CN109729540A - A kind of base station parameter automatic optimization method neural network based - Google Patents
A kind of base station parameter automatic optimization method neural network based Download PDFInfo
- Publication number
- CN109729540A CN109729540A CN201910048705.3A CN201910048705A CN109729540A CN 109729540 A CN109729540 A CN 109729540A CN 201910048705 A CN201910048705 A CN 201910048705A CN 109729540 A CN109729540 A CN 109729540A
- Authority
- CN
- China
- Prior art keywords
- base station
- neuron
- ring
- cell
- switching problem
- 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
Landscapes
- Mobile Radio Communication Systems (AREA)
Abstract
The present invention relates to a kind of base station parameter automatic optimization methods neural network based, comprising the following steps: step S1: being scanned to full dose base station MRE data, output has the base station of ring switching problem to configure and configure without the base station of ring switching problem respectively;Step S2: being input by the base station config set of no ring switching problem, carries out model training using self organizing neural network Kohonen algorithm, obtains trained neural network model；Step S3: according to trained neural network model, the base station for having switching problem is clustered；Step S4: being found the base station configuration of immediate not switching problem in cluster, configured using this as configuration scheme, and problematic base station configuration is replaced.The present invention is the parameter configuration that problem base station automatically generates optimization, solves ring switching problem, improves the intelligent level of network optimization.
Description
Technical field
The present invention relates to mobile communication fields, and in particular to a kind of base station parameter Automatic Optimal side neural network based
Method.
Background technique
When a mobile communication equipment (such as mobile phone) is in the coverage area of multiple mobile base station cells, if
The configuration of associated base stations is problematic, may cause the cell that mobile device is frequently connected from one and disconnects, switching is connected to another
Outer cell.It is undying in several handover between cells even when mobile device is static, it forms ring cutting and changes, cause mobile device
Quality of connection is bad.As mobile operator, needs to optimize base station parameter setting, reduce ring switch instances to the greatest extent
It generates, improves the service quality of mobile communication.
In prior art, discovery to ring switching problem, mainly to drive test or to the property in the cell short time
Energy data are for statistical analysis, and the solution of ring switching problem, rely primarily on artificial optimization's configuration of network optimization engineer.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of base station parameter automatic optimization method neural network based,
The parameter configuration of optimization is automatically generated for problem base station, is solved ring switching problem, is improved the intelligent water of network optimization
It is flat.
To achieve the above object, the present invention adopts the following technical scheme:
A kind of base station parameter automatic optimization method neural network based, which comprises the following steps:
Step S1: being scanned full dose base station MRE data, and output has the base station of ring switching problem to configure and do not have respectively
There is the base station of ring switching problem to configure；
Step S2: being input by the base station config set of no ring switching problem, using self organizing neural network Kohonen
Algorithm carries out model training, obtains trained neural network model；
Step S3: according to trained neural network model, the base station for having switching problem is clustered；
Step S4: the base station configuration of immediate not switching problem is found in cluster, is configured using this as optimization and is matched
Scheme is set, problematic base station configuration is replaced.
Further, ring cutting can be formed specified in TS36.331 measurement event agreement of the step S1 according to 3GPP
The condition of parameter problem is changed, carries out changing problem differentiation whether there is or not ring cutting.
Further, the step S2 specifically:
Step S21: the standardization of zero-mean value, conversion formula are carried out one by one to each parameter of the base station of no ring switching problem
It is as follows:
WhereinFor the mean value of each one parameter in base station, σ is the standard deviation of each one parameter value in base station；
Step S22: one self organizing neural network Kohonen model of initialization, each neuron neuronkWeight
weightkFor an array set (w1,w2,…,wn), wherein n is the number of base station parameter, and each w initial value takes between 0 to 1
A random decimal；
Step S23: by a base station cell without ring switching problemnoringParameters input, calculate with it is each
Neuron neuronkEuclidean distance d (cellnoring,neuronk):
Wherein neuronkFor k-th of neuron,For i-th of configuration parameter of no ring switching problem base station, n
For number of parameters；
Step S24: d (cell is takennoring,neuronk) the smallest neuron neuronkFor the triumph neuron of the base station,
And marking cluster number belonging to the base station is k；
Step S25: neuron neuron is updatedkWeight weightkValue is (w1',w2',…,wn'):
wi'=wi+l×(ci-wi)
Wherein l is customized learning rate, and the number between [0,1] can be set as needed；
Step S26: each do not had into the base station cell of ring switching problemnoringConfiguration input, repeats step S23~step
Rapid S25；
Step S27: each neuron neuron ultimately generated is savedkWeight weightkValue, that is, save trained
Neural network model.
Further, the step S3 specifically:
Step S31: zero-mean value standardization (also referred to as standard deviation standardization), conversion are carried out one by one to each parameter of each base station
Formula is as follows:
WhereinFor the mean value of each one parameter in base station, σ is the standard deviation of each one parameter value in base station；
Step S32: one there is the base station cell of ring switching problemringParameters input, calculate with each nerve
First neuronkEuclidean distance d (cellring,neuronk):
Wherein neuronkFor k-th of neuron,To there is i-th of configuration parameter of ring switching problem base station, n is ginseng
Several numbers；
Step S33: d (cell is takenring,neuronk) the smallest neuron neuronkFor the triumph neuron of the base station,
And marking cluster number belonging to the base station is k；
Step S34: each is had to the base station cell of ring switching problemringConfiguration input, repeats step S32 and step
S33 completes the cluster to all base stations for having ring switching problem.
Further, the step S4 specifically:
Step S41: input one has the base station of ring switching problem, according to its cluster number k, in institute either with or without ring cutting
It changes problem and cluster number is also to calculate Euclidean distance d (cell in the base station of kring,cellnoring):
Wherein cellringTo have the base station of ring switching problem, cellringTo there is the base station of ring switching problem,To have
Some configuration parameter of ring switching problem base station,For some configuration parameter of no ring switching problem base station, n is ginseng
Several numbers；
Step S42: d (cell is takenring,cellnoring) the smallest base station without ring switching problem configures cellnoring,
Cell is substituted as optimization collocationnoringConfiguration；
Step S43: each is had to the base station cell of ring switching problemringConfiguration input, repeats step S42 and step
S43 completes to generate the prioritization scheme of all base stations for having ring switching problem.
Compared with the prior art, the invention has the following beneficial effects:
1, the present invention automatically generates reasonable base station parameter configuration scheme, and optimization quality is more stable.
2, the present invention can carry out rapid detection, output prioritization scheme to the whole network ring switching problem, greatly improve effect
Rate.
Detailed description of the invention
Fig. 1 is the method for the present invention flow chart.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
Fig. 1 is please referred to, the present invention provides a kind of base station parameter automatic optimization method neural network based, is problem base
Station automatically generates the parameter configuration of optimization, specifically includes the following steps:
1. a pair full dose base station MRE data (event mode measurement report) are scanned, thing is measured according to the TS36.331 of 3GPP
The condition that ring cutting changes parameter problem can be formed specified in part agreement, output has the base station of ring switching problem to configure and do not have respectively
There is the base station of ring switching problem to configure.
2. with the base station config set of no ring switching problem be input, using self organizing neural network Kohonen algorithm into
Row model training, and base station is clustered.Specific step is as follows:
2.1. zero-mean value standardization (also referred to as standard deviation standardization), conversion formula are carried out one by one to each parameter of each base station
It is as follows:
WhereinFor the mean value of each one parameter in base station, σ is the standard deviation of each one parameter value in base station.
2.2. a self organizing neural network Kohonen model, each neuron neuron are initializedkWeight weightk
For an array set (w1,w2,…,wn), wherein n is the number of base station parameter, each w initial value take one between 0 to 1 with
Machine decimal.
2.3. by a base station cell without ring switching problemnoringParameters input, calculate with each nerve
First neuronkEuclidean distance d (cellnoring,neuronk):
Wherein neuronkFor k-th of neuron,For i-th of configuration parameter of no ring switching problem base station, n
For number of parameters.
2.4. d (cell is takennoring,neuronk) the smallest neuron neuronkFor the triumph neuron of the base station, and mark
Remember that cluster number belonging to the base station is k.
2.5. neuron neuron is updatedkWeight weightkValue is (w1',w2',…,wn'):
wi'=wi+l×(ci-wi)
Wherein l is customized learning rate, and the number between [0,1] can be set as needed.
2.6. each is not had the base station cell of ring switching problemnoringThe step of configuration input, repetition 2.3~2.5.
2.7. each neuron neuron ultimately generated is savedkWeight weightkValue, that is, save trained nerve
Network model.
3. being clustered with trained model to the base station for having switching problem.
3.1. zero-mean value standardization (also referred to as standard deviation standardization), conversion formula are carried out one by one to each parameter of each base station
It is as follows:
WhereinFor the mean value of each one parameter in base station, σ is the standard deviation of each one parameter value in base station.
3.2. one there is the base station cell of ring switching problemringParameters input, calculate with each neuron
neuronkEuclidean distance d (cellring,neuronk):
Wherein neuronkFor k-th of neuron,To there is i-th of configuration parameter of ring switching problem base station, n is
Number of parameters.
3.3. d (cell is takenring,neuronk) the smallest neuron neuronkFor the triumph neuron of the base station, and mark
Remember that cluster number belonging to the base station is k.
3.4., each is had to the base station cell of ring switching problemringIt is the step of configuration input, repetition 3.2~3.3, complete
The cluster of pairs of all base stations for having ring switching problem.
4. finding the base station configuration of immediate not switching problem in cluster, configured using this as the side of distributing rationally
Case replaces problematic base station configuration.
4.1. the base station for having ring switching problem is inputted, according to its cluster number k, changes and asks either with or without ring cutting in institute
Topic and cluster number are also to calculate Euclidean distance d (cell in the base station of kring,cellnoring):
Wherein cellringTo have the base station of ring switching problem, cellringTo there is the base station of ring switching problem,To there is ring
Some configuration parameter of switching problem base station,For some configuration parameter of no ring switching problem base station, n is parameter
Number.
4.2. d (cell is takenring,cellnoring) the smallest base station without ring switching problem configures cellnoring, as
Optimization collocation substitutes cellnoringConfiguration.
4.3., each is had to the base station cell of ring switching problemringIt is the step of configuration input, repetition 4.1~4.2, complete
The prioritization scheme of pairs of all base stations for having ring switching problem generates.
The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent
With modification, it is all covered by the present invention.
Claims (5)
1. a kind of base station parameter automatic optimization method neural network based, which comprises the following steps:
Step S1: being scanned full dose base station MRE data, and output has the base station of ring switching problem to configure and without ring cutting respectively
Change the base station configuration of problem；
Step S2: by the base station config set of no ring switching problem be input, using self organizing neural network Kohonen algorithm into
Row model training obtains trained neural network model；
Step S3: according to trained neural network model, the base station for having switching problem is clustered；
Step S4: the base station configuration of immediate not switching problem is found in cluster, is configured using this as the side of distributing rationally
Case replaces problematic base station configuration.
2. a kind of base station parameter automatic optimization method neural network based according to claim 1, it is characterised in that: institute
The condition that ring cutting changes parameter problem can be formed by stating specified in TS36.331 measurement event agreement of the step S1 according to 3GPP, be carried out
Problem is changed whether there is or not ring cutting to distinguish.
3. a kind of base station parameter automatic optimization method neural network based according to claim 1, it is characterised in that: institute
State step S2 specifically:
Step S21: the standardization of zero-mean value is carried out one by one to each parameter of the base station of no ring switching problem, conversion formula is as follows:
WhereinFor the mean value of each one parameter in base station, σ is the standard deviation of each one parameter value in base station；
Step S22: one self organizing neural network Kohonen model of initialization, each neuron neuronkWeight weightk
For an array set (W1, W2..., Wn), wherein n is the number of base station parameter, each w initial value take one between 0 to 1 with
Machine decimal；
Step S23: by a base station cell without ring switching problemnoringParameters input, calculate with each neuron
neuronkEuclidean distance d (cellnoring, neuronk):
Wherein neuronkFor k-th of neuron,For i-th of configuration parameter of no ring switching problem base station, n is ginseng
Several numbers；
Step S24: d (cell is takennoring, neuronk) the smallest neuron neuronkFor the triumph neuron of the base station, and mark
Remember that cluster number belonging to the base station is k；
Step S25: neuron neuron is updatedkWeight weightkValue is (W1', w2' ..., Wn'):
wi'=wi+l×(ci-wi)
Wherein I is customized learning rate, and the number between [0,1] can be set as needed；
Step S26: each do not had into the base station cell of ring switching problemnoringConfiguration input, repeats step S23~step
S25；
Step S27: each neuron neuron ultimately generated is savedkWeight weightkValue, that is, save trained nerve
Network model.
4. a kind of base station parameter automatic optimization method neural network based according to claim 1, it is characterised in that: institute
State step S3 specifically:
Step S31: zero-mean value standardization (also referred to as standard deviation standardization), conversion formula are carried out one by one to each parameter of each base station
It is as follows:
WhereinFor the mean value of each one parameter in base station, σ is the standard deviation of each one parameter value in base station；
Step S32: one there is the base station cell of ring switching problemringParameters input, calculate with each neuron
neuronkEuclidean distance d (cellring, neuronk):
Wherein neuronkFor k-th of neuron,To there is i-th of configuration parameter of ring switching problem base station, n is parameter
Number；
Step S33: d (cell is takenring, neuronk) the smallest neuron neuronkFor the triumph neuron of the base station, and mark
Cluster number is k belonging to the base station；
Step S34: each is had to the base station cell of ring switching problemringConfiguration input repeats step S32 and step S33, complete
The cluster of pairs of all base stations for having ring switching problem.
5. a kind of base station parameter automatic optimization method neural network based according to claim 1, it is characterised in that: institute
State step S4 specifically:
Step S41: input one has the base station of ring switching problem, according to its cluster number k, changes problem either with or without ring cutting in institute
And cluster number is also to calculate Euclidean distance d (cell in the base station of kring, cellnoring):
Wherein cellringTo have the base station of ring switching problem, cellringTo there is the base station of ring switching problemTo there is ring cutting to change
Some configuration parameter of problem base station,For some configuration parameter of no ring switching problem base station, n is number of parameters；
Step S42: d (cell is takenring, cellnoring) the smallest base station without ring switching problem configures cellnoring, as most
Distribute substitution cell rationallynoringConfiguration；
Step S43: each is had to the base station cell of ring switching problemringConfiguration input repeats step S42 and step S43, complete
The prioritization scheme of pairs of all base stations for having ring switching problem generates.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910048705.3A CN109729540B (en) | 2019-01-18 | 2019-01-18 | Base station parameter automatic optimization method based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910048705.3A CN109729540B (en) | 2019-01-18 | 2019-01-18 | Base station parameter automatic optimization method based on neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109729540A true CN109729540A (en) | 2019-05-07 |
| CN109729540B CN109729540B (en) | 2022-05-17 |
Family
ID=66298494
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910048705.3A Active CN109729540B (en) | 2019-01-18 | 2019-01-18 | Base station parameter automatic optimization method based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109729540B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111954206A (en) * | 2019-05-17 | 2020-11-17 | 株式会社Ntt都科摩 | Terminal and base station |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101635949A (en) * | 2008-07-25 | 2010-01-27 | 三星电子株式会社 | Switching method in wireless cellular network |
| CN102149114A (en) * | 2011-05-10 | 2011-08-10 | 北京邮电大学 | Femto base station network control system and method |
| CN102740335A (en) * | 2012-06-13 | 2012-10-17 | 北京邮电大学 | Self-optimizing method for switching femtocell network |
| CN103385022A (en) * | 2011-01-31 | 2013-11-06 | 诺基亚西门子通信公司 | Method and apparatus for managing neighbour relations |
| US20150098393A1 (en) * | 2013-10-09 | 2015-04-09 | At&T Intellectual Property I, L.P. | Network operating system client architecture for mobile user equipment |
| JPWO2016076293A1 (en) * | 2014-11-11 | 2017-08-24 | シャープ株式会社 | Base station apparatus, location management apparatus, terminal apparatus, and control method |
| CN107645393A (en) * | 2016-07-20 | 2018-01-30 | 中兴通讯股份有限公司 | Determine the method, apparatus and system of the black-box system input and output degree of association |
| CN108134979A (en) * | 2017-12-04 | 2018-06-08 | 东南大学 | Small base station switch control method based on deep neural network |
-
2019
- 2019-01-18 CN CN201910048705.3A patent/CN109729540B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101635949A (en) * | 2008-07-25 | 2010-01-27 | 三星电子株式会社 | Switching method in wireless cellular network |
| CN103385022A (en) * | 2011-01-31 | 2013-11-06 | 诺基亚西门子通信公司 | Method and apparatus for managing neighbour relations |
| CN102149114A (en) * | 2011-05-10 | 2011-08-10 | 北京邮电大学 | Femto base station network control system and method |
| CN102740335A (en) * | 2012-06-13 | 2012-10-17 | 北京邮电大学 | Self-optimizing method for switching femtocell network |
| US20150098393A1 (en) * | 2013-10-09 | 2015-04-09 | At&T Intellectual Property I, L.P. | Network operating system client architecture for mobile user equipment |
| JPWO2016076293A1 (en) * | 2014-11-11 | 2017-08-24 | シャープ株式会社 | Base station apparatus, location management apparatus, terminal apparatus, and control method |
| CN107645393A (en) * | 2016-07-20 | 2018-01-30 | 中兴通讯股份有限公司 | Determine the method, apparatus and system of the black-box system input and output degree of association |
| CN108134979A (en) * | 2017-12-04 | 2018-06-08 | 东南大学 | Small base station switch control method based on deep neural network |
Non-Patent Citations (1)
| Title |
|---|
| ZTE: ""The Restriction of Cell Reselection and Handover Parameters"", 《3GPP TSG RAN WG3 #66BIS R3-100183》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111954206A (en) * | 2019-05-17 | 2020-11-17 | 株式会社Ntt都科摩 | Terminal and base station |
| CN111954206B (en) * | 2019-05-17 | 2024-04-09 | 株式会社Ntt都科摩 | Terminal and base station |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109729540B (en) | 2022-05-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10911318B2 (en) | Future network condition predictor for network time series data utilizing a hidden Markov model for non-anomalous data and a gaussian mixture model for anomalous data | |
| CN106658422B (en) | For the network side localization method and system of the sparse WiFi data of height | |
| CN107211294A (en) | The analysis auxiliary of intelligent self-organizing network, multiagent, self study, easily Self management, extension and adaptive framework | |
| US11451971B2 (en) | Networking method, networking apparatus, network access method, and user equipment for coordinated multiple points transmission/reception | |
| US10600001B2 (en) | Determining a target device profile including an expected behavior for a target device | |
| WO2021211282A1 (en) | Architecture for machine learning (ml) assisted communications networks | |
| WO2020048594A1 (en) | Procedure for optimization of self-organizing network | |
| US20180227176A1 (en) | Determining whether to include or exclude device data for determining a network communication configuration for a target device | |
| US11716240B2 (en) | Method, device and computer program product for anomaly detection | |
| CN104981002A (en) | Position determining method of convergent node in wireless sensor network | |
| Chamola et al. | Outage estimation for solar powered cellular base stations | |
| Liao et al. | Network state awareness and proactive anomaly detection in self-organizing networks | |
| CN105578486A (en) | Capacity and coverage combined optimization method in heterogeneous dense network | |
| Xu et al. | Data-cognition-empowered intelligent wireless networks: Data, utilities, cognition brain, and architecture | |
| Lee et al. | Affinity propagation for energy-efficient BS operations in green cellular networks | |
| CN111162888A (en) | Distributed antenna system, remote access unit, power distribution method, and medium | |
| WO2018184667A1 (en) | Apparatus and method for performing network optimization | |
| CN105704797B (en) | Wireless access point energy-saving control method in WLAN based on software defined network | |
| CN110062410A (en) | A kind of cell outage detection localization method based on adaptive resonance theory | |
| CN105334493A (en) | WLAN-based indoor positioning method | |
| CN109729540A (en) | A kind of base station parameter automatic optimization method neural network based | |
| Yu et al. | Self-organized cell outage detection architecture and approach for 5G H-CRAN | |
| CN106486768A (en) | Electrical tilt antenna parameter value adjustment method and device | |
| CN103095431A (en) | Method and device for processing reference signal | |
| DE102016005639B4 (en) | Near field communication (NFC) enabled peripheral device |
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 |