Low-Complexity Uplink Time-Varying Interference Cancellation Method in Multi-Cell Cellular Networks

Abstract

A low-complexity uplink time-varying interference cancellation method in a multi-cell cellular network. When in the interference time window, calculate the difference degree of the active state information of all interfered users at the current moment and the previous moment, that is, the proportion of newly added interfering users to the original interfering users. When the difference degree is too high, perform semi-blind interference detection on the newly added interfered users to obtain the interference data signal; then obtain the channel parameters between the target data signal and all interference data signals and the base station through channel estimation, and use the minimum mean square error interference rejection combining algorithm (MMSE-IRC) to improve the accuracy of MIMO equalization and recover the target user data signal, so as to achieve the interference cancellation of the target user data signal. The present invention makes full use of cell information and has a low system calculation complexity.

Description

Technical Field

[0001] The present invention relates to a technology in the field of wireless communication, specifically a low-complexity uplink time-varying interference cancellation method in a multi-cell cellular network. Background Art

[0002] In the existing LTE system, it is quite common that the base station is severely interfered by the edge users of adjacent cells. Through interference cancellation technology, the interference signal is demodulated or even decoded at the receiving end in a manner similar to the reception of the useful signal, and then the demodulated interference signal is subtracted from the received signal, and then the desired signal of itself is decoded from the remaining received signal with less interference. This method can significantly improve the system performance at the cell edge when the signal-to-interference-plus-noise ratio is large. However, in the existing interference cancellation schemes, usually only the interference received at a certain moment is considered, and attempts are made to avoid and eliminate it, and this scheme is repeated at each moment. However, in an actual communication system, the interference often changes with time, and there may be a certain similarity between the interfering users at adjacent moments. Repeatedly detecting interference at each moment will waste a large amount of resources. Summary of the Invention

[0003] When the existing interference coordination technology is used for the interaction of inter-cell information in the present invention, it will increase the signaling overhead and system complexity of the system, and also has higher requirements for delay. Although the blind interference cancellation receiver does not require the help of other adjacent base stations or cells, when generating all possible RS for cross-correlation, due to the huge candidate set, the search complexity will increase. Therefore, a low-complexity uplink time-varying interference cancellation method in a multi-cell cellular network is proposed, which makes full use of cell information and has low system calculation complexity.

[0004] The present invention is realized through the following technical solutions:

[0005] The present invention relates to a low-complexity uplink time-varying interference cancellation method. When in the interference time window, calculate the difference degree of the active state information of all interfered users at the current moment and the previous moment, that is, the proportion of the newly added interfering users to the original interfering users. When the difference degree is too high, perform semi-blind interference detection on the newly added interfered users to obtain the interference data signal; then obtain the channel parameters between the target data signal and all interference data signals and the base station through channel estimation, and use the minimum mean square error interference rejection combining algorithm (MMSE-IRC) to improve the accuracy of MIMO equalization, and recover the target user data signal, thereby realizing the interference cancellation of the target user data signal.

[0006] Preferably, by predicting the future interfering user state information and comparing it with the current moment, new interference is ignored to a certain extent, and only the same interfering users are considered. Therefore, the relevant interference information at the current moment can be directly used at the next moment without performing semi-blind interference detection, thus greatly reducing the computational complexity of the system.

[0007] The present invention relates to a system for implementing the above method, including: a prediction module, a semi-blind interference parameter detection module, and an interference cancellation module, where: the prediction module uses the historical data of all user states to establish an LSTM model for each user, predicts whether the user's state is active in a period of time after prediction, that is, whether it interacts with the base station or other devices, and then selects an appropriate time window T according to the prediction result to cancel the interference received by the target user within the time window; the semi-blind interference parameter detection module performs cross-correlation between the received signal and the candidate set of demodulation reference signals (DMRS) generated by the base station side, and uses the cross-correlation value to judge the existence of strong interference by combining hypothesis testing and successive interference cancellation, and obtains the corresponding strong interference information; the interference cancellation module uses the detected interference information as prior information for channel estimation and MIMO equalization, so as to recover a more accurate target signal and achieve the effect of interference cancellation.

[0008] The said strong interference information includes: the position and length of the resource block, symbol length, type, transmit port, etc.

[0009] The said prediction module includes: a training unit and a prediction unit, where: the training unit divides the historical state data of a single user into a training set and a test set for training the LSTM model. Since the user state is divided into active users and inactive users, it is represented by 1 and 0, so the data does not need to be normalized when training the model and can be directly used. The prediction unit inputs the historical data according to the LSTM model obtained by the training unit and obtains the prediction result.

[0010] The described semi-blind interference parameter detection module includes: an alternative set generation unit, an interference presence detection unit, and a strong interference detection unit, where: The alternative set generation unit generates pilots in each cell according to the pilot generation method in the 3GPP protocol and using the ID number information of the target cell and adjacent cells. Since there are differences in the resource blocks used by each user, all possible pilot sequences are generated according to the size and position of the possible resource blocks used, thereby obtaining an alternative set. The interference presence detection unit calculates the power based on the received pilot signal and compares it with the power when only additive white Gaussian noise (AWGN) exists to determine whether there is interference in the received signal. The strong interference detection unit performs a cross-correlation operation on the received pilot signal with the alternative set and compares it with a set threshold to obtain the pilot sequence used by the strong interference.

[0011] The described interference cancellation module includes: a channel estimation unit and MIMO equalization, where: The channel estimation unit performs least squares (LS) channel estimation based on the strong interference pilot information obtained by the semi-blind interference parameter detection module to obtain the channel parameters between the interfering user and the target base station. The MIMO equalization unit uses the minimum mean square error interference rejection combining (MMSE-IRC) algorithm to recover the data signal sent by the target user according to the channel parameters and the received data signal.

[0012] Technical effects

[0013] In the present invention, for uplink time-varying interference, an LSTM prediction model based on user status is combined with semi-blind interference detection and cancellation, and the similarity of interfering users at adjacent times is used to achieve low-complexity interference cancellation. Compared with the prior art, the present invention reduces the computational complexity of the system while ensuring an ideal mean square error (MSE) level, improves the trade-off level between performance and complexity, and increases the speed of interference cancellation. Description of the drawings

[0014] Figure 1 It is a schematic diagram of the system of the present invention;

[0015] Figure 2 It is a flowchart of the embodiment;

[0016] Figure 3 It is a schematic diagram of the missed detection probability when the signal-to-interference-plus-noise ratio is -5 to 15 dB;

[0017] Figure 4 It is a schematic diagram of the false alarm probability when the signal-to-interference-plus-noise ratio is -5 to 15 dB;

[0018] Figure 5 It is a schematic diagram of the performance comparison when the signal-to-interference-plus-noise ratio is -5 to 15 dB. Detailed implementation manners

[0019] Such asFigure 2 As shown, this embodiment involves a low-complexity uplink time-varying interference elimination method. When located in the interfered time window, the difference in active state information of all interfered users at the current moment and the previous moment is calculated, that is, the proportion of newly added interfering users to the original interfering users. When the difference is too high, semi-blind interference detection is performed on the newly added interfered users to obtain the interference data signal; then, the channel parameters between the target data signal and all the interfering data signals and the base station are obtained through channel estimation, and the minimum mean square error interference suppression combining algorithm (MMSE-IRC) is used to improve the accuracy of MIMO equalization, and the target user data signal is restored, thereby achieving interference elimination of the target user data signal.

[0020] The interference time window is obtained in the following way: the base station side establishes an LSTM model for each user based on the historical data of all user states to predict its active state information, and for the target user to eliminate interference, determines its interference time window T={T i (s i ,l i )|i∈N *}, where: s1 = n1, l1=m1-n1, s i It is T i The starting index of the time window, l i It is T i The length of the time window, ρ0, the state information of the target user, is expanded to ρ0 = [ρ0(1), ρ0(2), …, ρ0(j), …, ρ0(N)], 0 <j,k≤N,j∈Z,k∈Z,第一个时间窗口的计算公式如上所示,第二个的时间窗口计算时j=n1+m1,得到s2=n2,计算l2时k=n2,剩下的计算依此类推。

[0021] The candidate set of pilot DMRS is obtained by the following method: on the base station side, the inherent information of the cell and the adjacent cell, that is, the physical cell number PCI, is used to generate all pseudo-random sequences, and the pseudo-random sequences are used to generate reference signals and mapped to resource blocks. Finally, the candidate set of pilot DMRS is generated based on the reference signals appearing on all possible RBs, specifically: first, a pseudo-random sequence is generated according to 3GPP protocol 38.211: c(n)=(x1(n+N c )+x2(n+N C ))mod2,x1(n+31)=(x1(n+3)+x1(n))mod2,x2(n+31)=(x2(n+3)+x2(n+2)+x2(n+1)+x2(n))mod2, where: Nc =1600, x1(0)=1, x1(n)=0, n=1,2,...,30, x2(n) is given by c init Decided, in: represents the number of symbols in each time slot, When the subcarrier spacing is μ, the number of time slots in each frame is n SCID Given by DCI or 0. The reference signal is generated as follows: At this time, j represents an imaginary number. According to the mapping rule, it is mapped to RB, that is, j=0,1,…,怸-1, where: These parameters are used to determine the mapping position, and the specific values can be obtained by querying the table in the protocol. Finally, the DMRS candidate set Z is generated considering the resource block size and position that the base station may allocate, for example, only allocating the first RB for user use, or allocating the last two RBs to the user.

[0022] The semi-blind interference detection comprises:

[0023] Step 1. Determine whether there is interference: The base station side calculates the average RB power when only AWGN (Additive white Gaussian noise) exists in the received signal, and calculates the average RB power of the received signal after receiving the signal. When the average RB power of the received signal is greater than the average RB power when only AWGN exists, there is an interference signal in the received signal and step 2 is executed, otherwise there is no interference.

[0024] Step 2, determine whether it is strong interference: after cross-correlating the received signal with the candidate set, the maximum value and the threshold, that is, the received signal with only AWGN is cross-correlated with the candidate set and the cross-correlation value obtained when the cumulative distribution function is 99.9% is compared. When it is greater than the threshold, there is strong interference in the received signal and step 3 is executed, specifically: cross-correlate the candidate set with the received signal with only AWGN, and the cross-correlation value obtained is Where: N r is the number of receiving antennas, N RB is the number of occupied RBs, is the number of subcarriers in an RB, is the signal containing only AWGN received by the ath antenna, Z iThe i-th pilot sequence in the set, calculate the probability density function and cumulative distribution function of C0, and use the cross-correlation value when the cumulative distribution function is 99.9% as the threshold. After determining the threshold, when receiving a signal, calculate the cross-correlation value between the received signal and the candidate set Wherein: Is the received pilot signal, compare the maximum value with the threshold to determine whether it is strong interference.

[0025] Step 3, elimination of interfering pilot signals: Perform channel estimation according to the DMRS sequence corresponding to the information of the strong interference obtained, and eliminate it from the pilot signal of the received signal. Specifically, when the pilot sequence of the strong interference obtained in step 2 is Z s , obtain the channel parameters between this interference and the base station through channel estimation Then eliminate it from the received signal to obtain the remaining interfering pilot signal

[0026] Preferably, by repeating steps 1 to 3, multiple interferences in the received signal can be eliminated in order from strong to weak.

[0027] Preferably, when there is no interference in the received signal but the result obtained in step 1 is judged to have interference, update the false alarm probability

[0028] Preferably, when there is interference in the received signal but the result obtained in step 1 is judged to have no interference, update the miss detection probability

[0029] The interference cancellation of the data signal is achieved by the following method: According to the pilot sequence Z obtained by semi-blind interference parameter detection s , perform channel estimation Then perform the MMSE-IRC algorithm, that is, the recovered target data signal is Wherein: Y da Is the received data signal, N I廸E Is the number of interfering users, Is the estimated channel parameter, σ 2 Is the variance of the noise.

[0030] After specific practical experiments, in a cellular network with 7 cells in total, the center is the target cell, the physical cell number is set to 0, and the surrounding 6 adjacent cells are interference cells, with cell numbers set to 1-6 respectively. The cells all use the same frequency, so the edge users of the adjacent cells may cause uplink interference to the base station of the target cell. The base stations in the cells are deployed in the center of the cells, with 8 antennas, and the number of antennas for all users is set to 1. The system bandwidth is set to 20M, and the number of resource blocks is 100. According to the mapping rules of 3GPP, when the configuration type is type 1, there are 4 antenna ports; when the configuration is type 2, there are 6 antenna ports.

[0031] Since the signal received by the target base station can obtain the relevant information of strong interference and the false alarm and missed detection probability through the semi-blind interference detection module, the false alarm and missed detection probability of the experimental test module is designed first, such as Figure 3 As shown in the figure, when the interference-to-noise ratio is -5 to 15dB, the missed detection probability of three interference signals at different power levels is detected. The power of interference signal 1, interference signal 2, and interference signal 3 in the figure is set from large to small, which obviously conforms to the rule that the greater the interference power, the smaller the missed detection probability. Figure 4 As shown in FIG. 1 , it is the false alarm probability at -5 to 15 dB, that is, the probability of detecting interference when there is no interference.

[0032] The present invention is compared with other solutions. The horizontal axis is the signal-to-interference-to-noise ratio, and the vertical axis is the performance-complexity ratio. When the MSE meets the requirements, the lower the system's computational complexity FLOP, the better, that is, the larger the vertical axis, the better. Figure 5 As shown, it is a performance comparison when the signal to interference noise ratio is -5 to 15 dB. The present invention is compared with the interference elimination scheme based on semi-blind detection and the exhaustive search scheme respectively. Experiments prove that the performance of the present invention is more superior.

[0033] Compared with the prior art, the present invention makes full use of the user status information in the cell, performs prediction through the LSTM model, designs an experimental scheme to reduce the computational complexity of the system as much as possible while satisfying the minimum mean square error (MSE) of the recovered target signal, and can obtain better performance under the same SINR.

[0034] The above-mentioned specific implementation can be partially adjusted in different ways by those skilled in the art without departing from the principle and purpose of the present invention. The protection scope of the present invention shall be based on the claims and shall not be limited by the above-mentioned specific implementation. Each implementation scheme within its scope shall be subject to the constraints of the present invention.

Claims

1. A method for uplink time-varying interference cancellation based on low complexity, characterized in that When it is within the interference time window, calculate the difference degree of the active state information of all interfered users at the current moment and the previous moment, that is, the proportion of newly added interfered users to the original interfered users. When the difference degree is too high, perform semi-blind interference detection on the newly added interfered users to obtain interference data signals; then obtain the channel parameters between the target data signal and all interference data signals and the base station through channel estimation, and use the minimum mean square error interference suppression combining algorithm to improve the accuracy of MIMO equalization and recover the target user data signal, thereby achieving interference cancellation of the target user data signal. The disturbed time window is obtained in the following manner: The base station side establishes an LSTM model for each user based on the historical data of all user states to predict their active state information. For the target user to eliminate interference, the time window during which it is disturbed is determined according to the active state information. , where: is the start index of the th time window, is the length of the th time window; for the first time window when calculating , , , the state information of the target user is expanded into , , ; When calculating the time window of the second , obtain , calculate when , and the remaining calculations follow the same pattern.

2. The uplink time-varying interference cancellation method based on low complexity according to claim 1, characterized in that By predicting the future interference user state information and comparing it with the current moment, ignore the newly added interference and only consider the interference users that are the same as the current moment; then directly use the relevant interference information at the current moment in the follow-up without performing semi-blind interference detection, thereby greatly reducing the computational complexity of the system.

3. The uplink time-varying interference cancellation method based on low complexity according to claim 1, characterized in that, The semi-blind interference detection described above includes: Step 1, determine whether there is interference: The base station side calculates the average RB power when there is only AWGN in the received signal, and calculates the average RB power of the received signal after receiving the signal. When the average RB power of the received signal is greater than the average RB power when there is only AWGN, there is an interference signal in the received signal and step 2 is executed, otherwise there is no interference. Step 2: Determine whether there is strong interference. After performing cross-correlation between the received signal and the candidate set of pilot DMRSs, compare the maximum value among them with the threshold, that is, perform cross-correlation calculation between the received signal with only AWGN and the candidate set of pilot DMRSs and compare it with the cross-correlation value when the cumulative distribution function is 99.9%. When it is greater than the threshold, there is strong interference in the received signal and proceed to Step 3. Specifically: Perform cross-correlation between the candidate set of pilot DMRSs and the received signal with only AWGN, and the obtained cross-correlation value is , where: is the number of receiving antennas, is the number of occupied RBs, is the number of subcarriers in one RB, is the signal containing only AWGN received on the th antenna, refers to the th pilot sequence in the set, calculate the probability density function and cumulative distribution function of , and use the cross-correlation value when the cumulative distribution function is 99.9% as the threshold. When a signal is received, calculate the cross-correlation value between the received signal and the candidate set of pilot DMRSs, where: is the received pilot signal, compare the maximum value with the threshold to determine whether there is strong interference; Step 3, elimination of interference pilot signals: Channel estimation is performed according to the DMRS sequence corresponding to the obtained information of strong interference, and it is eliminated from the pilot signals of the received signal. Specifically: When the pilot sequence of strong interference obtained in Step 2 is , the channel parameters between this interference and the base station are obtained through channel estimation , and then it is eliminated from the received signal to obtain the remaining interference pilot signal .

4. The method for eliminating uplink time-varying interference based on low complexity according to claim 3, characterized in that, The candidate set of the pilot DMRS is obtained through the following method: At the base station side, generate all pseudo-random sequences from the inherent information of this cell and adjacent cells, that is, the physical cell identifier PCI. The pseudo-random sequences then generate reference signals and map them to resource blocks. Finally, generate the candidate set of the pilot DMRS according to the fact that the reference signals will appear on all possible RBs.

5. The uplink time-varying interference cancellation method based on low complexity according to claim 3, characterized in that, By repeating step 1 to step 3, multiple interferences in the received signal can be eliminated one by one from strong to weak.

6. The method for eliminating uplink time-varying interference based on low complexity according to claim 3, characterized in that When there is no interference in the received signal but the result obtained in step 1 is judged to have interference, update the false alarm probability ; When there is interference in the received signal and the result obtained in step 1 is judged as no interference, update the probability of missed detection .

7. The uplink time-varying interference cancellation method based on low complexity according to claim 3, characterized in that The interference cancellation of the data signal is achieved by the following method: according to the pilot sequence obtained by detecting the semi-blind interference parameters , channel estimation is performed , and then the MMSE-IRC algorithm is performed, and the recovered target data signal is obtained as , where: is the received data signal, is the number of interfering users, is the estimated channel parameter, is the variance of the noise.

8. A system for implementing the uplink time-varying interference cancellation method based on low complexity according to any one of claims 1-7, characterized in that, It includes: A prediction module, a semi-blind interference parameter detection module, and an interference cancellation module, where: The prediction module uses the historical data of all user states to establish an LSTM model for each user, predicts whether the state of the user will be active in a period of time after prediction, that is, whether it interacts with the base station or other devices, and then selects an appropriate time window T according to the prediction result to eliminate the interference received by the target user within the time window; The semi-blind interference parameter detection module performs cross-correlation between the received signal and the candidate set of the demodulation reference signal generated by the base station side, and uses the method of combining hypothesis testing and successive interference cancellation to judge the existence of strong interference using the cross-correlation value and obtain the corresponding strong interference information; The interference cancellation module uses the detected interference information as prior information for channel estimation and MIMO equalization, thereby recovering a more accurate target signal and achieving the effect of interference cancellation. The strong interference information described above includes: the position and length of the resource block, the symbol length, the type, and the transmission port.

9. The system according to claim 8, characterized in that, The described semi-blind interference parameter detection module includes: an alternative set generation unit, an interference presence detection unit, and a strong interference detection unit, where: The alternative set generation unit uses the ID number information of the target cell and adjacent cells to generate pilots in each cell, and generates all possible pilot sequences according to the size and position of the resource blocks that may be used, so as to obtain an alternative set of pilot DMRS; The interference presence detection unit calculates the power based on the received pilot signal, and compares it with the power when only Gaussian white noise exists to determine whether there is interference in the received signal; The strong interference detection unit performs a cross-correlation operation on the received pilot signal with the alternative set of pilot DMRS, and compares it with the set threshold to obtain the pilot sequence used by the strong interference.

Images