CN113784372A - Joint optimization method for terminal multi-service model - Google Patents

Joint optimization method for terminal multi-service model Download PDF

Info

Publication number
CN113784372A
CN113784372A CN202110918183.5A CN202110918183A CN113784372A CN 113784372 A CN113784372 A CN 113784372A CN 202110918183 A CN202110918183 A CN 202110918183A CN 113784372 A CN113784372 A CN 113784372A
Authority
CN
China
Prior art keywords
terminal
service
model
access point
optimization
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
Application number
CN202110918183.5A
Other languages
Chinese (zh)
Inventor
王德胜
高成
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.)
Huazhong University of Science and Technology
Original Assignee
Huazhong University of Science and Technology
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 Huazhong University of Science and Technology filed Critical Huazhong University of Science and Technology
Priority to CN202110918183.5A priority Critical patent/CN113784372A/en
Publication of CN113784372A publication Critical patent/CN113784372A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/02—Arrangements for optimising operational condition
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/06—Testing, supervising or monitoring using simulated traffic

Abstract

The invention discloses a joint optimization method for a terminal-oriented multi-service model, which belongs to the field of wireless communication and comprises the following steps: establishing a combined optimization model P1 according to preset constraint conditions by taking terminal access selection, a service unloading decision and a resource deployment strategy as variables and aiming at minimizing system weighted energy consumption; the constraint conditions include: each terminal can only access one access point in the same time slot, and a plurality of tasks in the same terminal can be independently selected to be unloaded to the connected access point for processing or be executed locally; decomposing the combined optimization model P1 to separate integer variables from continuous variables to obtain submodels P2 and P3; and solving the submodel P2 and the submodel P3 to obtain the optimization results of the terminal access selection, the service unloading decision and the resource deployment strategy. The invention can establish a terminal multi-service model which truly reflects the service generation condition of the mobile terminal and carry out combined optimization on terminal access selection and service unloading decision.

Description

Joint optimization method for terminal multi-service model
Technical Field
The invention belongs to the field of wireless communication, and particularly relates to a joint optimization method for a terminal multi-service model.
Background
With the gradual maturity of novel time-delay-sensitive applications such as virtual reality, augmented reality, digital holography and the like in a future smart city, the traditional cloud computing is more and more difficult to meet the requirements of real-time performances such as computing and storage due to a long transmission link. And edge computing, by moving the server from the cloud to the edge of the wireless access network closer to the terminal layer, not only can the transmission delay be effectively reduced, but also extra computing power can be provided for the terminal. The change of the network structure provides a wide application potential for novel services with large calculation amount and harsh time delay, and simultaneously provides a new challenge for resource deployment: due to the characteristics of unbalanced spatial distribution of the access base station and the service terminal and heterogeneous computing capability of the edge server in the edge computing scene, how to optimize terminal access selection, service offloading decision, resource deployment strategy and the like to adapt to environmental changes has very important significance and value.
In a multi-cell scene, each mobile terminal can select to Access to one of a plurality of Access Points (APs) in a time slot, and the selection is called terminal Access selection; for the service on the mobile terminal, the service can be processed locally, or the service can be transmitted to the edge server in the AP to which the service is connected, and the service is processed by the edge server, and this selection is called as a service offloading decision; in the whole multi-cell scenario, the data transmission power of the mobile terminal and the base station, and the calculation frequency allocation of the terminal and the edge server are called as a resource deployment strategy. At present, under the multi-cell-multi-terminal edge calculation scene, most documents adopt a single-service unloading model, namely, the service generated by a terminal is regarded as a whole, the processing of all services must be completed on one device, and the model cannot truly reflect the service generation condition of the mobile terminal; in addition, in order to reduce the difficulty in solving the optimization problem, the existing single-service offloading model does not consider the joint optimization problem of terminal access selection and service offloading decision, but this also results in poor optimization effect and is not easy to apply to actual scenes.
Disclosure of Invention
Aiming at the defects and improvement requirements of the prior art, the invention provides a terminal multi-service model-oriented joint optimization method, aiming at establishing a terminal multi-service model capable of truly reflecting the service generation condition of a mobile terminal and carrying out joint optimization on terminal access selection and service unloading decisions so as to solve the technical problems that the optimization result solved by the existing optimization method cannot optimize the system performance and is not easy to apply to an actual scene.
To achieve the above object, according to an aspect of the present invention, there is provided a joint optimization method for a terminal-oriented multi-service model, including:
establishing a combined optimization model P1 according to preset constraint conditions by taking terminal access selection, a service unloading decision and a resource deployment strategy as variables and aiming at minimizing system weighted energy consumption; the preset constraint conditions comprise: each terminal can only access one access point in the same time slot, and a plurality of tasks in the same terminal can be independently selected to be unloaded to the connected access point for processing or be executed locally;
decomposing the combined optimization model P1 to separate integer variables from continuous variables to obtain a sub-model P2 for performing combined optimization on access selection service and service unloading decisions of a terminal and a sub-model P3 for optimizing a resource deployment strategy;
and solving the submodel P2 and the submodel P3 to obtain the optimization results of the terminal access selection, the service unloading decision and the resource deployment strategy, thereby realizing the joint optimization of the terminal access selection, the service unloading decision and the resource deployment strategy.
The invention can respectively set up corresponding service unloading strategies for a plurality of services in each terminal by using that each terminal can only be accessed to one access point in the same time slot, a plurality of tasks in the same terminal can be independently selected and unloaded to the connected access point for processing or locally executed as two constraint conditions of a combined optimization model, compared with a single service unloading model which takes the services generated by the terminal as a whole and uniformly sets up the service unloading strategies, the invention establishes a terminal multi-service model which can truly reflect the service generation condition of the mobile terminal; the method and the device simultaneously take the terminal access selection, the service unloading decision and the resource deployment strategy as variables, establish a joint optimization model, decompose the joint optimization model according to the integer variable and the continuous variable and then respectively solve the decomposed model, can realize the joint optimization of the terminal access selection, the service unloading decision and the resource deployment strategy, simultaneously reduce the solving difficulty of the model, and are easy to be applied to practical application scenes.
Further, the preset constraint condition further includes:
the calculation frequency of the terminal is not negative and cannot exceed the maximum calculation frequency;
a computation frequency constraint of the edge calculator;
the unit bit transmission delay of the uplink and downlink channels cannot be smaller than the transmission delay under the maximum power;
the device accessed by each access point cannot exceed the maximum available subcarrier number;
the service processing delay of the terminal cannot exceed the unit time slot length.
Further, the resource deployment policy includes: the unit bit transmission time of an uplink channel, the unit bit transmission time of a downlink channel, the terminal calculation frequency and the edge server calculation frequency distribution variable; and, the joint optimization model P1 is:
Figure BDA0003206433290000031
Figure BDA0003206433290000032
Figure BDA0003206433290000033
Figure BDA0003206433290000034
Figure BDA0003206433290000035
Figure BDA0003206433290000036
Figure BDA0003206433290000037
Figure BDA0003206433290000038
Figure BDA0003206433290000041
Figure BDA0003206433290000042
Figure BDA0003206433290000043
wherein U represents a utility function of the system; i. j and k represent an access point number, a terminal number, and a task number in the terminal, respectively, M and N represent a total number of access points and a total number of terminals, respectively,
Figure BDA0003206433290000044
and
Figure BDA0003206433290000045
respectively representing a set of access point numbers and a set of terminal numbers, alphajRepresenting the total number of tasks in the jth terminal; rhoij∈{0,1}，ρ ij1 denotes that the terminal j accesses the first access point i, ρ ij0 means that the terminal j does not access the access point i; m isjk∈{0,1}，m jk1 indicates that a terminal j offloads traffic k therein to a connected access point for processing, m jk0 denotes that terminal j is in this bookProcessing the k-th service therein;
Figure BDA0003206433290000046
represents the unit bit transmission time of the uplink channel between access point i and terminal j,
Figure BDA0003206433290000047
representing the unit bit transmission time of the downlink channel between access point i and terminal j,
Figure BDA0003206433290000048
representing the local calculation frequency, f, of terminal jijThe CPU calculation frequency which is distributed to the terminal j by the edge server of the access point i is represented; ρ, m, τul,τdl,flocAnd f is respectively rhoij、mjk、
Figure BDA0003206433290000049
And fijThe set of (2) respectively represents terminal access selection, service unloading decision, unit bit transmission time of an uplink channel, unit bit transmission time of a downlink channel, terminal calculation frequency and edge server calculation frequency distribution variable;
Figure BDA00032064332900000410
represents the maximum local calculation frequency, F, of terminal ji maxRepresenting the maximum computation frequency of the edge server of access point i,
Figure BDA00032064332900000411
represents the maximum transmission power, P, of terminal ji maxRepresenting the maximum transmit power, N, of the access point i on each subchannelmaxRepresenting the number of available subcarriers of the access point; wulAnd WdlRespectively representing the unit subcarrier bandwidths of the uplink and downlink channels,
Figure BDA00032064332900000412
and
Figure BDA00032064332900000413
representing the channel gain, N, on the uplink and downlink channels between access point i and terminal j0A power spectral density representing additive white gaussian noise;
Figure BDA00032064332900000414
indicating the delay of the traffic of terminal j processed locally,
Figure BDA00032064332900000415
represents the total delay, T, for terminal j to offload traffic to the edge serversIndicating the slot length.
In each constraint condition of the joint optimization model P1 established by the invention: c1 indicates that the calculation frequency of the terminal is not negative and cannot exceed the maximum calculation frequency; c2, C3 represent the compute frequency constraints of the edge server; c4 and C5 indicate that the unit bit transmission delay of the uplink and downlink channels cannot be smaller than the transmission delay at the maximum power; c6, C7 show that each terminal can only access one AP in the same time slot; c8 indicates that the device accessed by each AP may not exceed its maximum number of available subcarriers; c9 is a variable 0-1 integer indicating that task k in terminal j can choose to be offloaded to a connected access point for processing or executed locally; c10 indicates that the traffic processing delay of the terminal cannot exceed the unit slot length. The setting of the constraint bars can consider the unloading condition of the terminal multi-service, and realize the joint optimization of the terminal access selection, the service unloading decision and the resource deployment strategy under the service delay constraint.
Further, the submodel P2 for jointly optimizing the access selection service and the service offloading decision of the terminal is:
Figure BDA0003206433290000051
Figure BDA0003206433290000052
Figure BDA0003206433290000053
Figure BDA0003206433290000054
Figure BDA0003206433290000055
the submodel P3 for optimizing the resource deployment policy is:
Figure BDA0003206433290000056
Figure BDA0003206433290000057
Figure BDA0003206433290000058
Figure BDA0003206433290000059
Figure BDA00032064332900000510
Figure BDA00032064332900000511
Figure BDA00032064332900000512
in submodels P2 and P3 obtained by decomposing a combined optimization model P1, P2 is a model only containing integer variables, and a heuristic algorithm can be used for seeking a suboptimal solution; p3 is a model containing only continuous variables, and a global optimal solution can be obtained by using a traditional convex optimization method.
Further, the solution submodel P3 includes:
the submodel P3 is decomposed to make the variable flocAnd { τul,τdlF separation to obtain the frequency f used for calculating the terminallocLocal processing energy consumption minimization optimization model P31 for optimization, and unit bit transmission time tau for uplink channelulUnit bit transmission time tau of downlink channeldlA traffic offload energy consumption minimization optimization model P32 for performing joint optimization with the edge server computing frequency distribution variable f;
decomposing constraint C10 into constraint C11:
Figure BDA0003206433290000061
and constraint C12:
Figure BDA0003206433290000062
the constraint conditions of the local processing energy consumption minimization optimization model P31 comprise constraint conditions C11, and the constraint conditions of the traffic offload energy consumption minimization optimization model P32 comprise constraint conditions C12;
and respectively solving the local processing energy consumption minimization optimization model P31 and the service unloading energy consumption minimization optimization model P32 to obtain the unit bit transmission time of the uplink channel, the unit bit transmission time of the downlink channel, the terminal calculation frequency and the optimization result of the calculation frequency distribution variable of the edge server.
The analysis of the sub-model P3 shows that f is used for determining the access selection and the traffic unloading of the fixed terminallocAnd { τul,τdlF, no coupling relation exists between the two parts, and the solution can be carried out independently; the invention further decomposes the sub-model P3 to make flocAnd { τul,τdlF, separating to obtain a local processing energy consumption minimized optimization model P31 and a business unloading energy consumption minimized optimization model P32, and then respectively solving, so that the solving precision of the sub model P3 can be ensuredThe solving difficulty is effectively reduced, and the solving efficiency is improved.
Further, the local processing energy consumption minimization optimization model P31 is:
Figure BDA0003206433290000063
Figure BDA0003206433290000064
Figure BDA0003206433290000065
the traffic offload energy consumption minimization optimization model P32 is as follows:
Figure BDA0003206433290000071
Figure BDA0003206433290000072
Figure BDA0003206433290000073
Figure BDA0003206433290000074
Figure BDA0003206433290000075
Figure BDA0003206433290000076
wherein u isjk、cjkAnd djkRespectively representThe size of input data volume required for executing the service, the number of CPU cycles required for executing the service and the data volume of the service calculation result;
Figure BDA0003206433290000077
is a constant, k, associated with the chip type of terminal jiIs a constant related to the access point i chip type;
Figure BDA0003206433290000078
and
Figure BDA0003206433290000079
respectively representing the channel gains on an uplink channel and a downlink channel between a terminal j and an access point i;
Figure BDA00032064332900000710
and
Figure BDA00032064332900000711
respectively representing the single bit transmission time delay of uplink transmission and downlink transmission between the terminal j and the access point i.
In the constraint conditions of the local processing energy consumption minimization optimization model P31, the constraint condition C11 decomposed by the original constraint condition C10 indicates that the total processing time delay of all local processing services of the terminal j does not exceed the time slot length; as can be seen from the objective function of the local processing energy consumption minimization optimization model P31, under the determined traffic offloading mode, the local computing energy consumption of the terminal is proportional to the computing frequency; the time delay of the terminal service in local processing is inversely proportional to the calculation frequency, so that the optimal solution of the local calculation frequency and the local minimum calculation energy consumption can be determined to be
Figure BDA00032064332900000712
And
Figure BDA00032064332900000713
in the above traffic offload energy consumption minimization optimization model P32, the constraint condition C12 decomposed from the original constraint condition C10 indicates that the total processing delay of all the traffic offloaded to the edge server at the terminal j cannot exceed the slot length; the traffic offload energy consumption minimization optimization model P32 is already a convex optimization problem, and can be solved by using a traditional convex optimization method.
Further, the solution submodel P2 includes:
(S0) taking a set of terminal access selection and traffic offload decisions as an individual, each individual corresponding to a double-stranded encoded chromosome, the double strands being a terminal access selection chain and a traffic offload decision chain, respectively;
(S1) generating an initial population using a random method, and calculating a fitness of each individual; the coding adopts a double-chain coding structure, the terminal access selection chain adopts an integer coding mode, and the service unloading decision chain adopts a binary coding mode; the fitness of the individual is the weighted energy consumption of the system corresponding to the individual;
(S2) selecting a part of the parent chromosomes, and updating the crossover probability p of the parent chromosomes according to the smaller fitness value f in the fitness values of the parent chromosomescAnd the resulting individual variation probability pmSo that the crossover probability p of the parent chromosomes is smallercAnd the resulting individual variation probability pmAre all smaller; according to the updated cross probability pcAnd the probability of variation pmPerforming crossing and mutation operations to generate a plurality of new individuals, and generating a new population by combining an elite retention strategy;
(S3) decoding each individual in the new population, calculating the fitness value of each individual, and if the maximum iteration number is not reached, turning to the step (S2); otherwise, the individual with the minimum fitness value is used as the solution result, and the solution of the sub-model P2 is finished.
When the invention is used for solving the submodel P2 for carrying out combined optimization on the access selection service and the service unloading decision of the terminal, on the basis of the traditional genetic algorithm, the cross probability P of the parent chromosome is updated according to the smaller fitness value f in the fitness values of the parent chromosome before crossing and mutation each timecAnd the resulting individual variation probability pmSo that the crossover probability p of the parent chromosomes is smallercAnd the resulting individual variation probability pmThe probability of cross and variation of better individuals is smaller, so that the better individuals are kept in the solving process, and the optimization effect of the model is further ensured.
Further, in step (S2), after the update, the crossover probability p of the parent chromosome is updatedcAnd the resulting individual variation probability pmAnd satisfies the following conditions:
Figure BDA0003206433290000091
Figure BDA0003206433290000092
wherein f isminRepresenting the minimum value of fitness value in the population, favgRepresenting the average value of fitness values of the population; p is a radical ofcmaxAnd pcminRespectively representing the maximum and minimum of the cross probability, pmmaxAnd pmminRespectively representing the maximum and minimum of the mutation probability.
Further, during crossing, the terminal access selection chain adopts two-point crossing, and the service unloading decision chain adopts a uniform crossing mode.
According to another aspect of the present invention, there is provided a computer readable storage medium comprising a stored computer program; when being executed by the processor, the computer program controls the device on which the computer readable storage medium is positioned to execute the joint optimization method for the terminal multi-service model provided by the invention.
Generally, by the above technical solution conceived by the present invention, the following beneficial effects can be obtained:
(1) the invention can respectively establish corresponding service unloading strategies for a plurality of services in each terminal by using that each terminal can only access one access point in the same time slot, a plurality of tasks in the same terminal can independently select to be unloaded to the connected access point for processing or locally execute two constraint conditions serving as a combined optimization model, and a terminal multi-service model is established and can truly reflect the service generation condition of the mobile terminal; on the basis, a joint optimization model is established by taking the terminal access selection, the service unloading decision and the resource deployment strategy as variables, and the joint optimization model is decomposed according to the integer variables and the continuous variables and then respectively solved, so that joint optimization of the terminal access selection, the service unloading decision and the resource deployment strategy can be realized, the solving difficulty of the model is reduced, and the method is easy to apply to practical application scenes.
(2) When the submodel P3 for optimizing the resource deployment strategy is solved, the variables without coupling relation are further excavated, and the submodel P3 is further decomposed, so that the solving difficulty can be reduced and the solving efficiency can be improved under the condition of ensuring the solving precision of the model.
(3) When the submodel P2 for performing joint optimization on the access selection service and the service unloading decision of the terminal is solved, on the basis of the traditional genetic algorithm, the crossover probability and the individual mutation probability of the parent chromosomes can be determined before crossover and mutation are performed each time, so that the better individual crossover and mutation probability is smaller, better individuals are reserved in the solving process, and the optimization effect of the model is further ensured.
Drawings
Fig. 1 is a model diagram of a multi-base-station-multi-terminal-multi-service mobile edge scene according to an embodiment of the present invention;
FIG. 2 is a flowchart of a joint optimization method for a terminal-oriented multi-service model according to an embodiment of the present invention;
FIG. 3 is a flow chart of a conventional genetic algorithm;
FIG. 4 is a graph illustrating adaptive probability variation according to an embodiment of the present invention;
fig. 5 is a schematic diagram illustrating a variant operation of a terminal accessing a selection chain according to an embodiment of the present invention;
fig. 6 is a schematic diagram illustrating a variant operation of a traffic offload decision chain according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
In the present application, the terms "first," "second," and the like (if any) in the description and the drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.
In order to solve the technical problems that the existing single-service unloading model can not truly reflect the service generation condition of a mobile terminal and can not realize the joint optimization of terminal access selection and service unloading decision, so that the solved optimization result can not optimize the system performance and can not be easily applied to actual scenes, the invention provides a joint optimization method for a terminal multi-service model, and the overall thought is as follows: optimizing a terminal service model to enable the model to truly reflect the service generation condition of the terminal; under the constraint of a terminal multi-service model, an optimization model is established, meanwhile, a terminal access selection strategy, a service unloading decision strategy and a resource deployment strategy are used as variables of the optimization model, so that the joint optimization of the terminal access selection strategy, the service unloading decision strategy and the resource deployment strategy is realized, the optimization model is decomposed according to a digital variable and a continuous variable and then respectively solved, the solving difficulty of the model is reduced, and finally, an optimization result which is easy to apply to an actual scene is obtained.
Before explaining the technical solution of the present invention in detail, a moving edge calculation scenario related to the present invention is briefly introduced with reference to fig. 1:
as shown in fig. 1, the moving edge computation scenario includes a control layer, an edge computation layer, and a termination layer. The control layer includes a central controller capable of receiving network information from the edge computation layer and the termination layer. There are M Access Points (APs) in the edge computation layer, denoted as
Figure BDA0003206433290000111
Each access point corresponds to an edge server (MEC), and the maximum calculation frequency of the edge server of the access point i is Fi max(ii) a The maximum transmission power of the AP on each sub-channel is the same, and is Pi max. N mobile terminals are distributed in the terminal layer and are represented as
Figure BDA0003206433290000112
Each mobile terminal has multiple independent calculation intensive service waiting processes, and the number of services generated by terminal j is alphaj(ii) a The maximum local computation frequency of terminal j is
Figure BDA0003206433290000113
Maximum transmission power of
Figure BDA0003206433290000114
Traffic k on terminal j (k e {1,2, … … α)j}) use triplets ujk,cjk,djkDenotes ujk、cjkAnd djkRespectively indicating the size of input data required for executing the service, the number of CPU cycles required for executing the service, and the data size of the service calculation result.
In the traditional single traffic offload model, α at terminal jjThe services can be taken as a whole and are unloaded to the AP for processing in a unified way or are executed locally at the terminal j; in contrast, in the terminal multi-service model provided by the invention, a service unloading decision is made for each service in the terminal j, each service can be unloaded to the terminal A independently for processing, or can be executed locally at the terminal j independently, and the model truly reflects the service generation condition of the terminal.
In the following examples, in ρijE {0,1} represents whether the mobile terminal j is accessed to the access point i; when rhoijWhen 1, it means that the access point i is connected with the mobile terminal j; when rhoijWhen the value is 0, the access point i is not connected with the mobile terminal j; each terminal cannot switch within a time slotThe connected AP is changed, and the terminal can only be accessed to one AP at each moment, so that the access mode of the terminal meets the requirement
Figure BDA0003206433290000121
In mjkE {0,1} represents the unloading strategy of the service k on the terminal j; when m isjkWhen the traffic is 1, the terminal offloads the traffic to the connected AP for processing; when m isjkWhen 0, the terminal processes the traffic locally.
The following are examples.
Example 1:
a joint optimization method for a terminal-oriented multi-service model, as shown in fig. 2, includes:
establishing a combined optimization model P1 according to preset constraint conditions by taking terminal access selection, a service unloading decision and a resource deployment strategy as variables and aiming at minimizing system weighted energy consumption; the preset constraint conditions comprise: each terminal can only access one access point in the same time slot, and a plurality of tasks in the same terminal can be independently selected to be unloaded to the connected access point for processing or be executed locally;
decomposing the combined optimization model P1 to separate integer variables from continuous variables to obtain a sub-model P2 for performing combined optimization on access selection service and service unloading decisions of a terminal and a sub-model P3 for optimizing a resource deployment strategy;
and solving the submodel P2 and the submodel P3 to obtain the optimization results of the terminal access selection, the service unloading decision and the resource deployment strategy, thereby realizing the joint optimization of the terminal access selection, the service unloading decision and the resource deployment strategy.
In this embodiment, each terminal can only access to one access point in the same timeslot, and multiple tasks in the same terminal can independently select to be offloaded to the connected access point for processing or locally execute two constraint conditions serving as a joint optimization model, that is, constraints are provided for the established joint optimization model P1 according to the terminal multi-service model, so that the joint optimization model P1 can also optimize according to the actual situation generated by the service of the mobile terminal; the joint optimization model P1 established in this embodiment takes terminal access selection, service offloading decision, and resource deployment policy as variables, and by solving the model, joint optimization of terminal access selection, service offloading decision, and resource deployment policy can be achieved; in this embodiment, the resource deployment policy includes: the unit bit transmission time of an uplink channel, the unit bit transmission time of a downlink channel, the terminal calculation frequency and the edge server calculation frequency distribution variable;
the establishment process of the joint optimization model P1 is as follows:
(1) establishing an uplink channel model and a downlink channel model:
according to the Shannon formula, the channel capacity of the corresponding link can be obtained, and the channel capacity is used as the transmission rate of the data on the channel and the transmission rate of the data on the uplink channel
Figure BDA0003206433290000131
And the transmission rate of data on the downlink channel
Figure BDA0003206433290000132
Respectively expressed as:
Figure BDA0003206433290000133
Figure BDA0003206433290000134
wherein N is0Represents the power spectral density of Additive White Gaussian Noise (AWGN); wulAnd WdlRespectively representing unit subcarrier bandwidths of an uplink channel and a downlink channel; when the access point i is accessed to the terminal j, the transmission power of the terminal j and the transmission power of the access point i are respectively
Figure BDA0003206433290000135
And
Figure BDA0003206433290000136
the channel gains on the corresponding channels are respectively
Figure BDA0003206433290000137
And
Figure BDA0003206433290000138
when the sending power is used as a variable, the problem is still not convex under the condition that the access selection and the service unloading decision are determined, and the conventional method is difficult to solve; after the single-bit transmission delay variable is used for replacing the sending power as the optimization variable, under the condition that the access selection and the service unloading decision are determined, the problem is convex, and a convex optimization method can be used for solving, so that the introduction of the single-bit transmission delay variable as the optimization variable is beneficial to the subsequent optimization solution. Therefore, single bit transmission delay variable of uplink and downlink transmission is introduced
Figure BDA0003206433290000139
And
Figure BDA00032064332900001310
the single-bit transmission delay and the transmission rate are reciprocal relations, namely:
Figure BDA0003206433290000141
Figure BDA0003206433290000142
then, the transmission delay of the transmission power using unit bit is expressed as:
Figure BDA0003206433290000143
Figure BDA0003206433290000144
the time delay for the terminal j to transmit the service data to the server and the transmission time delay of the service processing result are respectively as follows:
Figure BDA0003206433290000145
Figure BDA0003206433290000146
the time delay for transmitting the service data to the server by the terminal j and the transmission energy consumption of the service processing result are respectively as follows:
Figure BDA0003206433290000147
Figure BDA0003206433290000148
(2) establishing a service processing model:
there are two processing modes for the service: and locally processing and unloading to an edge server.
The time delay and energy consumption of the local processing of the service of the terminal j are respectively as follows:
Figure BDA0003206433290000149
Figure BDA00032064332900001410
wherein the content of the first and second substances,
Figure BDA00032064332900001411
indicating the current local CPU computation frequency for terminal j,
Figure BDA00032064332900001412
is a constant associated with the terminal j chip type.
The processing delay and the energy consumption of the service k of the terminal j in the edge server are respectively as follows:
Figure BDA0003206433290000151
Figure BDA0003206433290000152
wherein f isijCPU calculation frequency, k, assigned to terminal j by edge server representing access point iiIs a constant associated with the access point i chip type.
(3) Establishing corresponding constraints:
in this embodiment, the constraint conditions of the joint optimization model P1 include the following constraint conditions, in addition to that each terminal mentioned above can only access to one access point in the same timeslot, and multiple tasks in the same terminal can independently choose to be offloaded to the connected access point for processing or be executed locally:
the calculation frequency of the terminal is not negative and cannot exceed the maximum calculation frequency;
a computation frequency constraint of the edge calculator;
the unit bit transmission delay of the uplink and downlink channels cannot be smaller than the transmission delay under the maximum power;
the device accessed by each access point cannot exceed the maximum available subcarrier number;
the service processing time delay of the terminal cannot exceed the unit time slot length;
the corresponding calculation for each constraint is as follows:
the known slot length is TsThe time delay after the terminal j finishes processing all services can be expressed as:
Figure BDA0003206433290000153
wherein the content of the first and second substances,
Figure BDA0003206433290000154
representing the total delay for terminal j to offload traffic to the edge server.
However, the total processing delay cannot exceed the slot length, so the delay constraint of terminal j is:
Figure BDA0003206433290000155
ρijthe access selection of the terminal is shown, at the same time, the terminal can only access to one base station, and the number of terminals accessed by one base station cannot exceed the number of available subcarriers of the base station, so that the following steps are provided:
Figure BDA0003206433290000161
Figure BDA0003206433290000162
the transmission power of the terminal and the base station cannot exceed the maximum transmission power, so the corresponding unit bit transmission delay constraint is as follows:
Figure BDA0003206433290000163
Figure BDA0003206433290000164
(4) taking the weighted energy consumption of the minimization system as an optimization target, wherein the utility function of the terminal j is as follows:
Figure BDA0003206433290000165
wherein the content of the first and second substances,
Figure BDA0003206433290000166
representing the energy consumption of the terminal j for processing the traffic,
Figure BDA0003206433290000167
Figure BDA0003206433290000168
representing the total power consumption of the operator,
Figure BDA0003206433290000169
βj∈[0,1]the preference of the terminal j to the energy conservation of the operator is shown, and when the electric quantity of the terminal is sufficient, larger beta can be setj(ii) a When the terminal is short of power, betajSmaller until the reduction is 0; since most mobile terminals are battery powered and should preferably reduce the processing power consumption of the terminal as much as possible, β is setj∈[0,1]The preference of the energy consumption of an operator can be ensured to be less than the energy consumption of terminal processing, and the energy consumption of the terminal tends to be reduced while the weighted energy consumption of the system is reduced;
the sum of the utility functions of the system is:
Figure BDA00032064332900001610
(5) constructing a weighted energy consumption problem of a minimization system, wherein the corresponding established joint optimization model P1 is as follows:
Figure BDA00032064332900001611
Figure BDA00032064332900001612
Figure BDA0003206433290000171
Figure BDA0003206433290000172
Figure BDA0003206433290000173
Figure BDA0003206433290000174
Figure BDA0003206433290000175
Figure BDA0003206433290000176
Figure BDA0003206433290000177
Figure BDA0003206433290000178
Figure BDA0003206433290000179
where ρ, m, τul,τdl,flocAnd f is respectively rhoij、mjk、
Figure BDA00032064332900001710
And fijThe set of (2) respectively represents terminal access selection, service unloading decision, unit bit transmission time of an uplink channel, unit bit transmission time of a downlink channel, terminal calculation frequency and edge server calculation frequency distribution variable;
Figure BDA00032064332900001711
represents the maximum local calculation frequency, F, of terminal ji maxRepresenting the maximum computation frequency of the edge server of access point i,
Figure BDA00032064332900001712
represents the maximum transmission power, P, of terminal ji maxRepresenting the maximum transmit power, N, of the access point i on each subchannelmaxRepresenting the number of available subcarriers of the access point;
among the constraints of the joint optimization model P1: c1 indicates that the calculation frequency of the terminal is not negative and cannot exceed the maximum calculation frequency; c2, C3 represent the compute frequency constraints of the edge server; c4 and C5 indicate that the unit bit transmission delay of the uplink and downlink channels cannot be smaller than the transmission delay at the maximum power; c6, C7 show that each terminal can only access one AP in the same time slot; c8 indicates that the device accessed by each AP may not exceed its maximum number of available subcarriers; c9 is a variable 0-1 integer indicating that task k in terminal j can choose to be offloaded to a connected access point for processing or executed locally; c10 indicates that the traffic processing delay of the terminal cannot exceed the unit slot length. The setting of the constraint bars can consider the unloading condition of the terminal multi-service, and realize the joint optimization of the terminal access selection, the service unloading decision and the resource deployment strategy under the service delay constraint.
The combined optimization model P1 is complex, the solving difficulty is high, the combined optimization model P1 is decomposed according to the digital variables and continuously, and then the solution is carried out, so that the model solving difficulty can be effectively reduced; in the variables of the joint optimization model P1, the terminal access selection and service offloading decision is a digital variable, the resource deployment policy is a continuous variable, and after the joint optimization model P1 is decomposed, the submodel P2 for performing joint optimization on the terminal access selection service and service offloading decision is:
Figure BDA0003206433290000181
Figure BDA0003206433290000182
Figure BDA0003206433290000183
Figure BDA0003206433290000184
Figure BDA0003206433290000185
the submodel P3 for optimizing the resource deployment policy is:
Figure BDA0003206433290000186
Figure BDA0003206433290000187
Figure BDA0003206433290000188
Figure BDA0003206433290000189
Figure BDA00032064332900001810
Figure BDA00032064332900001811
in submodels P2 and P3 obtained by decomposing a combined optimization model P1, P2 is a model only containing integer variables, and a heuristic algorithm can be used for seeking a suboptimal solution; p3 is a model containing only continuous variables, and a global optimal solution can be obtained by using a traditional convex optimization method.
And f, analyzing the sub-model P3 to find that after the fixed terminal access selection and the service unloading decision, flocAnd { τul,τdlF, no coupling relation exists between the two parts, and the solution can be carried out independently; in order to further reduce the difficulty in solving the model and improve the solving efficiency, as an optimal implementation manner, the present embodiment further decomposes the sub-model P3, and solves the solution based on the decomposition result; accordingly, solving the submodel P3 includes:
the submodel P3 is decomposed to make the variable flocAnd { τul,τdlF separation to obtain the frequency f used for calculating the terminallocLocal processing energy consumption minimization optimization model P31 for optimization, and unit bit transmission time tau for uplink channelulUnit bit transmission time tau of downlink channeldlA traffic offload energy consumption minimization optimization model P32 for performing joint optimization with the edge server computing frequency distribution variable f;
decomposing the constraint C10 into constraints C11 and C12:
Figure BDA0003206433290000191
the constraint condition C11 indicates that the total processing time delay of all local processing services of the terminal j does not exceed the time slot length, and the constraint condition of the local processing energy consumption minimization optimization model P31 comprises the constraint condition C11; the constraint condition C12 indicates that the total processing delay of all the traffic offloaded to the edge server by the terminal j cannot exceed the time slot length, and the constraint condition of the traffic offload energy consumption minimization optimization model P32 includes the constraint condition C12;
respectively solving a local processing energy consumption minimization optimization model P31 and a service unloading energy consumption minimization optimization model P32 to obtain the unit bit transmission time of an uplink channel and the unit bit transmission time of a downlink channel, and the optimization results of the terminal calculation frequency and the edge server calculation frequency distribution variable;
in this embodiment, the sub-model P3 is further decomposed to obtain a local processing energy consumption minimization optimization model P31 and a service offloading energy consumption minimization optimization model P32, and then the solutions are respectively performed, so that the solution difficulty can be effectively reduced and the solution efficiency can be improved under the condition that the solution accuracy of the sub-model P3 is ensured;
the local energy consumption minimization optimization problem is that after a service unloading mode of a terminal is determined, the terminal reaches the minimum value of local processing energy consumption on the premise of meeting time delay constraint by adjusting the local calculation frequency of the terminal; in this embodiment, the local processing energy consumption minimization optimization model P31 is:
Figure BDA0003206433290000201
Figure BDA0003206433290000202
Figure BDA0003206433290000203
when the local energy consumption minimization optimization problem has a solution, the terminal can process all generated services within the time slot length; as can be seen from the objective function of the local energy consumption minimization optimization model P31, under the determined traffic offloading mode, the local computing energy consumption of the terminal is in direct proportion to the computing frequency; the time delay of the terminal service in local processing is inversely proportional to the calculation frequency, and the optimal solution of the local calculation frequency and the local minimum calculation energy consumption can be determined as follows:
Figure BDA0003206433290000204
Figure BDA0003206433290000205
the optimization problem of minimizing the energy consumption of the service unloading is that after the service unloading mode of the terminal is determined, the weighted total energy consumption of the terminal data transmission and the energy consumption of an operator is minimized by optimizing the unit bit transmission delay (namely, equivalently, adjusting the transmission power) of the terminal and the base station and the frequency resource allocation of the base station; in this embodiment, the traffic offload energy consumption minimization optimization model P32 is:
Figure BDA0003206433290000206
Figure BDA0003206433290000207
Figure BDA0003206433290000208
Figure BDA0003206433290000209
Figure BDA00032064332900002010
Figure BDA00032064332900002011
after the terminal access selection and the traffic offload decision are determined, the traffic offload energy consumption minimization optimization problem is already a convex optimization problem, and accordingly, the traffic offload energy consumption minimization optimization model P32 can be solved by using a conventional convex optimization method.
After the local processing energy consumption minimization optimization model P31 and the service unloading energy consumption minimization optimization model P32 are respectively solved, the solving results of the two models are integrated, and then the solving result of the sub-model P3 can be obtained.
The submodel P2 may use heuristic algorithms such as genetic algorithm to seek sub-optimal solution, and the flow of the conventional genetic algorithm is shown in fig. 3, and mainly includes the following steps:
(S0) generating an initial population using a random method;
(S1) calculating fitness function values of all individuals in the population;
(S2) if the loop turns reach the maximum iteration times or the population stability condition is met, outputting the optimal individual and the corresponding variable value, otherwise, continuing to execute the step S3;
(S3) generating a new population through cross mutation and selection operations, and returning to step S1.
In order to further improve the optimization effect, the embodiment improves the traditional genetic algorithm, and proposes an adaptive genetic algorithm for solving the sub-model P2, and accordingly the solving process includes:
(S0) taking a set of terminal access selection and traffic offload decisions as an individual, each individual corresponding to a double-stranded encoded chromosome, the double strands being a terminal access selection chain and a traffic offload decision chain, respectively;
(S1) generating an initial population using a random method, and calculating a fitness of each individual; the coding adopts a double-chain coding structure, the terminal access selection chain adopts an integer coding mode, and the service unloading decision chain adopts a binary coding mode; the fitness of the individual is the weighted energy consumption of the system corresponding to the individual;
(S2) selecting a part of the parent chromosomes, and updating the crossover probability p of the parent chromosomes according to the smaller fitness value f in the fitness values of the parent chromosomescAnd the resulting individual variation probability pmSo that the crossover probability p of the parent chromosomes is smallercAnd the resulting individual variation probability pmAre all smaller; according to the updated cross probability pcAnd the probability of variation pmPerforming crossing and mutation operations to generate a plurality of new individuals, and generating a new population by combining an elite retention strategy;
the embodiment updates the cross probability and the mutation probability according to the above manner, and can adapt to the solving situation of the model, so that the better individual has smaller probability of cross and mutation, thereby keeping the better individual in the solving process and further ensuring the optimization effect of the model;
as an alternative, in this embodiment, in step (S2), after updating, the crossover probability p of the parent chromosomecAnd the resulting individual variation probability pmAnd satisfies the following conditions:
Figure BDA0003206433290000221
Figure BDA0003206433290000222
wherein f isminRepresenting the minimum value of fitness value in the population, favgRepresenting the average value of fitness values of the population; p is a radical ofcmaxAnd pcminRespectively representing the maximum and minimum of the cross probability, pmmaxAnd pmminRespectively representing the maximum value and the minimum value of the variation probability; accordingly, the cross probability pcAnd the probability of variation pmThe variation curves of (a) are all consistent with the variation curve shown in fig. 4;
as an optional implementation manner, during the crossing, the terminal access selection chain and the service offloading decision chain adopt different crossing manners, specifically, the terminal access selection chain adopts two-point crossing, as shown in fig. 5; the traffic offload decision chain adopts a uniform crossing manner, as shown in fig. 6; the cross mode combination is matched with the data characteristics of terminal access selection and service unloading decision, so that the model solving precision can be improved;
(S3) decoding each individual in the new population, calculating the fitness value of each individual, and if the maximum iteration number is not reached, turning to the step (S2); otherwise, the individual with the minimum fitness value is used as the solution result, and the solution of the sub-model P2 is finished.
In this embodiment, after solutions of the sub-model P2 and the sub-model P3 are respectively completed, the solution results of the two models are integrated, so that the solution result of the joint optimization model P1 can be obtained, and joint optimization of terminal access selection, service offloading decision and resource deployment strategy is realized. And then, the central controller transmits the joint optimization result to the edge computing layer and the terminal layer through a control instruction, so that the terminal and the edge computing device are specifically executed according to corresponding strategies, and the optimization of the system energy consumption can be realized.
Example 2:
a computer readable storage medium comprising a stored computer program; when being executed by the processor, the computer program controls the device on which the computer-readable storage medium is located to execute the joint optimization method for the terminal multi-service model provided in embodiment 1.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims (10)

1. A joint optimization method for a terminal multi-service model is characterized by comprising the following steps:
establishing a combined optimization model P1 according to preset constraint conditions by taking terminal access selection, a service unloading decision and a resource deployment strategy as variables and aiming at minimizing system weighted energy consumption; the preset constraint conditions comprise: each terminal can only access one access point in the same time slot, and a plurality of tasks in the same terminal can be independently selected to be unloaded to the connected access point for processing or be executed locally;
decomposing the combined optimization model P1 to separate integer variables from continuous variables to obtain a sub-model P2 for performing combined optimization on access selection service and service unloading decisions of a terminal and a sub-model P3 for optimizing a resource deployment strategy;
and solving the submodel P2 and the submodel P3 to obtain the optimization results of the terminal access selection, the service unloading decision and the resource deployment strategy, thereby realizing the joint optimization of the terminal access selection, the service unloading decision and the resource deployment strategy.
2. The terminal-oriented multi-service model joint optimization method of claim 1, wherein the preset constraint condition further comprises:
the calculation frequency of the terminal is not negative and cannot exceed the maximum calculation frequency;
a computation frequency constraint of the edge calculator;
the unit bit transmission delay of the uplink and downlink channels cannot be smaller than the transmission delay under the maximum power;
the device accessed by each access point cannot exceed the maximum available subcarrier number;
the service processing delay of the terminal cannot exceed the unit time slot length.
3. The terminal-oriented multi-service model joint optimization method of claim 2, wherein the resource deployment strategy comprises: the unit bit transmission time of an uplink channel, the unit bit transmission time of a downlink channel, the terminal calculation frequency and the edge server calculation frequency distribution variable; and, the joint optimization model P1 is:
Figure FDA0003206433280000021
Figure FDA0003206433280000022
Figure FDA0003206433280000023
Figure FDA0003206433280000024
Figure FDA0003206433280000025
Figure FDA0003206433280000026
Figure FDA0003206433280000027
Figure FDA0003206433280000028
Figure FDA0003206433280000029
Figure FDA00032064332800000210
Figure FDA00032064332800000211
wherein U represents a utility function of the system; i. j and k represent an access point number, a terminal number, and a task number in the terminal, respectively, M and N represent a total number of access points and a total number of terminals, respectively,
Figure FDA00032064332800000212
and
Figure FDA00032064332800000213
respectively representing a set of access point numbers and a set of terminal numbers, alphajRepresenting the total number of tasks in the jth terminal; rhoij∈{0,1}，ρijTable 1 (the attached drawings)Showing that the terminal j accesses the first access point i, rhoij0 means that the terminal j does not access the access point i; m isjk∈{0,1}，mjk1 indicates that a terminal j offloads traffic k therein to a connected access point for processing, mjk0 means that the terminal j processes the k-th service therein locally;
Figure FDA00032064332800000214
represents the unit bit transmission time of the uplink channel between access point i and terminal j,
Figure FDA00032064332800000215
representing the unit bit transmission time of the downlink channel between access point i and terminal j,
Figure FDA00032064332800000216
representing the local calculation frequency, f, of terminal jijThe CPU calculation frequency which is distributed to the terminal j by the edge server of the access point i is represented; ρ, m, τul,τdl,flocAnd f is respectively rhoij、mjk、
Figure FDA00032064332800000217
And fijThe set of (2) respectively represents terminal access selection, service unloading decision, unit bit transmission time of an uplink channel, unit bit transmission time of a downlink channel, terminal calculation frequency and edge server calculation frequency distribution variable;
Figure FDA00032064332800000218
represents the maximum local calculation frequency, F, of terminal ji maxRepresenting the maximum computation frequency of the edge server of access point i,
Figure FDA0003206433280000031
represents the maximum transmission power, P, of terminal ji maxRepresenting the maximum transmit power, N, of the access point i on each subchannelmaxUsable subcarriers representing access pointCounting; wulAnd WdlRespectively representing the unit subcarrier bandwidths of the uplink and downlink channels,
Figure FDA0003206433280000032
and
Figure FDA0003206433280000033
representing the channel gain, N, on the uplink and downlink channels between access point i and terminal j0A power spectral density representing additive white gaussian noise;
Figure FDA0003206433280000034
indicating the delay of the traffic of terminal j processed locally,
Figure FDA0003206433280000035
represents the total delay, T, for terminal j to offload traffic to the edge serversIndicating the slot length.
4. The terminal-oriented multi-service model joint optimization method of claim 3, wherein the submodel P2 for joint optimization of access selection service and service offloading decisions for the terminal is:
Figure FDA0003206433280000036
Figure FDA0003206433280000037
Figure FDA0003206433280000038
Figure FDA0003206433280000039
Figure FDA00032064332800000310
the submodel P3 for optimizing the resource deployment policy is:
Figure FDA00032064332800000311
Figure FDA00032064332800000312
Figure FDA00032064332800000313
Figure FDA00032064332800000314
Figure FDA00032064332800000315
Figure FDA00032064332800000316
Figure FDA0003206433280000041
5. the terminal-oriented multi-service model joint optimization method of claim 4, wherein solving the submodel P3 comprises:
to the aboveThe submodel P3 is decomposed to make the variable flocAnd { τul,τdlF separation to obtain the frequency f used for calculating the terminallocLocal processing energy consumption minimization optimization model P31 for optimization, and unit bit transmission time tau for uplink channelulUnit bit transmission time tau of downlink channeldlA traffic offload energy consumption minimization optimization model P32 for performing joint optimization with the edge server computing frequency distribution variable f;
decomposing the constraint C10 into constraints
Figure FDA0003206433280000042
And constraint conditions
Figure FDA0003206433280000043
The constraint conditions of the local processing energy consumption minimization optimization model P31 comprise the constraint condition C11, and the constraint conditions of the traffic offload energy consumption minimization optimization model P32 comprise the constraint condition C12;
and respectively solving the local processing energy consumption minimization optimization model P31 and the service unloading energy consumption minimization optimization model P32 to obtain the unit bit transmission time of an uplink channel, the unit bit transmission time of a downlink channel, the terminal calculation frequency and the optimization result of the edge server calculation frequency distribution variable.
6. The terminal-oriented multi-service model joint optimization method of claim 5, wherein the local processing energy consumption minimization optimization model P31 is:
Figure FDA0003206433280000044
Figure FDA0003206433280000045
Figure FDA0003206433280000046
the traffic offload energy consumption minimization optimization model P32 is as follows:
Figure FDA0003206433280000047
Figure FDA0003206433280000048
Figure FDA0003206433280000049
Figure FDA0003206433280000051
Figure FDA0003206433280000052
Figure FDA0003206433280000053
wherein u isjk、cjkAnd djkRespectively representing the size of input data volume required for executing the service, the number of CPU cycles required for executing the service and the data volume of the service calculation result;
Figure FDA0003206433280000054
is a constant, k, associated with the chip type of terminal jiIs a constant related to the access point i chip type;
Figure FDA0003206433280000055
and
Figure FDA0003206433280000056
respectively representing the channel gains on an uplink channel and a downlink channel between a terminal j and an access point i;
Figure FDA0003206433280000057
and
Figure FDA0003206433280000058
respectively representing the single bit transmission time delay of uplink transmission and downlink transmission between the terminal j and the access point i.
7. The terminal-oriented multi-service model joint optimization method of claim 4, wherein solving the submodel P2 comprises:
(S0) taking a set of terminal access selection and traffic offload decisions as an individual, each individual corresponding to a double-stranded encoded chromosome, the double strands being a terminal access selection chain and a traffic offload decision chain, respectively;
(S1) generating an initial population using a random method, and calculating a fitness of each individual; the coding adopts a double-chain coding structure, the terminal access selection chain adopts an integer coding mode, and the service unloading decision chain adopts a binary coding mode; the fitness of the individual is the weighted energy consumption of the system corresponding to the individual;
(S2) selecting a part of the parent chromosomes, and updating the crossover probability p of the parent chromosomes according to the smaller fitness value f in the fitness values of the parent chromosomescAnd the resulting individual variation probability pmSo that the crossover probability p of the parent chromosomes is smallercAnd the resulting individual variation probability pmAre all smaller; according to the updated cross probability pcAnd the probability of variation pmPerforming crossing and mutation operations to generate a plurality of new individuals, and generating a new population by combining an elite retention strategy;
(S3) decoding each individual in the new population, calculating the fitness value of each individual, and if the maximum iteration number is not reached, turning to the step (S2); otherwise, the individual with the minimum fitness value is taken as a solution result, and the solution of the submodel P2 is finished.
8. The terminal-oriented multi-service model joint optimization method of claim 7, wherein in the step (S2), the crossover probability p of the parent chromosomes is updatedcAnd the resulting individual variation probability pmAnd satisfies the following conditions:
Figure FDA0003206433280000061
Figure FDA0003206433280000062
wherein f isminRepresenting the minimum value of fitness value in the population, favgRepresenting the average value of fitness values of the population; p is a radical ofcmaxAnd pcminRespectively representing the maximum and minimum of the cross probability, pmmaxAnd pmminRespectively representing the maximum and minimum of the mutation probability.
9. The method for joint optimization towards terminal multi-service model according to claim 7 or 8, characterized in that, in the step (S2), during crossing, the terminal access selection chain adopts two-point crossing, and the service offloading decision chain adopts uniform crossing.
10. A computer-readable storage medium comprising a stored computer program; the computer program, when executed by a processor, controls an apparatus on which the computer-readable storage medium is located to perform the joint optimization method for a terminal multi-service model according to any one of claims 1 to 9.
CN202110918183.5A 2021-08-11 2021-08-11 Joint optimization method for terminal multi-service model Pending CN113784372A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110918183.5A CN113784372A (en) 2021-08-11 2021-08-11 Joint optimization method for terminal multi-service model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110918183.5A CN113784372A (en) 2021-08-11 2021-08-11 Joint optimization method for terminal multi-service model

Publications (1)

Publication Number Publication Date
CN113784372A true CN113784372A (en) 2021-12-10

Family

ID=78837243

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110918183.5A Pending CN113784372A (en) 2021-08-11 2021-08-11 Joint optimization method for terminal multi-service model

Country Status (1)

Country Link
CN (1) CN113784372A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115565656A (en) * 2022-10-12 2023-01-03 广州智链科技有限公司 Bidirectional doctor seeing recommendation algorithm based on artificial intelligence

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109814951A (en) * 2019-01-22 2019-05-28 南京邮电大学 The combined optimization method of task unloading and resource allocation in mobile edge calculations network
CN111148134A (en) * 2019-12-19 2020-05-12 南京大学 Multi-user multi-task unloading method based on mobile edge calculation
CN113242568A (en) * 2021-06-08 2021-08-10 重庆邮电大学 Task unloading and resource allocation method in uncertain network environment

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109814951A (en) * 2019-01-22 2019-05-28 南京邮电大学 The combined optimization method of task unloading and resource allocation in mobile edge calculations network
CN111148134A (en) * 2019-12-19 2020-05-12 南京大学 Multi-user multi-task unloading method based on mobile edge calculation
CN113242568A (en) * 2021-06-08 2021-08-10 重庆邮电大学 Task unloading and resource allocation method in uncertain network environment

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
吴大鹏;吕吉;李职杜;王汝言;: "移动性感知的边缘服务迁移策略", 《通信学报》 *
闫伟;申滨;刘笑笑;: "基于自适应遗传算法的MEC任务卸载及资源分配", 《电子技术应用》, pages 4 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115565656A (en) * 2022-10-12 2023-01-03 广州智链科技有限公司 Bidirectional doctor seeing recommendation algorithm based on artificial intelligence

Similar Documents

Publication Publication Date Title
CN109814951B (en) Joint optimization method for task unloading and resource allocation in mobile edge computing network
CN112492626B (en) Method for unloading computing task of mobile user
Zhao et al. Energy-saving offloading by jointly allocating radio and computational resources for mobile edge computing
CN111585816B (en) Task unloading decision method based on adaptive genetic algorithm
CN111930436B (en) Random task queuing unloading optimization method based on edge calculation
CN109951869B (en) Internet of vehicles resource allocation method based on cloud and mist mixed calculation
CN111918339B (en) AR task unloading and resource allocation method based on reinforcement learning in mobile edge network
CN111372314A (en) Task unloading method and task unloading device based on mobile edge computing scene
CN111475274B (en) Cloud collaborative multi-task scheduling method and device
CN113296845B (en) Multi-cell task unloading algorithm based on deep reinforcement learning in edge computing environment
CN111010684B (en) Internet of vehicles resource allocation method based on MEC cache service
CN111132191A (en) Method for unloading, caching and resource allocation of joint tasks of mobile edge computing server
CN106791887B (en) Distributed caching and transmission optimization method for videos in wireless network
CN110392079A (en) The node calculating task dispatching method and its equipment calculated towards mist
CN111800812B (en) Design method of user access scheme applied to mobile edge computing network of non-orthogonal multiple access
CN112491957B (en) Distributed computing unloading method and system under edge network environment
CN112202847B (en) Server resource allocation method based on mobile edge calculation
CN110719641A (en) User unloading and resource allocation joint optimization method in edge computing
CN113573363A (en) MEC calculation unloading and resource allocation method based on deep reinforcement learning
CN115473896A (en) Electric power internet of things unloading strategy and resource configuration optimization method based on DQN algorithm
Dou et al. Mobile edge computing based task offloading and resource allocation in smart grid
CN112235387B (en) Multi-node cooperative computing unloading method based on energy consumption minimization
CN113784372A (en) Joint optimization method for terminal multi-service model
CN111580943B (en) Task scheduling method for multi-hop unloading in low-delay edge calculation
CN114615705B (en) Single-user resource allocation strategy method based on 5G network

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