CN105744561B - Various dimensions measurement report indoor and outdoor separation method - Google Patents
Various dimensions measurement report indoor and outdoor separation method Download PDFInfo
- Publication number
- CN105744561B CN105744561B CN201610127602.2A CN201610127602A CN105744561B CN 105744561 B CN105744561 B CN 105744561B CN 201610127602 A CN201610127602 A CN 201610127602A CN 105744561 B CN105744561 B CN 105744561B
- Authority
- CN
- China
- Prior art keywords
- value
- data
- grid
- room
- outdoor
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/10—Scheduling measurement reports ; Arrangements for measurement reports
Abstract
The invention discloses a kind of various dimensions measurement report indoor and outdoor separation methods, effectively compensate for the limitation of MR rasterizing appraisal procedure, by being based on room sub-signal source partition method, being based on outdoor test feature partition method, based on user mobility partition method, various dimensions indoor and outdoor MR separation method is formed to MR data inside and outside divided chamber, and then MR grid inside and outside forming chamber, to assess the quality of wireless network of the indoor scenes such as residential building, office building, hotel and various outdoor road scenes.
Description
Technical field
The present invention relates to a kind of various dimensions measurement report indoor and outdoor separation methods.
Background technique
Measurement report MR is that user initiates the measurement report generated during business and base station communication, it is round-the-clock, complete
The radio environment measurements data of period are the foundations of wireless resource scheduling, directly reaction quality of service and user's perception, it has
Mass data feature, sample variance is small, and accuracy is high, and the advantages such as procurement cost is low, therefore MR is to instruct operator's planning, construction
And the effective means of optimization.Traditional appraisal procedure have (1) cell and MR evaluation measures, i.e., by statistics cell covering
Rate, matter rate, uplink and downlink receive the indexs such as unbalanced power ratio to evaluate the quality of cell.(2) by MR grid physical and chemical, ground
The quality of wireless network of each grid is evaluated in reason with smaller granularity.
(1) the MR appraisal procedure of cell-level can only determine the quality of wireless network of current area, but the covering of a cell
Range be it is very wide, it is bad which specific position signal cannot be oriented.
(2) although MR rasterizing appraisal procedure can evaluate the geographical location quality of wireless network more refined, due to
Current MR, which specifically positions longitude and latitude, certain error, so with specific grid come depth in agent's room or the wireless network in street
Quality can have inaccuracy.
Summary of the invention
It is an object of the invention to overcome the deficiencies of the prior art and provide a kind of various dimensions measurement report indoor and outdoor separation sides
Method is formed by being based on room sub-signal source partition method, being based on outdoor test feature partition method, based on user mobility partition method
Various dimensions indoor and outdoor MR separation method is to MR data inside and outside divided chamber, and then MR grid inside and outside forming chamber, to assess resident
The quality of wireless network of the indoor scenes such as building, office building, hotel and various outdoor road scenes.
The purpose of the present invention is achieved through the following technical solutions: various dimensions measurement report indoor and outdoor separation method,
It includes following sub-step:
S1: data acquisition, from server by acquisition interface acquisition MR data, drive test data, space division basic data and
GIS data；
S2: MR data are cleaned, and remove invalid data and noise；
S3: MR data are subjected to the separation of room divided data and are calculated, the room for calculating MR data to be detected splits reliability a, judges room
The relationship of reliability a Yu room point threshold value A are split, if room splits reliability a less than room point threshold value A, separation terminates, jump procedure
S6, if otherwise jump procedure S4；
S4: MR data are subjected to user mobility separation and are calculated, the mobility confidence level b of MR data to be detected is calculated, sentences
The relationship of disconnected indoor and outdoor separation confidence level b and threshold value thre, if indoor and outdoor separation confidence level b is higher than threshold value thre1, and small
In threshold value thre2, then separation terminates, jump procedure S6, otherwise jump procedure S5；
S5: carrying out outdoor test feature separation for MR data and calculate, calculate the test feature confidence level of MR data to be detected,
Test feature confidence level c input neural network classifier is judged that then separation terminates, jump procedure S6, if not can be carried out
Separation, it is impossible to which the MR data of differentiation discard；
S6: current MR separating resulting, the i.e. result that output MR belongs to indoor or outdoors are exported；The threshold value A value
It is 432, the value of the threshold value thre1 is 1.2, and the value of the threshold value thre2 is 0.8.
It includes following sub-step that room divided data separation in the step S3, which calculates:
S31: collected MR data are segmented by the session of IMSI；
S32: obtain main serving cell Cell1 according to current MR data and detect each adjacent area Cell2,
Cell3……CellN}；
S33: set { Cell1, Cell2 ... CellN } is matched with room divided data library, if CellN is room branch website,
Then it is put into set { Celli1, Celli2 ... CelliN }；
S34: longitude and latitude track { { lon1, lat1 }, { lon2, the lat2 } ... of single user are obtained according to step S31
{ lonn, latn } }；Obtain each room branch website in room divided data library longitude and latitude { hlon1, hlat1 }, hlon2,
Hlat2 } ... { hlon, hlatm } }；
S35: according to step S34, obtaining the distance set { d1, d2 ... ... dn } of user trajectory Yu nearest room branch website,
Then average distance avgdist=(d1+d2+ ... dn)/n of user and room branch website are calculated；
S36: if average distance avgdist less than a thresholding A, and set { Celli1, the Celli2 ... of step S33
CelliN } it is not empty set, then the MR that active user's session generates is located at interior, otherwise enters step S4 and further judges.
It includes following sub-step that user mobility separation in the step S4, which calculates:
S41: being divided into grid by N*N meters for collected MR data in morning in the evening, launches by longitude and latitude to different grid
In lattice；
S42: in each grid in traversal step S41 each user longitude and latitude track { lon1, lat1 }, lon2,
Lat2 } ... { lonn, latn } }, the rate set { vk1, vk2 ... vkn } and variance of each user is obtained, in this, as every
A grid evening mobility feature, as classifier；
S43: collected MR on daytime data are segmented by IMSI and user conversation, as test object；
S44: according to S43 obtain single user longitude and latitude track { lon1, la1 }, { lon2, la2 } ... lonn,
Lan } }, calculate its rate { v1, v2 ... vk } and variance；
S45: in the rate mean value in S44 and the classifier in variance input S42, the indoor and outdoor of each user is exported
Separate confidence level b；
S46: if indoor and outdoor separation confidence level b is greater than a thresholding thre1, judge the MR that active user's session on daytime generates
For interior, if indoor and outdoor separates confidence level b less than a thresholding thre2, judge MR that user conversation generates be it is outdoor, otherwise into
The judgement of row step S5.
It includes following sub-step that outdoor test feature separation in the step S5, which calculates:
S51: being divided into grid by N*N meters for the DT/CQT data of outdoor test, launches by longitude and latitude to different grids
In, form M grid altogether, each grid determine its center longitude { lon1, la1 }, { lon2, la2 } ... lonm,
lam}}
S52: it chooses BP artificial neural network algorithm and establishes disaggregated model, by each sampled point of each DT/CQT grid
The first six strong RSCP, EC/IO value uses 1 or 0 to represent indoor and outdoors as output layer parameter, each DT/ as input layer parameter
CQT grid is trained to a classifier, shares M classifier；
S53: collected MR on daytime data are segmented by IMSI and user conversation, as test object；
S54: being divided into grid by N*N meters for the MR of each session of each user, launches by longitude and latitude to different grid
In lattice, K grid is formed altogether, and each grid determines its center longitude { mrlon1, mrla1 }, { mrlon2, mrla2 } ...
{ mrlonk, mrlak }；
S55: the corresponding neural network classifier of each MR grid in step S54 is found out according to minimum distance；
S56: traversing the MR grid in each S54, extracts every MR in MR grid, obtains the first six strong RSCP and EC/
12 input parameters of the NO as classifier；
S57: 12 parameters are input in the corresponding neural network classifier of current MR grid, every MR is obtained after operation
Indoor and outdoor attribute.
The calculation method of the average distance of user and room branch website includes: the warp for setting Target cell A in the step S35
Latitude is (LonA, LatA), and the longitude and latitude of Target cell B is (LonB, LatB), according to the benchmark of 0 degree of warp, east longitude degree of learning from else's experience
Positive value, west longitude degree of learning from else's experience negative value, north latitude takes 90- latitude value, and south latitude takes 90+ latitude value, then by it is above-mentioned it is processed after two
Point is counted as (MLonA, MLatA) and (MLonB, MLatB)；It is derived according to triangle, it is available to calculate the as follows of two o'clock distance
Formula:
C=sin (MLatA) * sin (MLatB) * cos (MLonA-MLonB)+cos (MLatA) * cos (MLatB)
Dist=R*Arccos (C) * Pi/180
Dist is Target cell at a distance from abutting subdistrict；
The method for calculating user trajectory and any room branch website average distance:
Ci=sin (lati) * sin (hlatk) * cos (loni-hlonk)+cos (lati) * cos (hlatk) (1 <=i
<=n)
K value represents k-th of Room point in above formula, and Ci represents in user trajectory ith sample point at a distance from the k of room minute,
Avgdist indicates the average distance of entire user trajectory Yu k-th of Room branch website；
According to user trajectory at a distance from any room point, minimum distance then can be obtained are as follows:
Minavgdist=min (avgidst1, avgdist2......avgdistk).
It includes following sub-step that DT/CQT test sample point grid, which is launched, in the step S51:
S511: longitude and latitude is uniformly processed, and retains 5 decimals, if digit is more than 5 after collected longitude and latitude decimal point,
Digit after then dispensing the 5th, if mending 0 below less than 5；
S512: determining the grid identifier of every DT/CQT data, by taking 50 meters of * 50 meters of grids as an example, intercepts longitude and latitude decimal
The 4th is assumed to be a after point, and the value range of a is 0≤a≤9, and it is 0 that a, which unifies value, if a≤5；If a > 5, a value are 5；
First 7 of longitude are added with a the grid identifier for being added with first 6 of latitude with a and obtaining every DT/CQT；
S513: determining the grid identifier of each DT/CQT sampled point, then determines each sampled point and which grid phase again
Corresponding, i.e. the grid of completion DT/CQT test sample point is launched.
The method that BP artificial neural network algorithm establishes disaggregated model in the step S52 includes following sub-step:
S521: data normalization processing, normalization algorithm are as follows:
Y=(x-min)/(max-min) (0 <=<=1 y)
X in above formula is RSCP the or Ec/No value of specific MR sampled point N pilot tone, and min is the minimum of RSCP or Ec/No
Value, respectively -112 and -24；Max is the maximum value of RSCP or Ec/No, and respectively -40 and -1, y is after normalizing
Value, is mapped as 0 to 1 value range；
S522: the weight array iptHidWeights [12] [6] and hidden layer of initialization input layer to hidden layer are to defeated
The weight array hidOptWeights [6] [2] of layer out, initial value are generated using random function；
S523: weighted value that input layer propagates to hidden layer and hidden layer are obtained to the weighted value of output layer, such as following formula:
yi=f (neti)
Wij is input layer to hidden layer and hidden layer to the weighted value of output layer in above formula, and θ indicates a threshold value default
Take 0, xij be i-th input layer to j-th hidden layer input value or i-th of hidden layer to j-th of output layer input value,
yiFor the output valve of neuron i；F (x) is transfer function；The indoor and outdoor attribute that each MR can be obtained by above-mentioned three formula is defeated
Out；
S524: according to the weight array iptHidWeights [12] of the reality output in theoretical output calibration step S423
[6] and hidOptWeights [6] [2], continuous iterative learning.
The beneficial effects of the present invention are: the present invention provides a kind of various dimensions measurement report indoor and outdoor separation methods, effectively
The limitation for compensating for MR rasterizing appraisal procedure, by being separated based on room sub-signal source partition method, based on outdoor test feature
Method is based on user mobility partition method, forms various dimensions indoor and outdoor MR separation method to MR data inside and outside divided chamber, and then shape
At indoor and outdoor MR grid, to assess the wireless of the indoor scenes such as residential building, office building, hotel and various outdoor road scenes
Network quality.
Detailed description of the invention
Fig. 1 is various dimensions measurement report indoor and outdoor separation method schematic diagram；
Fig. 2 is that room divided data separates calculation method schematic diagram；
Fig. 3 is that user mobility separates calculation method schematic diagram；
Fig. 4 is that outdoor test feature separates calculation method schematic diagram.
Specific embodiment
Technical solution of the present invention is described in further detail with reference to the accompanying drawing, but protection scope of the present invention is not limited to
It is as described below.
As shown in Figure 1, various dimensions measurement report indoor and outdoor separation method, it includes following sub-step:
S1: data acquisition, from server by acquisition interface acquisition MR data, drive test data, room divide basic data and
GIS data；Collect using the MR data on daytime as detection；Using evening MR data, drive test data, room divided data as training set；
S2: MR data are cleaned, and remove invalid data and noise；
S3: MR data are subjected to the separation of room divided data and are calculated, the room for calculating MR data to be detected splits reliability a, judges room
The relationship of reliability a Yu room point threshold value A are split, if room splits reliability a less than room point threshold value A, separation terminates, jump procedure
S6, if otherwise jump procedure S4；
S4: MR data are subjected to user mobility separation and are calculated, the mobility confidence level b of MR data to be detected is calculated, sentences
The relationship of disconnected indoor and outdoor separation confidence level b and threshold value thre, if indoor and outdoor separation confidence level b is higher than threshold value thre1, and small
In threshold value thre2, then separation terminates, jump procedure S6, otherwise jump procedure S5；
S5: carrying out outdoor test feature separation for MR data and calculate, calculate the test feature confidence level of MR data to be detected,
Test feature confidence level c input neural network classifier is judged that then separation terminates, jump procedure S6, if not can be carried out
Separation, it is impossible to which the MR data of differentiation discard；
S6: current MR separating resulting, the i.e. result that output MR belongs to indoor or outdoors are exported；
The threshold value A value is 432, and the value of the threshold value thre1 is 1.2, the threshold value thre2
Value be 0.8.
As shown in Fig. 2, it includes following sub-step that the room divided data separation in the step S3, which calculates:
S31: collected MR data are segmented by the session of IMSI；
S32: obtain main serving cell Cell1 according to current MR data and detect each adjacent area Cell2,
Cell3……CellN}；
S33: set { Cell1, Cell2 ... CellN } is matched with room divided data library, if CellN is room branch website,
Then it is put into set { Celli1, Celli2 ... CelliN }；
S34: longitude and latitude track { { lon1, lat1 }, { lon2, the lat2 } ... of single user are obtained according to step S31
{ lonn, latn } }；Obtain each room branch website in room divided data library longitude and latitude { hlon1, hlat1 }, hlon2,
Hlat2 } ... { hlon, hlatm } }；
S35: according to step S34, obtaining the distance set { d1, d2 ... ... dn } of user trajectory Yu nearest room branch website,
Then average distance avgdist=(d1+d2+ ... dn)/n of user and room branch website are calculated；
S36: if average distance avgdist less than a thresholding A, and set { Celli1, the Celli2 ... of step S33
CelliN } it is not empty set, then the MR that active user's session generates is located at interior, otherwise enters step S4 and further judges.
As shown in figure 3, it includes following sub-step that the user mobility separation in the step S4, which calculates:
S41: being divided into grid by N*N meters for collected MR data in morning in the evening, launches by longitude and latitude to different grid
In lattice；
S42: in each grid in traversal step S41 each user longitude and latitude track { lon1, lat1 }, lon2,
Lat2 } ... { lonn, latn } }, the rate set { vk1, vk2 ... vkn } and variance of each user is obtained, in this, as every
A grid evening mobility feature, as classifier；
S43: collected MR on daytime data are segmented by IMSI and user conversation, as test object；
S44: according to S43 obtain single user longitude and latitude track { lon1, la1 }, { lon2, la2 } ... lonn,
Lan } }, calculate its rate { v1, v2 ... vk } and variance；
S45: in the rate mean value in S44 and the classifier in variance input S42, the indoor and outdoor of each user is exported
Separate confidence level b；
S46: if indoor and outdoor separation confidence level b is greater than a thresholding thre1, judge the MR that active user's session on daytime generates
For interior, if indoor and outdoor separates confidence level b less than a thresholding thre2, judge MR that user conversation generates be it is outdoor, otherwise into
The judgement of row step S5.
As shown in figure 4, it includes following sub-step that the outdoor test feature separation in the step S5, which calculates:
S51: being divided into grid by N*N meters for the DT/CQT data of outdoor test, launches by longitude and latitude to different grids
In, form M grid altogether, each grid determine its center longitude { lon1, la1 }, { lon2, la2 } ... lonm,
lam}}
S52: it chooses BP artificial neural network algorithm and establishes disaggregated model, by each sampled point of each DT/CQT grid
The first six strong RSCP, EC/IO value uses 1 or 0 to represent indoor and outdoors as output layer parameter, each DT/ as input layer parameter
CQT grid is trained to a classifier, shares M classifier；
S53: collected MR on daytime data are segmented by IMSI and user conversation, as test object；
S54: being divided into grid by N*N meters for the MR of each session of each user, launches by longitude and latitude to different grid
In lattice, K grid is formed altogether, and each grid determines its center longitude { mrlon1, mrla1 }, { mrlon2, mrla2 } ...
{ mrlonk, mrlak }；
S55: the corresponding neural network classifier of each MR grid in step S54 is found out according to minimum distance；
S56: traversing the MR grid in each S54, extracts every MR in MR grid, obtains the first six strong RSCP and EC/
12 input parameters of the NO as classifier；
S57: 12 parameters are input in the corresponding neural network classifier of current MR grid, every MR is obtained after operation
Indoor and outdoor attribute.
The earth is the spheroid of an intimate standard, its equatorial radius is 6378.140 kms, and polar radius is
6356.755 km, 6371.004 km of mean radius.If on the basis of 0 degree of warp, according to earth surface any two
The surface distance that the longitude and latitude of point can calculate this point-to-point transmission (is ignored earth surface landform here and is missed to bring is calculated
Difference, only theoretic estimated value).The calculation method packet of the average distance of user and room branch website in the step S35
It includes: setting the longitude and latitude of Target cell A as (LonA, LatA), the longitude and latitude of Target cell B is (LonB, LatB), is passed through according to 0 degree
The benchmark of line, the positive value of east longitude degree of learning from else's experience, west longitude degree of learning from else's experience negative value, north latitude take 90 latitude values, and south latitude takes 90+ latitude value, then passes through
Two o'clock after crossing above-mentioned process is counted as (MLonA, MLatA) and (MLonB, MLatB)；It is derived according to triangle, it is available
Calculate the following formula of two o'clock distance:
C=sin (MLatA) * sin (MLatB) * cos (MLonA-MLonB)+cos (MLatA) * cos (MLatB)
Dist=R*Arccos (C) * Pi/180
Dist is Target cell at a distance from abutting subdistrict；
The method for calculating user trajectory and any room branch website average distance:
Ci=sin (lati) * sin (hlatk) * cos (loni hlonk)+cos (lati) * cos (hlatk) (1 <=i
<=n)
K value represents k-th of Room point in above formula, and Ci represents in user trajectory ith sample point at a distance from the k of room minute,
Avgdist indicates the average distance of entire user trajectory Yu k-th of Room branch website；
According to user trajectory at a distance from any room point, minimum distance then can be obtained are as follows:
Min avgdist=min (avgidst1, avgdist2. ... ..avgdistk).
It includes following sub-step that DT/CQT test sample point grid, which is launched, in the step S51:
S511: longitude and latitude is uniformly processed, and retains 5 decimals, if digit is more than 5 after collected longitude and latitude decimal point,
Digit after then dispensing the 5th, if mending 0 below less than 5；
S512: determining the grid identifier of every DT/CQT data, by taking 50 meters of * 50 meters of grids as an example, intercepts longitude and latitude decimal
The 4th is assumed to be a after point, and the value range of a is 0≤a≤9, and it is 0 that a, which unifies value, if a≤5；If a > 5, a value are 5；
First 7 of longitude are added with a the grid identifier for being added with first 6 of latitude with a and obtaining every DT/CQT, such as:
50 meters * 50 meters of the grid identifier of (106.34562,29.38127) is (106.3455,29.3810).
S513: determining the grid identifier of each DT/CQT sampled point, then determines each sampled point and which grid phase again
Corresponding, i.e. the grid of completion DT/CQT test sample point is launched.
The method that BP artificial neural network algorithm establishes disaggregated model in the step S52 includes following sub-step:
S521: data normalization processing, normalization algorithm are as follows:
Y=(x-min)/(max-min) (0 <=<=1 y)
X in above formula is RSCP the or Ec/No value of specific MR sampled point N pilot tone, and min is the minimum of RSCP or Ec/No
Value, respectively -112 and -24；Max is the maximum value of RSCP or Ec/No, and respectively -40 and -1, y is after normalizing
Value, is mapped as 0 to 1 value range；
S522: the weight array iptHidWeights [12] [6] and hidden layer of initialization input layer to hidden layer are to defeated
The weight array hidOptWeights [6] [2] of layer out, initial value are generated using random function；
S523:The weighted value and hidden layer for propagating to hidden layer to input layer are to the weighted value of output layer, such as following formula:
yi=f (neti)
Wij is input layer to hidden layer and hidden layer to the weighted value of output layer in above formula, and θ indicates a threshold value default
Take 0, xij be i-th input layer to j-th hidden layer input value or i-th of hidden layer to j-th of output layer input value,
yiFor the output valve of neuron i；F (x) is transfer function；The indoor and outdoor attribute that each MR can be obtained by above-mentioned three formula is defeated
Out；
S524: according to the weight array iptHidWeights [12] of the reality output in theoretical output calibration step S423
[6] and hidOptWeights [6] [2], continuous iterative learning.
The present invention is by being based on room sub-signal source partition method, being based on outdoor test feature partition method, based on user mobility
Partition method forms various dimensions indoor and outdoor MR separation method to MR data inside and outside divided chamber, and then MR grid inside and outside forming chamber, uses
To assess the quality of wireless network of the indoor scenes such as residential building, office building, hotel and various outdoor road scenes.
Claims (6)
1. various dimensions measurement report indoor and outdoor separation method, it is characterised in that: it includes following sub-step:
S1: data acquisition divides basic data and GIS number by acquisition interface acquisition MR data, drive test data, room from server
According to；
S2: MR data are cleaned, and remove invalid data and noise；
S3: MR data are subjected to the separation of room divided data and are calculated, the room for calculating MR data to be detected splits reliability a, judges that room is split
The relationship of threshold value A is divided in reliability a and room, if room splits reliability a less than room point threshold value A, separation terminates, jump procedure S6,
If otherwise jump procedure S4；
S4: MR data are subjected to user mobility separation and are calculated, the mobility confidence level b of MR data to be detected is calculated, judges room
The relationship of inside and outside separation confidence level b and threshold value thre if indoor and outdoor separation confidence level b is higher than threshold value thre1, and are less than door
Limit value thre2, then separation terminates, jump procedure S6, otherwise jump procedure S5；
S5: MR data are subjected to outdoor test feature separation and are calculated, the test feature confidence level of MR data to be detected is calculated, will survey
Examination feature confidence level c input neural network classifier is judged that then separation terminates, jump procedure S6, if not can be carried out separation,
Undistinguishable MR data are discarded；
S6: current MR separating resulting, the i.e. result that output MR belongs to indoor or outdoors are exported；
The threshold value A value is 432, and the value of the threshold value thre1 is 1.2, and the threshold value thre2's takes
Value is 0.8；
It includes following sub-step that room divided data separation in the step S3, which calculates:
S31: collected MR data are segmented by the session of IMSI；
Each adjacent area { Cell2, the Cell3 ... that S32: obtaining main serving cell Cell1 according to current MR data and detects
CellN}；
S33: set { Cell1, Cell2 ... CellN } is matched with room divided data library, if CellN is room branch website, is put
Enter set { Celli1, Celli2 ... CelliN }；
S34: according to step S31 obtain single user longitude and latitude track { lon1, lat1 }, { lon2, lat2 } ... lonn,
latn}}；Obtain longitude and latitude { { hlon1, hlat1 }, { hlon2, the hlat2 } ... of each room branch website in room divided data library
{ hlon, hlatm } }；
S35: according to step S34, the distance set { d1, d2 ... ... dn } of user trajectory Yu nearest room branch website is obtained, then
Calculate user and room branch website average distance avgdist=(d1+d2+ ... dn }/n；
S36: if average distance avgdist less than a thresholding A, and set { Celli1, the Celli2 ... of step S33
CelliN } it is not empty set, then the MR that active user's session generates is located at interior, otherwise enters step S4 and further judges.
2. various dimensions measurement report indoor and outdoor separation method according to claim 1, it is characterised in that: the step S4
In user mobility separation calculate include following sub-step:
S41: being divided into grid by N*N meters for collected MR data in morning in the evening, launches by longitude and latitude into different grids；
S42: in each grid in traversal step S41 each user longitude and latitude track { lon1, lat1 }, lon2,
Lat2 } ... { lonn, latn } }, the rate set { vk1, vk2 ... vkn } and variance of each user is obtained, in this, as every
A grid evening mobility feature, as classifier；
S43: collected MR on daytime data are segmented by IMSI and user conversation, as test object；
S44: obtaining the longitude and latitude track { { lon1, la1 }, { lon2, la2 } ... { lonn, lan } } of single user according to S43, meter
Calculate its rate { v1, v2 ... vk } and variance；
S45: in the rate mean value in S44 and the classifier in variance input S42, the indoor and outdoor separation of each user is exported
Confidence level b；
S46: if indoor and outdoor separation confidence level b is greater than a thresholding thre1, judge the MR of active user's session on daytime generation for room
It is interior, if indoor and outdoor separates confidence level b less than a thresholding thre2, judge that the MR that user conversation generates for outdoor, is otherwise walked
The judgement of rapid S5.
3. various dimensions measurement report indoor and outdoor separation method according to claim 1, it is characterised in that: the step S5
In outdoor test feature separation calculate include following sub-step:
S51: being divided into grid by N*N meters for the DT/CQT data of outdoor test, launches by longitude and latitude into different grids, altogether
M grid is formed, each grid determines its center longitude { { lon1, la1 }, { lon2, la2 } ... { lonm, lam } }
S52: it chooses BP artificial neural network algorithm and establishes disaggregated model, by the first six of each sampled point of each DT/CQT grid
Strong RSCP, EC/IO value uses 1 or 0 to represent indoor and outdoors as output layer parameter, each DT/CQT grid as input layer parameter
Lattice are trained to a classifier, share M classifier；
S53: collected MR on daytime data are segmented by IMSI and user conversation, as test object；
S54: being divided into grid by N*N meters for the MR of each session of each user, launches by longitude and latitude into different grids,
K grid is formed altogether, and each grid determines its center longitude { mrlon1, mrla1 }, { mrlon2, mrla2 } ...
{ mrlonk, mrlak }；
S55: the corresponding neural network classifier of each MR grid in step S54 is found out according to minimum distance；
S56: traversing the MR grid in each S54, extracts every MR in MR grid, obtains the first six strong RSCP and EC/NO and makees
For 12 input parameters of classifier；
S57: 12 parameters are input in the corresponding neural network classifier of current MR grid, are obtained after operation in every room MR
Outdoor attribute.
4. various dimensions measurement report indoor and outdoor separation method according to claim 1, it is characterised in that: the step
The calculation method of the average distance of user and room branch website includes: to set the longitude and latitude of Target cell A as (LonA, LatA) in S35,
The longitude and latitude of Target cell B is (LonB, LatB), according to the benchmark of 0 degree of warp, the positive value of east longitude degree of learning from else's experience, west longitude degree of learning from else's experience
Negative value, north latitude take 90- latitude value, and south latitude takes 90+ latitude value, then by it is above-mentioned it is processed after two o'clock be counted as (MLonA,
) and (MLonB, MLatB) MLatA；It is derived according to triangle, the available following formula for calculating two o'clock distance:
C=sin (MLatA) * sin (MLatB) * cos (MLonA-MLonB)+coS (MLatA) * cos (MLatB)
Dist=R*Arccos (C) * Pi/180
Dist is Target cell at a distance from abutting subdistrict；
The method for calculating user trajectory and any room branch website average distance:
Ci=sin (lati) * sin (hlatk) * cos (loni-hlonk)+cos (lati) * cos (hlatk) (1 <=i <=
n)
K value represents k-th of Room point in above formula, and Ci represents in user trajectory that ith sample point is at a distance from the k of room minute, avgdist table
Show the average distance of entire user trajectory Yu k-th of Room branch website；
According to user trajectory at a distance from any room point, minimum distance then can be obtained are as follows:
Minavgdist=min (avgidst1, avgdist2......avgdistk).
5. various dimensions measurement report indoor and outdoor separation method according to claim 3, it is characterised in that: the step
It includes following sub-step that DT/CQT test sample point grid, which is launched, in S51:
S511: longitude and latitude is uniformly processed, and retains 5 decimals, if digit is more than 5 after collected longitude and latitude decimal point, saves
Digit after omitting the 5th, if mending 0 below less than 5；
S512: determining the grid identifier of every DT/CQT data, by taking 50 meters of * 50 meters of grids as an example, after intercepting longitude and latitude decimal point
4th is assumed to be a, and the value range of a is 0≤a≤9, and it is 0 that a, which unifies value, if a≤5；If a > 5, a value are 5；It will be through
First 7 of degree are added the grid identifier for being added with first 6 of latitude with a and obtaining every DT/CQT with a；
S513: determining the grid identifier of each DT/CQT sampled point, then determines that each sampled point is corresponding with which grid again,
The grid for completing DT/CQT test sample point is launched.
6. various dimensions measurement report indoor and outdoor separation method according to claim 3, it is characterised in that: the step
The method that BP artificial neural network algorithm establishes disaggregated model in S52 includes following sub-step:
S521: data normalization processing, normalization algorithm are as follows:
Y=(x-min)/(max-min) (0 <=<=1 y)
X in above formula is RSCP the or Ec/No value of specific MR sampled point N pilot tone, and min is the minimum value of RSCP or Ec/No,
Respectively -112 and -24；Max is the maximum value of RSCP or Ec/No, and respectively -40 and -1, y is the value after normalization, is reflected
Penetrate the value range for 0 to 1；
S522: the weight array iptHidWeights [12] [6] and hidden layer of initialization input layer to hidden layer to output layer
Weight array hidOptWeights [6] [2], initial value utilize random function generate；
S523: weighted value that input layer propagates to hidden layer and hidden layer are obtained to the weighted value of output layer, such as following formula:
yi=f (neti) (2)
Wij is input layer to hidden layer and hidden layer to the weighted value of output layer in above formula, and θ indicates that a threshold value default takes 0,
Xij is i-th input layer to the input value to j-th of output layer of input value or i-th of hidden layer of j-th of hidden layer, yiFor
The output valve of neuron i；F (x) is transfer function；The indoor and outdoor of each MR can be obtained by above-mentioned (1), (2) and (3) formula
Attribute output；
S524: according to the weight array iptHidWeights [12] [6] of the reality output in step S523 and
HidOptWeights [6] [2], continuous iterative learning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610127602.2A CN105744561B (en) | 2016-03-07 | 2016-03-07 | Various dimensions measurement report indoor and outdoor separation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610127602.2A CN105744561B (en) | 2016-03-07 | 2016-03-07 | Various dimensions measurement report indoor and outdoor separation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105744561A CN105744561A (en) | 2016-07-06 |
| CN105744561B true CN105744561B (en) | 2019-05-10 |
Family
ID=56249215
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610127602.2A Active CN105744561B (en) | 2016-03-07 | 2016-03-07 | Various dimensions measurement report indoor and outdoor separation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105744561B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105682136A (en) * | 2016-03-07 | 2016-06-15 | 四川亨通网智科技有限公司 | Indoor and outdoor separation method based on outdoor test characteristics |
| CN106211194B (en) * | 2016-07-28 | 2019-10-11 | 武汉虹信技术服务有限责任公司 | Separation method outside a kind of MR data room based on statistical model |
| CN107786985A (en) * | 2016-08-29 | 2018-03-09 | 中国移动通信集团广东有限公司 | Measurement report data sorting technique and device |
| CN108616900B (en) * | 2016-12-12 | 2021-06-11 | 中国移动通信有限公司研究院 | Method for distinguishing indoor and outdoor measurement reports and network equipment |
| CN109151845B (en) * | 2017-06-16 | 2021-11-09 | 中国移动通信集团云南有限公司 | Method, apparatus, electronic device and storage medium for identifying indoor hierarchical high-level cell |
| CN109429242B (en) * | 2017-08-21 | 2021-11-23 | 中国移动通信集团广西有限公司 | MR data indoor and outdoor separation method and device |
| CN108133001B (en) * | 2017-12-21 | 2020-03-27 | 重庆玖舆博泓科技有限公司 | MR indoor and outdoor separation method, device and medium |
| CN109996276B (en) * | 2017-12-29 | 2022-06-10 | 中国移动通信集团四川有限公司 | Network telephone traffic positioning evaluation method, device, equipment and medium |
| CN109769216B (en) * | 2018-12-28 | 2021-06-11 | 科大国创软件股份有限公司 | Method and device for grouping users in complex environment based on mobile phone signals |
| CN109769201B (en) * | 2018-12-28 | 2021-01-26 | 科大国创软件股份有限公司 | Smart city management platform capable of achieving accurate positioning of user |
| CN109714712B (en) * | 2018-12-28 | 2021-02-05 | 科大国创软件股份有限公司 | Method and device for dropping data to grid based on attribute matching |
| CN114363925B (en) * | 2021-12-16 | 2023-10-24 | 北京红山信息科技研究院有限公司 | Automatic network quality difference identification method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102769866A (en) * | 2012-06-18 | 2012-11-07 | 华为技术有限公司 | Method and equipment for distinguishing indoor business data from outdoor business data |
| CN103052104A (en) * | 2011-10-11 | 2013-04-17 | 华为技术有限公司 | Indoor distributed signal leakage check method and device |
| CN104244317A (en) * | 2013-06-08 | 2014-12-24 | 华为技术有限公司 | Method and device for determining indoor telephone traffic states and indoor telephone traffic amount |
| CN104427513A (en) * | 2013-08-30 | 2015-03-18 | 华为技术有限公司 | Identification method, device, network equipment and network system |
-
2016
- 2016-03-07 CN CN201610127602.2A patent/CN105744561B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103052104A (en) * | 2011-10-11 | 2013-04-17 | 华为技术有限公司 | Indoor distributed signal leakage check method and device |
| EP2747473A1 (en) * | 2011-10-11 | 2014-06-25 | Huawei Technologies Co., Ltd. | Method and device for detecting indoor distribution signal leakage |
| CN102769866A (en) * | 2012-06-18 | 2012-11-07 | 华为技术有限公司 | Method and equipment for distinguishing indoor business data from outdoor business data |
| CN104244317A (en) * | 2013-06-08 | 2014-12-24 | 华为技术有限公司 | Method and device for determining indoor telephone traffic states and indoor telephone traffic amount |
| CN104427513A (en) * | 2013-08-30 | 2015-03-18 | 华为技术有限公司 | Identification method, device, network equipment and network system |
Non-Patent Citations (1)
| Title |
|---|
| 基于MR特征分离技术的室分监控研究;郭景赞等;《邮电设计技术》;20120620;第3栏第7-8行，以及第2部分 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105744561A (en) | 2016-07-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105744561B (en) | Various dimensions measurement report indoor and outdoor separation method | |
| CN102149192B (en) | Cellular network wireless positioning method based on cooperation of mobile stations | |
| Gogolak et al. | Indoor fingerprint localization in WSN environment based on neural network | |
| Liu et al. | Deeplora: Learning accurate path loss model for long distance links in lpwan | |
| CN110650436B (en) | WiFi data-based position fingerprint database establishing and fitting method | |
| KR20190053470A (en) | Positioning system based on deep learnin and construction method thereof | |
| AU2019216706A1 (en) | Iterative ray-tracing for autoscaling of oblique ionograms | |
| CN103561463A (en) | RBF neural network indoor positioning method based on sample clustering | |
| CN104093205A (en) | Method for deploying anchor nodes of wireless positioning system based on received signal strength indication | |
| RU2013110010A (en) | ADAPTIVE METHOD FOR EVALUATING THE ELECTRONIC CONTENT OF THE IONOSPHERE | |
| CN103338509A (en) | WSN (wireless sensor network) indoor positioning method based on hidden markov models | |
| CN109313273A (en) | Record has the recording medium of indoor and outdoor determining program, indoor and outdoor to judge system, indoor and outdoor judgment method, mobile terminal and indoor and outdoor surroundings classification judging unit | |
| CN102685677A (en) | Indoor positioning method and device | |
| De Coensel et al. | Dynamic noise mapping based on fixed and mobile sound measurements | |
| CN110430578A (en) | The method for realizing cell Azimuth prediction based on mobile terminal data | |
| CN114936957A (en) | Urban PM25 concentration distribution simulation and scene analysis model based on mobile monitoring data | |
| CN105657726A (en) | PCI planning based base station location method | |
| CN104635202A (en) | Positioning and testing system and method for wireless multimedia sensor network | |
| Arai et al. | Color radiomap interpolation for efficient fingerprint wifi-based indoor location estimation | |
| Ding et al. | Regional propagation model based fingerprinting localization in indoor environments | |
| Zhang et al. | Cultivated Land Monitoring System Based on Dynamic Wake-Up UAV and Wireless of Distributed Storage. | |
| CN116980933A (en) | Antenna parameter adjustment method and device, electronic equipment and storage medium | |
| CN116471661A (en) | Method for positioning opportunistic signal analysis method of underground pipe gallery based on ray tracking | |
| CN105682136A (en) | Indoor and outdoor separation method based on outdoor test characteristics | |
| CN107682864B (en) | Base station construction method based on coverage rate evaluation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |