A secure beam alignment algorithm based on privacy-preserving question-answering learning

By using a secure beam alignment algorithm with privacy question-and-answer learning in the MIMO communication system, combined with sortPM and channel encoding algorithm, the problems of large calculation overhead and privacy leakage of traditional beam alignment algorithms are solved, and the rapid and fast between the base station and the legal user is achieved. Safe beam alignment.

CN117134811BActive Publication Date: 2024-11-05BEIHANG UNIV
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202311395834.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-10-26
Publication Date
2024-11-05
Estimated Expiration
2043-10-26

AI Technical Summary

Technical Problem

In large-scale MIMO communication systems, the traditional digital beam alignment algorithm based on channel estimation is too overhead to apply, and the adaptive beam alignment algorithm has the risk of privacy leakage.

Method used

A secure beam alignment algorithm based on privacy question-and-answer learning is proposed. By dividing the base station communication space into multiple subspaces of equal size, combining the sortPM adaptive beam alignment algorithm and the channel-encoded non-adaptive beam alignment algorithm, the implementation is achieved Fast and accurate beam alignment between the base station and the legitimate user, and accurate results cannot be heard by the eavesdropper.

Benefits of technology

It realizes fast and accurate beam alignment between the base station and the legal user, prevents eavesdroppers from obtaining accurate beam alignment results and ensuring communication security.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117134811B_ABST
    Figure CN117134811B_ABST
Patent Text Reader

Abstract

The present invention discloses a secure beam alignment algorithm based on privacy question-answering learning, comprising: step S1, dividing the base station communication space into a plurality of subspaces of equal size, making the beam alignment problem equivalent to searching for random variables on a unit interval divided into a plurality of subintervals, and executing an adaptive beam alignment algorithm based on sortPM and obtaining a primary subspace where a legitimate user is located; step S2, executing a non-adaptive beam alignment algorithm based on channel coding on the primary subspace, obtaining a secondary subspace where a legitimate user is located; step S3, synchronously executing an adaptive beam alignment algorithm based on sortPM on all secondary subspaces, obtaining a tertiary subspace where a legitimate user is located, and taking the midpoint of the tertiary subspace as the location of the legitimate user. The present invention can realize fast and accurate beam alignment between a base station and a legitimate user, and an eavesdropper cannot obtain an accurate beam alignment result by eavesdropping.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of communication security technology, and specifically relates to a secure beam alignment algorithm based on privacy question-answering learning. Background Art

[0002] Beamforming technology is a type of Space Division Multiple Access (SDMA) technology. This technology "guides" the antenna array beam into a directional beam with high gain in the desired direction rather than low gain or even no gain in the desired direction by weighted summing of the signals of each array element. In actual large-scale multiple input multiple output (MIMO) communication systems, the channel environment is very complex and changeable. The traditional digital beam alignment algorithm based on channel estimation has too high computational overhead and is difficult to apply. Therefore, large-scale MIMO communication systems mainly use analog beamforming technology. In analog beamforming technology, the process of finding the desired direction to obtain the maximum beamforming gain is called beam alignment, which can also be called beam training or beam search. This process is the first step of analog beamforming technology and is also a very critical step.

[0003] Among them, the adaptive beam alignment algorithm usually has adaptive gain and can achieve better beam alignment performance with fewer beam searches. However, if the beam alignment process is eavesdropped by illegal users, then compared with the non-adaptive beam alignment algorithm, the adaptive beam alignment algorithm has the risk of privacy leakage. Summary of the invention

[0004] In view of this, the purpose of the present invention is to propose a secure beam alignment algorithm based on privacy question-answering learning, which can achieve fast and accurate beam alignment between the base station and the legitimate user, and the eavesdropper cannot obtain the accurate beam alignment result through eavesdropping.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] The present invention provides a secure beam alignment algorithm based on privacy question-answering learning, including:

[0007] Step S1, dividing the base station communication space into multiple subspaces of equal size, so that the beam alignment problem is equivalent to searching for random variables on a unit interval divided into multiple subintervals, and executing an adaptive beam alignment algorithm based on sortPM to obtain the primary subspace where the legitimate user is located;

[0008] Step S2: Execute a non-adaptive beam alignment algorithm based on channel coding on the primary subspace to obtain a secondary subspace where the legitimate user is located;

[0009] Step S3: synchronously execute the sortPM-based adaptive beam alignment algorithm on all secondary subspaces to obtain the tertiary subspace where the legitimate user is located, and take the midpoint of the tertiary subspace as the location of the legitimate user.

[0010] Preferably, the adaptive beam alignment algorithm based on sortPM in step S1 includes:

[0011] Enter the target location , resolution , error probability And the signal power judgment threshold , the error probability is subject to the parameter Bernoulli distribution of ;

[0012] The space Divide into equal-sized non-overlapping subspace, set Indicates from 1 to For all integers, , No. The probability of a random variable existing in a subspace is initialized to ;

[0013] The target position will be estimated Assigned to , the total number of beam searches Assign a value of 0;

[0014] Initialize the probability of the existence of the target in each subspace ,in:

[0015]

[0016] Enter The adaptive beam search works and determines the Secondary adaptive beam search vector:

[0017]

[0018]

[0019]

[0020] in, Indicates The probability of random variables existing in each subspace sorted in descending order during the search and , Represents the probability of the existence of a random variable in each subinterval Sort in descending order, Indicates the first The sum of probabilities, Before The sum of the probabilities is closest ;

[0021] Select The front The subspace corresponding to the probability is taken as the The search space for the noise search is The binary beam search vector , among which Elements 1 means the The search space contains subspace, 0 means The search space does not contain subspaces;

[0022] The binary beam search vector is expressed as:

[0023]

[0024] in, It represents an indicator function, that is, it takes 1 when the condition in the brackets is true and takes 0 when the condition is false;

[0025] According to the binary beam search vector Determine the beam search area:

[0026]

[0027] Calculate the signal power based on the signal received by the base station ;

[0028] like Then determine the The result of the secondary beam search ,otherwise ;

[0029] Update the posterior probability corresponding to each subspace:

[0030]

[0031]

[0032] Determine the number of the subspace corresponding to the maximum a posteriori probability:

[0033]

[0034] like The estimated target position is returned and the serial number of the subspace in which it is located , and make ;

[0035] like Then And execute the adaptive beam search work for the tth time in a loop.

[0036] Preferably, step S1 comprises:

[0037] Will The size of the space is divided into first-order subspaces of equal size , which makes the beam alignment problem equivalent to being divided into Unit interval of subintervals Search for random variables on ;

[0038] According to the adaptive beam alignment algorithm based on sortPM, the resolution , Error Probability ;

[0039] Calculate the signal power based on the signal received by the base station and update the posterior probability corresponding to each subspace;

[0040] The subspace corresponding to the maximum posterior probability is determined to be the first-level subspace where the legitimate user is located.

[0041] Preferably, step S2 comprises:

[0042] Set the The first-level subspace where the legitimate users are located after the adaptive beam search ;

[0043] The first-level subspace Divide Secondary subspaces of equal size ;

[0044] In the first subspace A non-adaptive beam alignment algorithm based on channel coding is performed on After a non-adaptive beam search, the secondary subspace where the legitimate user is located is determined ;

[0045] If you pass After the first non-adaptive beam search, if the secondary subspace where the legitimate user is located cannot be further determined, the non-adaptive beam search is terminated, and it is determined that the beam alignment has failed.

[0046] Preferably, step S3 includes:

[0047] Sure Secondary subspaces of equal size ;

[0048] Perform resolution synchronously on all secondary subspaces , error probability Adaptive beam alignment algorithm based on sortPM;

[0049] When it is determined that the secondary subspace where the legitimate user is located has a tertiary subspace The posterior probability is greater than When , the adaptive beam search is terminated and the number of adaptive beam searches is recorded ;

[0050] Determine the third-level subspace This is the subspace where the legitimate user exists, and the midpoint of the third-level subspace is taken as the estimate of the legitimate user's direction .

[0051] The present invention has achieved at least the following beneficial effects:

[0052] 1. It can realize fast and accurate beam alignment between the base station and the legitimate user, and the eavesdropper cannot obtain the accurate beam alignment result through eavesdropping.

[0053] 2. The beam alignment problem is made equivalent to searching for random variables on a unit interval divided into multiple subintervals, and the subinterval where the legitimate user is located is quickly located through the above adaptive beam alignment algorithm based on sortPM.

[0054] Other advantages, objectives and features of the present invention will be described in the following description and will be apparent to those skilled in the art to some extent, or those skilled in the art may be taught from the practice of the present invention. The objectives and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to make the purpose, technical solution and beneficial effects of the present invention clearer, the present invention provides the following drawings for illustration:

[0056] Figure 1 This is a flowchart of the steps of a secure beam alignment algorithm based on privacy question-answering learning in an embodiment of the present invention;

[0057] Figure 2 This is a pseudo code for implementing a secure beam alignment algorithm based on privacy question-answering learning in an embodiment of the present invention;

[0058] Figure 3 is a flow chart of a non-adaptive beam alignment algorithm in an embodiment of the present invention;

[0059] Figure 4is a flow chart of an adaptive beam alignment algorithm in an embodiment of the present invention;

[0060] Figure 5 This is a schematic diagram of privacy leakage of an adaptive algorithm in an embodiment of the present invention;

[0061] Figure 6 Schematic diagram of a communication scenario among a base station, a legitimate user and an eavesdropper in an embodiment of the present invention;

[0062] Figure 7 A schematic diagram of a uniform planar array antenna according to an embodiment of the present invention;

[0063] Figure 8 Schematic diagram of antenna direction angle and elevation angle in an embodiment of the present invention;

[0064] Fig. 9 Schematic diagram of a binary symmetric channel in an embodiment of the present invention;

[0065] Fig.10 A schematic diagram of a beam search process in an embodiment of the present invention;

[0066] Fig.11 is a pseudo code for implementing the adaptive beam alignment algorithm based on sortPM in an embodiment of the present invention;

[0067] Fig.12 is a pseudo code for implementing a non-adaptive beam alignment algorithm based on channel coding in an embodiment of the present invention;

[0068] Fig.13 This is a simulation diagram comparing the performance of three beam alignment algorithms in an embodiment of the present invention;

[0069] Fig.14 This is a simulation diagram comparing the estimation of the legitimate user's location by the base station and the eavesdropper in an embodiment of the present invention. DETAILED DESCRIPTION

[0070] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0071] Considering that the actual communication environment is complex and changeable, and the beam alignment process is prone to deviations, the present invention mainly studies robust adaptive beam alignment algorithms. Traditional robust adaptive beam alignment algorithms are mainly divided into related algorithms based on steering vector correction and related algorithms based on interference plus noise covariance matrix reconstruction. However, in actual scenarios, it is impossible to accurately judge the factors causing the deviation, and it is impossible to rule out the possibility of multiple factors acting together, so the applicability of most robust adaptive beam alignment algorithms is relatively narrow. Therefore, the present invention considers combining the classic sortPM noise search algorithm in question-and-answer learning to achieve a fast and accurate adaptive beam alignment algorithm with strong applicability.

[0072] The question-answering learning process involves two parties: the questioner and the answerer. The answerer selects a random variable on a unit interval. The questioner asks the answerer as few binary yes-no questions as possible and obtains a binary yes-no answer. It is worth noting that the binary yes-no answer received by the questioner from the answerer may be wrong with a certain probability. The questioner has two inquiry strategies. One is to randomly generate questions for each inquiry and estimate the random variable selected by the answerer after a fixed number of inquiries, which is a non-adaptive algorithm. The corresponding beam alignment algorithm process is as follows: Figure 3 As shown in the figure, one is that each question is generated according to the previous question and its answer, and the question can be terminated when the estimation condition is met to estimate the random variable selected by the respondent, that is, the adaptive algorithm. The corresponding beam alignment algorithm process is as follows: Figure 4 Compared with the non-adaptive algorithm, the adaptive algorithm can adjust the query range in real time according to the previous query results, and gradually refine with fewer queries to obtain a higher estimation accuracy, which can better meet the needs of the actual communication system.

[0073] However, suppose there is an eavesdropper in the question-answering learning process, who can eavesdrop on the questioner's questioning strategy and all his questions, but cannot eavesdrop on any of the answerer's answers. From a security perspective, in a non-adaptive algorithm based on question-answering learning, due to the randomness of question generation, the eavesdropper cannot obtain any information from the question alone, so there is no risk of privacy leakage; in an adaptive algorithm based on question-answering learning, new questions are related to old questions and their results, and the eavesdropper can obtain certain information from the question, so there may be a risk of privacy leakage.

[0074] Taking the classic adaptive binary search algorithm as an example, suppose there are three parties: questioner Alice, answerer Bob, and eavesdropper Eve. Answerer Bob randomly selects a number from a finite subset of the set of positive integers. , the questioner Alice asks the answerer Bob a binary yes or no question each time and get a binary yes or no answer , the eavesdropper Eve can eavesdrop on all the questions that the questioner Alice asks the answerer Bob, but cannot eavesdrop on any of the answers from the answerer Bob. Figure 5 As shown, if The value range of is a positive integer from 1 to 8, and the respondent Bob selects . Questioner Alice's first question , the answerer Bob was the first to answer , indicating an affirmative answer; the second question of the questioner Alice , the second answer of the respondent ; Questioner Alice’s third question , the third respondent answered At this point, although the questioner Alice has accurately estimated the number to be 8 based on the previous inquiries and answers, the eavesdropper Eve, who has eavesdropped on all the questions, can also accurately estimate the number based on the questions. The query logic roughly estimates the number to be 7 or 8.

[0075] The present invention analogizes the above process to the adaptive beam alignment process based on question-answer learning, where the base station and the legal user play the roles of questioner and answerer respectively, and the illegal user plays the role of eavesdropper. When the base station estimates the direction of the legal user through the question-answer learning method, assuming that the eavesdropper can eavesdrop on the base station's inquiry process in some way (such as stealing the phase shifter data of the base station array antenna, etc.), then he can get the same or similar estimation result as the base station with a high probability when the legal user's answer is completely unknown, thereby bringing security risks to the normal and stable communication between the base station and the legal user (such as the eavesdropper sends a strong interference signal in the direction of the legal user, the eavesdropper disguises as a base station, and the eavesdropper obtains our location information in a battlefield environment, etc.).

[0076] Thus, a fast, accurate and secure beam alignment algorithm with strong applicability is implemented, which can not only ensure accurate beam alignment between the base station and the legitimate user, but also prevent eavesdroppers from obtaining accurate beam alignment results through eavesdropping, thereby ensuring subsequent normal and stable communication between the base station and the legitimate user. This is the beneficial effect that the present invention is to achieve.

[0077] In the implementation process of the present invention, it is assumed that the direction of the signal received by the base station is the direction of the legitimate user who sent the signal. Considering the complex and changeable communication environment, it can be considered that each judgment of the beam search is wrong with a certain probability. Therefore, the brief process of the algorithm proposed by the present invention is that the legitimate user first sends a signal to the base station. After the base station receives the signal, in the case of possible error in judgment, the question-answering learning method can not only accurately estimate the direction of the legitimate user with as few beam searches as possible, but also avoid the illegal user eavesdropping on the question-answering learning query process. Estimation of the direction of the legitimate user relative to the base station with a higher probability is avoided. The schematic diagram is shown in Figure 6 shown.

[0078] Consider a single base station and a single legitimate user communication scenario, assuming that the array antenna on the base station is a uniform planar array (UPA). Figure 7 As shown, the steering vector for a signal from a certain direction is:

[0079]

[0080] In the horizontal direction, there are Antennas, vertically Antenna, represents the wavelength of the millimeter wave signal, represents the antenna array spacing, and They are the azimuth and elevation angles in physical space respectively and they are independent of each other, such as Figure 8 As shown, , .

[0081] Assume that At this moment, the legitimate user transmits the signal from the direction The signal is , then the signal received by the base station antenna array is for:

[0082]

[0083] Signal processed by base station UPA for:

[0084]

[0085] in, represents the antenna weight vector of UPA.

[0086] like Figure 8 As shown. The space of size is divided into Subspaces of equal size , the direction of the legal user to the base station is , the resolution (division value) of the beam search is From the perspective of the sphere, the area corresponding to a single beam search process is as follows: Fig.10 As shown. In the beam search vector, if The value of the element is 1, which means that the beam search will determine whether the legitimate user is in the first In the subspace corresponding to the element, if the If the value of the element is 0, it means that the beam search will not make a judgment. Assume that the base station assumes that the illegal user who is the eavesdropper will The size of the space is divided into Subspaces of equal size Without loss of generality, assume , then for the eavesdropper, the resolution of its beam search is If the security level set by the base station is , the error probability is , and the base station estimates the random variable representing the legitimate user as , the eavesdropper's estimate of the random variable representing the legitimate user is , then the secure beam alignment algorithm used by the base station should meet the following three conditions:

[0087] a.

[0088] b.

[0089] c.

[0090] Based on the above communication model, it is not difficult to see that since the later questions are generated based on the earlier questions and their answers, for the eavesdropper, the amount of information in the later inquiry process is greater than that in the earlier inquiry process, which is more helpful for it to obtain the same or similar estimation results as the base station.

[0091] If there is only one direction in the space The corresponding steering vector is When the antenna weight vector is taken as hour, Therefore, the base station changes the direction angle and elevation angle according to the beam search codebook, finds the corresponding steering vector, then takes the conjugate transpose and multiplies the signal received by the antenna array, and calculates the output signal power. If the output value exceeds a certain set threshold, it means that there is a legitimate user in the area corresponding to the steering vector, otherwise there is no legitimate user.

[0092] Since the actual communication environment is complex and changeable, there may be two kinds of misjudgments when judging whether there is a legitimate user in the search area. One is that the output signal power is higher than the set threshold when the search area is wrong, and the other is that the output signal power is lower than the set threshold when the search area is correct. Considering the above two kinds of misjudgments, assume that the probability of misjudgment is , then we can use a probability of The binary symmetric channel (BSC) is used to represent the search area. The input is the real answer of whether there is a target in the search area, and the output is the actual answer estimated by the beam alignment algorithm, such as Fig. 9 shown.

[0093] Based on the above technical considerations, the present invention proposes the following technical solutions:

[0094] The present invention provides a secure beam alignment algorithm based on privacy question-answering learning, referring to Figure 1 and 2 ,include:

[0095] Step S1, dividing the base station communication space into multiple subspaces of equal size, so that the beam alignment problem is equivalent to searching for random variables on a unit interval divided into multiple subintervals, and executing an adaptive beam alignment algorithm based on sortPM to obtain the primary subspace where the legitimate user is located;

[0096] Step S2: Execute a non-adaptive beam alignment algorithm based on channel coding on the primary subspace to obtain a secondary subspace where the legitimate user is located;

[0097] Step S3: synchronously execute the sortPM-based adaptive beam alignment algorithm on all secondary subspaces to obtain the tertiary subspace where the legitimate user is located, and take the midpoint of the tertiary subspace as the location of the legitimate user.

[0098] The working principle and beneficial effects of the above technical solution are as follows: the base station and the space around it are abstracted into a spherical space, and the signal transmitted by the legitimate user is emitted from the spherical surface to the center of the sphere. The secure beam alignment algorithm based on privacy question-answering learning proposed in the present invention divides the beam alignment process into two stages, namely, the pure beam alignment stage and the secure beam alignment stage. In the pure beam alignment process, the base station temporarily does not consider the problem of privacy leakage, and performs an adaptive beam alignment process according to the classic sortPM noise search algorithm in question-answering learning to quickly determine the approximate location of the legitimate user; in the secure beam alignment process, the base station subdivides the approximate location of the legitimate user into several small search intervals, and repeats the above beam alignment process in each interval to confuse the eavesdropper, thereby achieving secure beam alignment.

[0099] In a specific embodiment, referring to Fig.11 , the adaptive beam alignment algorithm based on sortPM in step S1 includes:

[0100] Enter the target location , resolution , error probability And the signal power judgment threshold , the error probability is subject to the parameter Bernoulli distribution of ;

[0101] The space Divide into equal-sized non-overlapping subspaces (It is worth noting that the traditional sortPM noise search algorithm is to locate the target on a one-dimensional multi-line segment, which is transferred to the polar coordinate system. The target positioning in the multi-angle range in space is raised to the three-dimensional spherical coordinate system. Target positioning in two angle ranges in space, namely ,in are the azimuth angle and the pitch angle, respectively. Here we only demonstrate the positioning of the subspace corresponding to one of the angles. The subspace corresponding to the other angle can be deduced in the same way). Set Indicates from 1 to For all integers, , No. The probability of a random variable existing in a subspace is initialized to ;

[0102] The target position will be estimated Assigned to , the total number of beam searches Assign a value of 0;

[0103] Initialize the probability of the existence of the target in each subspace ,in:

[0104]

[0105] Enter The adaptive beam search works and determines the Secondary adaptive beam search vector:

[0106]

[0107]

[0108]

[0109] in, Indicates The probability of random variables existing in each subspace sorted in descending order during the search and , Represents the probability of the existence of a random variable in each subinterval Sort in descending order, Indicates the first The sum of probabilities, Before The sum of the probabilities is closest ;

[0110] Select The front The subspace corresponding to the probability is taken as the The search space for the noise search is The binary beam search vector , among which Elements 1 means the The search space contains subspace, 0 means The search space does not contain subspaces;

[0111] The binary beam search vector is expressed as:

[0112]

[0113] in, It represents an indicator function, that is, it takes 1 when the condition in the brackets is true and takes 0 when the condition is false;

[0114] According to the binary beam search vector Determine the beam search area:

[0115]

[0116] Calculate the signal power based on the signal received by the base station ;

[0117] like Then determine the The result of the secondary beam search ,otherwise ;

[0118] Update the posterior probability corresponding to each subspace:

[0119]

[0120]

[0121] Determine the number of the subspace corresponding to the maximum a posteriori probability:

[0122]

[0123] like The estimated target position is returned and the serial number of the subspace in which it is located , and make ;

[0124] like Then And loop through the Sub-adaptive beam search operation.

[0125] The beneficial effect of the above technical solution is: making the beam alignment problem equivalent to searching for random variables on a unit interval divided into multiple sub-intervals, and realizing rapid positioning of the sub-interval where the legitimate user is located through the above adaptive beam alignment algorithm based on sortPM.

[0126] In a specific embodiment, step S1 includes:

[0127] Will The size of the space is divided into first-order subspaces of equal size , which makes the beam alignment problem equivalent to being divided into Unit interval of subintervals Search for random variables on ;

[0128] According to the adaptive beam alignment algorithm based on sortPM, the resolution , Error Probability ;

[0129] Calculate the signal power based on the signal received by the base station and update the posterior probability corresponding to each subspace;

[0130] The subspace corresponding to the maximum posterior probability is determined to be the first-level subspace where the legitimate user is located.

[0131] The beneficial effects of the above technical solution are: in the pure beam alignment stage, the base station first The size of the space is divided into Subspaces of equal size , so the beam alignment problem can be equivalent to being divided into Unit interval of subintervals Search for random variables on . Referring to the classic sortPM noise search algorithm, the resolution of the beam alignment algorithm at this stage is , the error probability of the beam alignment algorithm In the During the secondary beam search, it is assumed that the base station array antenna receives the signal of the legitimate user. After that, the generated length is The binary beam search vector , and accordingly find the corresponding antenna weight vector The base station uses the antenna weight vector and signal Calculate the signal power and the signal power threshold set in advance Assume that the base station may be interfered with, causing the signal power it calculates to be The probability of the event being higher (lower) than the set threshold is subject to the parameter Bernoulli distribution. This achieves the transformation of the beam alignment problem and the rapid positioning of the first-level subspace where the legitimate user is located.

[0132] In a specific embodiment, referring to Fig.12 , step S2 comprises:

[0133] Set the The first-level subspace where the legitimate users are located after the adaptive beam search ;

[0134] The first-level subspace Divide Secondary subspaces of equal size ;

[0135] In the first subspace A non-adaptive beam alignment algorithm based on channel coding is performed on After a non-adaptive beam search, the secondary subspace where the legitimate user is located is determined ;

[0136] If you pass After non-adaptive beam search, or the number of times it has been performed When it is greater than a preset limit value, and the secondary subspace where the legitimate user is located cannot be further determined, the non-adaptive beam search is terminated, and it is determined that the beam alignment has failed.

[0137] The working principle and beneficial effects of the above technical solution are as follows: After the beam search, there is a subspace The posterior probability is greater than , then continue to divide the subspace into Secondary subspaces of equal size , the set of secondary subspaces Each subspace in can be further divided into Level 3 Subspace In the safety beam alignment phase, the base station first uses a non-adaptive beam alignment algorithm based on channel coding. After the second beam search, the secondary subspace set The approximate location of the legitimate user is further determined as a secondary subspace , and the eavesdropper cannot obtain useful information from this non-adaptive process. Thus, the eavesdropper can be shielded during the beam alignment process, achieving a safe alignment process.

[0138] In a specific embodiment, referring to Figure 2 , step S3 comprises:

[0139] Sure Secondary subspaces of equal size ;

[0140] Perform resolution synchronously on all secondary subspaces , error probability Adaptive beam alignment algorithm based on sortPM;

[0141] When it is determined that the secondary subspace where the legitimate user is located has a tertiary subspace The posterior probability is greater than When , the adaptive beam search is terminated and the number of adaptive beam searches is recorded ;

[0142] Determine the third-level subspace This is the subspace where the legitimate user exists, and the midpoint of the third-level subspace is taken as the estimate of the legitimate user's direction .

[0143] The working principle and beneficial effects of the above technical solution are as follows: the base station is in the secondary subspace Repeat the pure beam alignment stage operation, so that the resolution of the beam alignment algorithm in this stage is , the error probability of the beam alignment algorithm If you pass After the beam search, there is a three-level subspace The posterior probability is greater than , then the beam search is terminated, and the three-level subspace That is, the subspace where the legal user exists, and then the estimation of the direction of the corresponding legal user is obtained To confuse eavesdroppers and prevent them from obtaining useful information, the beam search query in this process will be The query space is repeated on the second-level subspaces. The union of three-level subspaces with the same relative position, the query result is The result of the OR operation of the search results on the second-level subspace is used to confuse the eavesdropper's monitoring range and ensure the security of the beam alignment work.

[0144] In order to verify the effectiveness of the technology of the present invention, the following simulation results are provided:

[0145] 1. Assume that the number of legitimate user subspaces M = 360, the number of eavesdropper subspaces N = 60, the single beam search error probability p = 0.1, the single beam search vector generation probability q = 0.5, and the error probability = 0.1, the experimental results are as follows Fig.13 As shown in the figure, taking the number of beam searches required to achieve the same arc error as the standard, the performance of the secure beam alignment algorithm based on privacy question-answering learning is between the adaptive beam alignment algorithm based on question-answering learning and the non-adaptive beam alignment algorithm, slightly worse than the purely adaptive beam alignment algorithm based on question-answering learning, and much better than the purely non-adaptive beam alignment algorithm based on question-answering learning. In addition, as the security level L increases, the performance of the secure beam alignment algorithm based on privacy question-answering learning gradually deteriorates.

[0146] 2. Assume that the number of legitimate user subspaces M = 360, the number of eavesdropper subspaces N = 60, the single beam search error probability p = 0.1, and the error probability =0.1, e.g. Fig.14 As shown in Figure 2, taking the arc error under the same number of beam searches as the standard, in the secure beam alignment algorithm based on privacy question-answering learning, as the number of beam searches increases, the arc error estimated by the legitimate user is gradually smaller than that of the eavesdropper, and the higher the security level L, the greater the arc error estimated by the eavesdropper.

[0147] It is not difficult to see through analysis and simulation that in the presence of eavesdroppers in the communication scenario, the secure beam alignment algorithm based on privacy question-answering learning proposed in the present invention, compared with the adaptive beam alignment algorithm based on question-answering learning, can achieve fast and accurate beam alignment between the base station and the legitimate user with slight performance loss and an acceptable degree of privacy leakage.

[0148] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made in form and details without departing from the scope defined by the claims of the present invention.

Claims

1. A secure beam alignment method based on private question-answering learning, It is characterized in that include: Step S1, dividing the base station communication space into multiple subspaces of equal size, so that the beam alignment problem is equivalent to searching for random variables on a unit interval divided into multiple subintervals, and executing an adaptive beam alignment algorithm based on sortPM to obtain the primary subspace where the legitimate user is located; Step S2: Execute a non-adaptive beam alignment algorithm based on channel coding on the primary subspace to obtain a secondary subspace where the legitimate user is located; Step S3, synchronously executing the sortPM-based adaptive beam alignment algorithm on all secondary subspaces to obtain the tertiary subspace where the legitimate user is located, and taking the midpoint of the tertiary subspace as the location of the legitimate user; The step S1 comprises: Will The size of the space is divided into first-order subspaces of equal size , which makes the beam alignment problem equivalent to being divided into Unit interval of subintervals Search for random variables on ;in, Indicates the direction angle of the wave direction from which the base station receives the legitimate user signal. Indicates the elevation angle of the direction of the wave from which the base station receives the legitimate user signal; According to the adaptive beam alignment algorithm based on sortPM, let , Error Probability ;in, represents the algorithm resolution, L represents the algorithm security level, and N represents the number of subspaces defined by the base station assuming that the eavesdropper is represents the probability of algorithm error; Calculate the signal power based on the signal received by the base station and update the posterior probability corresponding to each subspace; The subspace corresponding to the maximum posterior probability is determined to be the first-level subspace where the legitimate user is located.

2. According to claim 1, a secure beam alignment method based on private question-answering learning, It is characterized in that The step S2 comprises: Set the The first-level subspace where the legitimate users are located after the adaptive beam search ; The first-level subspace Divide Secondary subspaces of equal size ; In the first subspace A non-adaptive beam alignment algorithm based on channel coding is performed on After a non-adaptive beam search, the secondary subspace where the legitimate user is located is determined ; If you pass After the first non-adaptive beam search, if the secondary subspace where the legitimate user is located cannot be further determined, the non-adaptive beam search is terminated, and it is determined that the beam alignment has failed.

3. According to claim 1, a secure beam alignment method based on private question-answering learning, It is characterized in that The step S3 comprises: Sure Secondary subspaces of equal size ; Perform resolution synchronously on all secondary subspaces , error probability An adaptive beam alignment algorithm based on sortPM is proposed; where M represents the number of legal user subspaces defined by the base station; When it is determined that the secondary subspace where the legitimate user is located has a tertiary subspace The posterior probability is greater than When , the adaptive beam search is terminated and the number of adaptive beam searches is recorded ; Determine the third-level subspace This is the subspace where the legitimate user exists, and the midpoint of the third-level subspace is taken as the estimate of the legitimate user's direction ,in Indicates the direction angle of the incoming wave estimated by the base station, Indicates the elevation angle of the incoming wave direction estimated by the base station.

Citation Information

Patent Citations

  • Millimeter wave base station end beam alignment method based on position information reasoning

    CN115473554A

  • Self-adaptive millimeter wave beam searching method

    CN115882912A

  • Self-adaptive multi-beam alignment method based on question-answer type learning

    CN116701882A

  • KR20220073644A