A method and system for jointly optimizing computing offloading and resource allocation

Abstract

The present invention discloses a method and system for joint optimization of computing offloading and resource allocation, which uses mobile edge computing technology to offload computing tasks from IoT devices with computing tasks to idle devices and MEC servers for execution, thereby reducing the computing burden of active users; uses wireless power supply technology to power IoT devices; and uses user collaboration technology to allow IoT devices in idle states to share their computing resources, thereby improving the performance of active users. The mobile edge computing technology, wireless power supply technology, and user collaboration technology are organically combined in a modeling manner, and reasonable resource allocation and user pairing are performed in the network system in a joint optimization solution manner to form a resource allocation strategy and a user pairing strategy, thereby ensuring that IoT devices complete computing tasks within the maximum allowable delay, and at the same time, the energy consumption of IoT devices is further realized through reasonable resource optimization configuration and user pairing.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless communication networks, and in particular to a method and system for jointly optimizing computation offloading and resource allocation. Background Art

[0002] With the rapid development of the Internet of Things, sensors, cameras, wearable devices, etc. are able to perform data collection and data exchange tasks. However, with the continuous increase of mobile IoT devices, various IoT applications such as face recognition, online games, virtual reality, and e-health have explosively occupied people's lives.

[0003] On the one hand, due to the limited battery storage capacity, transmission conditions and computing power of IoT devices, it is difficult for them to complete computing tasks within the maximum allowable delay. Therefore, IoT devices can use mobile edge computing technology to offload part or all of the computing tasks to access points deployed with MEC servers while ensuring the quality of service of their applications. At the same time, computing offload consumes the energy of IoT devices, so wireless power supply technology can be used to charge IoT devices.

[0004] On the other hand, the explosive increase in IoT devices may lead to the exhaustion of computing resources. Even MEC servers with strong computing power can hardly meet the huge computing resource requirements. In addition, in the IoT, some IoT devices are idle most of the time, and their computing resources are not fully utilized. Therefore, user collaboration technology can be used to allow idle IoT devices to share their computing resources, thereby improving the performance of IoT devices with computing tasks.

[0005] In view of this, this application is hereby filed. Summary of the invention

[0006] The technical problem to be solved by the present invention is that in the current Internet of Things, it is difficult for IoT devices to complete computing tasks within the maximum allowable delay. The purpose is to provide a method and system for jointly optimizing computing offloading and resource allocation, by combining mobile edge computing technology, wireless power supply technology and user collaboration technology, to perform resource allocation and user pairing between IoT devices with computing tasks and idle devices and MEC servers, to complete the offloading of computing tasks to idle devices and MEC servers for execution, so as to enable IoT devices to complete computing tasks within the maximum allowable delay and further reduce the energy consumption of IoT devices.

[0007] The present invention is achieved through the following technical solutions:

[0008] In one aspect, the present invention provides a method for jointly optimizing computation offloading and resource allocation, comprising the following steps:

[0009] Build a wireless power supply mobile edge computing network system based on user collaboration; under the network system,

[0010] Establish a user pairing model to characterize the pairing relationship between active users and idle users;

[0011] Establishing an energy collection model for characterizing the energy collected by the active user and the idle user from the energy station;

[0012] Establishing a local computing model for characterizing the local execution delay and local execution energy consumption of the active user, and the local execution delay and local execution energy consumption of the idle user;

[0013] Establishing a computing offloading model to characterize the energy consumption and data transmission rate of the active user offloading the computing task to the idle user, and the energy consumption and data transmission rate of offloading the computing task to the MEC server;

[0014] Establishing a minimum energy consumption model of the network system according to the user pairing model, the energy collection model, the local computing model and the computing offloading model;

[0015] The minimum energy consumption model is optimized for resource allocation and optimal user matching to obtain the minimum energy consumption of the network system.

[0016] As a further description of the present invention, the network system includes: an energy station, an active user, an idle user and a MEC server; the energy station supplies power to the active user and the idle user through a wireless power supply link; the active user offloads computing tasks to the idle user through a D2D link; the active user offloads computing tasks to the MEC server through a cellular link.

[0017] As a further description of the present invention, the user pairing model is: i,j ∈{0,1}, Among them, α i,j represents the pairing coefficient between the i-th active user and the j-th idle user, M represents the set of active users, and K represents the set of idle users.

[0018] As a further description of the present invention, the energy harvesting model is: in, represents the energy collected by the i-th active user from the energy station, represents the energy collected by the jth idle user from the energy station, N represents the number of channels, t0 represents the duration of the power supply phase, η represents the energy conversion efficiency of the energy station supplying power to active users or idle users, P b,nrepresents the transmission power from the energy station to active users or idle users, h e,i represents the channel gain from the energy station to the active user, h e,j Represents the channel gain from the energy station to the idle user.

[0019] As a further description of the present invention, the local execution delay of the active user is: T i loc ≤T; where represents the number of bits of computing tasks completed locally by the i-th active user, C i represents the number of CPU cycles required for the i-th active user to complete the computing task locally, f i represents the computing power of the i-th active user; T i loc ≤T is the local execution delay constraint of the i-th active user, where T represents the maximum tolerable delay of the network system;

[0020] The local execution energy consumption of the active user is: Among them, α i,j represents the pairing coefficient between the i-th active user and the j-th idle user, κ i represents the energy conversion coefficient;

[0021] The local execution delay of the idle user is: in, represents the number of bits of computing tasks completed locally by the jth idle user, C j represents the number of CPU cycles required for the jth idle user to complete the computing task, f j represents the computing power of the jth idle user; The local execution delay constraint for the jth idle user;

[0022] The local execution energy consumption of the idle user is: Among them, κ j Represents the energy conversion coefficient.

[0023] As a further description of the present invention, the energy consumption of offloading the computing task to the idle user is: Among them, α i,j represents the pairing coefficient between the i-th active user and the j-th idle user, t i,j represents the time required to offload computing tasks from active users to idle users, p i,j represents the link transmission power from active users to idle users; Constraints that need to be met in order to offload the computing task to the idle user;

[0024] The data transmission rate of offloading the computing task to the idle user is: Where B represents the channel bandwidth, g i,j represents the channel state information between active users and idle users, p i,j represents the link transmission power from active users to idle users, N O represents the noise power generated in the channel between active users and idle users;

[0025] The energy consumption of offloading the computing task to the MEC server is: Among them, t i,a represents the time required to offload computing tasks from active users to MEC servers, p i,a represents the link transmission power for offloading computing tasks from active users to MEC servers; Constraints that need to be met in order to offload computing tasks to the MEC server;

[0026] The data transmission rate for offloading the computing task to the MEC server is: Among them, g i,a represents the channel state information between active users and MEC servers, p i,a represents the link transmission power from active users to MEC servers, N O represents the noise power generated in the channel between active users and the MEC server.

[0027] As a further description of the present invention, the minimum energy consumption model of the network system is:

[0028]

[0029]

[0030] in, p=[p i,j ,p i,a ]f=[f i ,f j ],t=[t0,t i,j ,t i,a ], f i max represents the maximum computing power of active users, Indicates the maximum computing capacity of idle users.

[0031] As a further description of the present invention, the method for optimizing resource allocation for the minimum energy consumption model is:

[0032] A1: Establish an optimized resource allocation model;

[0033] A2: Using the Lagrangian dual method to lock the power allocation variables and computing capacity allocation variables in the resource allocation model, an equivalent optimized resource allocation model is obtained;

[0034] A3: Using the CVX toolkit to lock the computing task allocation variables and the time allocation variables in the resource allocation model, and obtain an equivalent optimized resource allocation model;

[0035] A4: Using the block coordinate descent method to solve the equivalent optimal resource allocation model, an energy consumption matrix of the network system is obtained, which is used to represent the energy consumption of computing tasks from any active user to any idle user in the network system.

[0036] The method for performing optimal user matching on the minimum energy consumption model is:

[0037] B1: Establish the optimal user matching model;

[0038] B2: Using the Hungarian method and the energy consumption matrix, the optimal user matching model is solved to obtain the minimum energy consumption of the integrated system.

[0039] As a further description of the present invention, the A2 comprises:

[0040] A21: Establishing a Lagrangian function according to the equivalent optimal resource allocation model;

[0041] A22: Establishing the dual function of the Lagrangian function;

[0042] A23: Establishing a dual model with the goal of maximizing the dual function;

[0043] A24: Solve the dual model to obtain the computing task allocation variable and the time allocation variable;

[0044] A25: According to the computing task allocation variable and the time allocation variable, the multiplier of the Lagrangian function, the computing power of the active user, the computing power of the idle user, the link transmission power of offloading the computing task to the idle user, and the link transmission power of offloading the computing task to the MEC server are iteratively updated using a subgradient method until a convergence condition is met, thereby obtaining an optimal solution to the optimization resource allocation model;

[0045] A26: According to A1 to A25, the energy consumption of the computing task from any active user to any idle user is obtained based on the block coordinate descent method to form an energy consumption matrix of the network system.

[0046] In another aspect, the present invention provides a system for jointly optimizing computation offloading and resource allocation, comprising:

[0047] Network system building module, used to build a wireless power supply mobile edge computing network system based on user collaboration;

[0048] Model building module, used to build:

[0049] User pairing model, which represents the pairing relationship between active users and idle users;

[0050] Energy harvesting model, characterizing the energy harvested from the energy station by active and idle users;

[0051] A local computation model that characterizes the local execution delay and local execution energy consumption of active users, as well as the local execution delay and local execution energy consumption of idle users;

[0052] The computation offloading model characterizes the energy consumption and data transmission rate of computation tasks from active users to idle users, and the energy consumption and data transmission rate of computation tasks from active users to MEC servers;

[0053] Minimum energy consumption model, characterizing the minimum energy consumption of the network system;

[0054] An optimization resource allocation module, used for optimizing resource allocation for the minimum energy consumption model;

[0055] The optimal user matching module is used to perform optimal user matching on the minimum energy consumption model to obtain the minimum energy consumption of the network system.

[0056] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0057] (1) Through mobile edge computing technology and user collaboration technology, resource allocation and user pairing are performed between IoT devices with computing tasks and idle devices and MEC servers, so that part or all of the computing tasks can be offloaded to idle devices and MEC servers for execution, thereby reducing the computing pressure of IoT devices with computing tasks and enabling them to complete computing tasks within the maximum allowable delay;

[0058] (2) By using user collaboration technology to offload computing tasks to idle IoT devices for execution, the computing resources of IoT devices can be fully utilized in the IoT, solving the problem of wasting existing computing resources.

[0059] (3) Using wireless power supply technology to power IoT devices can make up for the limited energy storage capacity of existing IoT devices;

[0060] (4) Combining mobile edge computing technology, wireless power supply technology and user collaboration technology, it realizes the reasonable allocation of resources in the Internet of Things and solves the problem that it is difficult for IoT devices to complete computing tasks within the maximum allowable delay, while further reducing the energy consumption of IoT devices;

[0061] (5) The coordinate descent algorithm, Lagrange dual method and Hungarian algorithm are used to jointly optimize resource pairing and user pairing, which can quickly and accurately obtain the optimal solution of resource allocation strategy and user matching strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0063] Figure 1 This is a diagram of the architecture of a wireless power supply mobile edge computing network system based on user collaboration provided in Example 1 of the present invention;

[0064] Figure 2 This is the algorithm convergence diagram provided in Example 1 of the present invention.

[0065] Figure 3 A curve diagram showing the relationship between the total energy consumption of the system and the maximum delay allowed by the system provided in Example 1 of the present invention;

[0066] Figure 4 A curve diagram showing the relationship between the total energy consumption of the system and the size of the computing task provided in Example 1 of the present invention;

[0067] Figure 5 This is a curve diagram showing the relationship between the total energy consumption of the system and the bandwidth provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0068] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0069] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that these specific details are not required to practice the present invention.

[0070] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "one embodiment," "an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. In addition, it will be appreciated by those of ordinary skill in the art that the figures provided herein are for illustrative purposes and that the figures are not necessarily drawn to scale. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0071] Example 1

[0072] With the rapid development of the Internet of Things, the number of IoT devices is increasing. However, the current IoT has the following shortcomings: first, the battery storage capacity, computing power and transmission conditions of IoT devices are limited; second, computing resources are insufficient; and third, some IoT devices are idle for a long time and their computing resources are not fully utilized. The above shortcomings make it difficult for IoT devices to complete computing tasks within the maximum allowable delay.

[0073] In this regard, this embodiment provides a method for jointly optimizing computing offloading and resource allocation. On the one hand, in view of the defects of limited computing power and transmission conditions of IoT devices, it is considered to use mobile edge computing technology to offload computing tasks from IoT devices with computing tasks (hereinafter referred to as "active users") to idle devices (hereinafter referred to as "idle users") and MEC servers (MEC servers have strong computing power) for execution, thereby reducing the computing burden of active users; on the other hand, in view of the defects of limited battery storage capacity of IoT devices, it is considered to use wireless power supply technology to power IoT devices; on the other hand, in view of the phenomenon that computing resources are not fully utilized due to insufficient computing resources and a large number of idle users in the current IoT, it is considered to use user collaboration technology to allow idle IoT devices to share their computing resources, thereby improving the performance of active users. In addition, based on the above technical concept, the method establishes a wireless powered mobile edge computing network system based on user collaboration. In this network system, the above mobile edge computing technology, wireless power supply technology and user collaboration technology are organically combined in a modeling manner, and reasonable resource allocation and user pairing are performed in the network system in a joint optimization solution manner to form a resource allocation strategy and a user pairing strategy, while satisfying the unloading of computing tasks, energy supply of IoT devices and full utilization of idle resources, thereby ensuring that IoT devices complete computing tasks within the maximum allowable delay. At the same time, the energy consumption of IoT devices is further realized through reasonable resource optimization configuration and user pairing.

[0074] The method is specifically implemented by the following steps:

[0075] Step 1: Build Figure 1 The user-collaboration-based wireless powered mobile edge computing network system shown establishes an execution environment for the method of this embodiment.

[0076] The network system consists of an energy station, multiple active users, multiple idle users, and an access point with an MEC server deployed. The energy station supplies power to the active users and the idle users through a wireless power supply link; the active users offload computing tasks to the idle users through a D2D link; and the active users offload computing tasks to the MEC server through a cellular link.

[0077] Step 2: Under the above network system,

[0078] (1) Establish a user pairing model to characterize the pairing relationship between active users and idle users.

[0079] The user pairing model is: i,j ∈{0,1}, Among them, α i,jrepresents the pairing coefficient between the i-th active user and the j-th idle user, when α i,j =1, it means that the i-th active user is paired with the j-th idle user, otherwise, α i,j =0; M represents the set of active users, and K represents the set of idle users. are the constraints that the pairing coefficients must satisfy.

[0080] (2) Establishing an energy collection model to characterize the energy collected by the active users and the idle users from the energy station.

[0081] The energy harvesting model is: in, represents the energy collected by the i-th active user from the energy station, represents the energy collected by the jth idle user from the energy station, N represents the number of channels, t0 represents the duration of the power supply phase, η represents the energy conversion efficiency of the energy station supplying power to active users or idle users, P b,n represents the transmission power from the energy station to active users or idle users, h e,i represents the channel gain from the energy station to the active user, h e,j Represents the channel gain from the energy station to the idle user.

[0082] (3) Establishing a local computing model to characterize the local execution delay and local execution energy consumption of the active user, and the local execution delay and local execution energy consumption of the idle user.

[0083] in,

[0084] The local execution latency for active users is: T i loc ≤T; where represents the number of bits of computing tasks completed locally by the i-th active user, C i represents the number of CPU cycles required for the i-th active user to complete the computing task locally, f i represents the computing power of the i-th active user; T i loc ≤T is the local execution delay constraint of the i-th active user, where T represents the maximum tolerable delay of the network system;

[0085] The local execution energy consumption of an active user is: Among them, α i,j represents the pairing coefficient between the i-th active user and the j-th idle user, κ i represents the energy conversion coefficient;

[0086] The local execution latency for an idle user is: in, represents the number of bits of computing tasks completed locally by the jth idle user, C j represents the number of CPU cycles required for the jth idle user to complete the computing task, f j represents the computing power of the jth idle user; The local execution delay constraint for the jth idle user.

[0087] (4) A computing offloading model is established to characterize the energy consumption and data transmission rate of the active user offloading the computing task to the idle user, as well as the energy consumption and data transmission rate of offloading the computing task to the MEC server.

[0088] in,

[0089] The energy consumption of offloading the computing task to the idle user is: Among them, α i,j represents the pairing coefficient between the i-th active user and the j-th idle user, t i,j represents the time required to offload computing tasks from active users to idle users, p i,j represents the link transmission power from active users to idle users; Constraints that need to be met in order to offload the computing task to the idle user;

[0090] According to Shannon's formula, the data transmission rate of offloading the computing task to the idle user is:

[0091] Where B represents the channel bandwidth, g i,j represents the channel state information between active users and idle users, p i,j represents the link transmission power from active users to idle users, N O represents the noise power generated in the channel between active users and idle users;

[0092] The energy consumption of offloading the computing task to the MEC server is: Among them, t i,a represents the time required to offload computing tasks from active users to MEC servers, p i,a represents the link transmission power for offloading computing tasks from active users to MEC servers; Constraints that need to be met in order to offload computing tasks to the MEC server;

[0093] According to Shannon's formula, the data transmission rate of offloading computing tasks to the MEC server is:

[0094] Among them, g i,arepresents the channel state information between active users and MEC servers, p i,a represents the link transmission power from active users to MEC servers, N O represents the noise power generated in the channel between active users and the MEC server.

[0095] in addition,

[0096] The constraints that the computing tasks of active users need to satisfy are: Among them, I i represents the computing task of the i-th active user;

[0097] The energy collection time and unloading time must satisfy the following constraints: t0+t i,j +t i,a ≤T;

[0098] The energy consumption of active users must satisfy the following constraints:

[0099] The energy consumption of idle users must meet the following constraints:

[0100] Step 3: Establish a minimum energy consumption model of the network system according to the user pairing model, the energy collection model, the local computing model and the computing offloading model.

[0101] First, the energy consumed by active users plus the energy consumed by idle users is taken as the total energy consumption, which is expressed as:

[0102]

[0103] Then, according to the total energy consumption expression, the minimum energy consumption model of the network system is obtained as follows:

[0104]

[0105] stC1

[0106] C2

[0107] C3

[0108] C4

[0109] C5

[0110]

[0111] C6

[0112] C7

[0113] C8

[0114] C9

[0115] C10

[0116] C11

[0117]

[0118] in, p=[p i,j ,p i,a ]f=[f i ,f j ],t=[t0,t i,j ,t i,a ], f i max represents the maximum computing power of active users, Indicates the maximum computing capacity of idle users.

[0119] Step 4: Optimize resource allocation and optimal user matching for the minimum energy consumption model to obtain the minimum energy consumption of the network system.

[0120] The specific steps are as follows:

[0121] (1) Establish an optimal resource allocation model.

[0122] The objective function of the optimal resource allocation model is the matching decision α i,j =1, that is, when the i-th IoT device with computing tasks is paired with the j-th idle user, the energy consumed to complete the corresponding computing task. The optimized resource allocation model is as follows:

[0123]

[0124] stC1

[0125] C2

[0126] C3

[0127] C4

[0128] C5

[0129]

[0130] C6

[0131] C7

[0132] C8

[0133] C9

[0134] (2) Using block coordinate descent to fix the computational task allocation variables and time distribution variables t0,t i,j ,t i,a , then the equivalent expression of the optimal resource allocation model is as follows:

[0135]

[0136] st C3-C9.

[0137] (3) Using the Lagrange dual method to solve the equivalent optimal resource allocation model, the power allocation variables and computing capacity allocation variables of the network system are obtained. The specific solution steps are:

[0138] (3-1) The Lagrangian function is constructed as follows:

[0139]

[0140] in, λ=[λ1,λ2,λ3,λ4,λ5,λ6].

[0141] (3-2) The dual function of the Lagrangian function is constructed as follows:

[0142]

[0143] st 0<f i ≤f i max

[0144]

[0145]

[0146]

[0147] (3-3) The dual model with the goal of maximizing the dual function is established as follows:

[0148]

[0149] (3-4) Solve the dual model to obtain the power allocation variable and the computing capacity allocation variable.

[0150] First, we find the expression of the solution for the computing power f of active users and idle users, and the expression of the solution for the power p required to offload computing tasks to idle users and MEC servers as follows:

[0151]

[0152]

[0153]

[0154]

[0155] Then, given the corresponding Lagrange multipliers, the corresponding values ​​of f and p are obtained.

[0156] Next, we use the obtained values ​​of f and p to solve the calculation task allocation variable and time distribution variables t0,t i,j ,t i,a .

[0157] Next, construct the variable for solving the calculation task allocation and time distribution variables t0,t i,j ,t i,a The objective function is given by:

[0158]

[0159] st C1-C4

[0160] C6-C9.

[0161] In the next step, the CVX toolkit in Matlab is used to solve the corresponding computing task allocation variables and time allocation variables.

[0162] In the next step, the obtained computing task allocation variables and time allocation variables are used to update the Lagrange multiplier λ using the subgradient method, thereby updating the corresponding values ​​of f and p, and iterating until the convergence condition is reached and the iteration stops, obtaining the optimal solution of the objective function of solving the resource allocation strategy constructed in step 3.2.

[0163] Finally, the energy consumed by the i-th IoT device with computing tasks to complete the computing task when paired with the j-th idle user is solved according to the method of the above steps (1)-(3), thereby obtaining an i×j energy consumption matrix.

[0164] (4) Based on the Hungarian algorithm, the optimal matching strategy is solved to minimize the energy consumption of the entire system.

[0165] (4-1) The optimal user matching model is established as follows:

[0166]

[0167]

[0168]

[0169] (4-2) Using the Hungarian method and the energy consumption matrix, the optimal user matching model is solved to obtain the minimum energy consumption of the integrated system.

[0170] In summary, the method for jointly optimizing computing offloading and resource allocation provided in this embodiment seeks the optimal resource allocation scheme and matching decision when performing wireless energy power supply and computing offloading and user pairing for IoT devices in a wireless power supply edge computing system based on user collaboration, thereby minimizing the energy consumption of the entire system, and jointly optimizes resource allocation and user matching decisions through the block coordinate descent algorithm, Lagrange dual method and Hungarian algorithm, and finally obtains the optimal solution; realizes the optimal resource allocation strategy and user matching decision in the wireless power supply edge computing system based on user collaboration, and reduces the energy consumption of the system.

[0171] In addition, to further illustrate the feasibility and technical effect of the above-mentioned method of jointly optimizing computation offloading and resource allocation, the present embodiment conducted the following simulation experiment.

[0172] Figure 2 It is a graph of the convergence of the algorithm in the specific implementation mode of the present invention. It can be seen from the graph that when the number of iterations is 4, the algorithm proposed in this solution reaches convergence, which proves the effectiveness of the proposed algorithm.

[0173] Figure 3 The figure is a curve diagram showing the relationship between the total energy consumption of the system and the size of the computing task in the specific implementation of the present invention. It can be found that as the number of bits of the computing task increases, the total energy consumption of the system increases exponentially. In addition, when the number of bits of the computing task is 2×10 5 bits, the scheme proposed by the present invention saves about 50% of energy consumption compared with local calculation. At the same time, compared with the comparative scheme, the scheme proposed by the present invention always completes the corresponding calculation task with the minimum energy consumption.

[0174] Figure 4The figure is a curve diagram showing the relationship between the total energy consumption of the system and the size of the bandwidth in a specific implementation of the present invention. It can be seen that as the bandwidth increases, except for local computing, the energy consumed by all solutions shows a downward trend, and the downward trend tends to be gentle. It can be found that the solution proposed in the present invention has better performance when the bandwidth is the smallest, and can complete the corresponding computing task with the minimum energy consumption.

[0175] Example 2

[0176] Corresponding to a method for joint optimization of computing offloading and resource allocation provided in Example 1, this embodiment provides a system for joint optimization of computing offloading and resource allocation as a carrier of the method.

[0177] Further, the system for jointly optimizing computation offloading and resource allocation includes:

[0178] Network system building module, used to build a wireless power supply mobile edge computing network system based on user collaboration;

[0179] Model building module, used to build:

[0180] User pairing model, which represents the pairing relationship between active users and idle users;

[0181] Energy harvesting model, characterizing the energy harvested from the energy station by active and idle users;

[0182] A local computation model that characterizes the local execution delay and local execution energy consumption of active users, as well as the local execution delay and local execution energy consumption of idle users;

[0183] The computation offloading model characterizes the energy consumption and data transmission rate of computation tasks from active users to idle users, and the energy consumption and data transmission rate of computation tasks from active users to MEC servers;

[0184] Minimum energy consumption model, characterizing the minimum energy consumption of the network system;

[0185] An optimization resource allocation module, used for optimizing resource allocation for the minimum energy consumption model;

[0186] The optimal user matching module is used to perform optimal user matching on the minimum energy consumption model to obtain the minimum energy consumption of the network system.

[0187] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for jointly optimizing computation offloading and resource allocation, characterized in that: The following steps are involved: Build a wireless power supply mobile edge computing network system based on user collaboration; under the network system, Establish a user pairing model to characterize the pairing relationship between active users and idle users; Establishing an energy collection model for characterizing the energy collected by the active user and the idle user from the energy station; Establishing a local computing model for characterizing the local execution delay and local execution energy consumption of the active user, and the local execution delay and local execution energy consumption of the idle user; Establishing a computing offloading model to characterize the energy consumption and data transmission rate of the active user offloading the computing task to the idle user, and the energy consumption and data transmission rate of offloading the computing task to the MEC server; Establishing a minimum energy consumption model of the network system according to the user pairing model, the energy collection model, the local computing model and the computing offloading model; Optimizing resource allocation and optimal user matching for the minimum energy consumption model to obtain the minimum energy consumption of the network system; The user pairing model is: Among them, α i，j represents the pairing coefficient between the i-th active user and the j-th idle user, K represents the set of active users, and M represents the set of idle users; The energy harvesting model is: in, represents the energy collected by the i-th active user from the energy station, represents the energy collected by the jth idle user from the energy station, N represents the number of channels, t0 represents the duration of the power supply phase, η represents the energy conversion efficiency of the energy station supplying power to active users or idle users, P b，n represents the transmission power from the energy station to the active user or the idle user, h e，i represents the channel gain from the energy station to the active user, h e，j represents the channel gain from the energy station to the idle user; The local execution delay of the active user is: in, represents the number of bits of computing tasks completed locally by the i-th active user, C i represents the number of CPU cycles required for the i-th active user to complete the computing task locally, f i represents the computing power of the i-th active user; is the local execution delay constraint for the i-th active user, where T represents the maximum tolerable delay of the network system; The local execution energy consumption of the active user is: Among them, α i，j represents the pairing coefficient between the i-th active user and the j-th idle user, κ i represents the energy conversion coefficient; The local execution delay of the idle user is: in, represents the number of bits of computing tasks completed locally by the jth idle user, C j represents the number of CPU cycles required for the jth idle user to complete the computing task, f j represents the computing power of the jth idle user; The local execution delay constraint for the jth idle user; The local execution energy consumption of the idle user is: Among them, κ j represents the energy conversion coefficient; The energy consumption of offloading the computing task to the idle user is: Among them, α i，j represents the pairing coefficient between the i-th active user and the j-th idle user, t i，j represents the time required to offload computing tasks from active users to idle users, p i,j represents the link transmission power from active users to idle users; Constraints that need to be met in order to offload the computing task to the idle user; The data transmission rate of offloading the computing task to the idle user is: Where B represents the channel bandwidth, g i，j represents the channel state information between active users and idle users, p i，j represents the link transmission power from active users to idle users, N O represents the noise power generated in the channel between active users and idle users; The energy consumption of offloading the computing task to the MEC server is: Among them, t i，a represents the time required to offload computing tasks from active users to MEC servers, p i，a represents the link transmission power for offloading computing tasks from active users to MEC servers; Constraints that need to be met in order to offload computing tasks to the MEC server; The data transmission rate for offloading the computing task to the MEC server is: Among them, g i，a represents the channel state information between active users and MEC servers, p i，a represents the link transmission power from active users to MEC servers, N O represents the noise power generated in the channel between the active user and the MEC server; The minimum energy consumption model of the network system is: in, p=[p i,j ,p i,a ]f=[f i ,f j ],t=[t0,t i,j ,t i,a ], f i max represents the maximum computing power of active users, Indicates the maximum computing capacity of idle users.

2. A method for jointly optimizing computation offloading and resource allocation according to claim 1, characterized in that: The network system includes: an energy station, an active user, an idle user and an MEC server; the energy station supplies power to the active user and the idle user through a wireless power supply link; the active user offloads computing tasks to the idle user through a D2D link; the active user offloads computing tasks to the MEC server through a cellular link.

3. A method for jointly optimizing computation offloading and resource allocation according to claim 1, characterized in that: The method for optimizing resource allocation for the minimum energy consumption model is: A1: Establish an optimized resource allocation model; A2: Using the Lagrange dual method to lock the power allocation variables and computing capacity allocation variables in the resource allocation model, an equivalent optimized resource allocation model is obtained; A3: Using the CVX toolkit to lock the computing task allocation variables and the time allocation variables in the resource allocation model, and obtain an equivalent optimized resource allocation model; A4: solving the equivalent optimization resource allocation model using a block coordinate descent method to obtain an energy consumption matrix of the network system, which is used to represent the energy consumption of computing tasks from any active user to any idle user in the network system; The method for performing optimal user matching on the minimum energy consumption model is: B1: Establish the optimal user matching model; B2: Using the Hungarian method and the energy consumption matrix, solve the optimal user matching model to obtain the minimum energy consumption of the network system.

4. A method for jointly optimizing computation offloading and resource allocation according to claim 3, characterized in that: The A2 includes: A21: Establishing a Lagrangian function according to the equivalent optimal resource allocation model; A22: Establishing the dual function of the Lagrangian function; A23: establishing a dual model with the objective of maximizing the dual function; A24: Solve the dual model to obtain the computing task allocation variable and the time allocation variable; A25: According to the computing task allocation variable and the time allocation variable, the multiplier of the Lagrangian function, the computing power of the active user, the computing power of the idle user, the link transmission power of offloading the computing task to the idle user, and the link transmission power of offloading the computing task to the MEC server are iteratively updated using a subgradient method until a convergence condition is met, thereby obtaining an optimal solution to the optimization resource allocation model; A26: According to A1 to A25, the energy consumption of the computing task from any active user to any idle user is obtained based on the block coordinate descent method to form an energy consumption matrix of the network system.

5. A system for the method for joint optimization of computation offloading and resource allocation according to any one of claims 1 to 4, characterized in that: include: Network system building module, used to build a wireless power supply mobile edge computing network system based on user collaboration; Model building module, used to build: User pairing model, which represents the pairing relationship between active users and idle users; Energy harvesting model, characterizing the energy harvested from the energy station by active and idle users; A local computation model that characterizes the local execution delay and local execution energy consumption of active users, as well as the local execution delay and local execution energy consumption of idle users; The computation offloading model characterizes the energy consumption and data transmission rate of computation tasks from active users to idle users, and the energy consumption and data transmission rate of computation tasks from active users to MEC servers; Minimum energy consumption model, characterizing the minimum energy consumption of the network system; An optimization resource allocation module, used for optimizing resource allocation for the minimum energy consumption model; The optimal user matching module is used to perform optimal user matching on the minimum energy consumption model to obtain the minimum energy consumption of the network system.

Images