 Then, the base station can demap and decode the data of different users in parallel without interference between users.
[0175] Step five, sending data block X of the k-th user k The location of the pilot in is represented as and Its observation signal at the base station Can be written as:
[0176]
[0177] among The LS estimator can recover from the above formula. Have to:
[0178]
[0179] Accordingly, the secondary channel:
[0180]
[0181] In order to enhance the performance of data recovery, according to the main channel Approximate representation of and known Estimable Have to:
[0182]
[0183] Through the obtained channel parameter set in the current OTFS block Accurate information, the 3D channels with all lattice points can be reconstructed through the large-scale MIMO-OTFS primary and secondary channel models. and
[0184] Step six, consider that valid data block sent by the user. A data symbol in whose element index is. From the large-scale MIMO-OTFS channel model, we can derive the information about The received information of is in the collection Inside. Then, the data symbols P receiving grid points are listed as a P×1 vector. Have to:
[0185]
[0186] among Is a P×1 noise vector, and the third term on the right side of the formula is the sum of interferences. Use MRC's strategy for data recovery. The p-th path of the k-th user, that is Signal-to-interference-noise ratio of can be expressed as:
[0187]
[0188] According to the principle of MRC, the optimal weighting factor of the p-th path is:
[0189]
[0190] Define the weight vector β k ＝[β k，1 ，...，β k，P ] T , and combine P received signals with their own weighting factors. MRC received signals can be expressed as:
[0191]
[0192] It can be derived that the optimal signal-to-noise ratio of the combined signal is The optimal signal-to-noise ratio is obtained, and the transmitted data can be recovered by using classical algorithms, such as LS estimator. According to the characteristics of LS, the meansquareerror (MSE) of data recovery is
[0193] Step 7: For TDD system, due to the reciprocity of uplink and downlink channels in the angle-delay-Doppler domain, downlink channel parameters Same as the uplink. Furthermore, the uplink path scheduling strategy can be used for downlink multi-user services. Through the same angle grouping as in step 2, the received signal of the kth user mainly comes from the P layer, i.e. Accordingly, the received signal Can be written as:
[0194]
[0195] Where n k Is a noise matrix whose elements obey the mean and variance of 0. Gaussian distribution of.
[0196] Without loss of generality, the downlink effective observation area of the kth user Set the sending area allocated for uplink access. Within the 3D transmission resource space of the base station, the effective transmission area allocated to the kth user is For a given Unlike According to the path scheduling result in step 3, it can be ensured that:
[0197]
[0198] After the path division, the grid point of the kth user Will be observed from , which will respectively experience their own 3D channels. Before operation, give the following formula:
[0199]
[0200] In which P×1 vector b k，l，n Say yes Beamforming operation of observation point (L, N) in the middle, U k，l，n Indicates the valid data expected by the kth user. Thereby obtaining a user receiving signal:
[0201]
[0202] Among them b k，l，n Available MMSE beamforming frameworks are designed as follows:
[0203]
[0204] Note that the parameter Obtained by the reciprocity of uplink and downlink parameters, while Can be obtained by the nearest uplink OTFS block. Further, after beam forming, Equivalent channel with all lattice points in it Same. Therefore, only The equivalent channel can be implemented by sending the pilot of one grid point. Yes, the estimation method is the same as step 4.
[0205] Figure 5 (a) and Figure 5(b) The performance of channel reconstruction at lower speed and higher speed is demonstrated. The curve with triangle is the estimated three-dimensional main channel without considering the secondary channel, and the curve with circle and the curve with fork represent the estimated main channel and secondary channel. It can be seen that the MSE of the main channel without considering the secondary channel is much higher than that of other channels. In addition, in the high mobility scenario, its performance becomes worse, while the other two curves are closer to those in the low mobility scenario. This characteristic can be explained that the influence of sub-channel increases with the increase of Doppler frequency shift. In the case of high mobility, the Doppler frequency shift becomes larger, so the triangular curve becomes worse. For the other two curves, in addition, due to the approximate form of the estimation equation, the MSE of the secondary channel is always higher than that of the primary channel.
[0206] Figure 6 This paper describes the data recovery performance in two mobile situations, in which the sparsity of effective data blocks is considered. As expected, MSE decreases with the increase of signal-to-noise ratio and tends to converge; The latter case is due to interference in the data block. When the signal-to-noise ratio is low, the interference is small compared with the noise. However, in the case of high signal-to-noise ratio, the interference is far greater than the noise, which seriously affects the data recovery performance. The gap between MSE of sparse data block scheduling and MSE of dense data block distribution is worth noting. This is because the farther the distance between the main dispersion receiving grids is, the smaller the interference caused by energy leakage will be.
[0207] Figure 7 It shows the comparison of downlink data recovery performance and signal-to-noise ratio under different channel quality and user speed. It can be seen that the MSE in perfect channel is much lower than that in unmatched channel, while the MSE in unmatched channel is caused by inaccurate 3D channel estimation. In addition, MSE tends to converge at high signal-to-noise ratio, which is due to the following reasons Figure 6 Same.
[0208] The above is only the specific embodiment of the present invention, but the scope of protection of the present invention is not limited to this. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention by anyone familiar with the technical field should be covered in the scope of protection of the present invention.
