Channel Time-Varying Statistics-Assisted Alternating Training Method

Abstract

The present invention provides a channel time-varying statistics-assisted alternating training method, including system modeling, alternating training, redundant training, data transmission, and antenna training order setting phases. In the system modeling phase, a system transmission model and a channel time-varying model are constructed. In the alternating training phase, the training and feedback phases are interconnected. In the redundant training phase, when the system already meets the condition of no outage, the antennas are continuously trained. In the antenna training order setting phase, the order of antenna training at the next moment is assisted in setting. Compared with the existing alternating training scheme, the channel time-varying statistics-assisted alternating training scheme further reduces the training overhead of the system while not deteriorating the outage performance of the system.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communication, and particularly relates to an alternating training method assisted by channel time-varying statistics. Background Art

[0002] By configuring high-dimensional antennas at the base station and the user terminal, the data transmission rate, channel capacity, and spectrum efficiency can be significantly improved. Due to the further increase in the demand for data transmission rate in mobile communication, the current fifth-generation mobile communication system (5G) often configures a large-scale antenna array at the base station.

[0003] For a multiple-input single-output (MISO) system, since the number of antennas configured at the base station end has increased significantly, the number of channels to be estimated is also considerable. Under such realistic conditions, using traditional channel estimation schemes has the problem of a sharp increase in training overhead caused by the sharp increase in channel dimensions. How to reduce the huge training overhead and feedback overhead brought about by the significant increase in the number of base station antennas while ensuring that the system can obtain accurate channel state information (CSI) has become an urgent problem to be solved.

[0004] In some scenarios, the base station only needs to provide a transmission rate that meets the user's requirements, that is, the system takes the outage probability as the main index. At this time, a fixed-length training scheme aiming to obtain complete channel state information as much as possible may have training redundancy.

[0005] Different from traditional training schemes, the alternating training scheme connects the training and feedback processes. The base station can immediately obtain feedback information from the user terminal. When the trained channels are sufficient to prevent the system from experiencing an outage, the training process immediately terminates. By immediately judging whether the trained channels meet the condition that the system does not experience an outage, the adaptive adjustment of the training length is realized, and the average training length of the system is reduced.

[0006] However, existing alternating training schemes only consider independently and identically distributed (IID) channels, and when the trained antennas already meet the condition of not experiencing an outage, the system immediately stops training. In actual application scenarios, there is a certain correlation between channel coefficients at adjacent times. At the same time, the existence of redundant training can provide more auxiliary information for setting the channel training order at the next time. Considering the time correlation of channels and adding redundant training during the system training process, there is still room for further improvement in the training overhead performance of the system. Summary of the Invention

[0007] The object of the present invention is to provide a channel time-varying statistics-assisted alternating training method, which further explores the time correlation of the channel on the basis of the existing alternating training scheme and supplements it with redundant training to solve the technical problem of the sharp increase in training overhead caused by the sharp increase in the channel dimension in a large-scale MISO system.

[0008] To solve the above technical problems, the specific technical solution of the present invention is as follows:

[0009] A channel time-varying statistics-assisted alternating training method includes the following steps:

[0010] Step 1, system modeling stage: construct a system transmission model, model the time-varying characteristics of the channel, set the basis for system outage judgment, and define and mathematically represent various parameters;

[0011] Step 2, alternating training stage: the base station sends pilots to the user to train the antennas, the user estimates the channel state information (CSI) according to the pilots sent by the base station, judges whether the system meets the condition of not experiencing an outage, feeds back indication information, and the base station decides whether to interrupt the alternating training process based on the fed-back indication information;

[0012] Step 3, redundant training stage: when the system already meets the condition of not experiencing an outage, the base station sends pilots to the user to perform redundant training on the antennas, the user estimates the channel state information CSI according to the pilots sent by the base station, and when the number of antennas to be continuously trained meets the required number of redundant training times or there are no more antennas to be trained, the training terminates, and the user feeds back the channel state information CSI of all the antennas that have been trained currently;

[0013] Step 4, data transmission stage: the base station performs data transmission according to the channel state information CSI fed back by the user;

[0014] Step 5, antenna training order setting stage: set the order of antenna training for the next moment according to the results of antenna training at the current moment.

[0015] Further, in step 1, the system modeling stage includes the following specific steps:

[0016] Step 1.1, consider a MISO system, that is, the downlink of a large-scale system serving a single-antenna user, with N t transmit antennas configured at the base station end, each antenna equipped with a separate radio frequency link, the base station configured with a typical uniform array of high dimensions, and use h i to represent the channel coefficient between the i-th transmit antenna and the receive antenna, and use Denote the channel state of the entire MISO system under consideration. The transceiver equation is as follows:

[0017]

[0018] where \(P\) is the instantaneous total transmit power, which is also the normalized signal-to-noise ratio of the transmitted signal; is the beamforming vector transmitted by the base station; \(s\) represents the data stream with unit energy; is the local receiver noise;

[0019] Step 1.2: Model the time-varying characteristics of the channel. Use the first-order autoregressive model AR(1) to model the time-varying channel. The specific channel model is as follows:

[0020]

[0021] where \(n\) (t+1) represents the Gaussian noise introduced at time \(t + 1\) that is uncorrelated with \(h\) (t) , i.e., \(n\) (t+1) is statistically independent of \(h\) (t) , and \(h\) (0) follows a complex Gaussian distribution. \(\rho\) is the time correlation coefficient, which is used to characterize the strength of the correlation between channels at adjacent times; the correlation is only considered within the time slot range \(T\) l . When \(i\leq T\) l , \(h\) (t+i) is correlated with \(h\) (t) ; when \(i > T\) l , \(h\) (t+i) is independent of \(h\) (t) ;

[0022] Step 1.3: Set the system outage judgment criterion: Given the desired data transmission rate \(R\) th =\(\log_2(1 + \alpha P)\), where \(\alpha\) takes any value greater than 0. The signal-to-noise ratio (SNR) of the system is \(P||h\) T w|| 2 / n 2 , i.e., \(P||h\) T w|| 2 . Then the actual transmission rate of the system is \(\log_2(1 + ||h\) T w|| 2 P); if \(\log_2(1 + ||h\) T w|| 2 P)\(\leq R\) th , i.e., an outage will occur. Here, \(\alpha\) is the target normalized received SNR, and the outage probability is:

[0023]

[0024] If the base station knows the perfect equivalent channel vector, in order to minimize the outage probability, maximal ratio transmit (MRT) is used to design the baseband precoding: w = h * / ||h||; h * denotes the conjugate of h; the outage probability is further transformed into:

[0025] Pr(||h|| 2 ≤α),

[0026] The system outage judgment criterion is set as: If the received signal-to-noise ratio SNR provided by the antennas that have completed training is less than or equal to the target normalized received SNR, that is, ∑||h i || 2 ≤α, the system will experience an outage; otherwise, the system will not experience an outage.

[0027] Furthermore, step 2, the alternating training phase, includes a training phase and a feedback phase. Before the conditions for terminating training are met, the training steps and feedback steps are alternated, including the following specific steps:

[0028] Step 2.1: The base station sends pilots to the user to train the antennas.

[0029] Step 2.2: The user estimates the channel state information CSI based on the pilots sent by the base station and calculates the received SNR provided by the antennas that have completed training currently, and compares it with the target normalized received SNR. If the condition for the system not to experience an outage ∑||h i || 2 >α is satisfied, the user feeds back the indication information 1 and requests the base station to continue with δ redundant trainings; if the condition for the system not to experience an outage ∑||h i || 2 >α is not satisfied, the user feeds back the indication information 0 to request the base station to continue training the next antenna;

[0030] Step 2.3: The base station receives the information fed back by the user. If the indication information 1 is obtained and δ = 0, the user feeds back the channel state information CSI of all the antennas that have been trained currently and enters the data transmission phase; if the indication information 1 is obtained and δ>0, it enters the redundant training phase; if the indication information 0 is obtained and there are still antennas available for training, it goes to step 2.1; if the indication information 0 is obtained and there are no more antennas available for training, the user feeds back the channel state information CSI of all the antennas that have been trained currently and enters the antenna training order setting phase.

[0031] Furthermore, step 3, the redundant training phase, includes the following specific steps:

[0032] Step 3.1: The base station sends a pilot signal to the user to train the antenna;

[0033] Step 3.2: The user estimates the channel state information (CSI) based on the pilot signal sent by the base station. If the system has not completed the required number of redundant training times and there are still trainable antennas, go to step 3.1. If the system has completed the required number of redundant training times or there are no more trainable antennas, go to step 3.3.

[0034] Step 3.3: The user feeds back the channel state information (CSI) of all currently trained antennas.

[0035] Furthermore, step 5, antenna training sequence setting phase, includes the following specific steps:

[0036] Step 5.1: Sort the channel state information (CSI) of all antennas that have been trained at the current moment in descending order according to their modulus values, and update the antenna sequence number.

[0037] The present invention proposes an alternating training method assisted by channel time-varying statistics, which has the following advantages: the present invention proposes an alternating training scheme assisted by channel time-varying statistics, which further explores the time correlation of the channel on the basis of the existing alternating training scheme, utilizes the channel time-varying statistics to assist in the setting of the antenna training sequence, and considers the assistance of redundant training, thereby further reducing the system training overhead without losing the system interruption performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flow chart of an embodiment of the present invention;

[0039] Figure 2 An alternating training flow chart for use in alternating training phases and redundant training phases;

[0040] FIG3 is a comparison diagram of the average training length of the existing alternating scheme when considering IID channels and time-varying channels;

[0041] Figure 3(a) shows R th =3 bps / Hz, a comparison of the average training length of the existing alternating schemes considering IID channels and time-varying channels;

[0042] Figure 3(b) shows R th =6bps / Hz, a comparison of the average training length of the existing alternating schemes considering IID channels and time-varying channels;

[0043] FIG4 is a comparison diagram of the average training length of the alternating training scheme assisted by channel time-varying statistics and the existing alternating training scheme;

[0044] Figure 4(a) shows R thComparison chart of the average training lengths between the channel time-varying statistics-aided alternating training scheme and the existing alternating training scheme when R = 3 bps / Hz;

[0045] Figure 4(b) shows that for R th Comparison chart of the average training lengths between the channel time-varying statistics-aided alternating training scheme and the existing alternating training scheme when R = 6 bps / Hz. Detailed implementation manners

[0046] To better understand the purpose, structure, and function of the present invention, the following further describes in detail a channel time-varying statistics-aided alternating training method proposed by the present invention in conjunction with the accompanying drawings.

[0047] The following is the channel time-varying statistics-aided alternating training scheme proposed by the present invention. The specific steps for implementing the entire process are as follows:

[0048] Step 1: System modeling: Construct a system transmission model, model the time-varying characteristics of the channel, set the system outage judgment basis, and define and mathematically represent various parameters.

[0049] The system modeling stage includes the following specific steps:

[0050] Step S101: Consider a MISO system, that is, the downlink of a large-scale system serving a single-antenna user. The base station is configured with N t transmit antennas, and each antenna is equipped with a separate radio frequency link. The base station is configured with a typical uniform array of high dimensions. Let h i represent the channel coefficient between the i-th transmit antenna and the receive antenna, and use to represent the channel state of the entire MISO system under consideration. The transceiver equation is:

[0051]

[0052] where P = 0 dB is the instantaneous total transmit power, that is, the normalized signal-to-noise ratio of the transmitted signal. is the beamforming vector transmitted by the base station. s represents the data stream with unit energy. is the local receiver noise.

[0053] Step S102: Model the time-varying characteristics of the channel. Use the first-order autoregressive model AR(1) to model the time-varying channel. The specific channel modeling is as follows:

[0054]

[0055] where n (t+1) represents the introduced at time t + 1 and is related to h (t)Uncorrelated Gaussian noise, i.e., n (t+1) is statistically independent of h (t) and h (0) follows a complex Gaussian distribution. ρ is the time correlation coefficient, with values in the range [0, 0.4, 0.8, 0.99], used to characterize the strength of the correlation between channels at adjacent times. The correlation is only considered within the time slot range T l = 32. When i ≤ T l there is a correlation between h (t +i) and h (t) ; when i > T l h (t+i) is independent of h (t) .

[0056] Step S103: Set the system interruption judgment criterion: Given the expected data transmission rate R th = log2(1 + αP), where α can take any value greater than 0. The SNR of the system is P||h T w|| 2 / n 2 , i.e.: P||h T w|| 2 . Then the actual transmission rate of the system is log2(1 + ||h T w|| 2 P). If log2(1 + ||h T w|| 2 P) ≤ R th , i.e.: an interruption will occur, where α is the target normalized received SNR, and the interruption probability is:

[0057]

[0058] If the base station knows the perfect equivalent channel vector, then to minimize the interruption probability, MRT is used to design the baseband precoding: w = h * / ||h||; the interruption probability is further transformed into:

[0059] Pr(||h|| 2 ≤ α),

[0060] The system interruption judgment criterion is set as: If the received SNR provided by the trained antennas is less than or equal to the target normalized received SNR, i.e., ∑||h i || 2 ≤ α, then the system will experience an interruption. Otherwise, the system does not experience an interruption.

[0061] Step 2: In the alternating training phase: The base station sends pilots to the user to train the antennas. The user estimates the CSI based on the pilots sent by the base station, judges whether the system meets the condition of no outage, and feeds back indication information. The base station decides whether to interrupt the alternating training process based on the fed-back indication information.

[0062] The alternating training phase includes the following specific steps:

[0063] Step S201: The base station sends pilots to the user to train the antennas.

[0064] Step S202: The user estimates the CSI based on the pilots sent by the base station and calculates the received SNR that the currently trained antennas can provide, and compares it with the target normalized received SNR. If the condition for the system not to have an outage is met, i.e., ∑||h i || 2 >α, the user feeds back indication information 1, requesting the base station to continue with δ redundant trainings. If the condition for the system not to have an outage is not met, i.e., ∑||h i || 2 >α, the user feeds back indication information 0 to request the base station to continue training the next antenna.

[0065] Step S203: The base station receives the information fed back by the user. If indication information 1 is obtained and δ = 0, the user feeds back the channel state information CSI of all the antennas that have been trained currently, and transfers to the data transmission phase; if indication information 1 is obtained and δ>0, transfers to the redundant training phase; if indication information 0 is obtained and there are still antennas available for training, transfers to step S201; if indication information 0 is obtained and there are no more antennas to train, the user feeds back the CSI of all the antennas that have been trained currently, and transfers to the antenna training order setting phase.

[0066] Step 3: In the redundant training phase: When the system already meets the condition of no outage, the base station sends pilots to the user to perform redundant training on the antennas. The user estimates the channel state information CSI based on the pilots sent by the base station. When the number of antennas to be continuously trained meets the required number of redundant trainings or there are no more antennas to train, the training terminates, and the user feeds back the CSI of all the antennas that have been trained currently.

[0067] The redundant training phase includes the following specific steps:

[0068] Step S301: The base station sends pilots to the user to train the antennas.

[0069] Step S302: The user estimates the CSI based on the pilots sent by the base station. If the system has not completed the required number of redundant trainings and there are still antennas available for training, transfers to step S301. If the system has completed the required number of redundant trainings or there are no more antennas to train, transfers to step S303.

[0070] Step S303: The user feeds back the CSI of all the antennas that have been trained currently.

[0071] Step 4: In the data transmission phase: The base station performs data transmission according to the CSI fed back by the user.

[0072] Step 5: In the antenna training order setting phase: Assist in setting the order of antenna training for the next moment according to the results of antenna training at the current moment.

[0073] The antenna training order setting phase includes the following specific steps:

[0074] Step S501: Sort the channel state information CSI of all the antennas that have been trained at the current moment in descending order according to their modulus values, and update the antenna numbers.

[0075] The specific steps for verifying the effect of the alternating training scheme assisted by channel time-varying statistics include:

[0076] Step S601: When R th is 3 bps / Hz and 6 bps / Hz, use the existing alternating training scheme to train the IID channel and the time-varying channel respectively, and obtain the average training length in the corresponding cases. When R th is 3 bps / Hz, the value range of the number N of base station antennas t is: [10, 20, 30, 40, 50, 60, 70, 80, 90, 100]; when R th is 6 bps / Hz, the value range of the number N of base station antennas t is: [40, 60, 80, 100, 120, 140, 160, 180, 200].

[0077] Step S602: When R th is 3 bps / Hz and 6 bps / Hz, use the basic optimization scheme with δ = 0 and the alternating training scheme that jointly utilizes training redundancy and time correlation with δ = 1 (considering adding a redundant training once at each moment) to train the antennas respectively, and obtain the average training length in the corresponding cases. The value range of the number of base station antennas is the same as that in Step S601. [[ID=z6]]

[0078] Experimental results: As shown in Figure 3, for the channel considering time-varying characteristics, the training results of the existing alternating training scheme are the same as those of its training results considering the IID channel. Therefore, in this example, the simulation analysis results of the existing alternating scheme considering the IID channel are used to replace the simulation analysis results of the existing alternating scheme that is more complex and should have been used as the control standard but considers time correlation without using channel time-varying statistics to assist in training order setting, so as to achieve the purpose of simplifying the problem.

[0079] As shown in Figure 4, taking the existing alternating scheme considering the IID channel as a control method, this example compares the average training lengths of the system under different methods for different values of temporal correlation.

[0080] It can be seen that:

[0081] (1) Compared with the existing alternating scheme, the average training length of the basic optimization scheme with δ = 0 has been significantly reduced, and the greater the correlation, the more obvious this performance optimization is;

[0082] (2) Whether the alternating training scheme that jointly utilizes training redundancy and temporal correlation with δ = 1 and adds redundant training once at each moment can achieve further optimization is related to the magnitude of the correlation and the value of the desired data transmission rate R th of. The greater the correlation and the greater the value of R th , the more likely the alternating training scheme that jointly utilizes training redundancy and temporal correlation is to achieve further optimization.

[0083] It can be understood that the present invention is described through some embodiments. Those skilled in the art know that without departing from the spirit and scope of the present invention, various changes or equivalent replacements can be made to these features and embodiments. Additionally, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application belong to the scope protected by the present invention.

Claims

1. A channel time-varying statistics-assisted alternating training method, characterized in that It includes the following steps: Step 1, System Modeling Phase: Build a system transmission model, model the time-varying characteristics of the channel, set the basis for system outage judgment, and define and mathematically represent various parameters; Step 2, Alternating Training Phase: The base station sends pilots to the user to train the antennas. The user estimates the channel state information CSI based on the pilots sent by the base station, judges whether the system meets the condition of no outage, and feeds back indication information. The base station decides whether to interrupt the alternating training process based on the fed-back indication information; Step 3, Redundant Training Phase: When the system already meets the condition of no outage, the base station sends pilots to the user to perform redundant training on the antennas. The user estimates the channel state information CSI based on the pilots sent by the base station. When the number of antennas to be continuously trained meets the required number of redundant training times or there are no more antennas to be trained, the training terminates, and the user feeds back the channel state information CSI of all the antennas that have been trained currently; Step 4, Data Transmission Phase: The base station performs data transmission based on the channel state information CSI fed back by the user; Step 5, Antenna Training Order Setting Phase: Set the order of antenna training for the next moment according to the results of antenna training at the current moment; The system modeling phase described in Step 1 includes the following specific steps: Step 1.1: Consider a multiple-input single-output (MISO) system, i.e., the downlink of a large-scale system serving a single-antenna user. The base station is configured with N t transmit antennas, and each antenna is equipped with a separate radio frequency link. The base station is configured with a typical uniform array of high dimensions. Let h i represent the channel coefficient between the i-th transmit antenna and the receive antenna, and let represent the channel state of the entire MISO system under consideration. The transceiver equation is: where P is the instantaneous total transmission power, which is also the normalized signal-to-noise ratio of the transmitted signal; is the beamforming vector transmitted by the base station; s represents a data stream with unit energy; is the local receiver noise; Step 1.2, Model the time-varying characteristics of the channel. Use a first-order autoregressive model to model the time-varying channel. The specific channel modeling is as follows: where n (t+1) represents the Gaussian noise introduced at time t + 1 that is uncorrelated with h (t) , i.e., n (t+1) is statistically independent of h (t) , and h (0) follows a complex Gaussian distribution, and ρ is the time correlation coefficient used to characterize the strength of the correlation between channels at adjacent times; the correlation is only considered within the time slot range T l . When i ≤ T l , h (t+i) is correlated with h (t) ; when i > T l , h (t+i) is independent of h (t) . Step 1.

3. Set the system interruption judgment basis: Given the expected data transmission rate R to be achieved th = log2(1 + αP), where α takes any value greater than 0, and the signal-to-noise ratio SNR of the system is P||h T w|| 2 / n 2 , that is: P||h T w|| 2 , then the actual transmission rate of the system is log2(1 + ||h T w|| 2 P); If log2(1 + ||h T w|| 2 P) ≤ R th , that is: An interruption will occur, where α is the target normalized received signal-to-noise ratio SNR, and the interruption probability is: If the base station knows the perfect equivalent channel vector, to minimize the outage probability, the maximum ratio transmission (MRT) is used to design the baseband precoding: w = h * / ||h||; The outage probability is further transformed into: Pr(||h|| 2 ≤α), The system interruption judgment basis is set as: If the received signal-to-noise ratio SNR that the antenna that has completed training can provide is less than or equal to the target normalized received signal-to-noise ratio SNR, that is, ∑‖h i ‖ 2 ≤α, then the system will experience an interruption; otherwise, the system will not experience an interruption.

2. The channel time-varying statistics-assisted alternating training method according to claim 1, wherein The alternating training phase described in Step 2 includes a training phase and a feedback phase. Before the condition for terminating training is met, the training steps and the feedback steps are alternated, and it includes the following specific steps: Step 2.1, The base station sends pilots to the user to train the antennas; Step 2.2: The user estimates the channel state information CSI based on the pilot sent by the base station, calculates the received signal-to-noise ratio SNR that the currently trained antennas can provide, and compares it with the target normalized received signal-to-noise ratio SNR. If the condition for no system outage ∑‖h i ‖ 2 >α is satisfied, the user feeds back indication information 1 and requests the base station to continue with δ redundant trainings; if the condition for no system outage ∑‖h i ‖ 2 >α is not satisfied, the user feeds back indication information 0 to request the base station to continue training the next antenna; Step 2.3, The base station receives the information fed back by the user. If indication information 1 is obtained and δ = 0, the user feeds back the channel state information CSI of all the antennas that have been trained currently and transfers to the data transmission phase; If indication information 1 is obtained and δ > 0, transfer to the redundant training phase; if indication information 0 is obtained and there are still antennas available for training, transfer to Step 2.1; If indication information 0 is obtained and there are no more antennas to be trained, the user feeds back the channel state information CSI of all the antennas that have been trained currently and transfers to the antenna training order setting phase.

3. The channel time-varying statistics-assisted alternating training method according to claim 2, characterized in that The redundant training phase described in Step 3 includes the following specific steps: Step 3.1, The base station sends pilots to the user to train the antennas; Step 3.2, The user estimates the channel state information CSI based on the pilots sent by the base station. If the system has not completed the required number of redundant training times and there are still antennas available for training, transfer to Step 3.

1. If the system has completed the required number of redundant training times or there are no more antennas to be trained, transfer to Step 3.3; Step 3.3, The user feeds back the channel state information CSI of all the antennas that have been trained currently.

4. The channel time-varying statistics-assisted alternating training method according to claim 3, wherein The antenna training order setting phase described in Step 5 includes the following specific steps: Step 5.1, Sort the channel state information CSI of all the antennas that have been trained at the current moment in descending order according to their modulus values and update the antenna numbers.

Images