CN112188428B - Energy efficiency optimization method for Sink node in sensor cloud network - Google Patents
Energy efficiency optimization method for Sink node in sensor cloud network Download PDFInfo
- Publication number
- CN112188428B CN112188428B CN202011037390.1A CN202011037390A CN112188428B CN 112188428 B CN112188428 B CN 112188428B CN 202011037390 A CN202011037390 A CN 202011037390A CN 112188428 B CN112188428 B CN 112188428B
- Authority
- CN
- China
- Prior art keywords
- sensor
- sink node
- deep neural
- neural network
- sink
- 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
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000005457 optimization Methods 0.000 title claims abstract description 29
- 238000003062 neural network model Methods 0.000 claims abstract description 66
- 238000012360 testing method Methods 0.000 claims abstract description 28
- 239000013598 vector Substances 0.000 claims abstract description 26
- 238000012549 training Methods 0.000 claims abstract description 21
- 238000004891 communication Methods 0.000 claims abstract description 19
- 239000011159 matrix material Substances 0.000 claims description 27
- 238000004364 calculation method Methods 0.000 claims description 17
- 238000009826 distribution Methods 0.000 claims description 14
- 230000004913 activation Effects 0.000 claims description 9
- 238000005520 cutting process Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 230000001351 cycling effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 230000001186 cumulative effect Effects 0.000 description 3
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000005315 distribution function Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000001413 cellular effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/38—Services specially adapted for particular environments, situations or purposes for collecting sensor information
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- 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
- H04W52/04—TPC
- H04W52/38—TPC being performed in particular situations
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor 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)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention relates to the technical field of wireless communication, in particular to an energy efficiency optimization method of Sink nodes in a sensing cloud network, which comprises the following steps: s1, constructing a sensing cloud network, wherein a plurality of sensor nodes form a sensor area of a honeycomb structure in the bottom layer of the sensing cloud network, and each sensor area is provided with a Sink node; s2, constructing a reference algorithm model to obtain a test optimal beam forming vector of the Sink node; s3, constructing a deep neural network model, and training and testing the deep neural network model through data of the reference algorithm model; s4, optimizing energy efficiency, and obtaining an optimal beam forming vector of the Sink node through the trained and tested deep neural network model. The method and the device can improve the energy efficiency of Sink nodes, reduce the decision time of the sensing cloud network and improve the real-time performance of the system.
Description
Technical Field
The invention relates to the technical field of wireless communication, in particular to an energy efficiency optimization method for Sink nodes in a sensor cloud network.
Background
With popularization and development of communication engineering facilities, the requirements of people on high-energy-efficiency wireless communication are also higher and higher, and how to improve the energy efficiency of the wireless communication and the energy transmission again in the process of transmitting the information and the energy becomes a key problem of next-generation wireless communication.
At present, the sensing cloud technology is a sensor and application service platform integrating transmission, storage, acquisition, visualization, interface, APP and WeChat, and supports the privatization deployment of multiple mode platforms. Because of the strong hardware product access capability of the sensing cloud, the Internet of things access cost of the hardware product is effectively reduced, and the sensing cloud is a verification, display and application platform of the hardware products such as a sensor and a controller. The wireless network resource management strategy designed based on the traditional optimization method is often high in complexity and poor in real-time performance, is unfavorable for making an online decision, and is long in decision time of the sensing cloud network.
Disclosure of Invention
In order to solve the problems, the invention provides an energy efficiency optimization method for Sink nodes in a sensing cloud network, which can improve the energy efficiency of the Sink nodes, reduce the decision time of the sensing cloud network and improve the real-time performance of a system.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
an energy efficiency optimization method for Sink nodes in a sensing cloud network comprises the following steps:
s1, constructing a sensing cloud network, wherein a plurality of sensor nodes form a sensor area of a honeycomb structure in the bottom layer of the sensing cloud network, each sensor area is provided with a Sink node, and the Sink nodes are provided with multiple antennas so that the Sink nodes provide wireless energy-carrying communication services for the sensor nodes in the corresponding sensor areas;
s2, constructing a reference algorithm model, and inputting test channel state information of the Sink node to the reference algorithm model to obtain a test optimal beam forming vector of the Sink node;
s3, constructing a deep neural network model, and training and testing the deep neural network model through data of the reference algorithm model;
s4, optimizing energy efficiency, replacing the standard algorithm model with the trained and tested deep neural network model, and inputting the Sink node channel state information to the trained and tested deep neural network model to obtain an output optimal beam forming vector of the Sink node.
Thereby optimizing the energy efficiency of Sink nodes in the sensor cloud network.
Further, the construction method of the reference algorithm model comprises the following steps:
A1. initialization of
A2. Calculation of
A3. Calculation of
A4. Setting t=0, cycling through steps A1-A3, and accumulating 1 every time t is cycled;
A5. updating
A6. Updating
A7. Updating
A8. Calculation of
A9. When (when)When the Sink node is obtained, the experimental optimal beam forming matrix W of the Sink node is obtained i By testing the optimal beam forming matrix W of the Sink node i Performing eigenvector calculation to obtain experimental optimal beamforming vector +_of the ith Sink node>
Wherein,representing parameters of the Sink node in the ith sensor area in the iteration process of the reference algorithm model for three parameters, wherein t is the number of iterations; p (P) i Is the transmitting power of Sink node; ρ i The power cutting ratio is the power cutting ratio when the sensor node receives wireless energy carrying communication; η (eta) i The time length ratio of the downlink stage to one system period is calculated; h is a ik Data that is channel state; />Is random noise at the Sink node; alpha ik The weight coefficient of the kth sensor node in the ith sensor area is set as the weight coefficient of the kth sensor node in the ith sensor area; p (P) max Maximum transmit power for the sensor node; epsilon is an error parameter; t (T) i Representing the duration of one system cycle in the ith sensor area; w (W) i Trial beamforming matrix W for the Sink node i ；/>Trial optimal beamforming vector for the ith Sink node, +.>For 1 row M i Vector of columns.
Further, the Sink node tests the optimal beamforming matrix W i The calculation method comprises the following steps:
wherein,
E ik ＝p ik (1-η i )T i formula (10)
Further, the deep neural network model comprises an input layer, a plurality of hidden layers and an output layer, wherein the input layer, the hidden layers and the output layer are formed in a fully-connected mode; the activation function of the hidden layer adopts a ReLU function, and the activation function of the output layer is as follows:
y＝min(max(x,0),1)
formula (11)
Wherein x is the output of the hidden layer, and y is the output of the activation function of the output layer.
Further, the input of the deep neural network model is a model of scale (I, K, M i ) The output of the deep neural network model is the optimal beam forming matrix of the I Sink nodesAnd the data output by the deep neural network has the scale of (I, M) i ,M i ) Is a three-dimensional array of (c) in the array,
wherein I is the number of the sensor areas, K is the number of the sensor nodes existing in the sensor area where each Sink node is located, and M i Representing the number of Sink node antennas;
the training method of the deep neural network model comprises the following steps:
B1. standard normal distribution generation letter based on certaintyData of track status
B2. Setting epsilon value of step A9, and obtaining optimal beam forming matrix according to the reference algorithm model
B3. Repeating steps B1-B2 a plurality of times, and transmitting the data h of the channel state ik Conversion to the scale (I, K, M i ) Taking the three-dimensional array obtained in the step B2 as an input value of the deep neural network modelAs an output value of the deep neural network model;
B4. and B3, taking the input value and the output value of the deep neural network model in the step B3 as a data sample set, and slicing the data sample set into a training set and a testing set for training and testing the deep neural network model.
Further, a model of a difference between the beam forming matrix of the Sink node output by the deep neural network model and the test beam forming matrix of the Sink node output by the reference algorithm model is calculated to obtain a loss function, and when the loss function is lower than a threshold value, the deep neural network model is trained.
Further, the training set optimizes the weight of the deep neural network model by adopting a gradient descent algorithm.
Further, the Sink node is powered by a stable power supply, and the energy of the sensor node is collected from energy in the downlink wireless energy-carrying communication service process.
Further, the sensor node is equipped with a single antenna.
The beneficial effects of the invention are as follows:
1. the channel state information and the experimental optimal beam forming vector of the corresponding Sink node can be obtained by constructing a reference algorithm model, so that optimal input and output data can be provided for a subsequent deep neural network model; the data of the reference algorithm model is utilized to train and test the deep neural network model, and the reference algorithm model is replaced by the deep neural network model after training is finished so as to plan the operation of Sink nodes, so that the energy efficiency of the Sink nodes can be improved, the decision time of the sensing cloud network is shortened, the real-time performance of the system is improved, the energy efficiency optimization of the Sink nodes can be realized, and better real-time performance and lower complexity are provided.
2. In reference algorithm model construction, acquisition is performed according to channel statesThe error parameter of the reference algorithm model can be calculated by using the iteration process parameter of the reference algorithm model and the parameter of the Sink node according to the parameter in the iteration process of the expression reference algorithm model, and when the error parameter is smaller than the set error threshold, the test beam forming matrix W of the Sink node is proved i To test the optimal beamforming matrix W i By testing the beam forming matrix W of the Sink node i Performing eigenvector calculation to obtain experimental optimal beamforming vector +.>In order to improve the performance of the sensing cloud network, the Sink node is provided with multiple antennas, so that the channel state is a vector of 1 row Mi column, and therefore, the output of the reference algorithm model is also a vector of 1 row Mi column. Obtaining an optimal beam forming vector of a Sink node by using a reference algorithm model; the approximation of the reference algorithm model is realized by designing the corresponding deep neural network model, and the sensing cloud network energy efficiency distribution similar to the reference algorithm can be realized due to the good approximation effect of the deep neural network model on the reference algorithm model.
Drawings
Fig. 1 is a flowchart of an energy efficiency optimization method of Sink nodes in a sensor cloud network according to a preferred embodiment of the present invention.
Fig. 2 is a schematic diagram of a Sink node structure of an energy efficiency optimization method of a Sink node in a sensor cloud network according to a preferred embodiment of the present invention.
Fig. 3 is a timing diagram illustrating an energy efficiency optimization method of Sink nodes in a sensor cloud network according to a preferred embodiment of the present invention.
Fig. 4 is a schematic structural diagram of a deep neural network model of an energy efficiency optimization method of Sink nodes in a sensor cloud network according to a preferred embodiment of the present invention.
Fig. 5 is an energy efficiency cumulative comparison distribution diagram of an energy efficiency optimization method of Sink nodes in a sensor cloud network according to a preferred embodiment of the present invention.
In the figure, a 1-sensor cloud network, a 2-sensor area and 3-Sink nodes.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.
Referring to fig. 1 to fig. 4, the energy efficiency optimization method of Sink nodes in a sensor cloud network according to a preferred embodiment of the present invention includes the following steps:
s1, constructing a sensing cloud network, wherein a plurality of sensor nodes form a sensor area 2 with a honeycomb structure in the bottom layer of the sensing cloud network 1, each sensor area 2 is provided with a Sink node 3, and the Sink nodes 3 are provided with multiple antennas so that the Sink nodes 3 provide wireless energy-carrying communication services for the sensor nodes in the corresponding sensor areas; the sensor node is equipped with a single antenna.
In this embodiment, sink node 3 is powered by a stable power source, and the energy of the sensor node is collected from the energy in the downlink wireless energy-carrying communication service process.
As shown in fig. 1, the energy of the sensor nodes is all from energy collection during the wireless energy-carrying communication service. The sensing cloud network 1 is periodically divided into two stages of downlink wireless energy carrying communication and uplink WIT. In the downlink stage, sink node 3 provides wireless energy-carrying communication service for the sensor nodes in the sensor area 2, and the sensor nodes realize synchronous receiving of energy and task information in a Power Split (PS) working mode and assume that the duration of the downlink stage is sufficient for receiving the task information, and the sensor nodes in the uplink stage upload the acquired data to Sink node 3 by using the energy collected in the downlink stage.
As shown in fig. 2, wherein the system period of the ith cell is denoted as T i The duration factor of the downlink stage is eta i . In order to further improve the performance of the sensor cloud network 1, the Sink node 3 is provided with multiple antennas to enhance the wireless energy-carrying communication efficiency, and the sensor node is provided with a single antenna. And then, constructing a system energy efficiency optimization problem, and respectively designing a reference algorithm model and a deep neural network model to solve the optimal beamforming vector.
S2, constructing a reference algorithm model, and inputting test channel state information of the Sink node 3 to the reference algorithm model to obtain a test optimal beam forming vector of the Sink node 3.
S3, constructing a deep neural network model, and training and testing the deep neural network model through data of a reference algorithm model.
S4, optimizing energy efficiency, replacing the standard algorithm model with the trained and tested deep neural network model, and inputting the channel state information of the Sink node 3 through the trained and tested deep neural network model to obtain an output optimal beam forming vector of the Sink node 3.
The channel state information and the corresponding experimental optimal beam forming vector of the Sink node 3 can be obtained by constructing a reference algorithm model, so that optimal input and output data can be provided for a subsequent deep neural network model. The data of the reference algorithm model is utilized to train and test the deep neural network model, and the reference algorithm model is replaced by the deep neural network model after training is finished so as to plan the operation of the Sink node 3, so that the energy efficiency of the Sink node 3 can be improved, the decision time of the sensing cloud network 1 is shortened, the real-time performance of the system is improved, the energy efficiency optimization of the Sink node 3 can be realized, and better real-time performance and lower complexity are provided.
In the step S2 of the process of the present invention,
the construction method of the reference algorithm model comprises the following steps:
A1. initialization of
A2. Calculation of
A3. Calculation of
A4. Setting t=0, cycling through steps A1-A3, and accumulating 1 every time t is cycled;
A5. updating
A6. Updating
A7. Updating
A8. Calculation of
A9. When (when)In this case, a trial optimal beamforming matrix W for Sink node 3 is obtained i Optimal beamforming matrix W by trial on Sink node 3 i Performing eigenvector calculation to obtain experimental optimal beamforming vector +.>
Wherein,representing parameters of Sink node 3 in the ith sensor area 2 in the iteration process of the reference algorithm model for three parameters, wherein t is the number of iterations; p (P) i Is the transmit power of Sink node 3; ρ i The power cutting ratio is the power cutting ratio when the sensor node receives wireless energy carrying communication; η (eta) i The time length ratio of the downlink stage to one system period is calculated; h is a ik For the data of the channel state, h is because Sink node 3 is equipped with multiple antennas ik For 1 row M i Vectors of columns; />Is random noise at Sink node 3; alpha ik The weight coefficient of the kth sensor node in the ith sensor area 2; p (P) max Maximum transmit power for the sensor node; epsilon is an error parameter; t (T) i Representing the duration of one system cycle within the i-th sensor region 2; w (W) i Trial beamforming matrix W for Sink node 3 i ；/>Trial optimal beamforming vector for the ith Sink node 3,/v->For 1 row M i Vector of columns.
Experimental optimal beamforming matrix W for Sink node 3 i The calculation method comprises the following steps:
W i ＝h ik -1 E ik (h ik H ) -1 (P i ρ i η i T i ) -1 formula (8)
Wherein,
E ik ＝p ik (1-η i )T i formula (10)
In step S3, the deep neural network model includes an input layer, a plurality of hidden layers, and an output layer, where the input layer, the hidden layers, and the output layer are formed in a fully connected manner; the activation function of the hidden layer adopts a ReLU function, and the activation function of the output layer is as follows:
y=min (max (x, 0), 1) formula (11)
Where x is the output of the hidden layer and y is the output of the activation function of the output layer.
In this embodiment, the input of the deep neural network model is a model of scale (I, K, M) i ) Is a three-dimensional array of I Sink nodes 3, and the output of the deep neural network model is an optimal beam forming matrix of the I Sink nodes 3And the data scale of the deep neural network output is (I, M) i ,M i ) Is a three-dimensional array of (c) in the array,
wherein I is the number of sensor areas 2, K is the number of sensor nodes existing in the sensor area 2 where each Sink node 3 is located, M i Representing the number of Sink node 3 antennas;
the training method of the deep neural network model comprises the following steps:
B1. generating data of channel state based on standard normal distribution of certaintyThe present embodiment uses Rayleigh fading profiles of zero mean and unit variance.
B2. Setting epsilon value of step A9, obtaining optimal beam forming matrix according to the reference algorithm modelEpsilon of this example was set to 10 -3 。
B3. Repeating steps B1-B2 multiple times, and transmitting the data h of the channel state ik Conversion to the scale (I, K, M i ) Using the three-dimensional array obtained in the step B2 as an input value of a deep neural network modelAs an output value of the deep neural network model.
B4. And B3, taking the input value and the output value of the deep neural network model in the step B3 as a data sample set, and slicing the data sample set into a training set and a testing set for training and testing the deep neural network model.
In step B4, calculating a model of a difference between the beam forming matrix of the Sink node 3 output by the deep neural network model and the test beam forming matrix of the Sink node 3 output by the reference algorithm model to obtain a loss function, and when the loss function is lower than a threshold value, finishing training the deep neural network model.
The training set optimizes the weight of the deep neural network model by adopting a gradient descent algorithm, and in the embodiment, in order to improve the performance of the deep neural network model training process, the initialization of the weight and the side length obeys the truncated normal distribution, meanwhile, the attenuation rate is set to be 0.9, and the proper learning rate and the proper batch size are set through cross verification.
The implementation of the embodiment method can be achieved by replacing the channel vector with the channel coefficient for the Sink node equipped with the strong opening of the single antenna.
In this embodiment, the performance of this embodiment in terms of improving the energy efficiency of the sensing cloud network 1 is verified through four situations of a reference algorithm model (swift-SWMME), a random power algorithm, a maximum power algorithm, and a deep neural network model (DNN).
The system energy efficiency and its cumulative distribution function (Cumulative Distribution Function, CDF) achieved in 25000 computations for the four cases are shown in fig. 5, respectively.
The system strategy adopting the reference algorithm model realizes the best Sink node 3 energy efficiency; meanwhile, compared with two strategies of random power distribution and maximum power distribution, the energy efficiency of the system realized by the reference algorithm model algorithm has a broader distribution threshold value, because the interference management problem under different channel states is effectively considered in the algorithm calculation process, the compromise between the system performance and the energy consumption can be realized, otherwise, the random power distribution and the maximum power distribution algorithm cannot make corresponding decisions according to the channel states, namely, when the channel states are good, the transmitting power cannot be timely increased to improve the throughput of the system, and when the channel states are bad, the power cannot be timely reduced to reduce the energy transmission loss, so that the distribution range of the energy efficiency of the system is narrower.
Further, although the maximum power allocation strategy cannot adapt to the situation of poor channel state to a certain extent, the maximum power allocation strategy can improve the system performance under the situation of good channel state, so that the realized system energy efficiency value and the distribution range thereof are slightly larger than those of the random power allocation decision. On the other hand, the deep neural network model realizes a better approximation effect on the reference algorithm, and can realize the system energy efficiency and distribution similar to the reference algorithm. This demonstrates the effectiveness and performance of the Sink node 3 optimal energy efficiency method of the present embodiment.
As shown in Table 1, the comparison of the statistical data of the baseline algorithm model and the deep neural network model illustrates that the deep neural network model can achieve better real-time performance and lower complexity than the baseline algorithm model. Where K represents the number of sensor nodes present within the cellular sensor area 2 where each Sink node 3 is located. The comparison shows that the optimal energy efficiency algorithm based on the deep neural network model occupies lower CPU time than the optimal energy efficiency algorithm of the reference algorithm model, so that the complexity of the deep neural network model is lower and the shorter time is more beneficial to the on-line decision making.
TABLE 1 statistical data comparison of reference algorithm model and deep neural network model
Claims (7)
1. The energy efficiency optimization method for Sink nodes in the sensing cloud network is characterized by comprising the following steps of:
s1, constructing a sensing cloud network, wherein a plurality of sensor nodes form a sensor area of a honeycomb structure in the bottom layer of the sensing cloud network, each sensor area is provided with a Sink node, and the Sink nodes are provided with multiple antennas so that the Sink nodes provide wireless energy-carrying communication services for the sensor nodes in the corresponding sensor areas;
s2, constructing a reference algorithm model, and inputting test channel state information of the Sink node to the reference algorithm model to obtain a test optimal beam forming vector of the Sink node;
the construction method of the reference algorithm model comprises the following steps:
A1. initialization of
A2. Calculation of
A3. Calculation of
A4. Setting t=0, cycling through steps A1-A3, and accumulating 1 every time t is cycled;
A5. updating
A6. Updating
A7. Updating
A8. Calculation of
A9. When (when)When the Sink node is obtained, the experimental optimal beam forming matrix W of the Sink node is obtained i By testing the optimal beam forming matrix W of the Sink node i Performing eigenvector calculation to obtain experimental optimal beamforming vector +_of the ith Sink node>
Wherein,representing parameters of the Sink node in the ith sensor area in the iteration process of the reference algorithm model for three parameters, wherein t is the number of iterations; p (P) i Is the transmitting power of Sink node; ρ i The power cutting ratio is the power cutting ratio when the sensor node receives wireless energy carrying communication; η (eta) i The time length ratio of the downlink stage to one system period is calculated; h is a ik Data that is channel state; />Is random noise at the Sink node; alpha ik The weight coefficient of the kth sensor node in the ith sensor area is set as the weight coefficient of the kth sensor node in the ith sensor area; p (P) max Maximum transmit power for the sensor node; epsilon is an error parameter; t (T) i Representing the duration of one system cycle in the ith sensor area; w (W) i Trial beamforming matrix W for the Sink node i ；/>Trial optimal beamforming vector for the ith Sink node, +.>For 1 row M i Vectors of columns;
s3, constructing a deep neural network model, and training and testing the deep neural network model through data of the reference algorithm model;
the deep neural network model comprises an input layer, a plurality of hidden layers and an output layer, wherein the input layer, the hidden layers and the output layer are formed in a fully-connected mode; the activation function of the hidden layer adopts a ReLU function, and the activation function of the output layer is as follows:
y=min (max (x, 0), 1) formula (11)
Wherein x is the output of the hidden layer, and y is the output of the activation function of the output layer;
s4, optimizing energy efficiency, replacing the standard algorithm model with the trained and tested deep neural network model, and inputting the Sink node channel state information to the trained and tested deep neural network model to obtain an output optimal beam forming vector of the Sink node.
2. The energy efficiency optimization method for Sink nodes in a sensor cloud network according to claim 1, wherein the energy efficiency optimization method is characterized by comprising the following steps: experimental optimal beam forming matrix W of Sink node i The calculation method comprises the following steps:
W i ＝h ik -1 E ik (h ik H ) -1 (P i ρ i η i T i ) -1 formula (8)
Wherein,
E ik ＝p ik (1-η i )T i equation (10).
3. The energy efficiency optimization method for Sink nodes in a sensor cloud network according to claim 1, wherein the energy efficiency optimization method is characterized by comprising the following steps:
the input of the deep neural network model is a model with a scale of (I, K, M) i ) The output of the deep neural network model is the optimal beamforming matrix { W } of the I Sink nodes i opt And the data size of the deep neural network output is (I, M) i ,M i ) Is a three-dimensional array of (c) in the array,
wherein I is the number of the sensor areas, K is the number of the sensor nodes existing in the sensor area where each Sink node is located, and M i Representing the number of Sink node antennas;
the training method of the deep neural network model comprises the following steps:
B1. generating data of channel state based on standard normal distribution of certainty
B2. Setting epsilon value of step A9, and obtaining optimal beam forming matrix according to the reference algorithm model
B3. Repeating steps B1-B2 a plurality of times, and transmitting the data h of the channel state ik Conversion to the scale (I, K, M i ) Taking the three-dimensional array obtained in the step B2 as an input value of the deep neural network modelAs an output value of the deep neural network model;
B4. and B3, taking the input value and the output value of the deep neural network model in the step B3 as a data sample set, and slicing the data sample set into a training set and a testing set for training and testing the deep neural network model.
4. The energy efficiency optimization method for Sink nodes in a sensor cloud network according to claim 3, wherein the energy efficiency optimization method is characterized by comprising the following steps: and calculating a model of a difference between the beam forming matrix of the Sink node output by the deep neural network model and the test beam forming matrix of the Sink node output by the reference algorithm model to obtain a loss function, and when the loss function is lower than a threshold value, finishing training of the deep neural network model.
5. The energy efficiency optimization method for Sink nodes in a sensor cloud network according to claim 3, wherein the energy efficiency optimization method is characterized by comprising the following steps: and the training set adopts a gradient descent algorithm to optimize the weight of the deep neural network model.
6. The energy efficiency optimization method for Sink nodes in a sensor cloud network according to claim 1, wherein the energy efficiency optimization method is characterized by comprising the following steps: the Sink node is powered by a stable power supply, and the energy of the sensor node is collected in the downlink wireless energy-carrying communication service process.
7. The energy efficiency optimization method for Sink nodes in a sensor cloud network according to claim 1, wherein the energy efficiency optimization method is characterized by comprising the following steps: the sensor node is equipped with a single antenna.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011037390.1A CN112188428B (en) | 2020-09-28 | 2020-09-28 | Energy efficiency optimization method for Sink node in sensor cloud network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011037390.1A CN112188428B (en) | 2020-09-28 | 2020-09-28 | Energy efficiency optimization method for Sink node in sensor cloud network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112188428A CN112188428A (en) | 2021-01-05 |
| CN112188428B true CN112188428B (en) | 2024-01-30 |
Family
ID=73944365
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011037390.1A Active CN112188428B (en) | 2020-09-28 | 2020-09-28 | Energy efficiency optimization method for Sink node in sensor cloud network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112188428B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116111730B (en) * | 2023-04-13 | 2023-06-23 | 国网江西省电力有限公司信息通信分公司 | Power grid monitoring method based on electric power optical cable co-fiber transmission system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103841571A (en) * | 2014-03-20 | 2014-06-04 | 吉林大学 | Wireless sensor network beam forming transmission array node selecting method |
| CN106888430A (en) * | 2017-04-17 | 2017-06-23 | 华侨大学 | A kind of believable sensing cloud Data Collection appraisal procedure |
| CN107426826A (en) * | 2017-04-27 | 2017-12-01 | 成都瑞沣信息科技有限公司 | The MAC protocol for wireless sensor networks design method collected based on RF energy |
| CN108924897A (en) * | 2018-06-30 | 2018-11-30 | 北京工业大学 | A kind of mobile sink paths planning method based on deeply learning algorithm |
| CN109617584A (en) * | 2019-01-08 | 2019-04-12 | 南京邮电大学 | A kind of mimo system beamforming matrix design method based on deep learning |
| CN110300380A (en) * | 2019-07-30 | 2019-10-01 | 电子科技大学 | The method for tracking target of balance system energy consumption and tracking precision in mobile WSN |
| CN110769444A (en) * | 2019-10-25 | 2020-02-07 | 东北大学 | Transmission method of wireless energy-carrying communication based on power distribution |
| WO2020048594A1 (en) * | 2018-09-06 | 2020-03-12 | Nokia Technologies Oy | Procedure for optimization of self-organizing network |
| WO2020098256A1 (en) * | 2018-11-14 | 2020-05-22 | 平安科技（深圳）有限公司 | Speech enhancement method based on fully convolutional neural network, device, and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11343144B2 (en) * | 2019-03-12 | 2022-05-24 | Cisco Technology, Inc. | Downlink performance optimizations in wireless networks |
-
2020
- 2020-09-28 CN CN202011037390.1A patent/CN112188428B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103841571A (en) * | 2014-03-20 | 2014-06-04 | 吉林大学 | Wireless sensor network beam forming transmission array node selecting method |
| CN106888430A (en) * | 2017-04-17 | 2017-06-23 | 华侨大学 | A kind of believable sensing cloud Data Collection appraisal procedure |
| CN107426826A (en) * | 2017-04-27 | 2017-12-01 | 成都瑞沣信息科技有限公司 | The MAC protocol for wireless sensor networks design method collected based on RF energy |
| CN108924897A (en) * | 2018-06-30 | 2018-11-30 | 北京工业大学 | A kind of mobile sink paths planning method based on deeply learning algorithm |
| CN109936865A (en) * | 2018-06-30 | 2019-06-25 | 北京工业大学 | A kind of mobile sink paths planning method based on deeply learning algorithm |
| WO2020048594A1 (en) * | 2018-09-06 | 2020-03-12 | Nokia Technologies Oy | Procedure for optimization of self-organizing network |
| WO2020098256A1 (en) * | 2018-11-14 | 2020-05-22 | 平安科技（深圳）有限公司 | Speech enhancement method based on fully convolutional neural network, device, and storage medium |
| CN109617584A (en) * | 2019-01-08 | 2019-04-12 | 南京邮电大学 | A kind of mimo system beamforming matrix design method based on deep learning |
| CN110300380A (en) * | 2019-07-30 | 2019-10-01 | 电子科技大学 | The method for tracking target of balance system energy consumption and tracking precision in mobile WSN |
| CN110769444A (en) * | 2019-10-25 | 2020-02-07 | 东北大学 | Transmission method of wireless energy-carrying communication based on power distribution |
Non-Patent Citations (4)
| Title |
|---|
| 《基于云计算的无线传感网络资源分配策略研究》;徐荣;《大庆师范学院学报》;第37卷(第6期);全文 * |
| 《基于物联网技术的无线型建筑环境监测系统设计与实现》;李锐;《智能处理与应用》(第11期);全文 * |
| 《无线传感网中基于sink节点的目标位置选择移动算法》;饶芳，谭建军;《现代电子技术》;第38卷(第19期);全文 * |
| Arijit Roy ; Ayan Mondal ; Sudip Misra.《Connectivity Re-establishment in the Presence of Dumb Nodes in Sensor-Cloud Infrastructure: A Game Theoretic Approach》.《IEEE》.2015,全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112188428A (en) | 2021-01-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Downlink transmit power control in ultra-dense UAV network based on mean field game and deep reinforcement learning | |
| CN111800828B (en) | Mobile edge computing resource allocation method for ultra-dense network | |
| CN111507601B (en) | Resource optimization allocation decision method based on deep reinforcement learning and block chain consensus | |
| CN113222179B (en) | Federal learning model compression method based on model sparsification and weight quantification | |
| CN112598150B (en) | Method for improving fire detection effect based on federal learning in intelligent power plant | |
| CN110942205B (en) | Short-term photovoltaic power generation power prediction method based on HIMVO-SVM | |
| CN112788605B (en) | Edge computing resource scheduling method and system based on double-delay depth certainty strategy | |
| Dai et al. | Energy‐efficient resource allocation for device‐to‐device communication with WPT | |
| CN113473580B (en) | User association joint power distribution method based on deep learning in heterogeneous network | |
| CN110300417B (en) | Energy efficiency optimization method and device for unmanned aerial vehicle communication network | |
| CN112260733B (en) | Multi-agent deep reinforcement learning-based MU-MISO hybrid precoding design method | |
| CN107453396A (en) | A kind of Multiobjective Optimal Operation method that distributed photovoltaic power is contributed | |
| CN113596785A (en) | D2D-NOMA communication system resource allocation method based on deep Q network | |
| CN112188428B (en) | Energy efficiency optimization method for Sink node in sensor cloud network | |
| CN109309539A (en) | A kind of information fusion shortwave frequency-selecting method based on deeply study | |
| CN113783593A (en) | Beam selection method and system based on deep reinforcement learning | |
| Li et al. | Deep neural network based computational resource allocation for mobile edge computing | |
| Xu et al. | Deep reinforcement learning for communication and computing resource allocation in RIS aided MEC networks | |
| CN116050504A (en) | Wind power short-term prediction model based on deep learning | |
| CN105792218A (en) | Optimization method of cognitive radio network with radio frequency energy harvesting capability | |
| CN113472402B (en) | Parameter adjusting method in MIMO intelligent reflector transmission system | |
| Jin et al. | A Context-aware Task Offloading Scheme in Collaborative Vehicular Edge Computing Systems. | |
| CN113242066B (en) | Multi-cell large-scale MIMO communication intelligent power distribution method | |
| Wang et al. | Energy efficiency resource management for D2D-NOMA enabled network: A dinkelbach combined twin delayed deterministic policy gradient approach | |
| CN104361399A (en) | Solar irradiation intensity minute-scale predication method |
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 |