A Resource Allocation Method for Multi-Carrier Non-Orthogonal Multiple Access-Mobile Edge Computing System

By optimizing user power and subcarrier allocation in mobile edge computing systems, combining Lagrangian dual and branch bounding algorithms, the high energy consumption problem of multi-carrier non-orthogonal multi-access system in large-scale multi-user scenarios is solved, and the system energy consumption reduction and resource allocation are achieved.

CN116390234BActive Publication Date: 2025-07-11WUHAN UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202310347077.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-03-30
Publication Date
2025-07-11
Estimated Expiration
2043-03-30

AI Technical Summary

Technical Problem

The existing multi-carrier non-orthogonal multi-access-mobile edge computing system is not reasonably allocated enough when dealing with large-scale multi-user scenarios, resulting in high energy consumption. Especially when the power of mobile devices is limited, it is difficult to effectively reduce system energy consumption.

Method used

By deploying edge servers in wireless cellular cells, using K subcarriers to serve N user equipment, it is divided into two stages: communication and computing, optimized user power allocation and user-to-subcarrier allocation, using Lagrangian dual method and KKT conditions, combining convex optimization tools and branch bounding algorithms, communication and computing time are dynamically adjusted to reduce total energy consumption.

Benefits of technology

It effectively reduces the total energy consumption of the system, improves the flexibility and universality of resource allocation, can handle large-scale multi-user scenarios, adapt to the energy consumption needs of different scenarios, and has a green energy-saving effect.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116390234B_ABST
    Figure CN116390234B_ABST
Patent Text Reader

Abstract

The present invention provides a resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system. The present invention considers the scenario where a single edge server node and multiple users upload data through multi-carrier non-orthogonal multiple access. First, an energy consumption optimization problem on a single sub-carrier with the user transmission power as a variable is constructed. Through the Lagrangian dual method and the KKT conditions, a relational expression of the optimal user power allocation within the sub-carrier with respect to the communication time is obtained. Then, the allocation of each user to the sub-carrier is characterized by binary decision variables, and the relationship between communication and computing time under optimal energy consumption is characterized by the proof by contradiction. Based on this, a system total energy consumption optimization problem with binary decision variables and communication time as variables is constructed. Finally, the optimization problem is solved based on the branch and bound algorithm, and then the optimal resource allocation method of the system is obtained. The present invention can effectively reduce the system energy consumption and has high universality, flexibility and practical application significance.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of mobile edge computing, and particularly relates to a resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system. Background Art

[0002] With the development of the Internet of Things and communication technologies, mobile edge computing has become an effective solution for mobile devices to handle complex computing tasks. Mobile devices can offload difficult computing tasks to nearby edge nodes through wireless channels for computing, thereby achieving low-energy-consumption and low-latency computing.

[0003] Meanwhile, as a new generation of multiple access technology, non-orthogonal multiple access (NOMA) technology allows multiple users to share the same communication resource block in a non-orthogonal manner, significantly increasing network throughput and spectrum efficiency, accommodating more users, and further improving the performance of wireless communication systems. Multi-carrier non-orthogonal multiple access is a form of implementation, where the sub-carriers of the system are orthogonal to each other, and several users perform power-domain non-orthogonal multiple access on each sub-carrier, which is an improvement over traditional multi-carrier communication systems. Therefore, the combination of multi-carrier non-orthogonal multiple access and mobile edge computing systems, namely non-orthogonal multiple access-mobile edge computing systems, is a promising paradigm in the field of wireless communication.

[0004] However, in future mobile edge computing systems, a large number of mobile user devices will access and demand services, posing greater challenges to limited communication and computing resources. In existing research on multi-carrier non-orthogonal multiple access, only the simple case of fixing 2 users per carrier for NOMA has been considered, which is not conducive to serving more users with limited resources. In addition, in different scenarios, the energy consumption of mobile users and edge servers has different weights. For example, when the power of mobile devices is limited while the power of edge servers is sufficient, reducing the energy consumption of the mobile terminal is more important. Therefore, it is necessary to reasonably design resource allocation to meet the requirements in actual scenarios. Summary of the Invention

[0005] For the non-orthogonal multiple access-mobile edge scenario with a large number of multi-users in the future Internet of Things, to solve the above problems, the present invention aims to provide a resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system, aiming to reduce the energy consumption of the multi-carrier non-orthogonal multiple access-mobile edge computing system.

[0006] The present invention provides a resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system, including the following steps:

[0007] Step 1: In a wireless cell, deploy the edge server at the wireless base station. Meanwhile, serve N user equipments through K subcarriers, and divide an edge computing service frame into two phases: a communication phase and a computing phase;

[0008] Step 2: For the user power allocation on a single subcarrier in the communication phase, take the sum of the communication energy consumption of all users within the subcarrier as the optimization objective, and obtain the expression of the user rate with respect to its power according to the principle of NOMA; all users need to transmit all data within the communication time. Based on this, construct an energy consumption optimization problem on a single subcarrier with the user transmit power as the variable, and obtain the relationship between the optimal user power allocation within the subcarrier and the communication time through the Lagrangian dual method and the KKT conditions;

[0009] Step 3: On the basis of Step 2, for the allocation between users and subcarriers in the communication phase and the allocation of communication-computing time during the service process, characterize the allocation of each user to the subcarrier through binary decision variables, and then use the proof by contradiction to describe the relationship between communication and computing time under the optimal energy consumption. Based on this, construct a system total energy consumption optimization problem with binary decision variables and communication time as variables;

[0010] Step 4: For the optimization problem described in Step 3, verify its convexity under the given binary decision variables through the second derivative, and then use the branch-and-bound based algorithm to solve the optimization problem to obtain the optimal allocation between users and subcarriers and the communication time allocation. Furthermore, use the relationship between the user power allocation within a single subcarrier and the communication time and the relationship between communication and computing time under the optimal energy consumption to obtain the optimal user power allocation and communication-computing time allocation.

[0011] Furthermore, the N user equipments described in Step 1 are defined as a set The K subcarriers are defined as a set Each of the N user equipments collects a set of data with a size of D bits, and then uploads it to the edge server through multi-carrier non-orthogonal multiple access uplink transmission. After receiving the data of all users, the edge server uses this data to calculate a specific mobile task, and the computing load is c bits.

[0012] Furthermore, in the communication phase, all user equipments send the data required for computing to the edge server through uplink NOMA; in the computing phase, after the edge server receives the data sent by all users, it uses the collected user data to calculate the mobile task. The total delay constraint of the entire service process composed of the communication phase and the computing phase is denoted as T max , the time used in the communication phase is denoted as t r , the time used in the computing phase is denoted as t c .

[0013] Furthermore, the users on the single sub - carrier described in step 2 are denoted as where M k is the number of users allocated to this sub - carrier, and K is the index of this sub - carrier in the set ; the transmission power of any user m is denoted as p m , and the ratio of the channel gain to the power of additive white Gaussian noise on this sub - carrier is denoted as h m,k . Without loss of generality, the users in m,k are arranged in ascending order of the value of h m,k . The users on this sub - carrier transmit data through NOMA uplink. Therefore, the transmission rate of user m is

[0014]

[0015] where B is the sub - carrier bandwidth, p i (i = 1,..., m - 1) is the transmission power of interfering users, and h i,k is the ratio of the channel gain of interfering users to the power of additive white Gaussian noise on this sub - carrier. Then the communication energy consumption of user m is p m t r , and the communication energy consumption of all users on this sub - carrier is

[0016]

[0017] The amount of data transmitted by each user during the communication time needs to be greater than or equal to the amount of data collected, that is, R m t r ≥D.

[0018] Furthermore, the optimization problem within this sub - carrier is:

[0019]

[0020] where P max is the upper limit of the total transmit power of users within the sub - carrier.

[0021] Furthermore, when t r is given, the objective function is an affine function, and the second constraint condition is equivalent to

[0022]

[0023] which is also a convex constraint. It is easy to see that other constraint conditions are also convex constraints. Therefore, this problem is a convex optimization problem, and its Lagrangian dual function is

[0024]

[0025] According to the KKT conditions, we have

[0026]

[0027] It can be derived by combining the constraint conditions

[0028]

[0029] Furthermore, the optimal power allocation of user m within the subcarrier with respect to the communication time t r has the following relationship

[0030]

[0031] where p i , p j are the transmission powers of users i and j respectively, h j,k is the ratio of the channel gain of user j on this subcarrier to the additive white Gaussian noise power, and δ i , μ, λ i are all Lagrange multipliers.

[0032] Furthermore, the allocation between users and subcarriers in step 3 is characterized by an n×k-sized matrix Φ composed of binary variables, and its element φ n,k ∈{0, 1}, where 1 indicates that user n is allocated to subchannel k, and 0 indicates the opposite. Accordingly, the system communication energy consumption can be expressed as a function of Φ and t r ;

[0033] The specific process of communication-computation time allocation is that the energy consumption of the server is denoted as E c , which is given by the formula E c =κc 3 / (t c ), where κ is a constant determined by the hardware and c is the server computing load. It is proved by contradiction that when the system energy consumption is optimal, the communication time and the computation time satisfy the following relationship 2 t

[0034] t r +t c =T max

[0035] Therefore, the computation time can be expressed in terms of the communication time as t c =T max -t r . The total system energy consumption is defined as the weighted sum of the communication energy consumption and the computation energy consumption, with weights α and β respectively, and the weights can be flexibly adjusted according to different actual scenarios. Accordingly, the total optimization problem can be written as

[0036]

[0037] where M maxThe maximum allowable number of users for NOMA on each subcarrier.

[0038] Further, the process of verifying convexity described in step 4 is to treat all elements in Φ as constants, calculate the second derivative of the total optimization problem with respect to t r , and it is easy to verify that this second derivative is always greater than 0 within the feasible region of t r . Therefore, with Φ fixed, the total optimization problem is a convex optimization problem with respect to t r , and the optimal value of t r can be effectively solved by convex optimization tools.

[0039] Further, the steps of using the branch-and-bound-based algorithm to solve the optimization problem are as follows: First, find an arbitrary feasible solution (Φ, t r ) of the problem and calculate the objective function value corresponding to this feasible solution as the initial solution of the problem. Denote the allocation matrix and communication time represented by the initial solution as Φ init , t r init , and the corresponding objective function value as the initial upper bound of the problem, denoted as UB init . Take the initial upper bound as the current global upper bound; at the same time, initialize a queue of subproblems, denoted as L, and insert the above total optimization problem as the initial problem into this queue. When L is not empty, continuously select a subproblem from it, denoted as Perform branching and bounding on the binary variable matrix Φ of this subproblem until L is empty.

[0040] Further, the branching operation is as follows: Assume that the first,..., n - 1 rows of the matrix have been given by previous branching operations. Then, for the nth row of the matrix, let the kth (k = 1, 2,..., K) element of this row be 1 and the other elements in this row be set to 0. For the n + 1,..., N rows of the matrix, set these rows to undetermined status. In this way, K branch subproblems can be generated and all generated subproblems are inserted into L;

[0041] The bounding operation is to find a feasible solution of a problem Calculate the optimal value of the objective function with respect to t corresponding to this given feasible solution r as the upper bound of this subproblem. At the same time, retain all the fixed elements in the matrix Φ of the current subproblem and set the undetermined elements to 0 to obtain the matrix Calculate the optimal value of the objective function with respect to t corresponding to this given matrix r as the lower bound of this subproblem. Subsequently, judge whether to prune according to the calculated upper and lower bounds;

[0042] The pruning operation is as follows: if the constraint condition is not satisfied, the sub-problem has no solution, or the lower bound of the sub-problem is greater than the global upper bound, then the sub-problem is discarded and the next loop is directly entered. If the sub-problem cannot be pruned, it is judged according to the calculated upper bound. If the upper bound is less than the current global upper bound, the global upper bound is updated to the current upper bound, and the optimal solution of the problem is updated to the one corresponding to this upper bound and this optimal solution of the objective function with respect to communication time under the following After the judgment, the sub-problem is removed from the queue and the next loop is entered. Repeat this process until the loop ends, and output the obtained optimal user-to-subcarrier allocation strategy and optimal communication time. Accordingly, the final user power allocation, communication-computation time allocation, and user-to-subcarrier allocation strategy can be obtained according to the optimal communication-computation time relationship described in Step 3 and the relationship between the optimal power allocation within the subcarrier with respect to communication time described in Step 2

[0043] Compared with the prior art, the advantages and positive effects of the present invention are as follows

[0044] (1) It provides methods for user transmission power allocation, user-to-subchannel allocation, and communication-computation time resource allocation in a non-orthogonal multiple access-mobile edge computing system, and can effectively reduce the total system energy consumption under the condition of completing data transmission, with the effects of green energy saving and low-carbon environmental protection

[0045] (2) It considers any number of sub-channels and the case where the number of non-orthogonal multiple access users on a single sub-channel is arbitrary, and can handle resource allocation problems in large-scale multi-user non-orthogonal multiple access-mobile edge computing scenarios, with high universality

[0046] (3) By introducing the weights of communication energy consumption and computing energy consumption, it can adaptively adjust the resources allocated to mobile devices and edge server devices according to different scenarios, so as to reduce the energy consumption of high-energy consumption weight devices as much as possible in specific scenarios, with high flexibility BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a schematic diagram of the system scenario of a resource allocation method for a low-energy multi-carrier non-orthogonal multiple access-mobile edge computing system provided by an embodiment of the present application

[0048] Figure 2 is a flowchart of a resource allocation method for a low-energy multi-carrier non-orthogonal multiple access-mobile edge computing system provided by an embodiment of the present application

[0049] Figure 3 is a simulation relationship diagram of weighted energy consumption and the number of users under different time limits for a resource allocation method for a low-energy multi-carrier non-orthogonal multiple access-mobile edge computing system provided by an embodiment of the present application Detailed implementation manners

[0050] The present invention aims to provide a resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system, aiming to reduce the energy consumption of the multi-carrier non-orthogonal multiple access-mobile edge computing system.

[0051] The following further describes the present invention with reference to the accompanying drawings.

[0052] Figure 1 It is a schematic diagram of the system scenario of a low-energy multi-carrier non-orthogonal multiple access-mobile edge computing system resource allocation method provided by an embodiment of the present application. In a wireless cellular cell, an edge server is deployed at a wireless base station, and at the same time, N user equipments are served through K sub-carriers. An edge computing service frame is divided into two stages. The first stage is the communication stage, in which all user equipments send the data required for computing to the edge server through uplink NOMA. The second stage is the computing stage, in which after the edge server receives the data sent by all users, it uses these data to compute the mobile task;

[0053] As shown in Figure 2 Preferably, the N user equipments described in step 1 are defined as a set The K sub-carriers are defined as a set Each of all N user equipments collects a group of data with a size of D bits, and then transmits it to the edge server through multi-carrier non-orthogonal multiple access uplink. After receiving the data of all users, the edge server uses these data to compute a specific mobile task, and the computing load is c bits. The total delay constraint of the entire service process composed of the communication stage and the computing stage is denoted as T max , the time used in the communication stage is denoted as t r , the time used in the computing stage is denoted as t c .

[0054] Preferably, the user on a single sub-carrier described in step 2 is denoted as where M k is the number of users allocated to this sub-carrier, and k is the index of this sub-carrier in the set . Denote the transmission power of any user m as p m , and the ratio of the channel gain to the power of additive white Gaussian noise on this sub-carrier is denoted as h m,k . Without loss of generality, arrange the users in in ascending order of the value of h m,k . The users on this sub-carrier transmit data through NOMA uplink, so the transmission rate of user m is

[0055]

[0056] where \(B\) is the sub - carrier bandwidth, \(p\) i (\(i = 1,\cdots,m - 1\)) is the transmit power of interfering users, \(h\) i,k is the ratio of the channel gain of the interfering user on this sub - carrier to the power of additive white Gaussian noise. Then the communication energy consumption of user \(m\) is \(p\) m t r , and the communication energy consumption of all users on this sub - carrier is

[0057]

[0058] The amount of data transmitted by each user during the communication time needs to be greater than or equal to the amount of data collected, that is, \(R\) m t r \(\geq D\). Based on this, the optimization problem within this sub - carrier can be written as:

[0059]

[0060] where \(P\) max is the upper limit of the total transmit power of users within the sub - carrier.

[0061] When \(t\) r is given, the objective function is an affine function, and the second constraint condition is equivalent to

[0062]

[0063] which is also a convex constraint. It is easy to see that other constraint conditions are also convex constraints. Therefore, this problem is a convex optimization problem, and its Lagrangian dual function is

[0064]

[0065] According to the KKT conditions, we have

[0066]

[0067] Combined with the constraint conditions, it can be deduced that

[0068] Furthermore, the relationship between the optimal power allocation of user \(m\) within the sub - carrier with respect to the communication time \(t\) r is

[0069]

[0070] where \(p\) i , \(p\) j are the transmit powers of users \(i\) and \(j\), \(h\) j,k is the ratio of the channel gain of user \(j\) on this sub - carrier to the power of additive white Gaussian noise, \(\delta\) i , \(\mu\), \(\lambda\) i are Lagrange multipliers.

[0071] Preferably, the allocation between the user and the subcarrier in step 3 is characterized by an n×k-sized matrix Φ composed of binary variables, and its element φ n,k ∈{0,1}, where 1 indicates that user n is allocated to subchannel k, and 0 indicates the opposite. Accordingly, the system communication energy consumption can be expressed as a function of Φ and t r . The specific communication-computation time allocation is that the server computing energy consumption is given by E c = κc 3 / (t c ), where κ is a constant determined by the hardware and c is the server computing load. It is proved by contradiction that when the system energy consumption is optimal, the communication time and the computing time satisfy the following relationship 2

[0072] t r + t c = T max

[0073] Thus, the computing time can be expressed in terms of the communication time as t c = T max - t r . The total system energy consumption is defined as the weighted sum of the communication energy consumption and the computing energy consumption, with weights α and β respectively, and the weights can be flexibly adjusted according to different actual scenarios. Accordingly, the total optimization problem can be written as

[0074]

[0075] where M max is the maximum allowable number of users for NOMA on each subcarrier.

[0076] Preferably, the verification of convexity in step 4 is to regard all elements in Φ as constants, calculate the second derivative of the total optimization problem with respect to t r . It is easy to verify that this second derivative is always greater than 0 within the feasible region of t r . Therefore, when Φ is fixed, the total optimization problem with respect to t r is a convex optimization problem, and the optimal value of t r can be effectively solved by convex optimization tools. The steps of the branch-and-bound-based algorithm are as follows: First, find an arbitrary feasible solution (Φ, t r ) of the problem and calculate the objective function value corresponding to this feasible solution as the initial solution Φ init , t r init , and the corresponding objective function value as the initial upper bound UB init ​, the initial upper bound is used as the current global upper bound. At the same time, a queue L of sub-problems is initialized, and the above total optimization problem is inserted into this queue as the initial problem. When L is not empty, continuously select a sub-problem from it Perform branching and bounding on the binary variable matrix Φ of this sub-problem until L is empty. The branching operation is as follows: Assume that the first,..., n - 1 rows of the matrix have been determined by previous branching operations. Then, for the n-th row of the matrix, let the k-th (k = 1, 2,..., K) element of this row be 1, and the other elements in this row be set to 0. For the (n + 1)-th,..., N-th rows of the matrix, these rows are set to undetermined status. In this way, K branch sub-problems can be generated, and all generated sub-problems are inserted into L. The bounding operation is to find a feasible solution of a problem Calculate the given feasible solution When given, the optimal value of the objective function with respect to t r is used as the upper bound of this sub-problem. At the same time, all the fixed elements in the matrix Φ of the current sub-problem are retained, and the undetermined elements are set to 0 to obtain the matrix Calculate this matrix When given, the optimal value of the objective function with respect to t r is used as the lower bound of this sub-problem. Subsequently, it is judged whether to prune according to the calculated upper and lower bounds. The pruning operation is as follows: If the constraint conditions are not met, the sub-problem has no solution, or the lower bound of this sub-problem is greater than the global upper bound, then this sub-problem is discarded and directly enter the next loop. If this sub-problem cannot be pruned, it is judged according to the calculated upper bound. If this upper bound is less than the current global upper bound, then the global upper bound is updated to the current upper bound, and the optimal solution of the problem is updated to the one corresponding to this upper bound and this The optimal solution of the objective function with respect to the communication time under this After the judgment, this sub-problem is removed from the queue and enter the next loop. Repeat this way until the loop ends, and output the obtained optimal user-to-subcarrier allocation strategy and optimal communication time. Accordingly, the final user power allocation, communication-computation time allocation, and user-to-subcarrier allocation strategy can be obtained according to the optimal communication-computation time relationship described in step 3 and the relationship between the optimal power allocation within the subcarrier with respect to the communication time described in step 2.

[0077] Figure 3 It is a schematic diagram of the system weighted total energy consumption varying with the number of users under different service delay constraints. It can be seen that as the number of users increases, the system weighted total energy consumption gradually increases. And as the service delay constraint becomes shorter, the system weighted total energy consumption increases.

[0078] It should be understood that the above description of the preferred embodiment is relatively detailed, and it should not be considered as a limitation to the protection scope of the present invention patent. Under the inspiration of the present invention, those of ordinary skill in the art can also make substitutions or deformations without departing from the protection scope defined by the claims of the present invention, and all of them fall within the protection scope of the present invention. The scope of protection claimed by the present invention shall be subject to the appended claims.

Claims

1. A resource allocation method for a multi - carrier non - orthogonal multiple access - mobile edge computing system, characterized in that It includes the following steps: Step 1: In a wireless cell, deploy the edge server at the wireless base station. Meanwhile, serve subcarriers to user equipments, and divide an edge computing service frame into two phases: a communication phase and a computing phase. Step 2: For the user power allocation on a single subcarrier in the communication phase, the sum of the communication energy consumption of all users within the subcarrier is used as the optimization objective, and an expression of the rate of a user with respect to its power is obtained according to the principle of NOMA; All users need to transmit all data within the communication time. Based on this, an energy consumption optimization problem on a single subcarrier with the user transmit power as the variable is constructed, and a relational expression of the optimal user power allocation within the subcarrier with respect to the communication time is obtained through the Lagrange dual method and the KKT conditions; Step 3: On the basis of Step 2, for the allocation between users and subcarriers in the communication phase and the allocation of communication-computation time during the service process, the allocation of each user to a subcarrier is characterized by a binary decision variable, and then the relationship between communication and computation time under the optimal energy consumption is characterized by proof by contradiction. Based on this, a system total energy consumption optimization problem with the binary decision variable and the communication time as variables is constructed; Step 4: For the optimization problem described in Step 3, its convexity under the given binary decision variable is verified through the second derivative, and then an algorithm based on branch and bound is used to solve the optimization problem to obtain the optimal allocation between users and subcarriers and the communication time allocation. Furthermore, the optimal user power allocation and communication-computation time allocation are obtained by using the relationship between the user power allocation within a single subcarrier and the communication time and the relationship between communication and computation time under the optimal energy consumption; 2. The resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system according to claim 1, wherein As described in Step 1 user equipment are defined as a set , sub - carriers are defined as a set . All user equipment each collect a set of data with a size of bits, and then transmit it to the edge server through multi - carrier non - orthogonal multiple access uplink. After receiving the data of all users, the edge server uses this data to calculate a specific mobile task, and the computational load is bits.

3. A resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system according to claim 2, characterized in that During the communication phase, all user devices will calculate the required data and send it to the edge server via uplink NOMA. After receiving the data sent by all users during the computing phase, the edge server uses the collected user data to calculate the mobile tasks. The total delay constraint of the entire service process composed of the communication phase and the computing phase is denoted as , and the time used in the communication phase is denoted as , and the time used in the computing phase is denoted as .

4. A resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system according to claim 3, wherein The user on the single subcarrier described in Step 2 is denoted as , where is the number of users allocated to this subcarrier, is the index of this subcarrier in the set ; Denote the transmission power of any user as , and denote the ratio of the channel gain to the power of additive white Gaussian noise on this subcarrier as . Without loss of generality, arrange the users in in ascending order according to the value of . The users on this subcarrier transmit data through NOMA uplink. Therefore, according to the Shannon formula, the transmission rate of user is wherein is the subcarrier bandwidth, is the transmit power of the interfering user, is the ratio of the channel gain of the interfering user on this subcarrier to the power of additive white Gaussian noise, then the communication energy consumption of user is , and the communication energy consumption of all users on this subcarrier is , The amount of data transmitted by each user during the communication time needs to be greater than or equal to the amount of data collected, that is .

5. A resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system according to claim 4, characterized in that The optimization problem within the subcarrier: Among them, is the upper limit of the total transmit power of users within a subcarrier.

6. The resource allocation method of a multi-carrier non-orthogonal multiple access-mobile edge computing system according to claim 5, characterized in that When is given, the objective function is an affine function, and the second constraint is equivalent to is also a convex constraint. It is easy to see that other constraint conditions are also convex constraints. Therefore, this problem is a convex optimization problem, and its Lagrange dual function is According to the KKT conditions, there is Combined with the constraint conditions, it can be deduced that Further obtain the users within a subcarrier The optimal power allocation with respect to the communication time The relational expression is Among them, , are respectively the transmission power of the user , is the ratio of the channel gain of the user on this subcarrier to the additive white Gaussian noise power, , , are all Lagrange multipliers.

7. A resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system according to claim 6, characterized in that The allocation between the user and the subcarrier described in step 3 consists of a matrix of size characterized by elements where a value of 1 indicates that the user is assigned to the subchannel and a value of 0 indicates otherwise. Accordingly, the system communication energy consumption can be expressed as a function; The specific communication-computation time allocation process is that the server calculates the energy consumption, denoted as , which is given by the formula , where is a constant determined by the hardware, and is the server's computing load. It is proved by contradiction that when the system energy consumption is optimal, the communication time and the computing time satisfy the following relationship Thus, the computing time can be represented by the communication time as , the total system energy consumption is defined as the weighted sum of the communication energy consumption and the computing energy consumption, and the weights are respectively , and the weights can be flexibly adjusted according to different actual scenarios. Based on this, the total optimization problem can be written as Among them is the maximum allowable number of users for NOMA on each subcarrier.

8. A resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system according to claim 7, characterized in that The verification convexity process described in step 4 is to regard all elements in as constants, calculate the second derivative of the total optimization problem with respect to , and it is easy to verify that this second derivative is always greater than 0 within the feasible region. Therefore, when is fixed, the total optimization problem with respect to is a convex optimization problem and can be effectively solved by convex optimization tools to obtain the optimal value of .

9. A resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system according to claim 7, characterized in that The steps to solve the optimization problem using a branch-and-bound based algorithm are as follows: First, find an arbitrary feasible solution to the problem and calculate the objective function value corresponding to this feasible solution as the initial solution to the problem. Denote the allocation matrix and communication time represented by the initial solution as , and the corresponding objective function value as the initial upper bound of the problem, denoted as . Take the initial upper bound as the current global upper bound; at the same time, initialize a queue of sub-problems, denoted as . Insert the above total optimization problem as the initial problem into this queue. When is not empty, continuously select a sub-problem from it, denoted as , and perform branching and bounding on the binary variable matrix of this sub-problem until is empty.

10. A resource allocation method for a multi-carrier non-orthogonal multiple access-mobile edge computing system according to claim 9, characterized in that, The branching operation is as follows: Assume that the rows of the matrix have been given by previous branching operations. Then, for the th row of the matrix, let the -th ( ) element of this row be 1 and set the other elements in this row to 0. For the th row of the matrix, set these rows to the undetermined state. In this way, branching sub-problems can be generated, and all the generated sub-problems are inserted into . The delimiter operation is to find a feasible solution to a problem , calculate the objective function corresponding to this feasible solution when given, and use the optimal value with respect to as the upper bound of this sub-problem. At the same time, retain all the fixed elements in the matrix of the current sub-problem, set the undetermined elements to 0, and obtain the matrix . Calculate the optimal value of the objective function corresponding to this matrix when given, and use the optimal value with respect to as the lower bound of this sub-problem. Then, determine whether to prune according to the calculated upper and lower bounds; The pruning operation is as follows: if the constraint condition is not satisfied, the sub-problem has no solution, or the lower bound of the sub-problem is greater than the global upper bound, then the sub-problem is discarded and the next loop is directly entered. If the sub-problem cannot be pruned, it is judged according to the calculated upper bound. If the upper bound is less than the current global upper bound, the global upper bound is updated to the current upper bound, and the optimal solution of the problem is updated to the one corresponding to this upper bound and this optimal solution of the objective function with respect to the communication time . After the judgment, the sub-problem is removed from the queue and the next loop is entered. Repeat this process until the loop ends, and output the obtained optimal user-to-subcarrier allocation strategy and the optimal communication time; Based on this, the final user power allocation, communication-computation time allocation, and allocation strategy between users and subcarriers can be obtained according to the optimal communication-computation time relationship described in Step 3 and the relationship between the optimal power allocation within the subcarrier and the communication time described in Step 2.

Citation Information

Patent Citations

  • A power distribution method based on a multi-user non-orthogonal multiple access technology

    CN109699083A

  • Energy-saving resource allocation method based on SCMA system

    CN111315017A