A MR indoor and outdoor distinction method

Abstract

The present invention discloses a method for distinguishing indoor and outdoor MR data, comprising: setting coding rules and forming a classifier in an offline stage; distinguishing indoor and outdoor MR data in an online stage. The present invention has the following beneficial effects: compared with supervised learning, the algorithm only uses one label, and the other label is unknown, which has lower use cost; compared with unsupervised learning, the algorithm uses a label known in the original data, which has higher accuracy.

Description

Technical Field

[0001] The present invention belongs to the field of mobile communication big data, and relates to distinguishing indoor or outdoor MR (measurement report) data of 4G and 5G full-network wireless networks, and more specifically, to a method for distinguishing indoor and outdoor MR. Background Art

[0002] Indoor and outdoor positioning based on MR data can help operators optimize networks and realize data on the one hand, and can also be used in vertical industries, such as public security, epidemic prevention and control, to play a greater role. At present, the mainstream MR indoor and outdoor positioning algorithms in the industry all adopt a two-level architecture, that is, first distinguishing MR indoors and outdoors, and then using different algorithms to estimate the position of indoor and outdoor MR respectively. The reason for this is mainly because there is a wealth of GNSS data outdoors, which will be written into MR and reported to the base station together for collection, which can help the positioning algorithm learn the mapping relationship between outdoor signal features and outdoor positions. However, due to the severe attenuation of satellite signals in indoor environments, GNSS positioning is difficult to perform indoors, resulting in indoor MR lacking sufficient GNSS data to learn the mapping relationship between indoor signal features and positions, and then other algorithms need to be used to optimize indoor positioning results. Therefore, different positioning algorithms from outdoor ones are differentiated to optimize indoor positioning results in a targeted manner.

[0003] In summary, before the positioning algorithm, there needs to be an indoor and outdoor distinction module to classify the MR to be positioned as indoor or outdoor, and guide the subsequent process to use the corresponding indoor or outdoor positioning algorithm to complete the high-precision positioning of the MR data. In addition, the indoor and outdoor distinction algorithm needs to have extremely high accuracy to ensure that the corresponding MR is not misclassified into the positioning algorithm corresponding to other scenes.

[0004] The indoor and outdoor scenes mainly involve two entities: base stations and terminals. Terminals can be distributed indoors or outdoors, and the same is true for the distribution of base stations. Base stations located indoors are generally called indoor distribution stations. These stations are mainly used to solve the problem of indoor signal coverage, and the coverage range is small (generally about 20m). Therefore, most of the terminals accessing the indoor distribution station are located indoors, and only a very small number of terminals located outdoors will access the indoor distribution station. Since the indoor distribution station has a small coverage range, these outdoor terminals are also very close to the building. Therefore, it is assumed that the terminals accessing the indoor distribution station are all located indoors.

[0005] Base stations located outdoors are called outdoor macro stations. These stations can solve the coverage problems of indoor and outdoor signals at the same time. They have a large coverage range (generally 400m) and account for a high proportion of the total MR reported on the entire network. The terminals accessing these base stations may be located indoors or outdoors, and need to be distinguished by indoor and outdoor differentiation algorithms.

[0006] Traditional indoor and outdoor distinction algorithms can be divided into supervised and unsupervised. The supervised indoor and outdoor distinction algorithm collects manually labeled indoor and outdoor MR data through crowdsourcing testing, and inputs them into the supervised AI algorithm as training samples to obtain a binary classification model for indoor and outdoor distinction of the entire network data in the online stage. This method has high indoor and outdoor distinction accuracy in some scenarios, but its disadvantages are also obvious. On the one hand, the cost of manual indoor and outdoor MR labeling is very high. Since the wireless network environment is constantly changing, data labeling needs to be performed regularly. Otherwise, the accuracy of the model obtained by training with previous data will drop sharply, which brings great costs; on the other hand, manual sampling is difficult to cover all scenarios, and the sampled data is affected by factors such as terminal type and mobile phone posture. Only a sampling of the entire network MR data is obtained. The model trained in this way is not stable enough. It has high accuracy in some scenarios, but may not be used in some scenarios.

[0007] The unsupervised indoor and outdoor distinction algorithm is based on business features, automatically learns the distribution characteristics of indoor and outdoor signals under these features, and completes the indoor and outdoor estimation based on these distribution characteristics. Compared with the supervised algorithm, this algorithm does not require manual MR labeling, thus effectively reducing the commercial cost. However, the accuracy of the unsupervised algorithm is limited, and the business features it uses are expert experience. Considering the variability of the wireless environment, these expert experiences cannot cover all scenarios, so the accuracy in some scenarios is very poor. Summary of the invention

[0008] The present invention provides an MR indoor and outdoor distinction method, which solves the cost problem of traditional supervised learning and the accuracy problem of unsupervised learning.

[0009] In order to solve the above problems, on the one hand, the present invention provides an MR indoor and outdoor distinction method, comprising:

[0010] Setting encoding rules and forming classifiers through an offline phase;

[0011] Indoor and outdoor differentiation is performed on MR data via an online stage.

[0012] The step of setting the coding rules and forming the classifier in the offline stage includes:

[0013] Automatically annotate outdoor MR data;

[0014] Develop coding rules for use in the online phase;

[0015] The classifier is trained based on the incompletely supervised learning method.

[0016] The automatic annotation of outdoor MR data includes:

[0017] Obtain all MR data within a certain period of time;

[0018] Extracting MR data carrying GNSS data from the full amount of MR data;

[0019] The MR data with GNSS data are marked as outdoor, and the MR data without GNSS data are marked as indoor.

[0020] The forming of encoding rules for use in the online stage includes:

[0021] Grouping the full amount of MR data according to serving cells;

[0022] Extract the serving cell and neighboring cell lists from all MR data in each group, perform deduplication processing, and obtain a cell list set as a feature column;

[0023] For each serving cell and neighboring cell in each MR data, if they appear in the cell set, the RSRP value of the cell is filled in the position of the corresponding cell in the set to generate a sparse vector, where each column of the vector represents a cell, and the value of the column represents the RSRP value of the current MR in the column;

[0024] A matrix is ​​generated according to all MR data in the same group as the encoding result of the MR data of the group, until each group obtains an encoded matrix, thereby completing the encoding of all data.

[0025] The method of training a classifier based on incomplete supervised learning includes:

[0026] Set all data under the selected service cell to set A, the number is M, and use the set {X1,X2,...,X M} indicates that the i-th MR data is represented by X i , the label is defined as y i ;

[0027] The number of cells after deduplication is set to N, and the set {c1,c2,...,c N}express;

[0028] A matrix of M*(N+1) is formed, where the N+1th column represents the corresponding indoor or outdoor label; according to Bayes' theorem, we can get:

[0029]

[0030] Combined with Laplace smoothing, the indoor and outdoor labels y can be obtained j When it is known, cell c t The probability P(c t |y j )for:

[0031]

[0032] Among them, N(c t ,X i ) represents the number of times the RSRP of the t-th cell in the i-th MR in the full set A is a valid value. If its RSRP is not 0, then N(c t ,X i ) takes the value of 1, otherwise takes the value of 0;

[0033] Assume that the values ​​of N cells in each MR are independent and identically distributed. Then when X i When it is known, the probability P(y i =1|X i )for:

[0034]

[0035] We can get P(y i =0|X i )=1-P(y i =1|X i )

[0036] Build a classifier.

[0037] The constructing of the classifier comprises:

[0038] Extract the data labeled as outdoor in set A to form set O, and the other indoor data to form set M;

[0039] Initialize the model parameters P(y i =1|X i ):

[0040]

[0041] Set P(y i =1|X i ) is defined as P t (y i =1|X i ), and perform loop iteration;

[0042] When the loop terminates, save the last round of P(c k |y i =1), P(c k |y i =0), P(y i =1) and P(y i =0), where k = 1, 2, ..., N, i = 1, 2, ..., M, and these values ​​constitute the classifier of the cell.

[0043] Performing the loop iteration includes:

[0044] If |P t+1 (y i = 1|X i ) - P t (y i = 1|X i )| < T, execute the loop:

[0045] For each data X in the set M i :

[0046] Update the current round's P(y i-1 ), P(c t |y i-1 ) according to the previous round's P(y i ), P(c t |y i )

[0047] Calculate the current round's P(y i = 1|X i ) based on the update result;

[0048] Among them, for the data X i ∈M, P(y i = 1|X i ) = 1 does not change with the iteration process and always remains a constant.

[0049] The indoor and outdoor discrimination of MR data in the online phase includes:

[0050] Find the coding rule corresponding to the serving cell according to the selected serving cell, and convert the MR data into the corresponding vector form according to the coding rule;

[0051] Find the corresponding P i (c k |y i = 1), P i (c k |y i = 0), P i (y i = 1) and P i (y i = 0) values for the serving cell, and calculate the probability P(y i = 1|X i ) that the current MR data is determined to be outdoor according to the calculation formula of P(y i = 1|X i );

[0052] If P(y i = 1|Xi )>0.5, the MR data is judged as outdoor, otherwise it is judged as indoor.

[0053] The step of finding out a coding rule corresponding to the selected serving cell according to the serving cell, and converting the MR data into a corresponding vector form according to the coding rule, further comprises:

[0054] If the selected cell in the MR data is not in the encoding rule, the cell is not encoded.

[0055] On the one hand, a computer-readable storage medium is provided, wherein a plurality of instructions are stored in the storage medium, and the instructions are suitable for being loaded by a processor to execute the MR indoor and outdoor differentiation method as described above.

[0056] The beneficial effects of the present invention are:

[0057] Compared with supervised learning, this algorithm only uses one label, and the other label is unknown, which has a lower usage cost; compared with unsupervised learning, this algorithm uses a label that is known in the original data, which has higher accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0059] Figure 1 It is a flowchart of a method for distinguishing indoor and outdoor MR provided by an embodiment of the present invention;

[0060] Figure 2 It is a schematic diagram of MR data after encoding provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0062] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0063] In the present invention, the word "exemplary" is used to mean "used as an example, illustration or description". Any embodiment described as "exemplary" in the present invention is not necessarily to be construed as being preferred or advantageous over other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other instances, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0064] This case is based on an indoor and outdoor distinction algorithm based on incomplete supervised learning, which has lower costs than supervised learning and higher accuracy than unsupervised learning. The algorithm uses outdoor GNSS (Global Navigation Satellite System) data as outdoor MR tags, which can be automatically obtained from the MR data of the entire network without manual labeling. For indoor MR tags, automatic labeling is performed through outdoor tags and incomplete supervised learning algorithms. At the same time, the training of the indoor and outdoor distinction model can be completed for indoor and outdoor MR distinction in the online stage. Since there is no human participation in the whole process, the cost can be effectively reduced. In addition, this method automatically learns a model that is effective for indoor and outdoor distinction based on some outdoor tags of the entire network. The scene coverage is more comprehensive, which can effectively avoid the low precision phenomenon in certain scenarios caused by expert experience in unsupervised algorithms, and has higher accuracy than unsupervised learning.

[0065] Before explaining the algorithm, it is necessary to understand the MR data. This case stipulates that the MR data are all processed into the format shown in the following table. Among them, the (ci:val) pair represents the level value of a cell (also known as the RSRP value), where ci represents the cell number, and val represents the signal strength received by the MR to the cell, in dBm. The first data in each MR represents the service cell, and the other data represents the neighboring cells. The subsequent indoor distinction is based on the MR as the distinction granularity.

[0066] Table 1 MR data diagram

[0067] Serial number Format MR1 c1:-90,c2:-95,c4:-100,c5:-103 MR2 c6:-75,c1:-80.c2:-99,c10:-103 ...... ...... MRN c3:-80,c2:-90,c8:-105

[0068] See also Figure 1 , Figure 1 This is a flow chart of an MR indoor and outdoor distinction method provided by an embodiment of the present invention, which is generally divided into an offline stage and an online stage. The offline stage mainly completes the generation of coding rules and classifiers, and the online stage uses the products of the above offline stage to complete the indoor and outdoor distinction of MR data. The MR indoor and outdoor distinction method includes steps S1-S2:

[0069] S1, setting encoding rules and forming a classifier in an offline stage; Step S1 includes steps S11-S13:

[0070] S11. Automatically label outdoor MR data; wherein, a full amount of MR data within a certain time period is obtained; MR data carrying GNSS data in the full amount of MR data is extracted; the MR data carrying GNSS data is labeled as outdoor, and the MR data not carrying GNSS data is labeled as indoor.

[0071] In this embodiment, outdoor MR data is automatically labeled to generate data with only some outdoor labels, and the indoor and outdoor labels of other data are unknown. That is, all MR data in a certain period of time are extracted, and the MR data carrying GNSS data are extracted, and these MR data are marked as outdoor, and other data are temporarily marked as indoor. For simplicity, the outdoor and indoor labels are abbreviated as 1 and 0 respectively. As shown in the following table:

[0072] Table 2 Schematic diagram of indoor and outdoor MR annotation

[0073] data Is it GNSS? Annotation results MR1 yes 1 MR2 yes 1 MR3 no 0 ...... ...... ...... MRN no 0

[0074] S12, forming a coding rule for use in the online stage; Step S12 includes steps S121-S124:

[0075] S121, grouping the full amount of MR data according to serving cells;

[0076] S122, extracting the serving cell and neighboring cell lists from all MR data in each group, and performing deduplication processing to obtain a cell list set as a feature column;

[0077] In this embodiment, for all MR data in each group, the serving cell and neighboring cell list are extracted and deduplication is performed to obtain a cell list set as a feature column.

[0078] S123. For the serving cell and neighboring cell in each piece of MR data, if they appear in the cell set, fill in the RSRP value of the cell at the position of the corresponding cell in the set to generate a sparse vector, wherein each column of the vector represents a cell, and the value of the column represents the RSRP value of the current MR in the column;

[0079] In this embodiment, for the serving cell and neighboring cell in each MR data, if they appear in the cell set, the RSRP value of the cell is filled in the position of the corresponding cell in the set, otherwise it is filled in 0. Finally, a sparse vector is generated, each column of the vector represents a cell, and the value of the column represents the RSRP value of the current MR in the column.

[0080] S124 , generating a matrix based on all MR data in the same group as the encoding result of the group MR data, until each group obtains an encoded matrix, thereby completing the encoding of all data.

[0081] In this embodiment, step S123 is performed on all MR data in the group to generate a matrix as the encoding result of the MR data in the group. Assuming that the number of cells in the group after deduplication is M and the number of MRs is N, an N*M sparse matrix will eventually be generated.

[0082] Finally, steps S121 to S124 are performed for each cell, and each group will obtain a coded matrix, completing the coding of all data. For example, the result of coding the MR input in Table 1 using the above method is as follows: Figure 2 As shown, it is assumed here that M=10.

[0083] The mapping relationship between cells and column positions in each group is stored as the coding rule of the group to form a coding rule for use in the online stage.

[0084] S13, train the classifier according to the incomplete supervised learning method. In which, all the data under the selected service cell are set to set A, the number is M, and the set {X1, X2, ..., X M} indicates that the i-th MR data is represented by X i , the label is defined as y i ; The number of cells after deduplication is set to N, and the set {c1,c2,...,cN} represents; forming an M*(N+1) matrix, where the N+1th column represents the corresponding indoor or outdoor label.

[0085] In this embodiment, all data in a certain serving cell is defined as set A. Assuming that the number of data is M, the set {X1, X2, ..., X M}, the i-th MR is represented by X i , the label is defined as y i The number of cells after deduplication is N, and the set {c1,c2,...,c N After the labeling module and the data encoding module, an M*(N+1) matrix is ​​formed, where the N+1th column represents the corresponding indoor and outdoor labels.

[0086] According to Bayes' theorem, we can get:

[0087]

[0088] Combined with Laplace smoothing, the indoor and outdoor labels y can be obtained j When it is known, cell c t The probability P(c t |y j )for:

[0089]

[0090] Among them, N(c t ,X i ) represents the number of times the RSRP of the t-th cell in the i-th MR in the full set A is a valid value. If its RSRP is not 0, then N(c t ,X i ) takes the value of 1, otherwise takes the value of 0;

[0091] Assume that the values ​​of N cells in each MR are independent and identically distributed. Then when X i When it is known, the probability P(y i =1|X i )for:

[0092]

[0093] We can get P(y i =0|X i )=1-P(y i =1|X i )

[0094] Build a classifier.

[0095] The constructing of the classifier comprises:

[0096] Extract the data with the label of outdoor in set A to form set O, and the remaining indoor data to form set M;

[0097] Initialize the model parameter P(y i =1|X i ):

[0098]

[0099] Set the P(y i =1|X i ) defined after the t-th iteration as P t (y i =1|X i ), and perform loop iteration;

[0100] When the loop terminates, save the values of P(c k |y i =1), P(c k |y i =0), P(y i =1) and P(y i =0), where k = 1, 2,..., N, i = 1, 2,..., M, and these values constitute the classifier of this cell.

[0101] Among them, the said loop iteration includes:

[0102] If |P t+1 (y i =1|X i ) - P t (y i =1|X i )| < T, execute the loop:

[0103] For each data X i in set M:

[0104] Update the current round of P(y i-1 ), P(c t |y i-1 ) according to the previous round of P(y i ), P(c t |y i )

[0105] Calculate the current round of P(y i =1|X i ) according to the update result;

[0106] Among them, for the data X i ∈ M, P(y i =1|X i)=1 does not change with the iteration process and always remains a constant.

[0107] In this embodiment, for each data X in the set M i :According to the above formula and the results of the previous round, update the current round P(y i )、P(c t |y i ), calculate the current round P(y i =1|X i ). In the iterative process, for X i ∈M data, P(y i =1|X i )=1 does not change with the iteration process and always remains a constant.

[0108] When the loop terminates, save the last round of P(c k |y i =1), P(c k |y i =0), P(y i =1) and P(y i =0), where k = 1, 2, ..., N, i = 1, 2, ..., M, and these values ​​constitute the classifier of the cell.

[0109] After all the data grouped by serving cells have been processed through the above process, each serving cell will generate a corresponding set of P(c k |y i =1), P(c k |y i =0), P(y i =1) and P(y i =0), store it in Table 3 to form a classifier.

[0110] Table 3 Schematic diagram of the offline stage classifier structure

[0111]

[0112] S2. Distinguish indoor and outdoor MR data through the online stage.

[0113] In this embodiment, in the online stage, each MR that is not distinguished between indoor and outdoor is identified as indoor or outdoor. It is assumed that the format of the MR to be distinguished is as shown in Table 1 {c1:-88, c2:-93, c6:-102, c15:-103}, and its serving cell is c1.

[0114] Step S2 includes steps S21-S23:

[0115] S21. Find out the coding rule corresponding to the selected service cell according to the service cell, and convert the MR data into a corresponding vector form according to the coding rule; if the cell selected in the MR data is not in the coding rule, the cell is not encoded.

[0116] In this embodiment, according to the serving cell, the coding rule corresponding to the serving cell is found from the coding rule table generated in step S12, and the MR is converted into a corresponding vector form according to the coding rule. If a cell in the MR is not in the coding rule, the cell is not encoded.

[0117] S22, find the corresponding P according to the serving cell i (c k |y i =1), P i (c k |y i =0), P i (y i =1) and P i (y i =0) value, and according to P(y i =1|X i ) is used to calculate the probability P(y i =1|X i );

[0118] In this embodiment, according to the serving cell, the corresponding P is found from the classifier table generated in step S13. i (c k |y i =1), P i (c k |y i =0), P i (y i =1) and P i (y i =0) value. And according to P(y i =1|X i ) is used to calculate the probability P(y i =1|X i ).

[0119] S23. If P(y i =1|X i )>0.5, the MR data is judged as outdoor, otherwise it is judged as indoor.

[0120] In summary, an indoor and outdoor distinction algorithm based on incomplete supervised learning was invented. Compared with supervised learning, this algorithm only uses one label, and the other label is unknown, which has a lower cost; compared with unsupervised learning, this algorithm uses a label known in the original data, which theoretically has higher accuracy. The offline and online stages derived from this algorithm, as well as the label annotation methods, feature engineering, and model training methods included in these two stages, etc.

[0121] Those skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructions, or by controlling related hardware through instructions, and the instructions can be stored in a computer-readable storage medium and loaded and executed by a processor. To this end, an embodiment of the present invention provides a storage medium, which stores multiple instructions, and the instructions can be loaded by a processor to execute the steps in any one of the MR indoor and outdoor differentiation methods provided in the embodiments of the present invention.

[0122] The storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0123] Since the instructions stored in the storage medium can execute the steps in any one of the MR indoor and outdoor distinction methods provided in the embodiments of the present invention, the beneficial effects that can be achieved by any one of the MR indoor and outdoor distinction methods provided in the embodiments of the present invention can be achieved. For details, please refer to the previous embodiments and will not be repeated here.

[0124] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A MR indoor and outdoor distinction method, characterized in that: include: Setting encoding rules and forming classifiers through an offline phase; Indoor and outdoor differentiation of MR data is performed through an online phase; The step of setting the coding rules and forming the classifier in the offline stage includes: Automatically annotate outdoor MR data; Develop coding rules for use in the online phase; Train the classifier based on the incompletely supervised learning method; The automatic annotation of outdoor MR data includes: Obtain all MR data within a certain period of time; Extracting MR data carrying GNSS data from the full amount of MR data; The MR data with GNSS data are marked as outdoor, and the MR data without GNSS data are marked as indoor; The forming of encoding rules for use in the online stage includes: Grouping the full amount of MR data according to serving cells; Extract the serving cell and neighboring cell lists from all MR data in each group, perform deduplication processing, and obtain a cell list set as a feature column; For each serving cell and neighboring cell in each MR data, if they appear in the cell set, the RSRP value of the cell is filled in the position of the corresponding cell in the set to generate a sparse vector, where each column of the vector represents a cell, and the value of the column represents the RSRP value of the current MR in the column; Generate a matrix according to all MR data in the same group as the encoding result of the MR data of the group, until each group obtains an encoded matrix, thereby completing the encoding of all data; The method of training a classifier based on incomplete supervised learning includes: Set all data under the selected service cell to set A, the number is M, and use the set {X1,X2,...,X M } indicates that the i-th MR data is represented by X i , the label is defined as y i ; The number of cells after deduplication is set to N, and the set {c1,c2,...,c N }express; Form an M*(N+1) matrix, where the N+1th column represents the corresponding indoor or outdoor label; According to Bayes' theorem, we can get: Combined with Laplace smoothing, the indoor and outdoor labels y can be obtained j When it is known, cell c t The probability P(c t |y j )for: Among them, N(c t ,X i ) represents the number of times the RSRP of the t-th cell in the i-th MR in the full set A is a valid value. If its RSRP is not 0, then N(c t ,X i ) takes the value of 1, otherwise takes the value of 0; Assume that the values ​​of N cells in each MR are independent and identically distributed. Then when X i When it is known, the probability P(y i =1|X i )for: We can get P(y i =0|X i )=1-P(y i =1|X i ) Build a classifier.

2. The MR indoor and outdoor distinguishing method according to claim 1, characterized in that: The constructing of the classifier comprises: Extract the data labeled as outdoor in set A to form set O, and the other indoor data to form set M; Initialize the model parameters P(y i =1|X i ): Set P(y i =1|X i ) is defined as P t (y i =1|X i ), and perform loop iteration; When the loop terminates, save the last round of P(c k |y i =1), P(c k |y i =0), P(y i =1) and P(y i =0), where k = 1, 2, ..., N, i = 1, 2, ..., M, and these values ​​constitute the classifier of the cell.

3. The MR indoor and outdoor distinguishing method according to claim 2, characterized in that: The loop iteration comprises: If |P t+1 (y i = 1 | X i ) - P t (y i = 1 | X i ) | < T, execute the loop: For each data X in the set M i : According to the last round P(y i-1 )、P(c t |y i-1 ), update this round P(y i )、P(c t |y i ) According to the update result, calculate the current round P(y i =1|X i ); Among them, for X i ∈M data, P(y i =1|X i )=1 does not change with the iteration process and always remains a constant.

4. The MR indoor and outdoor distinguishing method according to claim 3, characterized in that: The method of distinguishing indoor and outdoor MR data in an online stage includes: Find out a coding rule corresponding to the selected serving cell according to the serving cell, and convert the MR data into a corresponding vector form according to the coding rule; Find the corresponding P according to the serving cell i (c k |y i =1), P i (c k |y i =0), P i (y i =1) and P i (y i =0) value, and according to P(y i =1|X i ) is used to calculate the probability P(y i =1|X i ); If P(y i =1|X i )>0.5, the MR data is judged as outdoor, otherwise it is judged as indoor.

5. The MR indoor and outdoor distinguishing method according to claim 4, characterized in that: The step of finding out a coding rule corresponding to the selected serving cell according to the serving cell, and converting the MR data into a corresponding vector form according to the coding rule, further comprises: If the selected cell in the MR data is not in the encoding rule, the cell is not encoded.

6. A computer-readable storage medium, characterized in that: The storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor to execute the MR indoor and outdoor distinguishing method according to any one of claims 1 to 5.

Images