CN114980128A - Energy efficiency optimization method in multi-user cognitive edge computing network - Google Patents
Energy efficiency optimization method in multi-user cognitive edge computing network Download PDFInfo
- Publication number
- CN114980128A CN114980128A CN202210566366.XA CN202210566366A CN114980128A CN 114980128 A CN114980128 A CN 114980128A CN 202210566366 A CN202210566366 A CN 202210566366A CN 114980128 A CN114980128 A CN 114980128A
- Authority
- CN
- China
- Prior art keywords
- user
- scene
- secondary user
- calculation
- optimal
- 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.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
- H04W16/14—Spectrum sharing arrangements between different networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/382—Monitoring; Testing of propagation channels for resource allocation, admission control or handover
-
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W72/00—Local resource management
- H04W72/04—Wireless resource allocation
- H04W72/044—Wireless resource allocation based on the type of the allocated resource
- H04W72/0453—Resources in frequency domain, e.g. a carrier in FDMA
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W72/00—Local resource management
- H04W72/50—Allocation or scheduling criteria for wireless resources
- H04W72/54—Allocation or scheduling criteria for wireless resources based on quality criteria
- H04W72/541—Allocation or scheduling criteria for wireless resources based on quality criteria using the level of interference
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W72/00—Local resource management
- H04W72/50—Allocation or scheduling criteria for wireless resources
- H04W72/54—Allocation or scheduling criteria for wireless resources based on quality criteria
- H04W72/543—Allocation or scheduling criteria for wireless resources based on quality criteria based on requested quality, e.g. QoS
-
- 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)
- Quality & Reliability (AREA)
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses an energy efficiency optimization method in a multi-user cognitive edge computing network, which mainly solves the problems that the frequency spectrum of a user is difficult to access and the computing energy efficiency is poor in the existing complex scene. The method comprises the following implementation steps: 1) building a cognitive edge computing network system model, and dividing scenes according to the number of main users; 2) firstly, spectrum sensing is carried out on a single master user scene of a plurality of users and a multi-master user scene working in a plurality of different frequency bands, and then, each secondary user is coordinated to access a spectrum in a time division multiple access mode according to a sensing result; for a scene with a plurality of main users working in the same frequency band, each sub-user directly adopts time division multiple access; 3) maximizing the energy efficiency of the secondary users, and searching the optimal CPU calculation frequency, the transmission power and the access frequency spectrum time slot length; 4) and setting system working parameters according to the optimizing result to realize optimization. The invention can effectively improve the utilization rate of the wireless frequency spectrum and the calculation efficiency of the secondary user, so that the network performance is optimal, and the method can be used for an edge calculation system.
Description
Technical Field
The invention belongs to the technical field of wireless communication, and further relates to an edge computing technology, in particular to an energy efficiency optimization method in a multi-user cognitive edge computing network, which can be used for a massive user access spectrum unloading task in an edge computing system.
Background
With the rapid development of information and communication technologies, the number of wireless intelligent devices is increasing, so that the wireless traffic volume is explosively increased. Compared with the prior communication system, the capacity of the 5G communication system is improved by thousands of times, the spectrum efficiency, the energy efficiency and the average throughput are also improved by tens of times, and meanwhile, the network delay is as low as millisecond level. In addition, the 5G objective also includes improving the performance of the conventional mobile device to meet the new wireless service requirements, such as high-complexity low-latency computing services that can be implemented on the mobile device, for example: face recognition, virtual reality, automatic driving, and the like. However, due to the limited computing power and storage capacity of the mobile device, it is difficult to independently perform the above-mentioned services with high complexity and low latency.
Moving edge computing is a computing power enhancement technique. Partial or all data of nodes with weak computing power, such as mobile equipment, are unloaded to a peripheral edge computing server for execution, so that the mobile equipment is assisted to complete computing, the energy consumption of the mobile equipment is reduced, the computing time delay is reduced, and the computing efficiency is improved. However, offloading of computing tasks by mobile devices to edge computing servers in mobile edge computing networks requires spectrum resources, making it difficult for unauthorized users to allocate frequency bands to offload data due to the shortage of available spectrum resources. Therefore, a spectrum access mechanism must be used to provide spectrum access opportunities for unlicensed mobile devices in a mobile edge computing network.
Cognitive radio is a dynamic spectrum access technology with active searching and access to idle spectrum. The working principle is as follows: the cognitive radio allows an unauthorized user to access the frequency band of the authorized user when the authorized user is idle, and switches to a new idle frequency band or reduces the transmission power to continue transmission under the condition that the interference caused to the authorized user is less than the interference threshold by actively adjusting the transmission parameters when the authorized user needs to re-access the frequency band. The cognitive radio technology can greatly improve the frequency spectrum utilization rate and the network capacity.
Si P, Liang H, WuW et al, in its published paper "Joint resource management in cognitive radio and edge computing based wireless networks" (GLOBECOM 2017 and IEEE Global Communications conference. IEEE,2017:1-6), propose a three-layer structure of cognitive radio-mobile edge computing framework, and use cognitive radio to find available radio spectrum of mobile devices. Li X, Fan R, Hu H et al studied a cognitive radio-Mobile Edge Computing system in its published article "Energy-efficiency allocation for Mobile Edge Computing by Multiple Relays" (IEEE Internet of Things Journal, 2021: 101-. Both the two systems consider a single-user single-master-user scene, however, in a real scene, a large number of scenes are often faced, so that the method is difficult to solve the practical problem.
Disclosure of Invention
The invention aims to provide an energy efficiency optimization method in a multi-user cognitive edge computing network aiming at the defects of the prior art. The optimal CPU calculation frequency, the sending power and the time slot length of an access frequency spectrum of each secondary user are searched by maximizing the energy efficiency of the secondary users under two scenes of a multi-user single primary user and a multi-user multi-primary user; therefore, the problems that the frequency spectrum of the secondary user is difficult to access and the calculation efficiency is poor in a real complex scene are solved, the utilization rate of the wireless frequency spectrum and the calculation efficiency of the secondary user are effectively improved, and the performance of the multi-user cognitive edge calculation network is optimal.
The basic idea for realizing the invention is as follows: under the scene of multiple users and a single master user, the cognitive secondary base station firstly carries out spectrum sensing on the special spectrum state of the master user and coordinates each secondary user to access the spectrum in a time division multiple access mode according to the sensing result; under the multi-user multi-primary-user scene, if the primary users work in a plurality of different frequency bands, the scene is equivalent to the scene of a single primary user, and if the primary users all work in the same frequency band, the spectrum sensing cannot be carried out on each primary user, so that the spectrum access is directly carried out on each secondary user in a time division multiple access mode, and the calculation task is unloaded from the secondary users to the cognitive secondary base station for auxiliary edge calculation.
In order to achieve the purpose, the technical scheme of the invention comprises the following steps:
(1) building a cognitive small cellular network consisting of M main users, K sub-users and a cognitive sub-base station carrying an edge computing server; where K is greater than 1, let Ω M 1,2, M, and Ω K The primary user set and the secondary user set are respectively represented by {1,2, ·, K,. and K }, and m and K respectively represent the mth primary user and the kth secondary user;
(2) judging the number of main users in the cognitive small cellular network:
if the number M of the master users is 1, determining that the multiple-user single-master-user scene is obtained, and continuing to execute the step (3);
if the number M of the main users is more than 1, judging that the multi-user multi-main-user scene is a multi-user multi-main-user scene, and directly executing the step (4);
(3) obtaining the optimal optimization variable under the scene of a plurality of times of users and a single master user:
(3.1) the cognitive secondary base station carries out spectrum sensing on a wireless channel environment, obtains a sensing channel state j belonging to {0,1} of a primary user special spectrum, and records that an actual channel state is i belonging to {0,1 }; wherein 0 represents that a main user channel is idle, and 1 represents that the main user channel is occupied;
(3.2) each secondary user accesses the primary user frequency spectrum by adopting a time division multiple access mechanism, and the k-th secondary user is assumed to have one Q k A bit calculation task, wherein one part of the calculation task is transmitted to the cognitive secondary base station for unloading calculation, and the other part of the calculation task is subjected to local calculation; obtaining the average calculation bit number R of the kth secondary user ave,k Average energy consumption E ave,k And average interference I caused to primary users ave,k (ii) a And R is ave,k ≥Q k ,I ave,k Less than or equal to interference margin I of primary user th ;
(3.3) obtaining the first scene task calculation efficiency E according to the following formula CE :
wherein the optimization variable is the transmission power of the first scene secondary userLocal computation of CPU frequency for first scene secondary userFirst scene secondary userTime length of access to spectrumP k,j 、f k,j And beta k,j Respectively representing the first scene transmission power, the local calculation CPU frequency and the time length of accessing the frequency spectrum of the kth secondary user under the condition that the sensing channel state is j;
setting the first scene secondary user to meet the following constraint conditions:
wherein e k Energy available for the kth secondary user is represented, T represents the time length of a time slot, and gamma is an energy consumption factor;β 0 representing the spectrum sensing time length of the secondary user; f is not less than 0 k,j ≤f k,max ,0≤P k,j ≤P k,max Wherein f is k,max And P k,max Maximum available CPU frequency and maximum transmission power of the kth secondary user respectively;
(3.5) maximizing first scenario task computational efficiency E CE Through variable replacement, Dinkelbach algorithm willConverting the expression into a convex expression, and then solving the convex expression by using a Lagrange dual decomposition algorithm and a sub-gradient algorithm to obtain the calculation efficiency of the optimal task of the first sceneThe corresponding first scene optimal optimization variable, i.e. the optimal transmission power P of the first scene secondary user * Optimal local computation of CPU frequency f * And the time length beta of the optimal access spectrum * ;
(3.6) setting system working parameters according to the first scene optimal optimization variables, and entering the step (5);
(4) obtaining optimal optimization variables under a multi-user multi-main-user scene:
(4.1) judging whether the master users all work in the same frequency band, if so, executing the step (4.2), otherwise, executing the step (3) after the situation is identical to a scene of a single master user of multiple users;
(4.2) each secondary user directly accesses the primary user frequency spectrum by adopting a time division multiple access mechanism, and the k-th secondary user is assumed to have one secondary user with Q k A bit calculation task, wherein one part of the calculation task is transmitted to the cognitive secondary base station for unloading calculation, and the other part of the calculation task is subjected to local calculation; obtaining the calculated bit number R of the kth secondary user k Energy consumption E k And average interference I caused to mth primary user k,m (ii) a And R is k ≥Q k ,I k,m Interference tolerance I less than or equal to mth primary user th,m ;
(4.3) obtaining the calculation efficiency E of the second scene task according to the following formula CE1 :
wherein the optimization variable is the transmission power of the second scene secondary userSecond scenario secondary user's local computation CPU frequencyTime length of accessing frequency spectrum by secondary user in second sceneWherein, P k 、f k And beta k Respectively representing the second scene transmitting power, the local computing CPU frequency and the time length of accessing the frequency spectrum of the kth secondary user;
setting a second scene secondary user to meet the following constraint conditions:
wherein e k Energy available for the kth secondary user is represented, T represents the time length of a time slot, and gamma is an energy consumption factor;0≤f k ≤f k,max ,0≤P k ≤P k,max wherein f is k,max And P k,max The maximum available CPU frequency and the maximum transmitting power of the kth secondary user are respectively;
(4.5) maximizing second scenario task computational efficiency E CE1 Through variable replacement, Dinkelbach algorithm willConverting the expression into a convex expression, and then solving the convex expression by using a Lagrange dual decomposition algorithm and a sub-gradient algorithm to obtain the computing efficiency of the optimal task of the second sceneCorresponding second scenario optimal optimization variable, i.e. optimal transmit power of second scenario sub-userOptimal local computation CPU frequencyAnd time length of optimal access spectrum
(4.6) setting system working parameters according to the second scene optimal optimization variables, and entering the step (5);
(5) the system operates under the selected working parameters, and the energy efficiency optimization of the system is realized.
Compared with the prior art, the invention has the following advantages:
firstly, because the spectrum is dynamically accessed in a cognitive radio mode, spectrum access opportunities are provided for secondary users by enabling the secondary users to share the spectrum with a primary user, so that the secondary users can unload and transmit computing tasks to an edge server with stronger computing power, and the problem of limited computing power of the secondary users is solved;
secondly, a network system in a real scene usually faces a mass of user scenes, while the existing cognitive edge computing system mostly considers a single-user scene, so that the practical problem is difficult to solve; compared with the prior art, the method considers a multi-user scene on the basis of introducing cognitive radio, and respectively adopts different technical means to complete the calculation task aiming at the scene of a single main user of multiple users and multiple main users of multiple users, namely the situation that the multiple sub-users need to share the frequency spectrum with the main user, thereby optimizing the system performance under the multi-user scene and effectively solving the problems of low frequency spectrum utilization rate and poor network performance under the actual complex scene.
Drawings
FIG. 1 is a schematic view of an application scenario of the method of the present invention;
fig. 2 is a schematic diagram of a timeslot structure of the present invention, where a is a schematic diagram of a timeslot structure in a multi-user single-master user scenario, and b is a schematic diagram of a timeslot structure in a multi-user multi-master user scenario;
FIG. 3 is a flow chart of an implementation of the method of the present invention;
FIG. 4 is a Dinkelbach algorithm convergence performance curve diagram in the method of the present invention;
FIG. 5 shows the channel attenuation index α and the minimum calculated bit number Q of the secondary user in the multi-user single-primary-user scenario in the method of the present invention k Impact on task computational efficiencyA situation simulation result graph;
FIG. 6 shows the minimum number of bits Q calculated for the primary user M and the secondary user in the multi-user multi-primary-user scenario in the method of the present invention k A simulation result diagram of the influence condition on the task calculation efficiency;
fig. 7 is a simulation result diagram of the influence of the time slot length T and the channel attenuation index α on the task calculation efficiency in the scenario of multiple users and multiple primary users in the method of the present invention.
Detailed Description
The implementation process of the technical scheme of the invention is described in detail below with reference to the attached drawings:
referring to fig. 1, a schematic diagram of a system model in the invention is shown, the system of the invention forms a cognitive small cellular network by M primary users, K secondary users and a cognitive small base station carrying an edge server, wherein K is greater than 1, and Ω is set to M 1,2, a., M, and Ω K The primary user set and the secondary user set are denoted by {1, 2., K }, respectively, where m and K denote the mth primary user and the kt secondary user, respectively.
Referring to fig. 2, the time slot structure diagram of the present invention considers two scenarios together: single master user (M ═ 1) and multiple master user (M > 1) scenarios.
In the scene that M is 1, the cognitive small base station can sense the frequency spectrum of the master user state, coordinates each user according to the sensing result and adopts a time division multiple access mode to access the frequency spectrum to upload the task to be calculated to the cognitive small base station for calculation. The system runs by taking time slot as unit, and the total length of the time slot is T, beta 0 Representing the length of the spectrum sensing time, beta k,j ,k∈Ω K J belongs to {0,1} and represents the time length of accessing the frequency spectrum by the kth secondary user when the sensing result is j, wherein j equals 0 to indicate that the sensing result is that the primary user is in an idle state, and j equals 1 to indicate that the sensing result is that the primary user is in an occupied state. In order to ensure the transmission performance of the master user, the fixed detection probability P is adopted in the embodiment d The energy detection mechanism (the probability that the main user can correctly detect when the main user is in an occupied state) has the following corresponding false alarm probability (the probability that the main user is in an idle state but the detection result is occupied):
wherein k represents the signal-to-noise ratio of PU signals received by the cognitive small base station, f s Representing the sampling rate, Q (-) is a complementary cumulative distribution function of a standard gaussian distribution variable.
In a scene where M is greater than 1, if a plurality of primary users all work in the same frequency band, spectrum sensing cannot be performed on each primary user, and if a plurality of primary users work in a plurality of different frequency bands, the problem is equivalent to the scene of a single primary user. Therefore, the invention considers the scene that a plurality of main users work in the same frequency band. In this scenario, the spectrum sensing technology is not applicable, each secondary user still performs spectrum access in a time division multiple access manner, but the interference power constraint of each primary user must be satisfied, and the total length of the time slot is T, β k ,k∈Ω K Indicating the access spectrum duration of the kth secondary user.
The first embodiment is as follows:
referring to fig. 3, the invention provides an energy efficiency optimization method in a multi-user cognitive edge computing network, which comprises the following specific implementation steps:
step 1: building a cognitive small cellular network consisting of M main users, K sub-users and a cognitive sub-base station carrying an edge computing server; where K is greater than 1, let Ω M 1,2, a., M, and Ω K The primary user set and the secondary user set are respectively represented by {1,2, ·, K,. and K }, and m and K respectively represent the mth primary user and the kth secondary user;
and 2, step: judging the number of main users in the cognitive small cellular network:
if the number M of the master users is 1, determining that the multiple-user single-master-user scene is obtained, and continuing to execute the step (3);
if the number M of the main users is more than 1, judging that the multi-user multi-main-user scene is a multi-user multi-main-user scene, and directly executing the step (4);
and step 3: obtaining the optimal optimization variable under the scene of a plurality of times of users and a single master user:
(3.1) cognitive Secondary base station PairSensing a frequency spectrum in a wireless channel environment, acquiring a sensing channel state j belonging to {0,1} of a master user-specific frequency spectrum, and recording an actual channel state as i belonging to {0,1 }; wherein 0 indicates that the primary user channel is idle and 1 indicates that the primary user channel is occupied. Here, the cognitive secondary base station performs spectrum sensing on the wireless channel environment, and the embodiment adopts a fixed detection probability of P d The corresponding false alarm probability of the energy detection mechanism is as follows:
wherein, k represents the signal-to-noise ratio of PU signals received by the cognitive small base station, f s Representing the sampling rate, Q (-) is a complementary cumulative distribution function of a standard gaussian distribution variable.
(3.2) each secondary user accesses the primary user frequency spectrum by adopting a time division multiple access mechanism, and the k-th secondary user is assumed to have one Q k A bit calculation task, wherein one part of the calculation task is transmitted to the cognitive secondary base station for unloading calculation, and the other part of the calculation task is subjected to local calculation; obtaining the average calculation bit number R of the kth secondary user ave,k Average energy consumption E ave,k And average interference I caused to primary users ave,k (ii) a And R is ave,k ≥Q k ,I ave,k Less than or equal to interference margin I of primary user th ;
Average number of bits R calculated for the k-th secondary user ave,k Average energy consumption E ave,k And average interference I caused to primary users ave,k Respectively according to the following calculation:
wherein the content of the first and second substances, respectively representing the calculated bit number, the energy consumption and the interference to the master user of the kth secondary user when the real state of the master user is i and the sensing result is j, B representing the access frequency spectrum bandwidth, C k Denotes CPU clock cycle, h k,C Represents the channel power gain, h, from the k-th secondary user to the cognitive secondary base station k,R Representing the channel power gain, δ, between the k-th secondary user and the primary user k The interference, sigma, caused by the signals sent by the master user when the kth secondary user carries out task unloading when the master user is in an occupied state k Representing the local noise, alpha, of the second kth secondary user in transmitting and receiving signals i,j The probability of occurrence when the actual channel state of the primary user is i and the perceived channel state is j is represented as follows:
α 0,0 =Pr 0 [1-P f (β 0 )],
α 0,1 =Pr 0 P f (β 0 ),
α 1,0 =Pr 1 (1-P d ),
α 1,1 =Pr 1 P d ,
wherein Pr 0 And Pr 1 Indicating the prior probability that the primary user channel is in idle and occupied states.
(3.3) obtaining the first scene task calculation efficiency E according to the following formula CE :
wherein the optimization variable is the transmission power of the first scene secondary userLocal computation of CPU frequency for first scene secondary userTime length of accessing frequency spectrum by secondary user in first sceneP k,j 、f k,j And beta k,j Respectively representing the first scene transmission power, the local computing CPU frequency and the time length of accessing the frequency spectrum of the kth secondary user under the condition that the perceived channel state is j;
setting the first scene secondary user to meet the following constraint conditions:
wherein e k Energy available for the kth secondary user is represented, T represents the time length of a time slot, and gamma is an energy consumption factor;β 0 representing the spectrum sensing time length of the secondary user; f is not less than 0 k,j ≤f k,max ,0≤P k,j ≤P k,max Wherein f is k,max And P k,max Maximum available CPU frequency and maximum transmission power of the kth secondary user respectively;
(3.5) maximizing first scenario task computational efficiency E CE Through variable replacement and Dinkelbach calculationMethod for makingConverting the expression into a convex expression, and then solving the convex expression by using a Lagrange dual decomposition algorithm and a sub-gradient algorithm to obtain the calculation efficiency of the optimal task of the first sceneThe corresponding first scene optimal optimization variable, i.e. the optimal transmission power P of the first scene secondary user * Optimal local computation of CPU frequency f * And the time length beta of the optimal access spectrum * ;
(3.6) setting system working parameters according to the first scene optimal optimization variables, and entering the step (5);
and 4, step 4: obtaining optimal optimization variables under a multi-user multi-main-user scene:
(4.1) judging whether the master users all work in the same frequency band, if so, executing the step (4.2), otherwise, the same as a multi-user single master user scene, and returning to the step (3);
(4.2) each secondary user directly accesses the primary user frequency spectrum by adopting a time division multiple access mechanism, and the k-th secondary user is assumed to have one secondary user with Q k A bit calculation task, wherein one part of the calculation task is transmitted to the cognitive secondary base station for unloading calculation, and the other part of the calculation task is subjected to local calculation; obtaining the calculation bit number R of the kth secondary user k Energy consumption E k And average interference I caused to mth primary user k,m (ii) a And R is k ≥Q k ,I k,m Interference tolerance I less than or equal to mth primary user th,m ;
Number of bits R calculated for the kth secondary user k Energy consumption E k Average interference I caused to the m-th primary user k,m Respectively calculated according to the following:
I k,m =P k h k,m β k ,m∈Ω M ,k∈Ω K
wherein h is k,m Representing the channel power gain between the kth secondary user and the mth primary user, B representing the access spectrum bandwidth, C k Representing CPU clock cycles, δ k The interference, sigma, caused by the signals sent by the master user when the kth secondary user carries out task unloading when the master user is in an occupied state k Representing the local noise when the kth secondary user is transmitting and receiving signals.
(4.3) obtaining the calculation efficiency E of the second scene task according to the following formula CE1 :
wherein the optimization variable is the transmission power of the second scene secondary userSecond scenario secondary user's local computation CPU frequencyTime length of second scenario secondary user accessing frequency spectrumWherein, P k 、f k And beta k Are respectively provided withRepresenting the sending power of a second scene of the kth secondary user, the local computing CPU frequency and the time length of accessing the frequency spectrum;
setting a second scene secondary user to meet the following constraint conditions:
wherein e k Energy available for the kth secondary user is represented, T represents the time length of a time slot, and gamma is an energy consumption factor;0≤f k ≤f k,max ,0≤P k ≤P k,max wherein f is k,max And P k,max The maximum available CPU frequency and the maximum transmitting power of the kth secondary user are respectively;
(4.5) maximizing second scenario task computational efficiency E CE1 Through variable replacement, Dinkelbach algorithm willConverting the expression into a convex expression, and then solving the convex expression by using a Lagrange dual decomposition algorithm and a sub-gradient algorithm to obtain the computing efficiency of the optimal task of the second sceneCorresponding second scenario optimal optimization variable, namely, the optimal transmission power of the second scenario secondary userOptimal local computation CPU frequencyAnd time length of optimal access spectrum
(4.6) setting system working parameters according to the second scene optimal optimization variables, and entering the step (5);
and 5: the system operates under the selected working parameters, and the energy efficiency optimization of the system is realized.
Example two:
the energy efficiency optimization method in the multi-user cognitive edge computing network provided by the embodiment has the same overall implementation steps as those in the first embodiment, and further description is made for a multi-user single-master user scene:
step a: building a cognitive small cellular network consisting of 1 master user, K secondary users and a cognitive small base station carrying an edge computing server to enable omega to be omega K K represents a secondary user set;
step b: and the cognitive secondary base station performs spectrum sensing on the wireless channel environment to obtain the channel state of the primary user special spectrum. However, under the influence of noise and various external interferences, spectrum sensing cannot obtain a perfect spectrum sensing result, i.e., a sensing error exists. The actual channel state is assumed to be i belongs to {0,1}, and the perceived channel state is j belongs to {0,1}, wherein 0 represents that a primary user channel is idle, and 1 represents that the primary user channel is occupied;
step c: each secondary user accesses the primary user frequency spectrum by adopting a time division multiple access mechanism, and the k-th secondary user is assumed to have a Q k A bit calculation task, wherein one part of the calculation task is transmitted to the cognitive secondary base station for unloading calculation, and the other part of the calculation task is subjected to local calculation; namely, a partial unloading mechanism is adopted: and carrying out local calculation on part of the services, and unloading part of the services to the cognitive small base station for calculation.
Order toi, j belongs to {0,1} and represents the task calculation bit number of the kth secondary user when the real state of the primary user is i and the sensing result is j, and the specific expression is as follows:
wherein B denotes the access spectrum bandwidth, C k Indicating the number of CPU cycles that need to be spent computing a 1-bit task, f k,j ,P k,j Respectively representing the local calculation CPU frequency and the transmission power of the kth secondary user under the condition that the sensing result is j, h k,C Represents the power gain, delta, from the kth secondary user to the cognitive small base station k The interference, sigma, caused by the signals sent by the master user when the kth secondary user carries out task unloading when the master user is in an occupied state k Which represents the local noise when the kth secondary user is transmitting and receiving signals the next time. Order toi, j belongs to {0,1} and represents the energy consumption of the kth secondary user when the real state of the primary user is that the sensing result of i is j, and the specific steps are as follows:
wherein gamma is an energy consumption factor. Due to imperfect sensing result, the transmission of the secondary user may affect the primary user, so thati, j belongs to {0,1} and represents that the real state of the primary user is i, and the transmission of the kth secondary user causes the primary user when the sensing result is jSpecifically, the following interference is:
wherein h is k,R Representing the channel power gain between the k-th secondary user and the primary user receiver. According to the false alarm probability and the detection probability, the probability of various combinations of the real state of the main user and the sensing result can be given, and alpha is made i,j And i, j is epsilon {0,1} represents the probability that the sensing result is j in the case that the real state of the main user is i, and the probability is specifically as follows:
α 0,0 =Pr 0 [1-P f (β 0 )]
α 0,1 =Pr 0 P f (β 0 )
α 1,0 =Pr 1 (1-P d )
α 1,1 =Pr 1 P d
wherein Pr 0 And Pr 1 The prior probability of the main user in the idle and occupied states is shown and can be obtained through long-term observation of the main user. From the probabilities, the average number of calculated bits R obtained by the k-th user can be known ave,k Average energy consumption E ave,k And average interference I caused to primary users ave,k (ii) a And R is ave,k ≥Q k ,I ave,k Less than or equal to interference margin I of primary user th ;
Step d: and solving to obtain the task calculation efficiency of the secondary user according to the following formula:
step e: the optimization problem of the scene of a single master user of multiple users is established as follows:
wherein the optimization variables areThe secondary user's energy causality constraint isj∈{0,1},k∈Ω K ,e k Energy available to the kth secondary user; interference constraint of primary user is I ave,k ≤I th ,k∈Ω K ,I th Interference tolerance for primary users; the transmission time length of the secondary user is constrained toj∈{0,1},k∈Ω K (ii) a The power and transmit power constraints of the secondary users are f k,j ≥0,P k,j ≥0,f k,j ≤f k,max ,P k,j ≤P k,max ,j∈{0,1},k∈Ω K ,f k,max And P k,max The maximum available CPU frequency and the maximum transmitting power of the kth secondary user are respectively; the minimum average number of computed bits for a secondary user is constrained to R ave,k ≥Q k ,k∈Ω K 。
Step f: and solving the formula <1.1> by using variable replacement, a Dinkelbach algorithm, a Lagrange dual decomposition algorithm and a sub-gradient algorithm to obtain the optimal parameters.
(6.1) formula since the objective function is in fractional form and there is a coupling of the constraint variables<1.1>Is a non-convex problem. To solve it, auxiliary variables are introducedSo that q is k,0 =P k,0 β k,0 And q is k,1 =P k,1 β k,1 Thus eliminating the variable coupling, the objective function translates into:
wherein:
(6.2) molecules due to the objective functionIs a concave function, denominatorIs a convex function, and therefore, the target function is still in a non-convex form. Based on the method, the Dinkelbach algorithm is adopted to solve the problem. Introducing a parameter eta > 0 and specifying an initial value eta ═ eta 0 Updating the constraints, and applying the formula<1>The following steps are changed:
whereinConstrainingj∈{0,1},k∈Ω K Become intoj∈{0,1},k∈Ω K Constraint of alpha 10 P k,0 h k,R β k,0 +α 11 P k,1 h k,R β k,1 ≤I th ,k∈Ω K Become alpha 10 h k,R q k,0 +α 11 h k,R q k,1 ≤I th ,k∈Ω K Restraint f k,j ≥0,P k,j ≥0,f k,j ≤f k,max ,P k,j ≤P k,max , j∈{0,1},k∈Ω K Become f k,j ≥0,f k,j ≤f k,max ,q k,j ≥0,q k,j ≤P k,max β k,j ,j∈{0,1},k∈Ω K About R ave,k ≥Q k ,k∈Ω K Become intok∈Ω K 。
(6.3) the formula <1.2> is a convex problem, and the Lagrangian dual method is adopted to solve the convex problem.
(6.3.1) the Lagrangian function of equation <1.2> is:
The dual function of equation <1.2> is:
(6.3.2) decomposing the formula <1.3> into the following sub-formulae by using a dual decomposition method, specifically as follows:
for variable f k,0 The corresponding sub-formula is:
wherein the constraint is 0 ≦ f k,0 ≤f k,max ,θ 1,k =η(α 0,0 +α 1,0 )(T-β 0 )γ+λ k,0 (T-β 0 ) And gamma. Formula (II)
<1.3.1> is a convex problem, and the optimal solution obtained by using the KKT condition is as follows:
for variable f k,1 The corresponding sub-formula is:
wherein the constraint is 0 ≦ f k,1 ≤f k,max ,θ 2,k =η(α 0,1 +α 1,1 )(T-β 0 )γ+λ k,1 (T-β 0 ) And gamma. Formula (II)
<1.3.2> is a convex problem, the optimal solution of which can be obtained by using the KKT condition is as follows:
for the variable beta k,0 And q is k,0 The corresponding sub-formula is:
wherein the constraint is q k,0 ≥0,β k,0 ≥0,θ 3,k =η(α 0,0 +α 1,0 )+λ k,0 +μ k α 10 h k,R +ρ k,0 . Formula (II)<1.3.3>For convex problems, the solution can be solved by using a block iterative algorithm, and beta is initialized firstly k,0 Is a feasible solution, then the formula<3.3>The following steps are changed:
wherein the constraint is q k,0 Is more than or equal to 0. By KKT condition at a given beta k,0 Under the condition of (a) q k,0 The optimal solution is as follows:
wherein the constraint is beta k,0 ≥0。
Let Ψ (. beta.) k,0 ) Expression formula<1.3.3.2>The objective function of (a), namely:
the first derivative and the second derivative are respectively:
from the form of its first derivative, the very difficult to obtain closed-form solution β k,0 . From its first and second derivativesAnd is
Based on this, the present embodiment adopts the dichotomy to obtain β k,0 The optimal solution of (a).
For the variable beta k,1 And q is k,1 The corresponding sub-formula is:
wherein the constraint is beta k,1 ≥0,q k,1 ≥0,θ 4,k =[η(α 0,1 +α 1,1 )+λ k,1 +μ k α 11 h k,R +ρ k,1 ]. This sub-formula may be fully in the form of<1.3.3>The solution of (a) is solved.
(6.3.3) after each sub-formula is solved, the dual problem of the formula <1.3> is solved, wherein the dual problem is as follows:
wherein the constraint isSince the Slater condition is satisfied, the formula<1.4>Optimum value and formula<1.3>The optimum values are the same. Can adopt a secondary gradient method to pair formulas<1.4>Solving, optimizing dual variable, and aiming at dual variable lambda k,0 、λ k,1 、 μ k 、ο 0 、ο 1 、ρ k,0 、ρ k,1 And upsilon k One possible sub-gradient vector is as follows:
(6.3.4) under the condition of eta, updating eta values after the formula <1.2> is solved:
(6.4) repeating the steps (6.1) - (6.3) until the algorithm converges to obtain the optimal parameter P * 、f * And beta * 。
Step g: and the system selects working parameters according to the optimal optimization variables, so that the system performance is optimal.
Example three: the energy efficiency optimization method in the multi-user cognitive edge computing network provided by this embodiment integrally implements the steps as in the first embodiment, and further describes a situation where multiple users and multiple primary users operate in the same frequency band:
step A: building a cognitive small cellular network consisting of M & gt 1 main users, K & gt 1 secondary users and a cognitive small base station carrying an edge computing server to enable omega to be omega M 1,2, M, and Ω K K represents a primary user set and a secondary user set;
and B: each secondary user directly accesses the frequency spectrum of the primary user by adopting a time division multiple access mechanism, and the calculated bit number, energy consumption and average interference caused to the mth primary user, which can be obtained by the kth secondary user, are respectively as follows:
I k,m =p k h k,m β k ,m∈Ω M ,k∈Ω K ,
wherein h is k,m And the channel power gain of the kth secondary user and the mth primary user is obtained.
And C: from R k And E k Obtain task computing efficiency E CE1 :
Step D: establishing an optimization problem of a multi-user multi-main-user scene:
wherein the optimization variables areThe causal constraint on energy of the secondary user isk∈Ω K (ii) a Interference constraint of primary user as I k,m ≤I th,m ,k∈Ω K ,m∈Ω M ,I th,m Interference tolerance for the mth primary user; the transmission time length of the secondary user is constrained tok∈Ω K (ii) a The power and transmit power constraints of the secondary user are f k ≥0,f k ≤f k,max ,P k ≥0,P k ≤P k,max ,k∈Ω K (ii) a The minimum average calculated bit number of the secondary user is constrained to R k ≥Q k ,k∈Ω K 。
Step E: similar to a multi-user single-master user scene, solving a formula <2.1> by using a variable replacement, a Dinkelbach algorithm, a Lagrangian dual decomposition algorithm and a sub-gradient algorithm to obtain an optimal parameter.
(5.1) introduction of the variable u k Andsimultaneously, the Dinkelbach algorithm is utilized, and the target function is changed into the target functionWherein
whereinConstrainingk∈Ω K Become tok∈Ω K Constraint p k h k,m β k ≤I th,m ,k∈Ω K ,m∈Ω M Become intok∈Ω,r∈Ω R Restraint f k ≥0,f k ≤f k,max , P k ≥0,P k ≤P k,max ,k∈Ω K Become f k ≥0,f k ≤f k,max ,u k ≥0,u k ≤P k,max β k ,k∈Ω K Constraint of R k ≥Q k ,k∈Ω K Become intok∈Ω K 。
(5.3) formula<2.2>Solving process and formula for convex problem<1.2>Similarly, the formula can be directly adopted<1.2>The Lagrange dual method is used for solving and searching the optimal optimization variable
Step F: and the system selects working parameters according to the optimal optimization variables, so that the system performance is optimal.
The effect of the present invention is further explained by combining simulation experiments as follows:
A. simulation conditions
Simulation was performed using computer simulation software unless specified otherwiseThe channel power gain setting situation adopted by the invention is as follows: h is x,y =|u x,y | 2 ,x∈Ω K ,y∈{C,R}∪Ω R WhereinL 0 =-30dB,d x,y Is the distance between x and y, α x,y Is the channel fading index. Unless otherwise specified, the simulation parameters used in this section are as follows: when M is 1, d k,R =6m,k∈Ω K ,α k,C =α=3,k∈Ω K ,α k,R =3,k∈Ω K ,Q k =10 4 bits, k∈Ω K . When M is greater than 1, without loss of generality, this section assumes that a plurality of primary users are converged in a dense area and have the same position, i.e. d k,z =6m,k∈Ω K ,z∈{R}∪Ω R ,Q k =10 3 bits,k∈Ω K ,α k,C =α=3, k∈Ω K ,α k,R =3,k∈Ω K And M is 5. The remaining simulation parameters are summarized in table 1.
TABLE 1 simulation parameters
B. Emulated content
Simulation 1: the Dinkelbach algorithm adopted by the invention has a convergence performance curve, and the simulation result is shown in FIG. 4;
simulation 2: channel attenuation index alpha and minimum calculation bit number Q of secondary user in multi-user single-master user scene k The influence on EE, and the simulation result are shown in FIG. 5;
simulation 3: main user number M and minimum calculation bit number Q of sub-users under multi-user and multi-main-user scene k The simulation result of the influence on the task calculation efficiency is shown in fig. 6;
and (4) simulation: the influence of the time slot length T and the channel attenuation index α on the task calculation efficiency in the multi-user multi-master user scenario is shown in fig. 7;
C. simulation result
As can be seen from fig. 4, the Dinkelbach algorithm adopted in the present invention has excellent convergence performance, and can achieve convergence within a small number of iterations, so that the algorithm complexity is low. And it can also be seen from fig. 3 that there is no direct relationship between the number of master users and the convergence speed of the algorithm.
As can be seen from fig. 5, the larger the channel attenuation index α is, the worse the link quality between the secondary user and the cognitive small base station is, and in order to meet the minimum task amount requirement, the secondary user needs to spend more energy, thereby resulting in the reduction of task calculation efficiency. Likewise, the minimum number of calculated bits Q k The larger the secondary user is, the more energy the secondary user needs to consume to increase the amount of the calculation task, so that the task calculation efficiency is reduced.
As can be seen from fig. 6, the larger the number M of primary users is, the larger the interference of the primary users to each secondary user is, and meanwhile, the more the number of primary users is, the more the number of primary user interference power constraints that the secondary users need to satisfy is, the larger the influence on the EE of the secondary user is. The minimum number of bits Q calculated is similar to a single master user scene k Increasing, each user needs to consume more energy, resulting in a decrease in task computing efficiency.
As can be seen from fig. 7, the longer the slot length T, the longer the computation and task offloading time that can be used by each user. The energy consumption of the secondary user and the calculation task amount are in a nonlinear relation, and the reduction of the CPU frequency and the unloading power of the secondary user can finally reduce the task calculation efficiency. The increase of the channel attenuation index alpha in a multi-primary user scene causes the quality of a channel between each secondary user and the known small base station to be reduced, and the task calculation efficiency is reduced.
The simulation analysis proves the correctness and the effectiveness of the method provided by the invention.
The invention has not been described in detail in part of the common general knowledge of those skilled in the art.
While the invention has been particularly shown and described with reference to a preferred embodiment, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention.
Claims (4)
1. A method for optimizing energy efficiency in a multi-user cognitive edge computing network comprises the following steps:
(1) building a cognitive small cellular network consisting of M main users, K sub-users and a cognitive sub-base station carrying an edge computing server; where K is greater than 1, let Ω M 1,2, a., M, and Ω K The primary user set and the secondary user set are respectively represented by {1,2, ·, K }, and m and K respectively represent the mth primary user and the kt secondary user;
(2) judging the number of main users in the cognitive small cellular network:
if the number M of the master users is 1, determining that the multiple-user single-master-user scene is obtained, and continuing to execute the step (3);
if the number M of the main users is more than 1, judging that the multi-user multi-main-user scene is a multi-user multi-main-user scene, and directly executing the step (4);
(3) obtaining the optimal optimization variable under the scene of a plurality of times of users and a single master user:
(3.1) the cognitive secondary base station carries out spectrum sensing on a wireless channel environment, obtains a sensing channel state j belonging to {0,1} of a primary user special spectrum, and records that an actual channel state is i belonging to {0,1 }; wherein 0 represents that a main user channel is idle, and 1 represents that the main user channel is occupied;
(3.2) each secondary user accesses the primary user frequency spectrum by adopting a time division multiple access mechanism, and the k-th secondary user is assumed to have one Q k A bit calculation task, wherein one part of the calculation task is transmitted to the cognitive secondary base station for unloading calculation, and the other part of the calculation task is subjected to local calculation; obtaining the average calculation bit number R of the kth secondary user ave,k Average energy consumption E ave,k And average interference I caused to primary users ave,k (ii) a And R is ave,k ≥Q k ,I ave,k Less than or equal to interference margin I of primary user th ;
(3.3) obtaining the first scene task calculation efficiency E according to the following formula CE :
wherein the optimization variable is the transmission power of the first scene secondary userLocal computation of CPU frequency for first scene secondary usersTime length of accessing frequency spectrum by secondary user in first sceneP k,j 、f k,j And beta k,j Respectively representing the first scene transmission power, the local computing CPU frequency and the time length of accessing the frequency spectrum of the kth secondary user under the condition that the perceived channel state is j;
setting the first scene secondary user to meet the following constraint conditions:
wherein e k Energy available for the kth secondary user is represented, T represents the time length of a time slot, and gamma is an energy consumption factor;β 0 representing spectrum of secondary usersA sensing time length; f is not less than 0 k,j ≤f k,max ,0≤P k,j ≤P k,max Wherein f is k,max And P k,max The maximum available CPU frequency and the maximum transmitting power of the kth secondary user are respectively;
(3.5) maximizing first scenario task computational efficiency E CE Through variable replacement, Dinkelbach algorithm willConverting the expression into a convex expression, and then solving the convex expression by using a Lagrange dual decomposition algorithm and a sub-gradient algorithm to obtain the calculation efficiency of the optimal task of the first sceneThe corresponding first scene optimal optimization variable, i.e. the optimal transmission power P of the first scene secondary user * Optimal local computation of CPU frequency f * And a time length beta of an optimal access spectrum * ;
(3.6) setting system working parameters according to the first scene optimal optimization variables, and entering the step (5);
(4) obtaining optimal optimization variables under a multi-user multi-main-user scene:
(4.1) judging whether the master users all work in the same frequency band, if so, executing the step (4.2), otherwise, returning to the step (3) when the situation is equal to a scene of a single master user of multiple users;
(4.2) each secondary user directly accesses the primary user frequency spectrum by adopting a time division multiple access mechanism, and the k-th secondary user is assumed to have a frequency with Q k A bit calculation task, wherein one part of the calculation task is transmitted to the cognitive secondary base station for unloading calculation, and the other part of the calculation task is subjected to local calculation; obtaining the calculated bit number R of the kth secondary user k Energy consumption E k And average interference I caused to mth primary user k,m (ii) a And R is k ≥Q k ,I k,m Interference tolerance I less than or equal to mth primary user th,m ;
(4.3) obtaining the calculation effect of the second scene task according to the following formulaRate E CE1 :
wherein the optimization variable is the transmission power of the second scene secondary userSecond scenario secondary user's local computation CPU frequencyTime length of accessing frequency spectrum by secondary user in second sceneWherein, P k 、f k And beta k Respectively representing the second scene transmitting power of the kth secondary user, the local computing CPU frequency and the time length of accessing the frequency spectrum;
setting a second scene secondary user to meet the following constraint conditions:
wherein e k Energy available for the kth secondary user is represented, T represents the time length of a time slot, and gamma is an energy consumption factor;0≤f k ≤f k,max ,0≤P k ≤P k,max wherein f is k,max And P k,max The maximum available CPU frequency and the maximum transmitting power of the kth secondary user are respectively;
(4.5) maximizing second scenario task computational efficiency E CE1 Through variable replacement, Dinkelbach algorithm willConverting the expression into a convex expression, and then solving the convex expression by using a Lagrange dual decomposition algorithm and a sub-gradient algorithm to obtain the computing efficiency of the optimal task of the second sceneCorresponding second scenario optimal optimization variable, i.e. optimal transmit power of second scenario sub-userOptimal local computation CPU frequencyAnd length of time for optimal access to spectrum
(4.6) setting system working parameters according to the second scene optimal optimization variables, and entering the step (5);
(5) the system operates under the selected working parameters, and the energy efficiency optimization of the system is realized.
2. The method of claim 1, wherein: in the step (3.1), the cognitive secondary base station carries out spectrum sensing on the wireless channel environment, and the fixed detection probability is P d The corresponding false alarm probability of the energy detection mechanism is as follows:
wherein, k represents the signal-to-noise ratio of PU signals received by the cognitive small base station, f s Representing the sampling rate, Q (-) is a complementary cumulative distribution function of a standard gaussian distribution variable.
3. The method of claim 1, wherein: the average calculation bit number R of the kth secondary user in the step (3.2) ave,k Average energy consumption E ave,k And average interference I caused to primary users ave,k Respectively calculated according to the following:
wherein the content of the first and second substances, respectively representing the calculated bit number, the energy consumption and the interference to the master user of the kth secondary user when the real state of the master user is i and the sensing result is j, B representing the access frequency spectrum bandwidth, C k Denotes the CPU clock period, h k,C Represents the channel power gain, h, from the kth secondary user to the cognitive secondary base station k,R Representing the channel power gain, δ, between the k-th secondary user and the primary user k The interference, sigma, caused by the signals sent by the master user when the kth secondary user carries out task unloading when the master user is in an occupied state k Indicating that the secondary kth secondary user transceives signalsLocal noise of time, α i,j The probability of occurrence when the actual channel state of the primary user is i and the perceived channel state is j is represented as follows:
α 0,0 =Pr 0 [1-P f (β 0 )],
α 0,1 =Pr 0 P f (β 0 ),
α 1,0 =Pr 1 (1-P d ),
α 1,1 =Pr 1 P d ,
wherein Pr 0 And Pr 1 Indicating the prior probability that the primary user channel is in idle and occupied states.
4. The method of claim 1, wherein: the number of bits R calculated for the kth secondary user in step (4.2) k Energy consumption E k Average interference I caused to the m-th primary user k,m Respectively according to the following calculation:
I k,m =P k h k,m β k ,m∈Ω M ,k∈Ω K
wherein h is k,m Representing the channel power gain between the kth secondary user and the mth primary user, B representing the access spectrum bandwidth, C k Representing CPU clock cycles, δ k The interference, sigma, caused by the signals sent by the master user when the kth secondary user carries out task unloading when the master user is in an occupied state k Representing the local noise when the kth secondary user is transmitting and receiving signals.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210566366.XA CN114980128A (en) | 2022-05-23 | 2022-05-23 | Energy efficiency optimization method in multi-user cognitive edge computing network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210566366.XA CN114980128A (en) | 2022-05-23 | 2022-05-23 | Energy efficiency optimization method in multi-user cognitive edge computing network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114980128A true CN114980128A (en) | 2022-08-30 |
Family
ID=82985938
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210566366.XA Pending CN114980128A (en) | 2022-05-23 | 2022-05-23 | Energy efficiency optimization method in multi-user cognitive edge computing network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114980128A (en) |
-
2022
- 2022-05-23 CN CN202210566366.XA patent/CN114980128A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109729528B (en) | D2D resource allocation method based on multi-agent deep reinforcement learning | |
Zhang et al. | Energy–latency tradeoff for computation offloading in UAV-assisted multiaccess edge computing system | |
Yang et al. | Optimal resource allocation in full-duplex ambient backscatter communication networks for wireless-powered IoT | |
Al-Shuwaili et al. | Joint uplink/downlink optimization for backhaul-limited mobile cloud computing with user scheduling | |
CN109640320B (en) | Safe unloading method of computing task of mobile edge computing system based on mixed NOMA | |
CN111314935B (en) | Method for minimizing downlink transmission delay based on NOMA-MEC system | |
Wang et al. | Unified offloading decision making and resource allocation in ME-RAN | |
CN110167176B (en) | Wireless network resource allocation method based on distributed machine learning | |
Lu et al. | A cross-layer resource allocation scheme for ICIC in LTE-Advanced | |
CN109861728B (en) | Joint multi-relay selection and time slot resource allocation method for large-scale MIMO system | |
CN107343268B (en) | Non-orthogonal multicast and unicast transmission beamforming method and system | |
Hu et al. | Energy-efficient cooperative spectrum sensing in cognitive satellite terrestrial networks | |
CN114051222A (en) | Wireless resource allocation and communication optimization method based on federal learning in Internet of vehicles environment | |
CN116390125A (en) | Industrial Internet of things cloud edge cooperative unloading and resource allocation method based on DDPG-D3QN | |
CN113596785A (en) | D2D-NOMA communication system resource allocation method based on deep Q network | |
Sun et al. | Unsupervised deep learning for optimizing wireless systems with instantaneous and statistic constraints | |
CN115866787A (en) | Network resource allocation method integrating terminal direct transmission communication and multi-access edge calculation | |
Qin et al. | When the user-centric network meets mobile edge computing: challenges and optimization | |
CN111741520B (en) | Cognitive underwater acoustic communication system power distribution method based on particle swarm | |
Yan et al. | QoE-based semantic-aware resource allocation for multi-task networks | |
Liu et al. | Reliability-oriented design framework in NOMA-assisted mobile edge computing | |
CN113507716A (en) | SWIPT-based CR-NOMA network interruption and energy efficiency optimization method | |
CN115622595B (en) | High-energy-efficiency networking method for realizing self-adaptive large-scale URLLC | |
CN111343722B (en) | Cognitive radio-based energy efficiency optimization method in edge calculation | |
Duan et al. | Energy consumption minimization for near-far server cooperation in NOMA-assisted mobile edge computing system |
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 |