A computation offloading method and system based on auction algorithm

Abstract

The present invention discloses a method and system for computing offloading based on an auction algorithm, and establishes a transmission model considering the scenario in which multiple eMBB and URLLC service users coexist in the same area and perform uplink transmission to multiple MEC edge servers. The necessary parameters are listed according to the transmission model, and then a two-way auction model is established according to the principle of the auction algorithm, considering the joint allocation of multiple server resources, and based on this, the cost and valuation are expressed, and bidding and server asking price strategies are proposed for two different types of service users. Finally, the server revenue, the delay and energy consumption of computing offloading are comprehensively considered to establish an optimization problem, and a multi-round two-way auction is carried out according to the constraints to match the most suitable edge server for the user's computing task. When the remaining resources in the MEC server are too few or all user tasks have been allocated to the corresponding server for calculation, the auction ends. Compared with the traditional resource allocation method, the present invention can enable the MEC server to obtain higher revenue, promote efficient resource utilization, and have better performance.

Description

Technical Field

[0001] The present invention relates to a calculation offloading method and system based on an auction algorithm, and belongs to the technical field of 5G service resource allocation. Background Art

[0002] The International Telecommunication Union pointed out at the 22nd meeting that 5G networks mainly cover three application scenarios: enhanced mobile broadband (eMBB), ultra-reliable low-latency communication (URLLC) and large-scale machine communication (mMTC). Among them, the purpose of eMBB business is to improve the communication experience between users, requiring the user experience rate to reach 1Gbps, and meeting the peak rate of 20Gbps in specific scenarios, covering 3D, ultra-high-definition video, AR / VR, cloud games and other services with ultra-large traffic mobile bandwidth and medium latency; URLLC business, as a new field, requires millisecond-level end-to-end latency and up to 99.999% transmission reliability, covering services such as smart grid, industrial automation, autonomous driving, mobile medical care, tactile Internet, etc. that require ultra-low latency and ultra-high reliability. With the rapid growth of the number of communication devices, computing-intensive applications are becoming an integral part of our daily lives. However, due to the limited computing power and device life of mobile devices, processing data on local devices has become an important issue that needs to be considered. Mobile cloud computing (MCC) is a solution to this problem. In MCC, computing-intensive tasks are offloaded to cloud servers through cellular networks, which can reduce the energy consumption of resource-constrained devices. The cloud server then performs the offloaded tasks and returns the output to the device. However, the cloud server is far away from the mobile device, which will cause a large transmission delay and cannot meet some low-latency business requirements. Therefore, the wireless communication industry and academic research community have introduced a new technology called mobile edge computing (MEC). According to the current development strategy, from the perspective of 5G communication development needs, mobile edge computing is one of the scenarios where URLLC and eMBB services coexist.

[0003] The eMBB service is committed to pursuing the ultimate communication experience between people, requiring a user experience rate of 1 Gbps and a peak rate of 20 Gbps in specific scenarios. It covers services with ultra-large traffic mobile bandwidth and medium latency, such as 3D, ultra-high-definition video, AR / VR, and cloud gaming. In order to meet the needs of eMBB services, 5G needs to improve spectrum efficiency, signaling efficiency, bandwidth, and coverage. As a new field, URLLC services require millisecond-level end-to-end latency and up to 99.999% transmission reliability, covering services such as smart grids, industrial automation, autonomous driving, mobile medical care, and tactile Internet that require ultra-low latency and ultra-high reliability.

[0004] MEC technology transfers services such as network control, storage, and mobile computing to the edge of the network so that user-side mobile devices with limited resources can perform complex and low-latency services. This will significantly reduce latency and mobile device energy consumption, reduce latency, save device energy, improve environmental perception, and enhance privacy / security.

[0005] Resource allocation in MEC systems is a key issue that is worth studying in multi-user MEC systems. Since different services have different requirements for quality of service (QoS), it is a challenge to support different services in MEC systems. When many mobile users offload computing tasks to MEC servers at the same time, serious interference problems will occur. As the number of mobile users in the system continues to increase, the number of mobile users who offload tasks will not increase, but will decrease, because there are many computing tasks on the MEC server, which will cause the rate to drop and cause serious interference. Therefore, under the premise of ensuring the quality of service, it is particularly important to allocate resources reasonably.

[0006] Considering that both URLLC users and eMBB users will generate computing-intensive services, and the computing power and energy limitations of mobile devices themselves, they need to be offloaded to MEC servers for computing, which will compete for spectrum resources and computing resources in the system. Therefore, the joint allocation of spectrum resources and computing resources in the MEC system is a key issue that is worth studying in multi-user MEC systems. Since different services have different QoS requirements, supporting different services in the MEC system is a challenge. Existing research on resource allocation for the coexistence of URLLC services and eMBB services in MEC systems is not perfect enough. The main problems are as follows: 1) Many current studies only consider user transmission delay and computing offload energy consumption, but the construction and maintenance of servers require fixed costs, so it is necessary to maximize server revenue; 2) The joint allocation of multiple server resources, such as communication resources, storage resources, and computing resources, is not considered. Summary of the invention

[0007] The present invention provides a method and system for computing offloading based on an auction algorithm. In the MEC scenario where eMBB and URLLC services coexist, the ultimate goal is to allocate resources through a computing offloading strategy, while maximizing the benefits of the MEC server while satisfying indicators such as user latency and energy consumption. The resource allocation methods in traditional MEC scenarios have the disadvantage that they only consider user latency and energy consumption requirements, but do not consider the construction cost and benefits of the MEC server, which will lead to the cost of building the server and waste of resources in the server. Therefore, the present invention takes into account the cost and benefits of the MEC server, establishes a many-to-many auction model, assumes the user device as the buyer, the MEC server as the seller, and the base station as the auctioneer, and proposes a computing offloading strategy based on a multi-round two-way auction, which dynamically adjusts the user bid and the server bid as the auction rounds progress to improve the auction success rate. It can both reasonably allocate various resources of the server and ensure that the benefits of the MEC server are maximized.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0009] A computation offloading method based on auction algorithm:

[0010] Establish a MEC server area transmission model where eMBB users and URLLC users coexist;

[0011] Based on the regional transmission model, a double auction model is established with the eMBB user and the URLLC user as the buyer and the MEC server as the seller;

[0012] To maximize the seller's profit, formulate the constraints of the double auction model;

[0013] The double auction model dynamically adjusts the buyer's bid and the seller's offer, and conducts multiple rounds of auctions according to the constraints to match MEC servers for eMBB users and URLLC users;

[0014] The computing tasks of eMBB users and URLLC users are offloaded to the matching MEC servers.

[0015] Furthermore, the regional transmission model has K base stations, M eMBB users and N URLLC users are randomly distributed in the area, each user has a computing task to be executed, and it is assumed that a large number of URLLC users will not request at the same time, so M>>N, each base station is equipped with an MEC server, and the user chooses to perform calculations on the local device or offload them to the MEC server for processing; parameters such as the MEC server resource amount, computing power and user task data volume are randomly distributed within a certain range.

[0016] Furthermore, the parameters of the regional transmission model include:

[0017] eMBB service rate

[0018] Among them, N m is the number of subcarriers allocated to eMBB user m, W is the subcarrier bandwidth of the frequency band resource, P m is the transmit power of eMBB user m, N0 is the additive white Gaussian noise power, g m is the small-scale Rayleigh fading of eMBB users, G m is the path loss of eMBB user m, expressed as: G m =128.1+37.6lg(d m );

[0019] Among them, d m is the distance from eMBB user m to the base station;

[0020] The local calculation latency of eMBB user m

[0021] Among them, c m To handle the computing power required for this task, It is local computing power;

[0022] eMBB user m local computing energy consumption

[0023] Among them, σ is the energy consumption factor, which is related to the chip performance of the eMBB user equipment;

[0024] The offloading computation delay of eMBB user m consists of two parts: task offloading delay and MEC server computation delay.

[0025] Among them, R m is the achievable rate of the eMBB service; m is the task bit amount of eMBB user m.

[0026] MEC server calculation delay

[0027] in, The computing resources allocated by the MEC server for the task of eMBB user m;

[0028] Total offloading calculation delay of eMBB user m

[0029] set up The binary parameter of eMBB user m is an allocation parameter used to determine whether the computing task of eMBB user m is offloaded to MEC server k for processing:

[0030]

[0031] The energy consumption required for computing the data offloaded by eMBB user m on MEC server k

[0032] The total delay of offloading calculation is changed to:

[0033] The total energy consumption of unloading calculation is changed to:

[0034] Among them, ξ m is the delay energy cost factor of eMBB user m;

[0035] The computation offloading delay of URLLC user n is:

[0036]

[0037] Where R n is the achievable rate of URLLC service; n is the task bit amount of URLLC user n, c n The computing power required to process the tasks of URLLC user n, is the computing resources allocated by the MEC server for the task of URLLC user n, MEC server computing delay in the offloading computing delay of URLLC user n;

[0038] set up The binary parameter for URLLC services is the allocation parameter, which is used to determine whether the computing task of URLLC user n is offloaded to MEC server k for processing:

[0039]

[0040] Furthermore, the method of establishing a two-way auction model with eMBB and URLLC users as buyers and MEC servers as sellers is as follows:

[0041] The cost of MEC server k is expressed as: C k =λ kw W k +λ kd D k +λ kf F k ;

[0042] Among them, W k , Dk , F k are the total communication, storage and computing resources of server k, respectively, kw ,λ kd ,λ kf These are the cost influencing factors of the three types of resources. The larger the server resources, the greater the cost of its construction and maintenance.

[0043] The resource valuation of eMBB user m is expressed as:

[0044] Among them, β m and 1-β m are the estimated weight parameters of service delay and energy consumption of user m, and their sum is 1; δ mT is the delay cost factor, δ mE is the energy cost factor;

[0045] Similarly, the resource valuation of URLLC user n is expressed as:

[0046] Among them, δ nT is the delay cost factor of URLLC user n.

[0047] The bid of eMBB user m in the t+1 round of auction is:

[0048] Among them, t represents the current auction round number, Sum represents the total number of auction rounds, Pro m is the priority of the user task, ε b To adjust the parameters of user bidding;

[0049] URLLC user n's bid for the t+1 round of auction is: b n (t+1)=η t Pro n b n (t)

[0050] Among them, Pro n is the URLLC user task priority, and η is the URLLC service bid index.

[0051] The auction asking price of MEC server k in the t+1 round is expressed as:

[0052] Among them, ε a To adjust the parameter of MEC server price, μ k (t) is the parameter that affects the quotation due to the remaining resources of the MEC server;

[0053]

[0054] Among them, w kt , f kt , d kt They represent the remaining communication resources, computing resources, and storage resources of server k in round t. As the auction progresses, the amount of resources gradually decreases, and μ k As (t) decreases, the asking price of the MEC server also decreases, allowing the server to maintain a certain advantage and improve the success rate of the auction.

[0055] Furthermore, with the goal of maximizing the seller's profit, the method of formulating the constraints of the double auction model is as follows:

[0056] The benefit that eMBB user m brings to MEC server k is expressed as the reward paid by eMBB user m minus the cost of MEC server k, that is:

[0057]

[0058] Similarly, the benefit that URLLC user n brings to MEC server k can be expressed as:

[0059] The total revenue of the MEC server is written as the sum of the revenue brought by each eMBB user m to the corresponding MEC server:

[0060]

[0061] The objective function to maximize the seller's profit is:

[0062]

[0063] Among them, constraints C1 and C2 indicate that if the task of each eMBB or URLLC user chooses to offload computing, it can only be completed by one MEC server k; constraint C3 indicates that each computing task of the user can only be offloaded as a whole or locally calculated; constraints C4, C5 and C6 indicate that the communication, storage and computing resources consumed by the two service users cannot exceed the total resource inventory of the server, where w m,k , w n,k , f m,k , f n,k , d m,k , d n,k They respectively represent the corresponding resource amounts provided by server k to user m or n; constraint C7 indicates that the delay energy consumption cost factor of eMBB user m is not less than 0.

[0064] Furthermore, the double auction model dynamically adjusts the buyer's bid and the seller's offer, and conducts multiple rounds of auctions according to the constraints. The method for matching MEC servers for eMBB users and URLLC users is as follows:

[0065] Step 1: If a URLLC user sends a request to MEC server k, the URLLC user is matched with the requested MEC server k;

[0066] Step 2: rank the preferences of the eMBB user m for the MEC server k, and select the MEC server k that meets the constraint conditions according to the constraint conditions;

[0067] Step 3: Assume that the base station is the auctioneer and the eMBB user m is the buyer. The eMBB user m provides the base station with service requirements, valuation of the required resources, preference ranking and bid for the MEC server k. In the first round of auction, the valuation of the required resources is used as the bid of the eMBB user m. The MEC server k is the seller and provides the base station with the total amount and asking price of various resources.

[0068] Step 4: The base station sorts the MEC servers k in descending order of preference according to the collected information, accesses the MEC servers k one by one in order of preference from high to low, selects the eMBB user m requesting the MEC server k with the preference, and if the bid of the eMBB user m is ≥ the asking price of the requested MEC server k, then the eMBB user m is added to the candidate list and the process goes to step 5; otherwise, the eMBB user m is added to the failure list and the process goes to step 6;

[0069] Step 5: traverse all eMBB users m who request the MEC server k with the preference, sort the bids of eMBB users m in the candidate list in descending order, and judge that the total amount of resources required by eMBB users m in the candidate list is ≤ the remaining amount of resources of the MEC server k with the preference. If so, the match is successful, and the remaining amount of resources of the MEC server k with the preference is updated; otherwise, the eMBB users m with low bids are added to the failure list in the order of the bids of eMBB users m in the candidate list from low to high, until the total amount of resources required by eMBB users m in the candidate list is ≤ the remaining amount of resources of the MEC server k with the preference, the match is successful, and the process goes to step 7; if all eMBB users m who request the MEC server k with the preference are not traversed, return to step 1;

[0070] Step 6: The eMBB user m added to the failure list sends a request to the MEC server k of the next preference, and the process returns to step 2. If the MEC server k of all the preference levels of the eMBB user m cannot be matched, the current auction ends and the next auction begins. In the next auction, the eMBB user m modifies the bid, and the MEC server k modifies the asking price, and the process returns to step 4.

[0071] Step 7: If the remaining resources in the MEC server k are insufficient to continue matching or all eMBB users m and URLLC users have been matched to the corresponding MEC server k, the auction ends; otherwise, return to step 1.

[0072] Accordingly, the present invention also provides a computation offloading system based on an auction algorithm, comprising:

[0073] The regional transmission module is used to establish the MEC server regional transmission relationship where eMBB users and URLLC users coexist;

[0074] The double auction model is used to dynamically adjust the buyer's bid and the seller's offer with eMBB users and URLLC users as buyers and MEC servers as sellers, and conduct multiple rounds of auctions according to constraints to match MEC servers for eMBB users and URLLC users. The computing tasks of eMBB users and URLLC users are offloaded to the matched MEC servers.

[0075] Furthermore, it also includes a computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, and when the instructions are executed by a computing device, the computing device executes any one of the above methods.

[0076] Furthermore, a computing device is also included, the computing device comprising:

[0077] One or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and are configured to be executed by the one or more processors, and the one or more programs include instructions for executing any of the methods described above.

[0078] The beneficial effects achieved by the present invention are:

[0079] The present invention proposes a method and system for computing offloading based on an auction algorithm. First, a transmission model is established, considering the scenario in which multiple eMBB and URLLC service users coexist in the same cell and perform uplink transmission to multiple MEC servers. Secondly, the necessary parameters are listed according to the established transmission model, including eMBB transmission rate, local computing delay, local computing energy consumption, offloading computing delay and offloading computing energy consumption. Then, a two-way auction model is established according to the auction algorithm principle to express the cost, valuation, and set parameters to adjust the bid and asking price. Finally, the server revenue, the delay and energy consumption of computing offloading are comprehensively considered to establish an optimization problem, and a multi-round two-way auction is performed according to the constraints to match the most suitable MEC server for the user's computing task. When the remaining resources in the MEC server are too few or all user tasks have been allocated to the corresponding server for calculation, the auction ends. Compared with the traditional resource allocation method, the present invention can enable the MEC server to obtain higher revenue, promote efficient resource utilization, and have better performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 It is a schematic diagram of the regional transmission model in the present invention;

[0081] Figure 2 It is a schematic diagram of the double auction model in the present invention;

[0082] Figure 3 It is a schematic diagram of the auction process of the double auction model in the present invention. DETAILED DESCRIPTION

[0083] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0084] Based on the multi-round double auction theory, the present invention combines the different latency and energy consumption requirements of eMBB and URLLC services, and comprehensively considers the cost and benefit of MEC servers to construct a computing offloading strategy based on resource joint scheduling in the MEC scenario where eMBB and URLLC services coexist.

[0085] like Figure 1 As shown, the present invention establishes a cell with multiple users and multiple MEC servers. In the model, there are K base stations, M eMBB users and N URLLC users randomly distributed in a certain range. Each user has a computing task to be executed. Since the URLLC service is bursty and scattered, it is almost impossible for a large number of URLLC users to request at the same time, so M>>N. Each base station is equipped with an MEC server, and users can choose to perform calculations on local devices or offload them to the MEC server for processing. Parameters such as the amount of MEC server resources, computing power, and the amount of user task data are randomly distributed in a certain range.

[0086] Based on the above theory, the present invention provides the following execution steps of the calculation offloading strategy based on the auction algorithm in the MEC scenario where eMBB and URLLC services coexist:

[0087] Step 1: According to the established regional transmission model, express the necessary parameters;

[0088] Step 2: Based on auction theory, establish a multi-round double auction model;

[0089] Step 3: Comprehensively consider latency, energy consumption, and server benefits to establish an optimization problem;

[0090] Step 4: Dynamically adjust user bids and server quotations, and conduct multiple rounds of auctions based on the objective function and constraints of the optimization problem in step 3.

[0091] For step 1, according to the established cell transmission model, the necessary parameters are expressed, and the establishment steps are as follows:

[0092] 1. The achievable rate of eMBB services can be expressed as:

[0093]

[0094] Among them, N m is the number of subcarriers allocated to eMBB user m, W is the subcarrier bandwidth of the frequency band resource, P m is the transmit power of eMBB user m, N0 is the additive white Gaussian noise power, g m is the small-scale Rayleigh fading of eMBB users, G m is the path loss of eMBB user m, which can be expressed as:

[0095] G m =128.1+37.6lg(d m );

[0096] where d m is the distance from eMBB user m to the base station.

[0097] 2. The local calculation delay of eMBB user m is:

[0098]

[0099] Among them, c m To handle the computing power required for this task, It is local computing power.

[0100] 3. The local computing energy consumption of eMBB user m can be expressed as:

[0101]

[0102] Among them, σ is the energy consumption factor, which is related to the chip performance of the eMBB user equipment.

[0103] 4. The offloading computation delay of eMBB user m consists of two parts: task offloading delay and server computation delay. Combined with the wireless transmission rate model, the task offloading delay can be written as:

[0104]

[0105] Among them, R m represents the task bit amount (bits) of eMBB user m. The time required for offloading calculation of this user can be written as:

[0106]

[0107] It indicates the computing resources allocated by the MEC server to the task of eMBB user m. Considering that the time required for uplink transmission is much longer than that for downlink transmission, the present invention only considers the data transmission from the user to the MEC server, and adds the transmission delay to the task calculation delay. Therefore, the total offload calculation delay of the user can be written as:

[0108]

[0109] 5. Set The eMBB user binary parameter is an allocation parameter used to determine whether the computing task of the eMBB user m is offloaded to the server k for processing:

[0110]

[0111] The energy consumption required for computing the data offloaded by eMBB user m on MEC server k is as follows:

[0112]

[0113] Considering that the amount of data uploaded to the server is much larger than the amount of data output from the server, the present invention only considers the energy consumption of uplink transmission.

[0114] 6. Considering that different eMBB users and different tasks have different requirements for time delay and energy consumption, in some low-latency scenarios, users may sacrifice some energy consumption in exchange for shorter latency; conversely, in scenarios where latency requirements are not high, users can also choose to sacrifice latency to reduce energy consumption and costs. Therefore, the total latency and energy consumption of offloading calculations can be rewritten as:

[0115]

[0116] Among them, ξ m is the cost factor of eMBB user m, which can represent the offloading cost that the user expects to achieve according to the specific requirements of the task.

[0117] 7. For URLLC services, since they have very high latency requirements, but the data volume of the services themselves is small, compared with eMBB services, their energy consumption is not the focus of current research and can be ignored. Referring to the eMBB latency, the calculation offloading latency of URLLC user n can be obtained as:

[0118]

[0119] Where R n is the achievable rate of URLLC service; n is the task bit amount of URLLC user n, c nThe computing power required to process the tasks of URLLC user n, is the computing resources allocated by the MEC server for the task of URLLC user n, MEC server computing delay in the offloading computing delay of URLLC user n;

[0120] set up The binary parameter for URLLC services is the allocation parameter, which is used to determine whether the computing task of URLLC user n is offloaded to MEC server k for processing:

[0121]

[0122] For step 2, based on auction theory, a multi-round double auction model is established. The steps are as follows:

[0123] 1. This invention adopts a computation offloading method based on auction theory to improve the allocation efficiency through competition and mutual game among users. Assume that the MEC server plays the role of the seller and the user is the buyer. Figure 2 The two-way auction model is established as shown in the figure. The MEC server provides computing resources to users to complete the user's offload computing tasks. At the same time, the user also has to pay a certain amount of compensation to the MEC server. The payment rules are determined by the auctioneer. In order to ensure the user-side utility, the user's final payment cannot be greater than the user's valuation of resource demand.

[0124] 2. For mobile edge computing, the cost is related to many factors, such as storage resource cost, communication resource cost, and computing resource cost. The larger the memory cost of the MEC server, the more communication resources, and the faster the computing speed, the higher the deployment and maintenance cost of the server. Therefore, the cost of MEC server k is expressed as:

[0125] C k =λ kw W k +λ kd D k +λ kf F k ;

[0126] Among them, W k , D k , F k are the total communication, storage and computing resources of server k, respectively, kw ,λ kd ,λ kf These are the cost influencing factors of the three types of resources. It can be seen that the larger the server resources, the greater the cost required;

[0127] 3. Valuation refers to the highest price that the user is willing to accept. Valuation is related to the user's business needs. The higher the business priority, the higher the price the user is willing to pay, that is, the higher the valuation. For the model established by the present invention, the service delay and energy consumption of eMBB users are mainly considered, which can be expressed as follows:

[0128]

[0129] Among them, β m and 1-β m are the estimated weight parameters of service delay and energy consumption of user m, respectively, and their sum is 1. mT and δ mE They are delay cost factor and energy consumption cost factor respectively. Users can adjust them according to actual business needs. The larger the factor, the smaller the valuation, which means the lower the priority of the task and the smaller the advantage in the auction.

[0130] Similarly, the resource valuation of URLLC user n is expressed as:

[0131] Among them, δ nT is the delay cost factor of URLLC user n. Since URLLC services are very sensitive to delay, the impact of delay on valuation needs to be considered and given priority in the auction. Therefore, it is necessary to set δ nT >>δ mT ,δ mE , in order to ensure v n >v m , ultimately enabling URLLC services to be handled with priority.

[0132] 4. Based on the auction compensation strategy, the present invention designs bidding strategies for two different types of services and users with different priorities.

[0133] First, due to the high reliability and low latency of URLLC services, URLLC users are generally given higher priority than eMBB users. When receiving a request from a URLLC user, the server should process it first. In addition, considering the burstiness and sporadic nature of URLLC services, it is almost impossible for multiple URLLC services to send requests at the same time and thus generate competition. Therefore, the bidding strategy is designed only for resource competition between eMBB users.

[0134] 5. For eMBB users, the demand for resources is relatively large, so users with different priorities need different bidding strategies for resource allocation. The user bids are as follows:

[0135]

[0136] Among them, t represents the current auction round number, Sum represents the total number of auction rounds, Pro m is the priority of the user task, ε b It is a parameter for adjusting user bids and is used to adjust the bid differences between tasks of different priorities.

[0137] For URLLC user n, the auction bid in round t+1 is: b n (t+1)=η t Pro n b n (t)

[0138] Among them, Pro n It is the URLLC user task priority and meets the requirements of Pro n >>Pro m , to ensure that URLLC users’ bids gain an advantage, η is the URLLC service bid index, which is adjusted according to the importance of the service. The exponential growth can ensure that URLLC users’ bids increase rapidly with the auction rounds.

[0139] 6. For servers, as the number of auction rounds progresses, the remaining resources and the number of users are getting smaller and smaller. In order to maintain competitiveness in the auction, it is necessary to appropriately lower the price to maintain a certain advantage. Therefore, the asking price of MEC servers is as follows:

[0140]

[0141] Among them, ε a To adjust the parameter of MEC server price, μ k (t) is the parameter that affects the quotation due to the remaining resources of the MEC server, as follows:

[0142]

[0143] Among them, w kt , f kt , d kt They represent the remaining communication resources, computing resources, and storage resources of server k in round t. As the auction progresses, the amount of resources gradually decreases, and μ k As (t) decreases, the asking price of the MEC server also decreases, allowing the server to maintain a certain advantage and improve the success rate of the auction.

[0144] For step three, consider latency, energy consumption, and server benefits, and establish an optimization problem. The steps are as follows:

[0145] 1. The benefit that eMBB user m can bring to MEC server k can be expressed as the user's payment minus the server cost, as shown below:

[0146]

[0147] Similarly, the benefit that URLLC user n brings to MEC server k can be expressed as:

[0148] 2. The total revenue of the MEC server can be written as the sum of the revenue brought to the corresponding server by each eMBB and URLLC user:

[0149]

[0150] 3. The present invention comprehensively considers the server revenue, the latency and energy consumption of computing offload, so the optimization problem is as follows:

[0151]

[0152] Among them, constraints C1 and C2 indicate that if the task of each eMBB or URLLC user chooses to offload computing, it can only be completed by one MEC server k; constraint C3 indicates that each computing task of the user can only be offloaded as a whole or locally calculated; constraint C4 indicates that the communication resources consumed by the two service users cannot exceed the total resource stock of the server; constraint C5 indicates that the storage resources consumed by the two service users cannot exceed the total resource stock of the server; constraint C6 indicates that the computing resources consumed by the two service users cannot exceed the total resource stock of the server; where w m,k , w n,k , f m,k , f n,k , d m,k , d n,k They respectively represent the communication, computing and storage resources provided by server k to user m or n; constraint C7 indicates that the delay energy consumption cost factor of eMBB user m is not less than 0.

[0153] For step 4, dynamically adjust the user bid and server offer, and conduct multiple rounds of auctions according to the objective function and constraints of the optimization problem in step 3. Figure 3 As shown:

[0154] Step 1: If a URLLC user sends a request to MEC server k, the URLLC user is matched with the requested MEC server k;

[0155] Step 2: rank the preferences of the eMBB user m for the MEC server k, and select the MEC server k that meets the constraint conditions according to the constraint conditions;

[0156] Step 3: Assume that the base station is the auctioneer and the eMBB user m is the buyer. The eMBB user m provides the base station with service requirements, valuation of the required resources, preference ranking and bid for the MEC server k. In the first round of auction, the valuation of the required resources is used as the bid of the eMBB user m. The MEC server k is the seller and provides the base station with the total amount and asking price of various resources.

[0157] Step 4: The base station sorts the MEC servers k in descending order of preference according to the collected information, accesses the MEC servers k one by one in order of preference from high to low, selects the eMBB user m requesting the MEC server k with the preference, and if the bid of the eMBB user m is ≥ the asking price of the requested MEC server k, then the eMBB user m is added to the candidate list and the process goes to step 5; otherwise, the eMBB user m is added to the failure list and the process goes to step 6;

[0158] Step 5: traverse all eMBB users m who request the MEC server k with the preference, sort the bids of eMBB users m in the candidate list in descending order, and judge that the total amount of resources required by eMBB users m in the candidate list is ≤ the remaining amount of resources of the MEC server k with the preference. If so, the match is successful, and the remaining amount of resources of the MEC server k with the preference is updated; otherwise, the eMBB users m with low bids are added to the failure list in the order of the bids of eMBB users m in the candidate list from low to high, until the total amount of resources required by eMBB users m in the candidate list is ≤ the remaining amount of resources of the MEC server k with the preference, the match is successful, and the process goes to step 7; if all eMBB users m who request the MEC server k with the preference are not traversed, return to step 1;

[0159] Step 6: The eMBB user m added to the failure list sends a request to the MEC server k of the next preference, and the process returns to step 2. If the MEC server k of all the preference levels of the eMBB user m cannot be matched, the current auction ends and the next auction begins. In the next auction, the eMBB user m modifies the bid, and the MEC server k modifies the asking price, and the process returns to step 4.

[0160] Step 7: If the remaining resources in the MEC server k are insufficient to continue matching or all eMBB users m and URLLC users have been matched to the corresponding MEC server k, the auction ends; otherwise, return to step 1.

[0161] Accordingly, the present invention also provides a computation offloading system based on an auction algorithm, comprising:

[0162] The regional transmission module is used to establish the MEC server regional transmission relationship where eMBB users and URLLC users coexist;

[0163] The double auction model is used to dynamically adjust the buyer's bid and the seller's offer with eMBB users and URLLC users as buyers and MEC servers as sellers, and conduct multiple rounds of auctions according to constraints to match MEC servers for eMBB users and URLLC users. The computing tasks of eMBB users and URLLC users are offloaded to the matched MEC servers.

[0164] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

[0165] A computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, which when executed by a computing device, cause the computing device to perform a computation offloading method based on an auction algorithm.

[0166] A computing device includes one or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and are configured to be executed by the one or more processors, and the one or more programs include instructions for executing a computation offloading method based on an auction algorithm.

[0167] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0168] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0169] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0170] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0171] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are included in the scope of the claims of the present invention to be approved.

Claims

1. A computation offloading method based on an auction algorithm, characterized in that: Establish a MEC server area transmission model where eMBB users and URLLC users coexist; Based on the regional transmission model, a double auction model is established with the eMBB user and the URLLC user as the buyer and the MEC server as the seller; To maximize the seller's profit, formulate the constraints of the double auction model; The double auction model dynamically adjusts the buyer's bid and the seller's offer, and conducts multiple rounds of auctions according to the constraints to match MEC servers for eMBB users and URLLC users; Offload the computing tasks of eMBB users and URLLC users to the matched MEC servers; The two-way auction model dynamically adjusts the buyer's bid and the seller's offer, and conducts multiple rounds of auctions according to the constraints. The method for matching MEC servers for eMBB users and URLLC users is as follows: Step 1: If a URLLC user sends a request to MEC server k, the URLLC user is matched with the requested MEC server k; Step 2: rank the preferences of the eMBB user m for the MEC server k, and select the MEC server k that meets the constraint conditions according to the constraint conditions; Step 3: Assume that the base station is the auctioneer and the eMBB user m is the buyer. The eMBB user m provides the base station with service requirements, valuation of the required resources, preference ranking and bid for the MEC server k. In the first round of auction, the valuation of the required resources is used as the bid of the eMBB user m. The MEC server k is the seller and provides the base station with the total amount and asking price of various resources. Step 4: The base station sorts the MEC servers k in descending order of preference according to the collected information, accesses the MEC servers k one by one in order of preference from high to low, selects the eMBB user m requesting the MEC server k with the preference, and if the bid of the eMBB user m is ≥ the asking price of the requested MEC server k, then the eMBB user m is added to the candidate list and the process goes to step 5; otherwise, the eMBB user m is added to the failure list and the process goes to step 6; Step 5: traverse all eMBB users m who request the MEC server k with the preference, sort the bids of eMBB users m in the candidate list in descending order, and judge that the total amount of resources required by eMBB users m in the candidate list is ≤ the remaining amount of resources of the MEC server k with the preference. If so, the match is successful, and the remaining amount of resources of the MEC server k with the preference is updated; otherwise, the eMBB users m with low bids are added to the failure list in the order of the bids of eMBB users m in the candidate list from low to high, until the total amount of resources required by eMBB users m in the candidate list is ≤ the remaining amount of resources of the MEC server k with the preference, the match is successful, and the process goes to step 7; if all eMBB users m who request the MEC server k with the preference are not traversed, return to step 1; Step 6: The eMBB user m added to the failure list sends a request to the MEC server k of the next preference, and the process returns to step 4. If the MEC server k of all the preference levels of the eMBB user m cannot be matched, the current auction ends and the next auction begins. In the next auction, the eMBB user m modifies the bid, and the MEC server k modifies the asking price, and the process returns to step 4. Step 7: If the remaining resources in the MEC server k are insufficient to continue matching or all eMBB users m and URLLC users have been matched to the corresponding MEC server k, the auction ends; otherwise, return to step 1.

2. The auction algorithm-based computation offloading method according to claim 1, characterized in that: The regional transmission model has K base stations, M eMBB users and N URLLC users randomly distributed in the area. Each user has a computing task to be executed. It is assumed that a large number of URLLC users will not request at the same time, so M>>N. Each base station is equipped with an MEC server. Users choose to perform calculations on local devices or offload them to the MEC server for processing.

3. The computation offloading method based on auction algorithm according to claim 1, characterized in that: The parameters of the regional transmission model include: eMBB service rate Among them, N m is the number of subcarriers allocated to eMBB user m, W is the subcarrier bandwidth of the frequency band resource, P m is the transmit power of eMBB user m, N0 is the additive white Gaussian noise power, g m is the small-scale Rayleigh fading of eMBB users, G m is the path loss of eMBB user m, expressed as: G m =128.1+37.6lg(d m ); Among them, d m is the distance from eMBB user m to the base station; The local calculation latency of eMBB user m Among them, c m To handle the computing power required for this task, It is local computing power; eMBB user m local computing energy consumption Among them, σ is the energy consumption factor, which is related to the chip performance of the eMBB user equipment; The offloading computation delay of eMBB user m consists of two parts: task offloading delay and MEC server computation delay. Among them, R m is the achievable rate of the eMBB service; m is the task bit amount of eMBB user m; MEC server calculation delay in, The computing resources allocated by the MEC server for the task of eMBB user m; Total offloading calculation delay of eMBB user m set up The binary parameter of the eMBB service is an allocation parameter, which is used to determine whether the computing task of the eMBB user m is offloaded to the MEC server k for processing: The energy consumption required for computing the data offloaded by eMBB user m on MEC server k The total delay of offloading calculation is changed to: The total energy consumption of unloading calculation is changed to: Among them, ξ m is the delay energy cost factor of eMBB user m; The computation offloading delay of URLLC user n is: Where R n is the achievable rate of URLLC service; n is the task bit amount of URLLC user n, c n The computing power required to process the tasks of URLLC user n, is the computing resources allocated by the MEC server for the task of URLLC user n, MEC server computing delay in the offloading computing delay of URLLC user n; set up The binary parameter for URLLC services is the allocation parameter, which is used to determine whether the computing task of URLLC user n is offloaded to MEC server k for processing:

4. The auction algorithm-based computation offloading method according to claim 3, characterized in that: The method of establishing a two-way auction model with eMBB users and URLLC users as buyers and MEC servers as sellers is as follows: The cost of MEC server k is expressed as: C k =λ kw W k +λ kd D k +λ kf F k ; Among them, W k , D k , F k are the total communication, storage and computing resources of server k, respectively, kw ,λ kd ,λ kf These are the cost influencing factors of the three types of resources. The larger the server resources, the greater the cost of its construction and maintenance. The resource valuation of eMBB user m is expressed as: Among them, β m and 1-β m are the estimated weight parameters of service delay and energy consumption of user m, and their sum is 1; δ mT is the delay cost factor of eMBB user m, δ mE is the energy consumption cost factor of eMBB user m; Similarly, the resource valuation of URLLC user n is expressed as: Among them, δ nT is the delay cost factor of URLLC user n; The bid of eMBB user m in the t+1 round of auction is: Among them, t represents the current auction round number, Sum represents the total number of auction rounds, Pro m is the priority of the eMBB user task, ε b To adjust the parameters of user bidding; URLLC user n's bid for the t+1 round of auction is: b n (t+1)=η t Pro n b n (t); Among them, Pro n is the URLLC user task priority, η is the URLLC service bid index; The auction asking price of MEC server k in the t+1 round is expressed as: Among them, ε a To adjust the parameter of MEC server price, μ k (t) is the parameter that affects the quotation due to the remaining resources of the MEC server; Among them, w kt , f kt , d kt They represent the remaining communication resources, computing resources, and storage resources of server k in round t respectively; as the auction progresses, the amount of resources gradually decreases, and μ k As (t) decreases, the asking price of the MEC server also decreases, allowing the server to maintain a certain advantage and improve the success rate of the auction.

5. The auction algorithm-based computation offloading method according to claim 4, characterized in that: With the goal of maximizing seller revenue, the method for formulating the constraints of the double auction model is: The benefit that eMBB user m brings to MEC server k is expressed as the reward paid by eMBB user m minus the cost of MEC server k, that is: Similarly, the benefit that URLLC user n brings to MEC server k can be expressed as: The total revenue of the MEC server is written as the sum of the revenue brought by each user to the corresponding MEC server: The objective function to maximize the seller's profit is: Among them, constraints C1 and C2 indicate that if the task of each eMBB or URLLC user chooses to offload computing, it can only be completed by one MEC server k; constraint C3 indicates that each computing task of the user can only be offloaded as a whole or locally calculated; constraints C4, C5 and C6 indicate that the communication, storage and computing resources consumed by the two service users cannot exceed the total resource inventory of the server, where w m,k , w n,k , f m,k , f n,k , d m,k , d n,k They respectively represent the corresponding resource amounts provided by server k to user m or n; constraint C7 indicates that the delay energy consumption cost factor of eMBB user m is not less than 0.

6. A computation offloading system based on an auction algorithm, characterized in that: The system is used to execute the auction algorithm-based computation offloading method according to any one of claims 1 to 5, comprising: The regional transmission module is used to establish the MEC server regional transmission relationship where eMBB users and URLLC users coexist; The double auction model is used to dynamically adjust the buyer's bid and the seller's offer with eMBB users and URLLC users as buyers and MEC servers as sellers, and conduct multiple rounds of auctions according to constraints to match MEC servers for eMBB users and URLLC users. The computing tasks of eMBB users and URLLC users are offloaded to the matched MEC servers.

7. The auction algorithm-based computation offloading system according to claim 6, characterized in that: Also included is a computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any one of the methods according to claims 1 to 5.

8. The auction algorithm-based computation offloading system according to claim 6, characterized in that: Also included is a computing device, the computing device comprising: One or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and are configured to be executed by the one or more processors, and the one or more programs include instructions for executing any of the methods according to claims 1 to 5.

Images