CN110941675B - Wireless energy supply edge calculation delay optimization method based on deep learning - Google Patents

Wireless energy supply edge calculation delay optimization method based on deep learning Download PDF

Info

Publication number
CN110941675B
CN110941675B CN201911176654.9A CN201911176654A CN110941675B CN 110941675 B CN110941675 B CN 110941675B CN 201911176654 A CN201911176654 A CN 201911176654A CN 110941675 B CN110941675 B CN 110941675B
Authority
CN
China
Prior art keywords
node
energy
transmission
link
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
Application number
CN201911176654.9A
Other languages
Chinese (zh)
Other versions
CN110941675A (en
Inventor
张兴军
付哲
于博成
纪泽宇
蔡梦华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian Jiaotong University
Original Assignee
Xian Jiaotong University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Xian Jiaotong University filed Critical Xian Jiaotong University
Priority to CN201911176654.9A priority Critical patent/CN110941675B/en
Publication of CN110941675A publication Critical patent/CN110941675A/en
Application granted granted Critical
Publication of CN110941675B publication Critical patent/CN110941675B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • Databases & Information Systems (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Computing Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

The invention discloses a wireless energy supply edge calculation delay optimization method based on deep learning, and belongs to the field of edge calculation. The optimization method comprises the following steps: 1) establishing a time delay model for unloading tasks to an edge server based on wireless energy supply edge calculation of node cooperative transmission; 2) converting the edge server time delay model into a linear integer programming model by adopting a piecewise linearization algorithm; 3) carrying out simulation experiment to verify the effectiveness of the optimization method, and if the time delay is optimized, turning to the step 4); otherwise, go to step 2); 4) solving the linear integer programming model to obtain a data set of a training deep confidence network; 5) and training the DBN network by using the generated data set until the error rate is less than a preset value, and obtaining a trained network model. The problem of the 'near-far' effect of the wireless energy supply edge computing network is solved, and the time delay of data unloading to the edge server is reduced.

Description

Wireless energy supply edge calculation delay optimization method based on deep learning
Technical Field
The invention belongs to the field of edge calculation, and particularly relates to a wireless energy supply edge calculation delay optimization method based on deep learning.
Background
As the proportion of mobile data in the total amount of global data increases year by year, mobile cloud computing that migrates a data processing task to the cloud end has been unable to meet the requirements of users for low time delay and high service quality. To address the above needs, migration of cloud servers to edge computing closer to the user has arisen. The user can unload complex computing tasks to the edge server to meet the requirements of low time delay, low energy consumption and high service quality. With the continuous improvement of the hardware level of the mobile terminal and the gradual improvement of the related communication technology, the user can not only serve as a resource demanding party, but also serve as a resource providing party to contribute own computing storage resources for the mobile edge computing, thereby becoming an important component of the mobile edge computing.
Currently, batteries are used as the main power source of wireless devices, and the development of edge computing is limited due to the limited energy and difficulty in replacing batteries in specific environments. In order to solve the problem of energy limitation, wireless energy supply is researched more and more, the traditional solar energy, wind energy and other energy are greatly influenced by regions and nature, and wireless spectrum signal energy collection is researched widely due to controllable and predictable energy. The purpose of wireless energy supply communication is to charge devices or support device communication by using wireless energy, however, the amount of energy collected by wireless devices is influenced by channel quality, the more energy collected by wireless devices closer to an energy supply station, the more information can be transmitted, and conversely, the less energy collected by wireless devices farther from the energy supply station, so that when a task needs to be unloaded to an edge server, the task may not be transmitted to the edge server or as soon as possible due to insufficient energy, which is called "near-far" effect.
Disclosure of Invention
The invention aims to solve the problem of the near-far effect of a wireless energy supply edge computing network, and provides a wireless energy supply edge computing delay optimization method based on deep learning.
In order to achieve the purpose, the invention adopts the following technical scheme to realize the purpose:
a wireless energy supply edge calculation delay optimization method based on deep learning comprises the following steps:
1) based on the wireless energy supply edge calculation of node cooperative transmission, adopting a cooperative transmission protocol after energy acquisition, dividing a time block into downlink energy transmission time and uplink information transmission time, establishing a downlink energy transmission model and an uplink information transmission model, and establishing a time delay model for unloading tasks to an edge server according to the two models;
2) converting the edge server time delay model into a linear integer programming model by adopting a piecewise linearization algorithm;
3) carrying out simulation experiment to verify the effectiveness of the optimization method, comparing the effectiveness with the time delay without the cooperative direct transmission mode, and if the time delay is optimized, turning to the step 4); otherwise, go to step 2);
4) solving the linear integer programming model to obtain a data set of a training deep confidence network;
5) and training the DBN network by using the generated data set until the error rate is less than a preset value, and obtaining a trained network model.
Further, the method also comprises the following steps:
6) and (4) performing node power and link selection prediction by using the trained network model, and removing redundant links.
Further, in step 1), a downlink energy transmission model is established based on establishment of energy conservation, and the specific process is as follows:
the energy received by the user node i at the downlink time is Ei＝ζP0hHit0；
Wherein: ζ is the energy conversion efficiency of the user node, P0Is the transmission power of the base station, t0Time, h, for a user node to obtain wireless energy in downlinkHiPower gain for downlink transmission of the link;
in the transmission phase, the total transmission energy consumption of the user node i is
Figure BDA0002290143190000021
In the formula: pcFor power losses due to additional electronics losses during transmission of the user node, PiFor the transmission power of user node i, tijIs a link LijDuration of uplink transmission, LijThe link formed between the node i and the node j is obtained, L is an activated link, and N is a set of all user nodes;
the total energy consumption spent by the user node i in receiving information is
Figure BDA0002290143190000031
In the formula: eelecEnergy consumption required for receiving 1-bit information, ruiFor node i via link LuiAmount of information received, LuiIs a link formed between node u and node i;
the self-sustaining operation of the user nodes over the whole time block follows an energy constraint, i.e. the total energy consumption is not greater than the energy acquired, i.e. Eri+Eti≤Ei。
Further, in step 1), an uplink information transmission model is established based on establishment of flow conservation, and the specific process is as follows:
Cij＝Wlog(1+SNRij)
in the formula: cijIs a link LijThe capacity of the link of (a) is,
Figure BDA0002290143190000032
w is the bandwidth, η is the noise signal power, PiIs the transmission power of node i, gijIs a link LijThe power gain of (d);
the amount of information r transmitted per unit time on each linkijSatisfy link capacity constraints, i.e.
Figure BDA0002290143190000033
riThe data quantity generated by the user node i is represented, and the flow constraint condition satisfied by the user node i is
Figure BDA0002290143190000034
In the formula: l isijFor the link formed between node i and node j, rijFor node i via link LijAmount of information transmitted, ruiFor node i via link LuiAmount of information accepted, LuiFor the link formed between node i and node u, tijIs a link LijDuration in the uplink transmission;
the base station satisfies the following flow conservation constraint conditions
Figure BDA0002290143190000035
Further, the delay model of task unloading to the edge server in step 1) is as follows:
min
Figure BDA0002290143190000036
s.t.Eri+Eti≤Ei (1)
Figure BDA0002290143190000041
Figure BDA0002290143190000042
Figure BDA0002290143190000043
rij≥0,Pi≥0,t0≥0,tij≥0(5)
wherein E isiIs t0Energy gained by the user node in time, EriEnergy consumed for receiving data for node i, EtiEnergy consumed for transmitting data for node i, L represents an activated link, r represents traffic of the node and the line, P represents power, t0And obtaining the time of wireless energy for the user node in the downlink.
Further, the training method in step 5) is as follows:
carrying out unsupervised pre-training by adopting a greedy algorithm layer by layer to give an initial weight to the whole network;
and then carrying out supervised tuning training.
Further, the preset value of the error rate in the step 5) is 5%.
Compared with the prior art, the invention has the following beneficial effects:
the invention relates to a wireless energy supply edge calculation delay optimization method based on deep learning, which adopts a node cooperative transmission method to help a far node to transmit data through data forwarding among nodes so as to solve the problem of a near-far effect in an edge calculation energy supply network; under the wireless energy supply edge computing scene based on node cooperative transmission, constraint conditions such as energy conservation and traffic conservation are combined to optimize data unloading minimum delay, compared with node information transmission delay when a user node directly transmits information to a base station without node cooperation, the delay of data unloading to an edge server is reduced by 4.9%; the deep learning is further applied to the selection of the predicted node power and the link in the wireless energy supply edge calculation, redundant links are eliminated, the network model calculation amount is reduced, the time delay is further reduced, and the requirement of the real-time performance of the edge calculation is met.
Drawings
FIG. 1 is a flow chart of an embodiment of the present invention;
FIG. 2 is a schematic diagram of a single-hop transmission without coordination;
FIG. 3 is a schematic diagram of node cooperation based routing transmission;
FIG. 4 is a schematic diagram of the wireless device node cooperative transmission of the mobile edge computing wireless energy supply network;
fig. 5 is a schematic diagram of a mobile edge computing wireless power supply network transmission protocol.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The invention adopts a node cooperative transmission method to assist the wireless equipment to transmit information, so that the wireless equipment data with poor channel state can reach the edge server as soon as possible, thereby improving the service quality of users; the invention also adopts a deep learning method, and predicts the power of each node later by using the data set generated before to obtain the node power, and converts the original problem into a linear programming problem. Resource management tasks such as transmission power control and user authority control are important research contents in future wireless energy supply edge calculation, and the traditional numerical optimization method is long in execution and difficult to meet the requirement of edge calculation real-time performance. The trained deep learning model has strong computational efficiency, and in the training stage, the deep neural network is trained by using the existing data set, and the network can be used for subsequent resource management such as power prediction after convergence.
The invention is described in further detail below with reference to the accompanying drawings:
FIG. 1 is a deep learning-based method for optimizing the computation delay of a wireless power supply edge, which comprises the following steps:
s1, analyzing a wireless energy supply edge calculation scene, wherein a user needs to process own tasks by means of an edge server, and the edge server returns results to the user after finishing processing the tasks, wherein the results are returned to the user with less data volume and negligible time, and it is crucial to send own data to the server as soon as possible;
s2, because the battery energy is limited and the energy is self-limited and the battery is not easy to replace in specific environment, the prior energy-first and energy-second cooperative transmission protocol is generally adopted to divide the time block into downlink energy transmission time and uplink information transmission time. In the downlink time, the base station provides energy for the wireless equipment, and in the uplink time, the wireless equipment uses the received energy to unload the information which needs to be unloaded to the edge server to the server. And respectively establishing a downlink energy transmission model and an uplink information transmission model according to energy conservation and flow conservation. And finally, establishing a time delay model for unloading the task to the edge server, wherein the specific process is as follows:
the dependency relationship between decision variables in the data forwarding process needs to be considered when optimizing the delay performance. For the user node, the transmission data comprises two parts: forwarding the data and the data generated by itself. These two data quantities are considered as decision variables. Since the energy of the user node is limited, energy is consumed for transmitting data, and when the forwarding data is excessive, energy for transmitting the data amount of the user node is reduced, which may cause the user node to become a bottleneck affecting the delay performance. Therefore, it is necessary to balance the transmission amount of the two kinds of data on each user node. In addition, the time delay consists of two parts, namely energy supply time and activation time of a link formed between nodes in uplink transmission, and the two have a constraint relation. The problem is modeled below.
Establishing an energy transmission model:
in the time block, the time when the user node acquires wireless energy in the downlink is denoted as t0By LHiIndicating the link L formed by the base station and the user node i in the downlinkHiThe power gain of the downlink transmission is denoted as hHiThe energy received by the user node at the downlink time is denoted as Ei＝ζP0hHit0；
In the formula: ζ -energy conversion efficiency of the user node, generally 50% -70%; p0-a transmission power of the base station;
in the transmission phase, the total transmission energy consumption of the user node i can be expressed as
Figure BDA0002290143190000071
In the formula: pc-user node transmission procedurePower loss due to extra electronic device loss, PiFor the transmission power of user node i, tijIs a link LijDuration of uplink transmission, LijThe link formed between the node i and the node j is obtained, L is an activated link, and N is a set of all user nodes;
the total energy consumption spent by user node i in receiving information is expressed as
Figure BDA0002290143190000072
In the formula: eelecEnergy consumption required for receiving 1 bit of information, ruiFor node i via link LuiAmount of information received, LuiIs a link formed between node u and node i;
the self-sustaining operation of the user nodes over the whole time block must comply with an energy constraint, i.e. the total energy consumption is not greater than the energy taken, which can be denoted as Eri+Eti≤Ei。
Establishing an information transmission model:
link LijLink capacity C ofijC can be calculated byij＝Wlog(1+SNRij)。
In the formula:
Figure BDA0002290143190000073
w-bandwidth, eta-noise signal power, PiIs the transmission power of node i, gijIs a link LijThe power gain of (d);
the amount of information r transmitted per unit time on each linkijThe link capacity constraint, i.e. r, needs to be satisfiedij≤Cijtij,
Figure BDA0002290143190000074
By riThe data quantity generated by the user node i is represented, and the flow constraint condition required to be met by the user node i is
Figure BDA0002290143190000081
In the formula: l isijFor the link formed between node i and node j, rijFor node i via link LijAmount of information transmitted, ruiFor node i via link LuiAmount of information accepted, LuiFor the link formed between node i and node u, tijIs a link LijDuration in the uplink transmission;
the base station must satisfy the following traffic conservation constraint conditions
Figure BDA0002290143190000082
。
Under satisfaction of the energy constraint and the flow constraint, the problem can be formalized as:
min
Figure BDA0002290143190000083
s.t.Eri+Eti≤Ei(1)
Figure BDA0002290143190000084
Figure BDA0002290143190000085
Figure BDA0002290143190000086
rij≥0,Pi≥0,t0≥0,tij≥0(5)
it can be seen that the model objective function is to minimize the delay, which is subject to multiple constraints such as energy, traffic conservation, and link capacity. Wherein E isiRepresents t0Energy gained by the user node in time, EriAnd EtiThe energy consumed for the reception and transmission of data for node i, L represents the active link, r represents the traffic of the node and the line, P represents the power,the constraint (1) ensures that the energy consumed by the node for receiving and forwarding the data is not more than the obtained energy, and the constraints (2), (3) and (4) ensure that the flow constraint and all the data on each link of the user are transmitted to the base station.
S3, the original model is a nonlinear integer programming model, the solving calculation is complex, the problem is difficult to directly solve through the existing algorithm basically, the main difficult reason is that the problem contains a logarithmic function about a transmission power decision variable, and therefore a piecewise linearization algorithm is adopted to convert the problem into a solvable linear integer programming model;
s4, verifying the effectiveness of the optimization method by simulation experiments, wherein the experiments are most compared with node information transmission time delays when the user nodes directly transmit information to the base station without node cooperation, and the method is called a cooperation-free direct transmission mode (DT for short); the model is an NP (non-linear) integer programming problem difficult to mix, and can not be directly solved by the existing method. The invention fits a logarithmic function with a continuous piecewise linear function. And converting the original problem into a solvable linear integer programming model, and obtaining an approximate solution of the original problem by solving the linear integer programming model. And predicting the power of each node later by using the generated data set by adopting a deep learning method to obtain the node power, converting the original problem into a linear programming problem, and reducing the time complexity from O (2^ n) to O (n ^3.5), wherein n is the number of the nodes.
If the simulation experiment result shows that the time delay without the node cooperation is smaller than the optimized time delay provided by the invention, the step 3) is needed, different linear functions are used for fitting again, and otherwise, the step 5) is carried out.
S5, solving the model to obtain a data set of a training Deep Belief Network (DBN), wherein the data set comprises a training set and a testing set, the training of the DBN requires a large amount of data, and the solution model obtains an approximate solution for different channels.
And S6, training a DBN network by using the generated data set, wherein the DBN adopts a greedy algorithm layer by layer, firstly carrying out unsupervised pre-training to endow the whole network with a better initial weight, and then carrying out supervised tuning training.
And S7, performing node power and link selection prediction by using the trained network, eliminating redundant links and reducing model solving time delay.
Fig. 2 is a data forwarding path obtained in a node-less cooperative manner, in the method, cooperative transmission is not considered, information transmitted by the user nodes 1, 3, and 4 needs to be transmitted to the base station over a long distance, which causes very large signal attenuation, and the time delay for the user nodes 1, 3, and 4 to unload to the edge server is not optimized, so that the total time delay is increased. The main approach to reduce the delay is to reduce the node delay far from the base station, so the high delay of the user nodes 1, 3, 4 becomes the bottleneck of delay optimization.
As shown in fig. 3, in the wireless power supply network, it is assumed that there are 6 user nodes deployed in a 10 m by 10 m square area. The triangle in the center represents the base station and all red dots represent user nodes. As can be seen from fig. 2, the method based on node cooperation provided by the present invention can well balance the relationship between transmission information and energy consumption. The user nodes 0 and 2 are close to the base station, the channel quality is good, and more energy can be obtained, so that the user nodes 1 and 3 are respectively helped to forward information. The user node 4 is far away from the base station, one part of information is forwarded through the user node 5, and the other part of information is directly transmitted to the base station. In fig. 3 the information of the subscriber nodes 1, 3, 4 can be forwarded via the subscriber nodes 0, 2, 5. Therefore, the method of the invention can solve the problem of overlong time delay caused by poor quality of the node channel in direct transmission.
As shown in fig. 4 and 5, in the wireless energy supply network, in order to perform energy and information transmission in order, the invention adopts a very widely applied protocol of energy-first and transmission-second. The protocol divides time into identical time blocks, each time block comprising three parts: scheduling phase, energy transmission phase and information transmission phase. The time spent in the scheduling phase is very small compared with the time spent in the energy and information transmission phases, and the time resource allocation in the latter two phases is mainly considered. In the information transmission stage, the protocol used by the invention adopts a time division multiple access method to reduce the interference generated by the information transmission between the user nodes. Time Division Multiple Access (TDMA) for dividing information transmission time into user nodesAnd each user node shares one time slot with the same number of time slots. The access mode can effectively prevent co-channel interference. At time slot t0In 3, the wireless devices acquire energy respectively. At time slot t1The medium radio device 1 offloads its own information to the base station, time slot t2The middle wireless device 2 forwards part of the information of the middle wireless device through the wireless device 3, the rest information is directly transmitted to the base station, and the time slot t3The intermediate wireless device 3 transmits its own information to the base station together with the information that the wireless device 2 needs to forward. The base station returns the result to each wireless device after processing the information, and the result return time is ignored because the data volume of the result is usually very small.
The above-mentioned contents are only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and any modification made on the basis of the technical idea of the present invention falls within the protection scope of the claims of the present invention.

Claims (5)

1. A wireless energy supply edge calculation delay optimization method based on deep learning is characterized by comprising the following steps:
1) based on the wireless energy supply edge calculation of node cooperative transmission, adopting a cooperative transmission protocol after energy acquisition, dividing a time block into downlink energy transmission time and uplink information transmission time, establishing a downlink energy transmission model and an uplink information transmission model, and establishing a time delay model for unloading tasks to an edge server according to the two models;
establishing an uplink information transmission model based on flow conservation establishment in the step 1), wherein the specific process is as follows:
Cij＝Wlog(1+SNRij)
in the formula: cijIs a link LijThe capacity of the link of (a) is,
Figure FDA0003336546880000011
w is the bandwidth, η is the noise signal power, PiIs the transmission power of node i, gijIs a link LijThe power gain of (d);
the amount of information r transmitted per unit time on each linkijSatisfy link capacity constraints, i.e.
Figure FDA0003336546880000012
riThe data quantity generated by the user node i is represented, and the flow constraint condition satisfied by the user node i is
Figure FDA0003336546880000013
In the formula: l isijFor the link formed between node i and node j, rijFor node i via link LijAmount of information transmitted, ruiFor node i via link LuiAmount of information accepted, LuiFor the link formed between node i and node u, tijIs a link LijDuration in the uplink transmission;
the base station satisfies the following flow conservation constraint conditions
Figure FDA0003336546880000014
The time delay model for unloading the task to the edge server in the step 1) is as follows:
Figure FDA0003336546880000015
s.t.Eri+Eti≤Ei (1)
Figure FDA0003336546880000016
Figure FDA0003336546880000017
Figure FDA0003336546880000018
rij≥0,Pi≥0,t0≥0,tij≥0 (5)
wherein E isiIs t0Energy gained by the user node in time, EriEnergy consumed for receiving data for node i, EtiEnergy consumed for transmitting data for node i, L represents an activated link, r represents traffic of the node and the line, P represents power, t0Time for the user node to obtain wireless energy in downlink;
2) converting the edge server time delay model into a linear integer programming model by adopting a piecewise linearization algorithm;
3) carrying out simulation experiment to verify the effectiveness of the optimization method, comparing the effectiveness with the time delay without the cooperative direct transmission mode, and if the time delay is optimized, turning to the step 4); otherwise, go to step 2);
4) solving the linear integer programming model to obtain a data set of a training deep confidence network;
5) and training the DBN network by using the generated data set until the error rate is less than a preset value, and obtaining a trained network model.
2. The method for optimizing the computation delay of the wireless energy supply edge based on the deep learning of claim 1, further comprising:
6) and (4) performing node power and link selection prediction by using the trained network model, and removing redundant links.
3. The deep learning-based wireless energy supply edge calculation delay optimization method according to claim 1, wherein the step 1) is implemented by establishing a downlink energy transmission model based on energy conservation, and the specific process is as follows:
the energy received by the user node i at the downlink time is Ei＝ζP0hHit0；
Wherein: ζ is the energy conversion efficiency of the user node, P0Is transmission power of a base station，t0Time, h, for a user node to obtain wireless energy in downlinkHiPower gain for downlink transmission of the link;
in the transmission phase, the total transmission energy consumption of the user node i is
Figure FDA0003336546880000021
In the formula: pcFor power losses due to additional electronics losses during transmission of the user node, PiFor the transmission power of user node i, tijIs a link LijDuration of uplink transmission, LijThe link formed between the node i and the node j is obtained, L is an activated link, and N is a set of all user nodes;
the total energy consumption spent by the user node i in receiving information is
Figure FDA0003336546880000031
In the formula: eelecEnergy consumption required for receiving 1-bit information, ruiFor node i via link LuiAmount of information received, LuiIs a link formed between node u and node i;
the self-sustaining operation of the user nodes over the whole time block follows an energy constraint, i.e. the total energy consumption is not greater than the energy acquired, i.e. Eri+Eti≤Ei。
4. The method for optimizing the computation delay of the wireless energy supply edge based on the deep learning of claim 1, wherein the training method in the step 5) is as follows:
carrying out unsupervised pre-training by adopting a greedy algorithm layer by layer to give an initial weight to the whole network;
and then carrying out supervised tuning training.
5. The method for optimizing the computation delay of the wireless energy supply edge based on the deep learning of claim 1, wherein the preset value of the error rate in the step 5) is 5%.
CN201911176654.9A 2019-11-26 2019-11-26 Wireless energy supply edge calculation delay optimization method based on deep learning Active CN110941675B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911176654.9A CN110941675B (en) 2019-11-26 2019-11-26 Wireless energy supply edge calculation delay optimization method based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911176654.9A CN110941675B (en) 2019-11-26 2019-11-26 Wireless energy supply edge calculation delay optimization method based on deep learning

Publications (2)

Publication Number Publication Date
CN110941675A CN110941675A (en) 2020-03-31
CN110941675B true CN110941675B (en) 2022-04-22

Family

ID=69908886

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911176654.9A Active CN110941675B (en) 2019-11-26 2019-11-26 Wireless energy supply edge calculation delay optimization method based on deep learning

Country Status (1)

Country Link
CN (1) CN110941675B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111641681A (en) * 2020-05-11 2020-09-08 国家电网有限公司 Internet of things service unloading decision method based on edge calculation and deep reinforcement learning
CN111967179B (en) * 2020-07-02 2024-02-09 江苏能来能源互联网研究院有限公司 Dynamic optimization matching method for energy units of energy internet
CN111918339B (en) * 2020-07-17 2022-08-05 西安交通大学 AR task unloading and resource allocation method based on reinforcement learning in mobile edge network
CN113115354B (en) * 2021-03-23 2022-08-16 三峡大学 Data unloading excitation method and device based on attenuation helper selection algorithm
CN113572804B (en) * 2021-04-29 2023-06-30 重庆工程职业技术学院 Task unloading system, method and device based on edge collaboration
CN113382477B (en) * 2021-05-14 2022-09-27 北京邮电大学 Method for modeling uplink interference between wireless network users

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017160723A1 (en) * 2016-03-15 2017-09-21 Northeastern University Distributed wireless charging system and method
CN109729175A (en) * 2019-01-22 2019-05-07 中国人民解放军国防科技大学 Edge cooperative data unloading method under unstable channel condition

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017160723A1 (en) * 2016-03-15 2017-09-21 Northeastern University Distributed wireless charging system and method
CN109729175A (en) * 2019-01-22 2019-05-07 中国人民解放军国防科技大学 Edge cooperative data unloading method under unstable channel condition

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Wireless Powered Cooperation-Assisted Mobile Edge Computing;Xiaoyan Hu;《IEEE》;20180430;第17卷(第4期);全文 *
基于射频能量收集与无速率编码的协作中继网络研究;邸晓飞;《中国优秀博硕士学位论文全文数据库(博士)信息科技辑》;20170615;正文第17-37页 *

Also Published As

Publication number Publication date
CN110941675A (en) 2020-03-31

Similar Documents

Publication Publication Date Title
CN110941675B (en) Wireless energy supply edge calculation delay optimization method based on deep learning
Neely Optimal energy and delay tradeoffs for multiuser wireless downlinks
Kasgari et al. Model-free ultra reliable low latency communication (URLLC): A deep reinforcement learning framework
EP3977783A1 (en) Uplink power control using deep q-learning
Di Lorenzo et al. Joint optimization of radio resources and code partitioning in mobile edge computing
CN114389779B (en) Channel state information reporting
Ahmed et al. Optimal stochastic power control for energy harvesting systems with delay constraints
Saxena et al. Contextual multi-armed bandits for link adaptation in cellular networks
Ng et al. Energy-efficient power allocation for M2M communications with energy harvesting transmitter
Salh et al. Energy-efficient federated learning with resource allocation for green IoT edge intelligence in B5G
Chen et al. Energy‐efficiency resource allocation for cognitive heterogeneous networks with imperfect channel state information
US8649269B2 (en) Method of controlling resource usage in communication systems
Di Lorenzo et al. Joint optimization of radio resources and code partitioning in mobile cloud computing
CN113473544B (en) Network slice configuration
WO2023039905A1 (en) Ai data transmission method and apparatus, device, and storage medium
Sinaie et al. QoS‐driven resource allocation in green OFDMA wireless networks
Ding et al. Energy-efficient power control for underlaying full-duplex device-to-device communications
Yu et al. Resource allocation in RF energy harvesting‐assisted underlay D2D communication
Wu et al. Optimal energy supplementary and data transmission schedule for energy harvesting transmitter with reliable energy backup
CN106068663A (en) For managing method, wireless device, radio base station and second network node of EPS carrying
Adamopoulou et al. Enhancing cognitive radio systems with robust reasoning
Li et al. Traffic‐aware resource allocation schemes for HetNet based on CDSA
US20240154914A1 (en) Network entity and method
Benchaabene et al. Ultra reliable communication: Availability analysis in 5G cellular networks
CN108235383B (en) Distributed energy control-based multi-mobile-operator rate allocation 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