CN112954715B - Wireless service node capacity estimation method based on transfer learning - Google Patents
Wireless service node capacity estimation method based on transfer learning Download PDFInfo
- Publication number
- CN112954715B CN112954715B CN202110134344.1A CN202110134344A CN112954715B CN 112954715 B CN112954715 B CN 112954715B CN 202110134344 A CN202110134344 A CN 202110134344A CN 112954715 B CN112954715 B CN 112954715B
- Authority
- CN
- China
- Prior art keywords
- wireless service
- service node
- capacity
- areas
- model
- 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
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/06—Testing, supervising or monitoring using simulated traffic
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W52/00—Power management, e.g. TPC [Transmission Power Control], power saving or power classes
- H04W52/02—Power saving arrangements
- H04W52/0203—Power saving arrangements in the radio access network or backbone network of wireless communication networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Evolutionary Biology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention provides a wireless service node capacity estimation method based on transfer learning, which comprises the following steps: step S1: classifying the areas in the wireless network by using a clustering algorithm according to wireless service node deployment modes in different areas, and establishing a capacity model of a large wireless service node and a micro wireless service node by using transfer learning; step S2: according to the real-time flow demand, a wireless service node capacity estimation model and a wireless service node dormancy algorithm, power consumption is reduced to the maximum extent in an area meeting network coverage and real-time flow demand. The wireless service node dormancy strategy provided by the invention tends to activate the optimal number of large wireless service nodes to provide basic network coverage and micro wireless service nodes to improve the throughput, and researches show that more energy can be saved in areas with more micro wireless service nodes or larger flow fluctuation.
Description
Technical Field
The invention belongs to the field of wireless communication and the technical field of computers, particularly relates to the aspects of clustering algorithm, transfer learning, convex optimization and the like, and particularly relates to a transfer learning-based wireless service node capacity estimation method.
Background
With the arrival of the artificial intelligence era, machine learning and deep learning are developing more and more rapidly, wherein a clustering algorithm, deep learning, wireless network optimization and the like become research hotspots in academic circles, a sleep strategy aiming at reducing the basic operation power of a wireless service node becomes an attractive energy-saving method, in recent years, a plurality of strategies for realizing dynamic switching of the wireless service node are provided, an accurate wireless service node capacity model is a key factor of the sleep strategy, and in an area where network coverage and real-time flow requirements are guaranteed, power consumption is reduced to the maximum extent through an intelligent wireless service node sleep algorithm. Therefore, research on optimization of wireless service node network resources has also received more and more attention.
Disclosure of Invention
In view of the above, the present invention provides a method for estimating a capacity of a wireless service node based on transfer learning, which divides a city into different types of areas by using a clustering algorithm, and analyzes and displays that the different types of areas have different traffic patterns and wireless service node deployment patterns; the capacity models of the large wireless service node and the micro wireless service node are established by using transfer learning, and then an intelligent wireless service node dormancy algorithm is designed according to the real-time traffic demand and the learned wireless node capacity model, so that the power consumption is reduced to the maximum extent in the area meeting the network coverage and the real-time traffic demand.
The method establishes a capacity model of the wireless service node according to the transfer learning, meanwhile, different areas correspond to different capacity modes, the influence of an external factor r on the capacity model can be extracted from multi-domain data in real life in a centralized mode, and in addition, an intelligent wireless service node dormancy algorithm is designed according to real-time flow requirements and the capacity model of the wireless service node. The wireless service node dormancy strategy provided by the invention tends to activate the optimal number of large wireless service nodes to provide basic network coverage and micro wireless service nodes to improve the throughput, and researches show that more energy can be saved in areas with more micro wireless service nodes or larger flow fluctuation.
The invention specifically adopts the following technical scheme:
a wireless service node capacity estimation method based on transfer learning aims at solving the problem of resource allocation in a wireless network by modeling the wireless service node capacity by using the transfer learning method, and is characterized by comprising the following steps:
step S1: classifying areas in a wireless network by using a clustering algorithm according to wireless service node deployment modes in different areas, and establishing capacity models of wireless service nodes (including large wireless service nodes and micro wireless service nodes) in different areas by using transfer learning;
step S2: according to the real-time flow demand, a wireless service node capacity estimation model and a wireless service node dormancy algorithm, power consumption is reduced to the maximum extent in an area meeting network coverage and real-time flow demand;
the wireless service node dormancy algorithm enables the energy consumption to be minimum on the premise of meeting the flow demand by activating the large and micro wireless service nodes with the optimal number.
The performance of the proposed wireless service node dormancy strategy is evaluated through a large number of experiments, and researches show that more energy can be saved in areas with more micro wireless service nodes or larger flow fluctuation.
Preferably, step S1 specifically includes the following steps:
step S11: the traffic peak value of the area is used as the capacity of the wireless node, the city is divided into areas of different types by using a clustering algorithm, and the big data analysis shows that the areas of different types have different traffic modes and wireless service node deployment modes;
step S12: and capturing the mode similarity and diversity among different regions by using a transfer learning algorithm to obtain a capacity model of the wireless service node suitable for different region characteristics.
Preferably, in step S11, the maximum traffic that the wireless service node can provide in different areas is analyzed; clustering is realized by using multi-domain data as features and adopting a k-means algorithm: the aim is to minimize the distance within the same cluster and maximize the distance between different clusters; the regions are divided into different types of regions, where square grid regions with similar characteristics are grouped into a cluster. From the results, the entire urban area can be roughly classified into four types: rural, suburban, urban, city center;
different types of areas have different traffic patterns and wireless service node deployments, for example, the number of rural wireless service nodes is about 3, the number of suburban wireless service nodes is about 12, the number of urban wireless service nodes is about 27, and the number of urban wireless service nodes is about 48. In practice, the number of wireless service nodes in rural areas is mostly large-scale wireless service nodes, a capacity model of the large-scale wireless service nodes is obtained by deep learning, and then the mode similarity and diversity among different areas are captured by a transfer learning algorithm, so that the capacity model of the wireless service nodes suitable for different area characteristics is obtained.
In step S12, a capacity model of the large wireless service node is obtained by deep learning, and then a migration learning algorithm is used to capture pattern similarity and diversity between different areas, so as to obtain an overall capacity model of the large service node in different areas.
Preferably, in step S1, the influence of the external factor r on the capacity model is extracted from the multi-domain dataset in real life;
firstly, the specific influence of the capacity on the service node capacity is analyzed by using the time delay correlation between the capacity and each domain data:where k (l) represents a capacity sequence, m (l) represents each domain data sequence, τ is a time delay, N is a sequence length, and σ is a standard deviation; factors having a strong correlation to the flow peak are included in the external factor r. Secondly, grouping the square grid areas with similar characteristics into a cluster by adopting a clustering algorithm, and dividing the whole urban area into four types of areas: rural, suburban, urban, city center;
acquiring characteristics of different areas by using transfer learning, training external environment characteristics, the number of large wireless service nodes and a peak flow model of rural areas by utilizing a multilayer perception network to obtain a limited capacity between the external characteristic r and the number of large wireless service nodesA function; then, taking the model as a source domain, and transferring parameters to the next target domain suburban area, so that the target domain model has characteristics of both rural areas and suburban areas. After the model is gradually migrated and learned among the four regions, the mapping relation among the network capacity, the external environment characteristics and the number of the large wireless service nodes is finally obtained, namelyWherein n ismThe number of large wireless service nodes in the area.
And obtaining the number of the micro wireless service nodes in the square grids in different areas according to the capacity model of the large wireless service node. In suburbs, cities and city centers, many micro wireless service nodes are arranged besides large wireless service nodes. By usingAfter the network capacity provided by the large wireless service node is deducted, the network capacity provided by the micro wireless service node can be obtained.
And then obtaining the mapping relation among the network capacity, the external factor r and the number of the micro wireless service nodes through deep learning, namelyThe number of micro wireless serving nodes.
Preferably, the wireless service node dormancy algorithm in step S2 specifically includes the following steps:
step S21: establishing a power consumption model of the wireless service node:
wherein, the fixed power consumption c of the wireless service node0The power consumption of the circuit elements in the antenna is represented by c1Denotes, M is the number of antennas, c2For each antenna power consumption, the power consumption of the traffic load is BμB is power coefficient, mu represents traffic load generated by wireless service node, and power consumption of power amplifierIs represented by the formula, wherein KρRepresenting the input power of the wireless service node antenna, η representing the efficiency of the power amplifier, and the total power of the wireless service node is represented as:it should be noted that the power parameters of the macro wireless service node and the micro wireless service node are greatly different, and we use PmAnd PsRespectively representing the total power of the large wireless service node and the total power of the micro wireless service node;
step S22: in order to meet the real-time flow demand, the power consumption of each square grid area is reduced to the minimum, the optimal number n of large wireless service nodes and micro wireless service nodes needing to be started is found by adopting a wireless service node dormancy strategy algorithm through a power model of the wireless service nodesmAnd ns(ii) a To minimize power loss in the area and to ensure network coverage and real-time network traffic.
This optimization problem can be mathematically expressed as:
s.t.0<ns≤Ns
where μ is the real-time traffic, δ is the traffic redundancy, NmAnd NsRespectively a large wireless service node and a micro wireless service node deployed in a square grid. The first two limiting conditions ensure that the number of opened service nodes is not larger than the number of deployed nodes, and the last limiting condition ensures that the opened wireless clothesThe service node may provide sufficient traffic.Rounding up is shown, and is intended to leave at least 1/2 large wireless serving nodes operational to ensure coverage of the network.
Preferably, in step S2, the capacity models of the macro wireless service node and the micro wireless service node are determinedFitting to a polynomial and then using optimization tools such as CVX, etc., to obtain the optimum nmAnd ns。
According to the scheme, performance simulation of the wireless service node dormancy strategy shows that as the density of the wireless service node increases, the node capacity increases rapidly and then increases in speed and decreases in speed, the micro wireless service node plays an important role in improving the throughput of the urban central area, the power consumption of the micro wireless service node is much smaller than that of the large wireless service node, and in order to achieve the purpose of energy conservation, the micro wireless service node can be activated to provide larger network throughput under the condition of ensuring the coverage rate.
Compared with the prior art, the invention and the optimal scheme establish a model for measuring the capacity of the wireless service nodes in different areas, utilize a transfer learning method to model the capacity of the wireless service nodes in different areas, provide a wireless service node dormancy strategy for minimizing the power consumption of a wireless network, and reduce the power consumption to the maximum extent under the condition of meeting the requirements of network coverage and real-time flow. Simulation results show that compared with the existing model, the wireless service node dormancy strategy provided by the invention is prone to maintaining a plurality of active large wireless service nodes to provide basic coverage and activating a micro wireless service node to improve the throughput. In areas with more micro wireless service nodes or larger flow fluctuation, more energy can be saved.
Its beneficial effect still lies in:
1) in order to improve the generalization capability of the wireless service node capacity model, an external factor r is introduced to describe relevant characteristics in the area, so that the capacity model is not only suitable for the problems provided by the invention, but also can be applied to other scenes.
2) In terms of power consumption of wireless service node dormancy strategy, it is proposed to maintain multiple active large wireless service nodes to provide basic network coverage and to activate a certain number of micro wireless service nodes to improve throughput.
Drawings
The invention is described in further detail below with reference to the following figures and detailed description:
FIG. 1 is a diagram illustrating transfer learning of different regions according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of wireless serving node distribution in different areas according to an embodiment of the present invention;
fig. 3 is a diagram illustrating simulation results in a suburban area according to an embodiment of the present invention.
Detailed Description
In order to make the features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail as follows:
the method for estimating the capacity of the wireless service node based on the transfer learning provided by the embodiment is specifically implemented according to the following steps,
step S1: firstly, classifying areas in a wireless network by using a clustering algorithm according to wireless service node deployment modes in different areas, and establishing capacity models of the wireless service nodes in the different areas by using transfer learning, wherein the influence of an external factor r on the capacity models can be extracted from multi-domain data in real life in a centralized manner;
1) in order to obtain an accurate deployment mode of the wireless service node, the embodiment analyzes the time delay correlation between the capacity and the data of each domain to analyze the specific influence of the capacity on the service node:where k (l) represents a capacity sequence, m (l) represents each domain data sequence, τ is a time delay, N is a sequence length, and σ is a standard deviation; factors having strong correlation with the flow peak are taken into the external factorsr in (c). Secondly, grouping the square grid areas with similar characteristics into a cluster by adopting a clustering algorithm, and dividing the whole urban area into four types of areas: rural, suburban, urban, city center.
2) Based on the similarity of deployment of large wireless service nodes and micro wireless service nodes in four areas, migration learning can be used for obtaining characteristics of different areas, most of the wireless service nodes in the actual rural and middle areas are large wireless service nodes, then model training is performed on external environment characteristics, the number of large wireless service nodes and peak flow of the rural areas by utilizing a multilayer sensing network, a limited function between the capacity under the external characteristic r and the number of the large wireless service nodes can be obtained, then the model is used as a source area, parameters are transmitted to the suburban area of the next target area, and the target area model has characteristics of rural areas and suburban areas. After the model is gradually migrated and learned among the four regions, the mapping relation among the network capacity, the external environment characteristics and the number of the large wireless service nodes can be finally obtained, namely the model is subjected to the gradual migration and learning, namelyWherein n ismThe number of large wireless service nodes in the area. Therefore, the present embodiment obtains the number of micro wireless service nodes in the square grid in different areas according to the capacity model of the large wireless service node. In suburbs, cities and city centers, many micro wireless service nodes are arranged besides large wireless service nodes. By usingAfter the network capacity provided by the large wireless service node is deducted, the network capacity provided by the micro wireless service node can be obtained. And then obtaining the mapping relation among the network capacity, the external factor r and the number of the micro wireless service nodes through deep learning, namelynsThe number of micro wireless serving nodes.
Step S2: according to the real-time flow demand and the learned wireless service node capacity model, an intelligent wireless service node dormancy algorithm is designed, and power consumption is reduced to the maximum extent in an area meeting the network coverage and real-time flow demand;
1) the power consumption of the wireless serving node includes: fixed power consumption c for wireless serving node0The power consumption of the circuit elements in the antenna is represented by c1Denotes, M is the number of antennas, c2For each antenna power consumption, the power consumption of the traffic load is BμB is power coefficient, mu represents traffic load generated by wireless service node, and power consumption of power amplifierIs represented by the formula, wherein KρRepresenting the input power of the wireless service node antenna, η represents the efficiency of the power amplifier, so the total power of the wireless service node can be expressed asIt should be noted that the power parameters of the macro wireless service node and the micro wireless service node are greatly different, and we use PmAnd PsRespectively representing the total power of the large wireless service node and the total power of the micro wireless service node.
2) Different coefficients are employed and denoted by subscripts to allow for heterogeneous deployment of large wireless service nodes and micro wireless service nodes. The invention aims to reduce the power consumption of each area to the maximum extent while meeting the real-time flow demand, and converts the optimization problem by substituting the power model of the wireless service node:
where μ is the real-time traffic, δ is the traffic redundancy, Nm,NsRespectively a large wireless service node and a micro wireless service node deployed in a square grid. The first twoThe limiting conditions ensure that the number of enabled service nodes is unlikely to be greater than the number of deployed nodes, and the last limiting condition ensures that the enabled wireless service nodes can provide sufficient traffic.The representation is rounded up in order to keep at least 1/2 large wireless service nodes operational to ensure coverage of the network.
3) In order to solve the optimization problem, the invention models the capacity of the large wireless service node and the micro wireless service nodeFitting to a polynomial and then using an optimization tool, such as CVX or the like, the optimum n can be foundmAnd ns。
Step S3: the performance of the proposed dynamic wireless service node dormancy strategy is evaluated through a large number of experiments, and researches show that more energy can be saved in areas with more micro wireless service nodes or larger flow fluctuation.
1) With the increase of the density of the large wireless service nodes and the micro wireless service nodes, the network capacity is increased rapidly and then increased in speed and decreased, the interference among cells is more serious due to the higher probability of line-of-sight transmission in suburbs, so that the throughput of the large wireless service nodes is smaller, the throughput of the large wireless service nodes is limited by the number of users in the coverage range of the large wireless service nodes due to the smaller population in suburbs, and it can be seen that the influence of the regional characteristics on the performance is obvious, and similar phenomena can be observed in the capacity modeling of the micro wireless service nodes.
2) Although wireless traffic is changed violently with time, the active large wireless service nodes are changed less, the main reason is that the main task of the mobile communication system is to ensure the cell coverage and handle the mobility of users, while the number of the active micro wireless service nodes is changed dynamically with the change of real-time traffic, that is, the throughput of the micro wireless service nodes in the urban central area is increased, and as the power consumption of the micro wireless service nodes is much less than that of the large wireless service nodes, the micro wireless service nodes should be activated first to provide a larger network throughput in order to achieve the purpose of energy saving.
3) The proposed wireless service node dormancy strategy may save 65% of energy during off-peak periods compared to the total power consumption of all large and small wireless service nodes.
In order to further understand the method for estimating the capacity of a wireless serving node based on the migration learning proposed by the present invention, the following detailed description is made with reference to specific embodiments. The embodiment is implemented on the premise of the technical scheme of the invention, and a detailed implementation mode and a specific operation process are given.
As shown in fig. 2, the distribution of wireless service nodes in different areas is schematically illustrated.
According to the result in the graph, the density of wireless service nodes in rural areas is 18/km2-91/km2Within the range, the wireless nodes in rural areas can be determined to be large wireless service nodes. In this case, the capacity of a large wireless service node can be modeled using traffic data and regional eigenvectors for the rural area. That is, the rural area may be viewed as a source domain labeled with large wireless service nodes.
Fig. 3 is a diagram illustrating simulation results of different methods in a suburban area.
According to the real-time flow demand, the algorithm realized by the method is compared with the baseline algorithm under the condition of minimum energy consumption.
The analysis shows that the method for estimating the capacity of the wireless service node based on the transfer learning can obtain a better wireless service node dormancy strategy than the existing method, can well improve the energy consumption problem, and has certain reference value and practical economic benefit.
The present invention is not limited to the above-mentioned preferred embodiments, and any other various methods for estimating the capacity of a wireless service node based on migration learning can be derived from the teaching of the present invention.
Claims (3)
1. A wireless service node capacity estimation method based on transfer learning is characterized by comprising the following steps:
step S1: classifying the areas in the wireless network by using a clustering algorithm according to wireless service node deployment modes in different areas, and establishing a capacity model of a large wireless service node and a micro wireless service node by using transfer learning;
step S2: according to the real-time flow demand, a wireless service node capacity estimation model and a wireless service node dormancy algorithm, in an area meeting the network coverage and the real-time flow demand, the power consumption is reduced to the maximum extent;
the wireless service node dormancy algorithm enables the energy consumption to be minimum on the premise of meeting the flow demand by activating the large and micro wireless service nodes with the optimal number;
step S1 specifically includes the following steps:
step S11: taking the peak value of the flow of the area as the capacity of the wireless node; dividing the city into different types of areas by using a clustering algorithm, and analyzing and displaying the different types of areas by using big data to obtain different flow modes and wireless service node deployment modes;
step S12: capturing mode similarity and diversity among different regions by using a transfer learning algorithm to obtain a capacity model of the wireless service node suitable for different region characteristics;
in step S1, the influence of the external environmental feature r on the volume model is extracted from the multi-domain dataset in real life;
firstly, the specific influence of the capacity and the domain data on the capacity of the service node is analyzed by using the time delay correlation between the capacity and the domain data:where k (l) represents a capacity sequence, m (l) represents each data sequence, τ is a time delay, N is a sequence length, and σ is a standard deviation; will have a strong correlation to the flow peakSexual factors, incorporated into the external factors r; secondly, grouping the square grid areas with similar characteristics into a cluster by adopting a clustering algorithm, and dividing the whole urban area into four types of areas: rural, suburban, urban, city center;
acquiring features of different regions using transfer learning; model training is carried out on external environment characteristics, the number of large wireless service nodes and peak flow of rural areas by utilizing a multilayer sensing network to obtain a finite function between the capacity under the external characteristics r and the number of the large wireless service nodes; then, taking the model as a source domain, and transmitting the parameters to the suburb of the next target domain, so that the target domain model has the characteristics of both rural areas and suburb areas; after the model is gradually migrated and learned among the four regions, the mapping relation among the network capacity, the external environment characteristics and the number of the large wireless service nodes is finally obtained, namely the model is used for solving the problem that the mapping relation is not suitable for the large wireless service nodesWherein n ismThe number of large wireless service nodes in the area;
obtaining the number of micro wireless service nodes in the square grids in different areas according to the capacity model of the large wireless service node; by usingDeducting the network capacity provided by the large wireless service node to obtain the network capacity provided by the micro wireless service node; then obtaining the mapping relation among the network capacity, the external factor r and the number of the micro wireless service nodes through deep learning, namelynsThe number of micro wireless service nodes;
in step S2, the wireless service node dormancy algorithm specifically includes the following steps:
step S21: establishing a power consumption model of the wireless service node:
wherein the fixed power of the wireless service nodeOverhead c0The power consumption of the circuit elements in the antenna is represented by c1Denotes, M is the number of antennas, c2For each antenna power consumption, the power consumption of the traffic load is BμB is power coefficient, mu represents traffic load generated by wireless service node, and power consumption of power amplifierIs represented by the formula, wherein KρRepresenting the input power of the wireless service node antenna, η representing the efficiency of the power amplifier, and the total power of the wireless service node is represented as:by PmAnd PsRespectively representing the total power of the large wireless service node and the total power of the micro wireless service node;
step S22: through a power model of the wireless service node, a wireless service node dormancy strategy algorithm is adopted to find out the optimal number n of the large wireless service nodes and the micro wireless service nodes which need to be startedmAnd ns(ii) a The power loss in the area is minimized, and the coverage of the network and the real-time flow of the network are ensured;
this optimization problem can be mathematically expressed as:
s.t.0<ns≤Ns
where μ is real-time traffic, δ is traffic redundancy, NmAnd NsRespectively a large wireless service node and a micro wireless service node deployed in the square grid; the first two limiting conditions ensure that the number of the opened service nodes is not larger than the number of the deployed nodes, and the last limiting condition ensures that the opened wireless service nodes can provide enough flow;the representation is rounded up in order to keep at least 1/2 large wireless service nodes operational to ensure coverage of the network.
2. The method of claim 1, wherein the method comprises:
in step S11, analyzing the maximum traffic that the wireless service node can provide in different areas; clustering is realized by using multi-domain data as features and adopting a k-means algorithm: dividing the regions into different types of regions, wherein square grid regions with similar characteristics are grouped into a cluster; the whole urban area is divided into four types: rural, suburban, urban, city center;
in step S12, a capacity model of the large wireless service node is obtained by deep learning, and then a migration learning algorithm is used to capture pattern similarity and diversity between different areas, so as to obtain an overall capacity model of the large service node in different areas.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110134344.1A CN112954715B (en) | 2021-02-01 | 2021-02-01 | Wireless service node capacity estimation method based on transfer learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110134344.1A CN112954715B (en) | 2021-02-01 | 2021-02-01 | Wireless service node capacity estimation method based on transfer learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112954715A CN112954715A (en) | 2021-06-11 |
CN112954715B true CN112954715B (en) | 2022-05-31 |
Family
ID=76240360
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110134344.1A Active CN112954715B (en) | 2021-02-01 | 2021-02-01 | Wireless service node capacity estimation method based on transfer learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112954715B (en) |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2901772A1 (en) * | 2012-09-26 | 2015-08-05 | Telefonaktiebolaget LM Ericsson (PUBL) | Energy efficient operation for wireless access nodes in a low-load service area within a dense deployment of wireless access nodes |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105050170B (en) * | 2015-06-30 | 2019-02-05 | 东南大学 | A kind of base station dormancy method based on volume forecasting in heterogeneous network |
CN110381523B (en) * | 2019-06-17 | 2023-04-07 | 盐城吉研智能科技有限公司 | Cellular base station network traffic prediction method based on TVF-EMD-LSTM model |
CN111246552B (en) * | 2020-01-14 | 2021-11-26 | 南京工业大学 | Base station dormancy method based on mobile network flow prediction |
-
2021
- 2021-02-01 CN CN202110134344.1A patent/CN112954715B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2901772A1 (en) * | 2012-09-26 | 2015-08-05 | Telefonaktiebolaget LM Ericsson (PUBL) | Energy efficient operation for wireless access nodes in a low-load service area within a dense deployment of wireless access nodes |
Also Published As
Publication number | Publication date |
---|---|
CN112954715A (en) | 2021-06-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Chang et al. | Energy saving technology of 5G base station based on internet of things collaborative control | |
Zhang et al. | A clustering routing protocol for energy balance of wireless sensor network based on simulated annealing and genetic algorithm | |
Tyagi et al. | An intelligent and optimal resource allocation approach in sensor networks for smart agri-IoT | |
CN111445111B (en) | Electric power Internet of things task allocation method based on edge cooperation | |
Cao et al. | A novel coverage optimization strategy for heterogeneous wireless sensor networks based on connectivity and reliability | |
CN111176929A (en) | Edge federal learning-oriented high-energy-efficiency calculation communication joint optimization method | |
CN110968426B (en) | Edge cloud collaborative k-means clustering model optimization method based on online learning | |
CN110351754A (en) | Industry internet machinery equipment user data based on Q-learning calculates unloading decision-making technique | |
CN107171701B (en) | Power distribution method of MassiveMIMO system based on hybrid energy acquisition | |
CN108667653B (en) | Cluster-based cache configuration method and device in ultra-dense network | |
CN112801411B (en) | Network traffic prediction method based on generation of countermeasure network | |
Prasath et al. | RMCHS: Ridge method based cluster head selection for energy efficient clustering hierarchy protocol in WSN | |
Khan et al. | On the application of agglomerative hierarchical clustering for cache-assisted D2D networks | |
CN112954715B (en) | Wireless service node capacity estimation method based on transfer learning | |
CN113676357A (en) | Decision method for edge data processing in power internet of things and application thereof | |
CN106954268A (en) | Access network resource distribution method under a kind of SDN frameworks | |
Sun et al. | Energy efficiency-driven mobile base station deployment strategy for shopping malls using modified improved differential evolution algorithm | |
CN114449610A (en) | Wireless sensing node clustering and cluster head election method based on network division | |
Huang et al. | Mobile traffic offloading with forecasting using deep reinforcement learning | |
CN112188600B (en) | Method for optimizing heterogeneous network resources by reinforcement learning | |
Li et al. | Clustering routing protocol for wireless sensor networks based on improved QPSO algorithm | |
CN104394599B (en) | A kind of design method of CSMA/CA agreements applied to M2M networks | |
Peng et al. | Edge-cloud collaborative computation offloading for federated learning in smart city | |
CN113852400B (en) | Carbon neutralization pre-coding method for wireless multi-point cooperative transmission system | |
CN115515144B (en) | Heterogeneous AIoT Ad hoc network signal full-coverage method and 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 |