CN108738032A - A kind of network work layout method based on gradient descent method - Google Patents
A kind of network work layout method based on gradient descent method Download PDFInfo
- Publication number
- CN108738032A CN108738032A CN201810380075.5A CN201810380075A CN108738032A CN 108738032 A CN108738032 A CN 108738032A CN 201810380075 A CN201810380075 A CN 201810380075A CN 108738032 A CN108738032 A CN 108738032A
- Authority
- CN
- China
- Prior art keywords
- sampled point
- running parameter
- indicate
- antenna
- capped
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000011478 gradient descent method Methods 0.000 title claims abstract description 34
- 230000000694 effects Effects 0.000 claims abstract description 67
- 238000009795 derivation Methods 0.000 claims abstract description 18
- 238000005457 optimization Methods 0.000 claims abstract description 15
- 230000006870 function Effects 0.000 claims description 14
- 230000005540 biological transmission Effects 0.000 claims description 13
- 238000005070 sampling Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 abstract description 7
- 230000008569 process Effects 0.000 description 5
- 238000010295 mobile communication Methods 0.000 description 4
- 241000208340 Araliaceae Species 0.000 description 3
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 3
- 235000003140 Panax quinquefolius Nutrition 0.000 description 3
- 235000008434 ginseng Nutrition 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008054 signal transmission Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000011438 discrete method Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
- H04W16/18—Network planning tools
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Radio Transmission System (AREA)
- Variable-Direction Aerials And Aerial Arrays (AREA)
Abstract
The present invention provides a kind of network work layout method based on gradient descent method, can improve the stability and accuracy of running parameter derivative vector result of calculation.The method includes:Obtain running parameter to be optimized;According to the running parameter to be optimized of acquisition, determine the whole coverage rate of all sampled points in target area to be optimized, the entirety coverage rate is equal to the mean value for the coverage effect that all sampled points are capped, successive value of the capped coverage effect of each sampled point between [0,1];According to the whole coverage rate of all sampled points the derivative vector of running parameter is determined using continuous Rule for derivation;According to the derivative of determining running parameter vector, the running parameter after optimization is determined.The optimization that the present invention is suitable for network running parameter operates.
Description
Technical field
The present invention relates to mobile communication fields, particularly relate to a kind of network work layout side based on gradient descent method
Method.
Background technology
Existing mobile communications network is based on cellular network, and by disposing multiple base stations in a network, these base stations are set
Standby multi-section (orientation) antenna that is assembled with emits wireless signal to different directions, realizes that the signal of communication in a region covers jointly
Lid.In each base station, generally comprises 3 (there is a situation where there are 2-4 individually) radio frequency units and antenna, every antenna possess
Respective network running parameter (abbreviation work ginseng), such as:Azimuth, angle of declination, transmission power etc..
The network planning and the network optimization are the important technical links in mobile communications network.The network planning is typically referred to first
The considerations of stage beginning is to rough estimate and the layout of network engineering in mobile communication;The network optimization is then by rough in planning
The parameter of estimation carries out tuning, to reach better coverage effect.The two is referred to collectively herein as the network optimization.
The Thoughts of the network optimization are to adjust some work ginsengs for being easy to adjustment, to reach better coverage effect.In face
Into the network optimization of coverage rate, coverage effect is based on meeting signal transmission matter in sampled point total in target area to be optimized
The quantity of the sampled point of amount and determination.Meet signal transmission quality judgment criteria be usually according to different covering demands and
Setting, generally has following several judging quotas (i.e.:The parameter of coverage effect):
1) Reference Signal Received Power (Reference Signal Receiving Power, RSRP);
2) Signal to Interference plus Noise Ratio (Signal to Interference plus Noise Ratio, SINR);
3) RSRP and SINR considers simultaneously.
In the prior art, network running parameter is optimized using the gradient descent method based on numerical solution, from covering,
Unlapped discrete state (0 or 1 problem) solves coverage effect, and work is found out using difference derivation method according to the coverage effect
The derivative vector of parameter, the difference derivation method are expressed as:
Wherein, C indicates coverage effect, is defined as the ratio that capped sampled point accounts for all sampled points,Represent kth
For the i item running parameters of antenna j,Indicate increment of the kth for the i item running parameters of antenna j.
In this way, running parameter derivative vector result of calculation can be caused unstable, the size of running parameter derivative vector depends on
In △ xi,jSize, △ xi,jIt then calculates greatly very much inaccurate;Since sampled point only covers and does not cover two kinds of effects, △
xi,jIt is too small, molecule will not be had an impact, it is 0 to lead to derivative, cannot update running parameter, and at boundary, derivative is excessive, and
And △ xi,jIt is also easy to produce calculating error when too small, will produce error value in undifferentiable point.
Invention content
The network work layout method based on gradient descent method that the technical problem to be solved in the present invention is to provide a kind of,
To solve the gradient descent method based on numerical solution present in the prior art, running parameter derivative vector result of calculation can be caused
Unstable problem.
In order to solve the above technical problems, to provide a kind of network running parameter based on gradient descent method excellent for the embodiment of the present invention
Change method, including:
Obtain running parameter to be optimized;
According to the running parameter to be optimized of acquisition, the whole covering of all sampled points in target area to be optimized is determined
Rate, the entirety coverage rate are equal to the mean value for the coverage effect that all sampled points are capped, the capped covering of each sampled point
Successive value of the effect between [0,1];
According to the whole coverage rate of all sampled points the derivative vector of running parameter is determined using continuous Rule for derivation;
According to the derivative of determining running parameter vector, the running parameter after optimization is determined.
Further, acquisition running parameter to be optimized includes:
Obtain Downtilt and antenna azimuth;
Calculate sampled point angle of declination and sampled point azimuth of the antenna relative to all sampled points;
According to the Downtilt of acquisition, antenna azimuth, and the sampled point angle of declination and sampled point orientation that are calculated
Angle determines horizontal angle and vertical angle.
Further, the horizontal angle is expressed as:Alpha=azimuth_g-P_azimuthk;
The vertical angle is expressed as:Beta=P_tiltk-tilt_g;
Wherein, alpha indicates horizontal angle, P_azimuthkIndicate the antenna azimuth of kth time iteration, azimuth_g tables
Show sampled point azimuth;Beta indicates vertical angle, P_tiltkIndicate that the Downtilt of kth time iteration, tilt_g indicate sampling
Point angle of declination.
Further, the running parameter to be optimized according to acquisition, determines all samplings in target area to be optimized
Putting capped coverage rate successive value includes:
Determine the parameter for judging coverage effect;
According to the running parameter to be optimized of acquisition, the value for judging coverage effect parameter is determined;
According to the value of determining judge coverage effect parameter all in target area to be optimized adopt is obtained in conjunction with S type functions
The capped coverage effect of sampling point, successive value of the capped coverage effect of each sampled point between [0,1];
The coverage effect being capped to all sampled points is averaged, and the whole coverage rate of all sampled points is obtained.
Further, it is determined that the parameter for judging coverage effect include:Reference Signal Received Power and signal with it is dry
Disturb plus noise ratio;
The running parameter to be optimized according to acquisition determines that all sampled points in target area to be optimized are capped
Coverage rate successive value includes:
According to the horizontal angle and vertical angle of acquisition, gain of the sampled point to antenna is determined;
Antenna transmission power is set, calculates antenna to the path loss between sampled point;
According to determining gain, the antenna transmission power of setting, the path loss that is calculated, Reference Signal Received Power is determined
And Signal to Interference plus Noise Ratio;
It is obtained to be optimized in conjunction with S type functions according to determining Reference Signal Received Power and Signal to Interference plus Noise Ratio
The capped coverage effect of all sampled points in target area, the capped coverage effect of each sampled point is between [0,1]
Successive value;
The coverage effect being capped to all sampled points is averaged, and the whole coverage rate of all sampled points is obtained.
Further, it is determined that the gain of sampled point to antenna be expressed as:
Gain=f1 (alpha)-(abs (alpha)/pi) * (f1 (pi)-f2 (pi-beta))-
(1-abs(alpha)/pi)*(f1(0)-f2(beta))
Wherein, Gain indicates that gain, abs indicate that ABS function, f1 and f2 are that the gain to horizontal angle and vertical angle is quasi-
Close function.
Further, the Reference Signal Received Power is expressed as:
RSSIi,j=Powerj+Gaini,j-Pathlossi,j
Wherein, RSRPiIndicate the Reference Signal Received Power of sampled point i, RSSIi,jIndicate sampled point i connecing to antenna j
The intensity instruction of the collection of letters number, PowerjIndicate the transmission power of antenna j, Gaini,jIndicate the gain of sampled point i to antenna j,
Pathlossi,jPath loss between expression antenna j to sampled point i.
Further, the Signal to Interference plus Noise Ratio is expressed as:
Wherein, SINRiIndicate the Signal to Interference plus Noise Ratio of sampled point i, NoiseiIndicate the noise of sampled point i,
RSRPiIndicate the Reference Signal Received Power of sampled point i, RSSIi,jIndicate that the intensity of the reception signal of sampled point i to antenna j refers to
Show.
Further, the capped coverage effect of each sampled point is expressed as:
Coverpoint=sigmoid (RSRP-Thrsrp)*sigmoid(SINR-Thsinr)
Wherein, coverpoint indicates that the capped coverage effect of sampled point, SINR indicate Signal to Interference plus Noise Ratio,
RSRP indicates Reference Signal Received Power, Thrsrp、ThsinrPre-set RSRP threshold values and SINR threshold values are indicated respectively,
Sigmoid (x) is S type functions,
Further, the derivative vector according to determining running parameter determines that the running parameter after optimization includes:
Utilize formula Pk+1=Pk+learning_rate*DkUpdate the running parameter P in+1 generation of kthk+1, until meeting default
Maximum iteration, wherein PkIndicate the running parameter in kth generation, DkIndicate the derivative vector of the running parameter in kth generation,
Learning_rate indicates learning rate.
The above-mentioned technical proposal of the present invention has the beneficial effect that:
In said program, running parameter to be optimized is obtained;According to the running parameter to be optimized of acquisition, determine to be optimized
The whole coverage rate of all sampled points in target area, the entirety coverage rate are equal to the coverage effect that all sampled points are capped
Mean value, successive value of the capped coverage effect of each sampled point between [0,1];According to the entirety covering of all sampled points
Rate determines the derivative vector of running parameter using continuous Rule for derivation;According to the derivative of determining running parameter vector, determine
Running parameter after optimization, in such manner, it is possible to coverage effect standard transformed to from covering, unlapped discrete state much general
The capped successive value of rate so that derivation process can be solved directly using continuous Rule for derivation, and work ginseng can be improved
The stability and accuracy of number derivative vector result of calculation, to solve to use difference derivation to change in independent variable in the prior art
△xi,jRunning parameter cannot update when smaller, at boundary, derivative is excessive, be also easy to produce calculating error, and independent variable changes △ xi,jCompared with
Running parameter result of calculation inaccurate problem when big.
Description of the drawings
Fig. 1 is that the flow of the network work layout method provided in an embodiment of the present invention based on gradient descent method is illustrated
Figure.
Specific implementation mode
To keep the technical problem to be solved in the present invention, technical solution and advantage clearer, below in conjunction with attached drawing and tool
Body embodiment is described in detail.
The present invention is directed to the existing gradient descent method based on numerical solution, and running parameter derivative vector can be caused to calculate knot
The unstable problem of fruit provides a kind of network work layout method based on gradient descent method.
As shown in Figure 1, the network work layout method provided in an embodiment of the present invention based on gradient descent method, packet
It includes:
S101 obtains running parameter to be optimized;
S102 determines the entirety of all sampled points in target area to be optimized according to the running parameter to be optimized of acquisition
Coverage rate, the entirety coverage rate are equal to the mean value for the coverage effect that all sampled points are capped, what each sampled point was capped
Successive value of the coverage effect between [0,1];
S103, according to the whole coverage rate of all sampled points, using continuous Rule for derivation, determine the derivative of running parameter to
Amount;
S104 determines the running parameter after optimization according to the derivative of determining running parameter vector.
The network work layout method based on gradient descent method described in the embodiment of the present invention, obtains work to be optimized
Make parameter;According to the running parameter to be optimized of acquisition, the whole coverage rate of all sampled points in target area to be optimized is determined,
The entirety coverage rate is equal to the mean value for the coverage effect that all sampled points are capped, the capped coverage effect of each sampled point
For the successive value between [0,1];According to the whole coverage rate of all sampled points running parameter is determined using continuous Rule for derivation
Derivative vector;According to the derivative of determining running parameter vector, the running parameter after optimization is determined, in such manner, it is possible to will covering
Effect standard transforms to the capped successive value of much probability from covering, unlapped discrete state so that derivation process can
Directly to be solved using continuous Rule for derivation, the stability of running parameter derivative vector result of calculation and accurate can be improved
Property, to solve to use difference derivation in independent variable variation △ x in the prior arti,jRunning parameter cannot update, on side when smaller
Derivative is excessive when boundary, is also easy to produce calculating error, and independent variable changes △ xi,jRunning parameter result of calculation is inaccurate when larger asks
Topic.
In the present embodiment, when solving the coverage effect that each sampled point is capped, departure process serialization can obtain
Successive value between [0,1], so that derivation process can directly apply mechanically the Derivative Formula of continuous derivation to solve, example
Such as, the Derivative Formula of lnx functions isAnd it is solved without calculus of finite differences.
In the specific implementation mode of the aforementioned network work layout method based on gradient descent method, further,
It is described to obtain running parameter to be optimized and include:
Obtain Downtilt and antenna azimuth;
Calculate sampled point angle of declination and sampled point azimuth of the antenna relative to all sampled points;
According to the Downtilt of acquisition, antenna azimuth, and the sampled point angle of declination and sampled point orientation that are calculated
Angle determines horizontal angle and vertical angle.
In the present embodiment, Downtilt refers to the angle between antenna and horizontal line (for example, due east direction), sampled point
Angle of declination refer to due to angle caused by the height above sea level difference between antenna and sampled point, vertical angle be sampled point angle of declination with
Angle between Downtilt.
In the present embodiment, antenna azimuth is the angle between antenna and due east direction, sampled point azimuth be antenna with
Angle between sampled point between line and due east direction, horizontal angle are the folders between sampled point azimuth and antenna azimuth
Angle.
In the specific implementation mode of the aforementioned network work layout method based on gradient descent method, further,
The horizontal angle is expressed as:Alpha=azimuth_g-P_azimuthk;
The vertical angle is expressed as:Beta=P_tiltk-tilt_g;
Wherein, alpha indicates horizontal angle, P_azimuthkIndicate the antenna azimuth of kth time iteration, azimuth_g tables
Show sampled point azimuth;Beta indicates vertical angle, P_tiltkIndicate that the Downtilt of kth time iteration, tilt_g indicate sampling
Point angle of declination.
In the specific implementation mode of the aforementioned network work layout method based on gradient descent method, further,
The running parameter to be optimized according to acquisition determines that the coverage rate that all sampled points are capped in target area to be optimized connects
Continuous value includes:
Determine the parameter for judging coverage effect;
According to the running parameter to be optimized of acquisition, the value for judging coverage effect parameter is determined;
According to the value of determining judge coverage effect parameter all in target area to be optimized adopt is obtained in conjunction with S type functions
The capped coverage effect of sampling point, successive value of the capped coverage effect of each sampled point between [0,1];
The coverage effect being capped to all sampled points is averaged, and the whole coverage rate of all sampled points is obtained.
In the specific implementation mode of the aforementioned network work layout method based on gradient descent method, further,
The determining parameter for judging coverage effect includes:Reference Signal Received Power and Signal to Interference plus Noise Ratio;
The running parameter to be optimized according to acquisition determines that all sampled points in target area to be optimized are capped
Coverage rate successive value includes:
According to the horizontal angle and vertical angle of acquisition, gain of the sampled point to antenna is determined;
Antenna transmission power is set, calculates antenna to the path loss between sampled point;
According to determining gain, the antenna transmission power of setting, the path loss that is calculated, Reference Signal Received Power is determined
And Signal to Interference plus Noise Ratio;
It is obtained to be optimized in conjunction with S type functions according to determining Reference Signal Received Power and Signal to Interference plus Noise Ratio
The capped coverage effect of all sampled points in target area, the capped coverage effect of each sampled point is between [0,1]
Successive value;
The coverage effect being capped to all sampled points is averaged, and the whole coverage rate of all sampled points is obtained.
In the specific implementation mode of the aforementioned network work layout method based on gradient descent method, further,
The gain of determining sampled point to antenna is expressed as:
Gain=f1 (alpha)-(abs (alpha)/pi) * (f1 (pi)-f2 (pi-beta))-
(1-abs(alpha)/pi)*(f1(0)-f2(beta))
Wherein, Gain indicates that gain, abs indicate that ABS function, f1 and f2 are that the gain to horizontal angle and vertical angle is quasi-
Close function.
In the present embodiment, will calculate beneficiating process can macro, can allow subsequent derivative solution procedure using connect
Continuous Rule for derivation obtains analytic solutions, derivative results accurate stable.
In the present embodiment, f1 and f2 are obtained using Fourier space fitting according to the gain table of actual antennas.
In the specific implementation mode of the aforementioned network work layout method based on gradient descent method, further,
The Reference Signal Received Power is expressed as:
RSSIi,j=Powerj+Gaini,j-Pathlossi,j
Wherein, RSRPiIndicate the Reference Signal Received Power of sampled point i, RSSIi,jIndicate sampled point i connecing to antenna j
The intensity instruction of the collection of letters number, PowerjIndicate the transmission power of antenna j, Gaini,jIndicate the gain of sampled point i to antenna j,
Pathlossi,jPath loss between expression antenna j to sampled point i.
In the specific implementation mode of the aforementioned network work layout method based on gradient descent method, further,
The Signal to Interference plus Noise Ratio is expressed as:
Wherein, SINRiIndicate the Signal to Interference plus Noise Ratio of sampled point i, NoiseiIndicate the noise of sampled point i,
RSRPiIndicate the Reference Signal Received Power of sampled point i, RSSIi,jIndicate that the intensity of the reception signal of sampled point i to antenna j refers to
Show.
In the specific implementation mode of the aforementioned network work layout method based on gradient descent method, further,
The capped coverage effect of each sampled point is expressed as:
Coverpoint=sigmoid (RSRP-Thrsrp)*sigmoid(SINR-Thsinr)
Wherein, coverpoint indicates that the capped coverage effect of sampled point, SINR indicate Signal to Interference plus Noise Ratio,
RSRP indicates Reference Signal Received Power, Thrsrp、ThsinrPre-set RSRP threshold values and SINR threshold values are indicated respectively,
Sigmoid (x) is S type functions,
In the present embodiment, sigmoid (x) infinite can will be corresponded to just infinite input one 0 to 1 from negative
Output.
In the specific implementation mode of the aforementioned network work layout method based on gradient descent method, further,
The derivative vector according to determining running parameter determines that the running parameter after optimization includes:
Utilize formula Pk+1=Pk+learning_rate*DkUpdate the running parameter P in+1 generation of kthk+1, until meeting default
Maximum iteration, wherein PkIndicate the running parameter in kth generation, DkIndicate the derivative vector of the running parameter in kth generation,
Learning_rate indicates learning rate.
In the present embodiment, learning_rate indicates learning rate, refers to the newer speed of running parameter, belongs to hyper parameter, surpasses
Parameter is the parameter with effect by user oneself setup parameter value according to actual demand.
The network work layout method based on gradient descent method described in embodiment for a better understanding of the present invention,
By taking certain city as an example, it is assumed that running parameter to be optimized is antenna azimuth and Downtilt, and the parameter of coverage effect is
RSRP and SINR consider (can also only consider, RSRP or SINR, method and consider that RSRP is similar with SINR simultaneously) simultaneously, to
The network work layout method based on gradient descent method described in inventive embodiments is described in detail:
A1, initialization
A) maximum iteration interation is set, the antenna amount for including in target area to be optimized is num_
Sector, target area to be optimized is (i.e.:Certain described city) in include sampled point quantity be num_samples;
B) it is tilt (Downtilt), azimuth (antenna azimuth) to obtain running parameter to be optimized;
C) setting running parameter vector P_tilt1、P_azimuth1, wherein subscript 1 indicates the 1st iteration, running parameter
Vectorial P_tilt1、P_azimuth1Length be num_sector;
D) sampled point angle of declination tilt_g and sampled point azimuth azimuth_ of the antenna relative to all sampled points are calculated
g;Wherein, the size of tilt_g and azimuth_g is [num_samples*num_sector].
A2, loop iteration
If this is on behalf of kth time cycle, k=1,2,3 ..., interation
A) calculated level angle alpha and vertical angle beta:
Alpha=azimuth_g-P_azimuthk(alpha independent variable ranges:- pi~pi)
Beta=P_tiltk-tilt_g
Wherein, P_azimuthkIndicate the antenna azimuth of kth time iteration, P_tiltkUnder the antenna for indicating kth time iteration
Inclination angle.
B) gain G ain is calculated according to alpha and beta:
Gain=f1 (alpha)-(abs (alpha)/pi) * (f1 (pi)-f2 (pi-beta))-
(1-abs(alpha)/pi)*(f1(0)-f2(beta))
Wherein, abs is ABS function, and f1 and f2 are the gain fitting functions of azimuthal and angle of declination, according to reality
The gain table of antenna is obtained using Fourier space fitting.
In the present embodiment, the size of Gain is [num_samples*num_sector].
C) antenna transmission power is set, calculate antenna to the path loss between sampled point.
D) according to determining gain, the antenna transmission power of setting, the path loss that is calculated, determine that reference signal receives work(
Rate and Signal to Interference plus Noise Ratio:
RSSIi,j=Powerj+Gaini,j-Pathlossi,j
Wherein, RSRPiIndicate the Reference Signal Received Power of sampled point i, RSSIi,jIndicate sampled point i connecing to antenna j
The intensity instruction of the collection of letters number, PowerjIndicate the transmission power of antenna j, Gaini,jIndicate the gain of sampled point i to antenna j,
Pathlossi,jPath loss between expression antenna j to sampled point i, SINRiIndicate the Signal to Interference plus Noise Ratio of sampled point i,
NoiseiIndicate the noise of sampled point i.
In the present embodiment, it is num_sector that RSRP and SINR, which are length,.
E) coverage effect of all sampled points, each sampled point capped coverage effect vector coverpoint are calculated
It indicating, each element value of coverpoint is poorer closer to 0 coverage effect between [0,1] (successive value), closer to
1 coverage effect is better:
Coverpoint=sigmoid (RSRP-Thrsrp)*sigmoid(SINR-Thsinr)
Wherein, Thrsrp、ThsinrIndicate that pre-set RSRP threshold values and SINR threshold values, sigmoid (x) are S type letters respectively
Number,
In the present embodiment, sigmoid (x) Chinese meanings are S type functions, infinite will be corresponded to from negative to just infinite input
Output between one 0 to 1.
It should be noted that in discrete method, it is believed that when the value of RSRP is greater than or equal to RSRP threshold values and the value of SINR
When more than or equal to SINR threshold values, the sampled point is capped;In the present embodiment, coverpoint indicates coverage effect closer to 0
It is poorer, indicate that coverage effect is better closer to 1.
F) average value of all elements in the whole coverage rate coverarea=coverpoint of the sampled point is calculated
G) running parameter tilt is found out using continuous Rule for derivation according to coverareakDerivative D_tiltkJoin with work
Number azimuthkDerivative D_azimuthk, and update tilt and azimuth;
The formula of update tilt is expressed as:P_tiltk+1=P_tiltk+learning_rate*D_tiltk, wherein work as k
When=1, D_tilt1For full 0 vector;
The formula of update azimuth is expressed as:P_azimuthk+1=P_azimuthk+learning_rate*D_
azimuthk, wherein as k=1, D_azimuth1For full 0 vector.
A3, as k=interation, the value of output services parameter azimuth and tilt, i.e. last time interative computation
The result of generation.
It should be noted that herein, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.
The above is the preferred embodiment of the present invention, it is noted that for those skilled in the art
For, without departing from the principles of the present invention, several improvements and modifications can also be made, these improvements and modifications
It should be regarded as protection scope of the present invention.
Claims (10)
1. a kind of network work layout method based on gradient descent method, which is characterized in that including:
Obtain running parameter to be optimized;
According to the running parameter to be optimized of acquisition, the whole coverage rate of all sampled points in target area to be optimized, institute are determined
Mean value of the whole coverage rate equal to the coverage effect that all sampled points are capped is stated, the capped coverage effect of each sampled point is
[0,1] successive value between;
According to the whole coverage rate of all sampled points the derivative vector of running parameter is determined using continuous Rule for derivation;
According to the derivative of determining running parameter vector, the running parameter after optimization is determined.
2. the network work layout method according to claim 1 based on gradient descent method, which is characterized in that described
Obtaining running parameter to be optimized includes:
Obtain Downtilt and antenna azimuth;
Calculate sampled point angle of declination and sampled point azimuth of the antenna relative to all sampled points;
According to the Downtilt of acquisition, antenna azimuth, and the sampled point angle of declination being calculated and sampled point azimuth, really
Determine horizontal angle and vertical angle.
3. the network work layout method according to claim 2 based on gradient descent method, which is characterized in that described
Horizontal angle is expressed as:Alpha=azimuth_g-P_azimuthk;
The vertical angle is expressed as:Beta=P_tiltk-tilt_g;
Wherein, alpha indicates horizontal angle, P_azimuthkIndicate that the antenna azimuth of kth time iteration, azimuth_g indicate sampling
Point azimuth;Beta indicates vertical angle, P_tiltkIndicate that the Downtilt of kth time iteration, tilt_g indicate that sampled point has a down dip
Angle.
4. the network work layout method according to claim 1 based on gradient descent method, which is characterized in that described
According to the running parameter to be optimized of acquisition, the coverage rate successive value that all sampled points are capped in target area to be optimized is determined
Including:
Determine the parameter for judging coverage effect;
According to the running parameter to be optimized of acquisition, the value for judging coverage effect parameter is determined;
According to the value of determining judge coverage effect parameter all sampled points in target area to be optimized are obtained in conjunction with S type functions
Capped coverage effect, successive value of the capped coverage effect of each sampled point between [0,1];
The coverage effect being capped to all sampled points is averaged, and the whole coverage rate of all sampled points is obtained.
5. the network work layout method according to claim 3 based on gradient descent method, which is characterized in that determine
The parameter for judging coverage effect include:Reference Signal Received Power and Signal to Interference plus Noise Ratio;
The running parameter to be optimized according to acquisition determines the covering that all sampled points are capped in target area to be optimized
Rate successive value includes:
According to the horizontal angle and vertical angle of acquisition, gain of the sampled point to antenna is determined;
Antenna transmission power is set, calculates antenna to the path loss between sampled point;
According to determining gain, the antenna transmission power of setting, the path loss that is calculated, Reference Signal Received Power and letter are determined
Number with interference plus noise ratio;
According to determining Reference Signal Received Power and Signal to Interference plus Noise Ratio target to be optimized is obtained in conjunction with S type functions
The capped coverage effect of all sampled points in region, the capped coverage effect of each sampled point are continuous between [0,1]
Value;
The coverage effect being capped to all sampled points is averaged, and the whole coverage rate of all sampled points is obtained.
6. the network work layout method according to claim 5 based on gradient descent method, which is characterized in that determine
The gain of sampled point to antenna be expressed as:
Gain=f1 (alpha)-(abs (alpha)/pi) * (f1 (pi)-f2 (pi-beta))-(1-abs (alpha)/pi) *
(f1(0)-f2(beta))
Wherein, Gain indicates that gain, abs indicate that ABS function, f1 and f2 are the fitting letters of the gain to horizontal angle and vertical angle
Number.
7. the network work layout method according to claim 5 based on gradient descent method, which is characterized in that described
Reference Signal Received Power is expressed as:
RSSIi,j=Powerj+Gaini,j-Pathlossi,j
Wherein, RSRPiIndicate the Reference Signal Received Power of sampled point i, RSSIi,jIndicate the reception signal of sampled point i to antenna j
Intensity instruction, PowerjIndicate the transmission power of antenna j, Gaini,jIndicate the gain of sampled point i to antenna j,
Pathlossi,jPath loss between expression antenna j to sampled point i.
8. the network work layout method according to claim 5 based on gradient descent method, which is characterized in that described
Signal to Interference plus Noise Ratio is expressed as:
Wherein, SINRiIndicate the Signal to Interference plus Noise Ratio of sampled point i, NoiseiIndicate the noise of sampled point i, RSRPiTable
Show the Reference Signal Received Power of sampled point i, RSSIi,jIndicate the intensity instruction of the reception signal of sampled point i to antenna j.
9. the network work layout method according to claim 5 based on gradient descent method, which is characterized in that each
The capped coverage effect of sampled point is expressed as:
Coverpoint=sigmoid (RSRP-Thrsrp)*sigmoid(SINR-Thsinr)
Wherein, coverpoint indicates that the capped coverage effect of sampled point, SINR indicate Signal to Interference plus Noise Ratio, RSRP
Indicate Reference Signal Received Power, Thrsrp、ThsinrPre-set RSRP threshold values and SINR threshold values, sigmoid are indicated respectively
(x) it is S type functions,
10. the network work layout method according to claim 1 based on gradient descent method, which is characterized in that institute
The derivative vector according to determining running parameter is stated, determines that the running parameter after optimization includes:
Utilize formula Pk+1=Pk+learning_rate*DkUpdate the running parameter P in+1 generation of kthk+1, until meet it is preset most
Big iterations, wherein PkIndicate the running parameter in kth generation, DkIndicate the derivative vector of the running parameter in kth generation,
Learning_rate indicates learning rate.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810380075.5A CN108738032B (en) | 2018-04-25 | 2018-04-25 | Network working parameter optimization method based on gradient descent method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810380075.5A CN108738032B (en) | 2018-04-25 | 2018-04-25 | Network working parameter optimization method based on gradient descent method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108738032A true CN108738032A (en) | 2018-11-02 |
CN108738032B CN108738032B (en) | 2020-11-13 |
Family
ID=63939840
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810380075.5A Expired - Fee Related CN108738032B (en) | 2018-04-25 | 2018-04-25 | Network working parameter optimization method based on gradient descent method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108738032B (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103024894A (en) * | 2011-09-21 | 2013-04-03 | 华为技术有限公司 | Method and equipment for positioning interference sources of multiple base stations |
US8436766B1 (en) * | 2009-11-06 | 2013-05-07 | Technology Service Corporation | Systems and methods for suppressing radar sidelobes using time and spectral control |
CN104581741A (en) * | 2013-10-25 | 2015-04-29 | 普天信息技术研究院有限公司 | Network coverage optimization method |
CN105848177A (en) * | 2015-01-15 | 2016-08-10 | 中国移动通信集团四川有限公司 | Method and system for obtaining antenna attitude information |
CN106686707A (en) * | 2017-01-11 | 2017-05-17 | 东南大学 | Tilt-adjusting dynamic sector hibernating method based on centralized self-organizing network architecture |
CN107277833A (en) * | 2017-08-10 | 2017-10-20 | 北京胜普多邦通信技术有限公司 | The multi-mode multiple target LTE base station antenna optimization method of limited parameter adjustment |
-
2018
- 2018-04-25 CN CN201810380075.5A patent/CN108738032B/en not_active Expired - Fee Related
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8436766B1 (en) * | 2009-11-06 | 2013-05-07 | Technology Service Corporation | Systems and methods for suppressing radar sidelobes using time and spectral control |
CN103024894A (en) * | 2011-09-21 | 2013-04-03 | 华为技术有限公司 | Method and equipment for positioning interference sources of multiple base stations |
CN104581741A (en) * | 2013-10-25 | 2015-04-29 | 普天信息技术研究院有限公司 | Network coverage optimization method |
CN105848177A (en) * | 2015-01-15 | 2016-08-10 | 中国移动通信集团四川有限公司 | Method and system for obtaining antenna attitude information |
CN106686707A (en) * | 2017-01-11 | 2017-05-17 | 东南大学 | Tilt-adjusting dynamic sector hibernating method based on centralized self-organizing network architecture |
CN107277833A (en) * | 2017-08-10 | 2017-10-20 | 北京胜普多邦通信技术有限公司 | The multi-mode multiple target LTE base station antenna optimization method of limited parameter adjustment |
Non-Patent Citations (4)
Title |
---|
ANTENEH A. GEBREMARIAM等: "Techniques of Candidate Cell Selection for Antenna Tilt Adaptation in LTE-Advanced", 《EUROPEAN WIRELESS 2014; 20TH EUROPEAN WIRELESS CONFERENCE》 * |
PREMKUMAR KARTHIKEYAN等: "5G-Optimizing Network Coverage in Radio Self Organizing Networks by M/L Based Beam Tilt Algorithm", 《SMARTCITY 360°》 * |
王西点等: "人工智能及其在网络优化运维中的应用", 《中国通信学会会议论文集》 * |
邹晓辉: "基于Logistic回归的数据分类问题研究", 《智能计算机与应用》 * |
Also Published As
Publication number | Publication date |
---|---|
CN108738032B (en) | 2020-11-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105375959B (en) | Based on the matched distributed disturbance coordination method of beam shape in 3D-MIMO systems | |
CN106937306B (en) | A kind of network quality appraisal procedure and device | |
CN101022297B (en) | Method and apparatus for processing transmit power control (TPC) commands in a wideband CDMA (WCDMA) network | |
CN108601033A (en) | A kind of network work layout method based on Adam | |
CN108306699A (en) | A kind of antenna feeder optimization method estimated based on gain | |
KR20010099929A (en) | Signal processing method and apparatus for computing an optimal weight vector of an adptive antenna array system | |
CN100459793C (en) | Method and apparatus for whole channel primary searching | |
CN101964774A (en) | Automatic gain control method and control circuit suitable for orthogonal frequency division multiplexing (OFM) system | |
CN102136850A (en) | Method and device for realizing automatic frequency control | |
CN102158272A (en) | Method, device and system for calibrating radio-frequency channels | |
CN103747496B (en) | Improve method, device and the user equipment of uplink signal-channel quality in heterogeneous network | |
CN105072640B (en) | Determine the method and device of antenna-feed parameter | |
CN107809273A (en) | A kind of multiple antennas compensation method and its device, radio-frequency apparatus | |
CN108414974B (en) | Indoor positioning method based on ranging error correction | |
CN104703272A (en) | Automatic gain control method suitable for OFDM system | |
CN106851665A (en) | The downdip adjusting method of antenna and base station | |
CN103095357B (en) | Smart antenna self-adapting Beamforming Method | |
CN110913401A (en) | Method for optimizing imbalance of network same coverage load based on user data | |
CN108601034A (en) | A kind of network work layout method based on stochastic gradient descent method | |
CN107277833A (en) | The multi-mode multiple target LTE base station antenna optimization method of limited parameter adjustment | |
CN113015192A (en) | Antenna weight determination method, device, equipment and storage medium | |
CN103596193B (en) | Signal Regulation method, Home eNodeB and the terminal of a kind of Home eNodeB | |
CN108738032A (en) | A kind of network work layout method based on gradient descent method | |
CN103475395A (en) | Antenna correction method, device and system | |
CN104735707B (en) | A kind of failure antenna positioning method, device and electronic equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20201113 |
|
CF01 | Termination of patent right due to non-payment of annual fee |