Service offloading method and system based on multi-level offloading network

Abstract

The application provides a service offloading method and system based on a multi-level offloading network, and the method comprises the following steps: a resource pricing scheme of a device is proposed, a multi-level task offloading algorithm MOS is designed to maximize the income of a user completing a task; the multi-level task offloading algorithm MOS takes the user as the center and takes the maximum completion delay of the task as the constraint to construct a multi-level task offloading network, and through layer-by-layer pricing, the user can determine the final task offloading scheme. The application solves the technical problems of resource waste, low utilization rate of auxiliary computing devices in the community and low income of task requesters.

Description

Technical Field

[0001] This invention relates to the field of service computing and D2D (Device-to-Device) communication, specifically to a service offloading method and system based on a multi-level offloading network. Background Technology

[0002] As the communication and computing capabilities of smart devices become increasingly powerful, more and more people need to use these devices to access online information and handle tasks. This rapid increase in traffic has brought numerous challenges to communication systems, with edge servers facing significant difficulties such as overload and scarce spectrum resources. However, Data-to-Device (D2D) can effectively solve these problems. Therefore, for future sixth-generation (6G) mobile networks, increasing research is attempting to introduce the mobile terminal layer into a cloud-edge two-layer computing paradigm. D2D can help edge servers respond to user requests, such as task computing requests and resource requests, thereby alleviating the immense pressure on edge servers.

[0003] In the D2D computation offloading problem, task requesters transmit their tasks to resource-rich idle devices (referred to as auxiliary computing devices) via direct device connections, requesting computational assistance. The auxiliary computing devices then complete the task computation using their own resources. Since most devices are private and exhibit selfishness, how to incentivize devices to participate in offloading is a widely studied problem. Without sufficient motivation and incentives, auxiliary computing devices are likely to avoid participating in the offloading network, resulting in resource waste. Furthermore, due to the limited communication range of devices, the auxiliary computing devices around the task requester only represent a portion of all auxiliary computing devices in that community (e.g., residential area, industrial zone). Therefore, using a single-hop direct connection method will inevitably lead to low utilization of auxiliary computing devices within the community. Therefore, a computation offloading mode needs to be designed to incentivize more auxiliary computing devices to participate in task offloading, improve the utilization of idle devices, and optimize the rewards for task requesters.

[0004] The existing invention patent application document CN111343595A, entitled "A D2D Task Offloading System and a Method for Transaction of Multiple Types of Resources," describes a method that includes: first, grouping mobile users according to the connected components of the network topology to improve resource allocation efficiency; then, pricing resources using a McAfee bilateral auction; and finally, dividing mobile users into multiple virtual traders and using a maximum matching method to obtain the final resource allocation scheme. From the specific implementation details of this existing scheme, it is clear that each mobile user reports their resource demand / supply and their bid / ask price for each type of resource. The goal of this existing scheme is to maximize the number of successful service transactions through an auction, where the final pricing for multiple resources is jointly determined by the bids / asks of all buyers and sellers. The number of successful transactions is maximized by determining the final resource transaction price. This auction method requires a server or other device to act as the auctioneer, a central control node with global information. This existing scheme carries the risk of a single point of failure.

[0005] The existing invention patent application document CN111107153A, entitled "A MEC Pricing Offloading Method Based on D2D Communication in Power Internet of Things," describes a method that includes: uniformly distributing tasks within a team using D2D data communication technology; each paired device determining its own data offloading amount; determining a pricing set and sorting the pricing set; the MEC server broadcasting a pricing scheme from the pricing set in sequence; inspection devices determining their own offloading scheme and uploading it to the MEC server; a cloud server determining its own revenue based on the inspection devices' calculation schemes and determining whether to update the pricing scheme according to the revenue maximization principle; and the cloud server iteratively outputting the optimal pricing scheme. The key focus of this existing solution is resource pricing, which directly affects the task offloading decisions of the devices. Furthermore, this existing solution uses a game theory approach to determine pricing and offloading decisions, meaning that both the price and offloading decisions are influenced by both the resource provider and the requester.

[0006] The existing literature, "Research on D2D-Assisted Computation Offloading Strategy in Mobile Edge Computing Networks," argues that since computing and communication resources in a network are always limited, reasonable resource allocation based on network conditions can improve the overall performance of the computation offloading system and ensure the quality of service for all users. Resource allocation during computation offloading aims to minimize task completion latency, energy consumption, or a trade-off between latency and energy consumption when allocating server computing and network communication resources. In the computation resource allocation problem, given limited server computing resources, it is typically necessary to find a reasonable scheme to allocate these limited resources to all users, thereby optimizing the objective function. The existing literature describes a task offloading scheme that combines D2D and edge servers, requiring the edge server to function as a central control node with a global perspective. Regarding user incentives, this literature employs a combination of social influence and reward incentives to encourage device participation in offloading; that is, the current participation in offloading affects the device's subsequent offloading actions. This literature focuses on multi-requester scenarios and does not consider single-user task offloading applications. While it uses a game theory approach to solve the problem, the game theory is highly complex and the algorithm execution time is long, especially in large-scale application scenarios.

[0007] In summary, existing technologies suffer from technical problems such as resource waste, low utilization of auxiliary computing devices within the community, and low returns for task requesters. Summary of the Invention

[0008] The technical problem to be solved by this invention is: how to solve the problems of resource waste, low utilization rate of auxiliary computing equipment in the community, and low benefits for task requesters in the prior art.

[0009] This invention solves the above-mentioned technical problems by employing the following technical solution: A service offloading method based on a multi-level offloading network includes:

[0010] S1. Construct a network model, where the devices in the network model include: user equipment and service provider equipment;

[0011] S2. Abstract the network model by using pre-defined quadruples to describe the devices in the network model and the sets of tasks generated by those devices:

[0012] S3. Execute the pricing algorithm of the quantitative resource model, wherein the demand function is obtained based on transaction history information, the requested price and the demand rate of n resources are calculated, so as to fit the relationship between the price and demand of a single resource, and the resource pricing problem is transformed into the problem of maximizing the expected revenue of the equipment, so as to obtain the optimal revenue of the equipment and the optimal price of the resource.

[0013] S4. Centered on the user equipment and constrained by the maximum task completion delay, construct a multi-level task offloading network, execute the multi-level offloading network task offloading algorithm MOS to perform layer-by-layer quotation operation, and obtain the applicable task offloading scheme.

[0014] This invention, based on a multi-layered offloading network, introduces a demand function to enable service pricing to adapt to changing demand, and designs a D2D service offloading method. The computational offloading mode designed in this invention incentivizes more auxiliary computing devices to participate in task offloading, improves the utilization rate of idle devices, and optimizes the revenue of task requesters.

[0015] The goal of this invention is to maximize the benefits of user resource offloading. Resource pricing is determined solely by the device itself, and is related to the device's remaining resource quantity and remaining sales time. Therefore, resource prices are not affected by other devices and remain unchanged once determined.

[0016] In a more specific technical solution, S2 includes:

[0017] S21. Perform device definition operation, where a pre-defined quadruple (d,s,p,r) is used to describe the device, where d is the device number; s represents the connection vector between devices; p represents the price list; and r represents the task set.

[0018] S22. Perform task set definition operation. The task set consists of no less than 2 sub-tasks. The task set is described by a pre-defined quadruple (d,n,q,td), where d represents the unique identifier of the task set, n represents the size of the task set, q is the resource list, and td represents the deadline for the task set to be completed.

[0019] In a more specific technical solution, S3 includes:

[0020] S31. Using the following logic, map the task set to the request type vector:

[0021]

[0022] This represents the amount of resources 1, 2, ..., n required by each task in the task set S at each time step;

[0023] S32. Based on the transaction history information, the demand function is obtained:

[0024] σ=Ω(P)

[0025] In the formula, σ and P represent the request arrival rate and service price, respectively;

[0026] S33. Use the following logic to obtain the requested price:

[0027]

[0028] In the formula, q τ,λ This represents the price of resource λ at time slice τ;

[0029] S34. Based on the request type vector, demand function, and request price, determine the demand rate for n resources using the following logic:

[0030]

[0031] In the formula, (S1,S2,...,S m () represents the m types of services deployed on the device. It is the arrival rate of request i;

[0032] S35. Based on the request type vector, demand function, request price, and demand rates for n resources, the relationship between the price and demand of a single resource is fitted:

[0033] φ λ =Θ λ (q)

[0034] S36. To formalize the resource pricing issue of equipment;

[0035] S37. Determine the optimal revenue of the equipment using the following logic:

[0036]

[0037] S38. Determine the optimal price of a resource using the following logic:

[0038] q τ,λ =V * (λ,t,n)-V * (λ,t,n-1)+1

[0039] This invention proposes a resource pricing scheme for devices, aiming to maximize profits for users when offloading tasks in D2D communication. The invention designs a multi-level task offloading algorithm, MOS, to maximize the user's revenue from task completion.

[0040] This invention integrates market-based concepts into a service supply framework, pricing equipment resources. The interconnection between devices creates an offloading network, within which users seek the optimal resource purchasing options to maximize profits.

[0041] In a more specific technical solution, S36 includes:

[0042] S361. Express the expected revenue of the equipment using the following logic:

[0043]

[0044] S362. Using the following logic, the resource pricing problem can be transformed into a problem of maximizing the expected revenue of equipment:

[0045]

[0046] S363. Using the following demand function logic, the demand for resource λ is modeled as following a Poisson distribution with intensity Φ. λ stochastic processes:

[0047]

[0048] In the formula, both k and α are constants.

[0049] This invention, through the development of a service offloading algorithm, allows users to purchase resources from idle devices to complete tasks. A pricing function based on the device's own resource availability is also established, ensuring that all devices receive satisfactory returns.

[0050] In a more specific technical solution, the task offloading algorithm MOS in the multi-level offloading network of S4 includes:

[0051] S41. Construct a multi-level offloading network, in which: select suitable auxiliary computing devices from the devices; identify the predecessor and successor devices of the auxiliary computing devices to determine the task transfer direction;

[0052] S42. A user benefit function was developed to represent the revenue of each intermediate auxiliary computing device. The price payable by the intermediate auxiliary device for purchasing resources from subsequent devices was determined. The goal of maximizing user revenue was transformed into the goal of maximizing the revenue of each device, so as to minimize its own cost.

[0053] S43. Implement a multi-level offloading strategy, sort the pricing of intermediate auxiliary computing devices, determine and formulate offloading task allocation decisions based on the resource sales prices in the bidding table.

[0054] In a more specific technical solution, S41 includes:

[0055] S411. Based on the task size C, determine the total task execution time and transmission time:

[0056] t e +t i,j ≤τ j

[0057]

[0058] In the formula, t e It is the total execution time of the task, t i,j It is the transmission time, τ jThe delay constraint is the time limit for device j to receive the task, where I and O are the input and output sizes of a single task, respectively, r is the transmission rate, and j satisfies the delay constraint.

[0059] S412, Current device i subtracts t from the latency constraint in the message. i,j Then increment the hop count by 1 and send the message to the next device j;

[0060] S413. When none of the successor devices of the current device i can meet the delay constraint, the current device i is determined to be the last layer node of the multi-level offloading network.

[0061] The multi-level task offloading algorithm provided by this invention constructs a multi-level task offloading network with the user as the center and the maximum task completion delay as the constraint. The user decides the final task offloading scheme through a layer-by-layer bidding process.

[0062] In a more specific technical solution, S42 includes:

[0063] S421. Using the following logic, formulate the user benefit function:

[0064]

[0065] In the formula, This represents the value of resources required for a user to complete the task alone, while Cost' represents the actual value of the user's own resources consumed in completing the task. V represents the incentive given to device j when a user unloads a task to device j, and V represents the benefit that the user can obtain by completing the task. This invention studies the scenario of a single user unloading a task.

[0066] S422. Obtain the expected rate of return (ERR) for each intermediate auxiliary computing device. When an intermediate auxiliary computing device receives a bid table from a successor node, it multiplies the bid table by the current intermediate auxiliary computing device's expected rate of return (ERR). Based on the resource pricing table and the expected rate of return, it formulates its current bid table. The revenue of each intermediate auxiliary computing device is determined using the following logic:

[0067] W i =R i -Q i

[0068] The revenue generated from the auxiliary computing equipment at the last layer of the multi-level offloading network will be used as the resource price.

[0069] W′=Q′

[0070] S423. Based on the bidding table of device i, determine the reward R that the predecessor node will give to device i. iBased on the following logic, determine the price Q payable by the intermediate auxiliary computing device for purchasing resources from the subsequent device. i :

[0071]

[0072] In the formula, q represents the total cost of all resources used by device j to complete its assigned task. λ,τ Let λ represent the price of resource λ during time period τ, o represent the quantity of all resources, Su(i) represent the set of successor devices of device i, and Se(j) represent the set of time slices for selecting to purchase resource j.

[0073] S424. Using the following logic, the goal of maximizing user revenue is transformed into the goal of each device maximizing its own revenue, thereby minimizing its own cost:

[0074] minQ i

[0075] stj∈Su(i)

[0076]

[0077] In the formula, C is the total size of the task. It is the amount of tasks executed locally on device i, C i This is the amount of tasks received by device i.

[0078] This invention uses only a reward mechanism to incentivize devices to participate in unloading. The unloading behavior is a one-time event and will not affect subsequent unloading behaviors.

[0079] In a more specific technical solution, S43, the price list organization operation includes:

[0080] S431. Within the preset time constraints, calculate the time-slice execution price of a task based on the types and quantities of resources required by a task in each time slice.

[0081] S432. Based on the time slice with the lowest execution price, select the time slice with the lowest price, calculate the number of tasks m that can be processed in the lowest time slice, fill the price p required for the lowest price time slice into the price table, add the price p to the price of the previous row to get the price of the min(m-1,n-1) row; remove the current lowest price time slice from the preset time constraint range.

[0082] S433. Repeat step S432 until the price table is filled to n rows, where n represents the number of tasks.

[0083] In a more specific technical solution, S43 includes the following operations for calculating the price list:

[0084] S431' Receive bid lists from all successor devices;

[0085] S432' Multiply the bid table value received by the subsequent equipment by the expected rate of return (ERR) of the intermediate auxiliary calculation equipment;

[0086] S433', Set |Su(i)|+1 pointers p k , making p k =1, k∈Su(i). Initialize variables S=0, l=1;

[0087] S434': Based on the value pointed to by the pointer of the current |Su(i)|+1 table row, select the smallest y, and at the same time record the table number corresponding to table a, and perform the update operation using the following logic:

[0088] S = S + y, T'(l) = S;

[0089] S435', Update Bidding Table T a Value:

[0090] T a (m)=T a (m)-y,m∈{p a +1,p a +2,...,n}

[0091] Update table T a The pointers and task labels: p = p + 1, l = l + 1;

[0092] S436', Repeat steps S434' and S435' until l>n.

[0093] This invention generates an offloading network through device interconnection, allowing users to complete tasks at a lower cost and improving resource utilization across a wider range of devices. The invention also measured its scalability, revealing that the algorithm exhibits good scalability as the system scales up.

[0094] The scenario described in this invention is a fully distributed one, requiring no server involvement. Devices only need and can obtain status information from devices directly connected to them. The focus of this invention is to expand the offloading network and reduce offloading costs for users through interconnection and communication between devices.

[0095] In more specific technical solutions, service offloading systems based on multi-level offloading networks include:

[0096] The network model building module is used to build network models, where devices in the network model include: user equipment and service provider equipment;

[0097] The network device description module is used to abstract the network model. It uses pre-defined four-tuples to describe the devices in the network model and the task sets generated by the devices. The network device description module is connected to the network model construction module.

[0098] The pricing algorithm module is used to execute the pricing algorithm of the quantitative resource model. It obtains and obtains the demand function based on transaction history information, calculates the requested price and the demand rate of n resources, fits the relationship between the price and demand of a single resource, and transforms the resource pricing problem into the problem of maximizing the expected revenue of the equipment, so as to find the optimal revenue of the equipment and the optimal price of the resource.

[0099] The task offloading scheme acquisition module is used to construct a multi-level task offloading network centered on the user equipment and constrained by the maximum task completion delay, to execute the multi-level offloading network task offloading algorithm MOS to perform layer-by-layer pricing operations and obtain applicable task offloading schemes. The task offloading scheme acquisition module is connected to the pricing algorithm module and the network device description module.

[0100] The present invention has the following advantages over the prior art:

[0101] This invention, based on a multi-layered offloading network, introduces a demand function to enable service pricing to adapt to changing demand, and designs a D2D service offloading method. The computational offloading mode designed in this invention incentivizes more auxiliary computing devices to participate in task offloading, improves the utilization rate of idle devices, and optimizes the revenue of task requesters.

[0102] This invention proposes a resource pricing scheme for devices, aiming to maximize profits for users when offloading tasks in D2D communication. The invention designs a multi-level task offloading algorithm, MOS, to maximize the user's revenue from task completion.

[0103] This invention integrates market-based concepts into a service supply framework, pricing equipment resources. The interconnection between devices creates an offloading network, within which users seek the optimal resource purchasing options to maximize profits.

[0104] This invention, through the development of a service offloading algorithm, allows users to purchase resources from idle devices to complete tasks. A pricing function based on the device's own resource availability is also established, ensuring that all devices receive satisfactory returns.

[0105] The multi-level task offloading algorithm provided by this invention constructs a multi-level task offloading network with the user as the center and the maximum task completion delay as the constraint. The user decides the final task offloading scheme through a layer-by-layer bidding process.

[0106] This invention generates an offloading network through device interconnection, allowing users to complete tasks at a lower cost and improving resource utilization across a wider range of devices. The invention also measured its scalability, revealing that the algorithm exhibits good scalability as the system scales up.

[0107] This invention solves the technical problems of resource waste, low utilization rate of auxiliary computing equipment in the community, and low benefits for task requesters in the prior art. Attached Figure Description

[0108] Figure 1 This is a schematic diagram of the basic steps of the service offloading method based on a multi-level offloading network in Embodiment 1 of the present invention;

[0109] Figure 2 This is a schematic diagram of an application scenario for a service offloading system based on a multi-level offloading network according to Embodiment 1 of the present invention.

[0110] Figure 3 This is a schematic diagram illustrating the specific steps of the task offloading algorithm MOS in the multi-level offloading network of Embodiment 1 of the present invention;

[0111] Figure 4 This is a schematic diagram illustrating the specific steps of compiling a price list in Embodiment 1 of the present invention;

[0112] Figure 5 This is an example diagram of the pricing table for Embodiment 1 of the present invention;

[0113] Figure 6 This is a schematic diagram illustrating the specific steps of calculating the price list in Embodiment 1 of the present invention;

[0114] Figure 7 This is an example diagram of user task uninstallation in Embodiment 2 of the present invention;

[0115] Figure 8 This is a first schematic diagram illustrating the effectiveness evaluation of the unloading algorithm in Embodiment 2 of the present invention;

[0116] Figure 9 This is a second schematic diagram illustrating the effectiveness evaluation of the unloading algorithm in Embodiment 2 of the present invention. Detailed Implementation

[0117] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0118] Example 1

[0119] like Figure 1 and Figure 2 As shown, the service offloading method based on a multi-level offloading network provided by this invention includes the following basic steps:

[0120] S1. Construct a network model, in which roles include, but are not limited to: users and service providers;

[0121] In this embodiment, the roles involved in the network model are:

[0122] Users: Task owners and producers, whose mobile device resources are limited and insufficient to support their task completion. They will purchase resources from other devices based on an uninstallation decision algorithm to complete the task.

[0123] Service Provider: A mobile device that has sufficient resources in addition to maintaining its own operation, and can earn revenue by selling its own resources.

[0124] S2. Perform abstraction operations on the network model:

[0125] In this embodiment, a device definition operation is performed, where both the user and the service provider are devices. Each device can be described by a quadruple (d, s, p, r), where each element has the following meaning:

[0126] (1) The device number;

[0127] (2) s connection vector, representing the connection status between the device and other devices;

[0128] (3) p Price list, which shows the price of each resource per unit of the equipment in the future;

[0129] (4)r task set: All devices may generate task sets. The device that generates the task set becomes the user and needs to purchase resources from other service provider devices to complete the task set.

[0130] In this embodiment, a task set definition operation is performed: a task set consists of multiple subtasks, and each subtask is equal. A task set can be described by a quadruple (d, n, q, td), where the meaning of each element is as follows:

[0131] (1) d is a unique identifier for the task set, which is used here to identify the user who created the task;

[0132] (2) n represents the size of the task set;

[0133] (3) q is a resource list, representing the number of different resource types required by each subtask in each time slice;

[0134] (4) td represents the deadline for the task set to be completed.

[0135] S3. Execute the pricing algorithm of the quantitative resource model, which introduces a demand function, calculates the requested price and the demand rate of n resources, fits the relationship between the price and demand of a single resource, and transforms the resource pricing problem into the problem of maximizing the expected revenue of the equipment, and finds the optimal revenue of the equipment and the optimal price of the resource.

[0136] In this embodiment, the auxiliary computing devices in the offloading network have sufficient resources, and selling these resources is their primary way of generating revenue. In this embodiment, the value of the resources is determined by both the remaining amount of resources and the remaining time available for sale. Since the task set proposed by the user contains multiple identical tasks, a task set can be mapped to a request type, vector:

[0137]

[0138] This represents the quantity of resources 1, 2, ..., n required by each task within the task set S at each moment. For price requests, the concept of a demand function from economics is introduced into the pricing algorithm of the quantified resource model in step S3.

[0139] In this embodiment, the aforementioned demand function can be obtained from transaction history:

[0140] σ=Ω(P)

[0141] Where σ and P represent the request arrival rate and service price, respectively. Thus, the request price is expressed as follows:

[0142]

[0143] Where, q τ,λ Let λ represent the price of resource λ at time slice τ. Combining the above formula, the demand rates for n resources are expressed as follows:

[0144]

[0145] Among them, (S) 11 ,S2,...,S m ) represents the m types of services deployed on this device. It is the arrival rate of request i.

[0146] By combining the four formulas above, we can fit the relationship between the price and demand of a single resource:

[0147] Φ λ =Θ λ (q)

[0148] Once the unit price of resources and the demand relationship are determined, the resource pricing problem of formalized equipment will be addressed.

[0149] In this embodiment, for a given time t and resource type λ, n = ξ λ,t This indicates that at time t, there are n units of idle resource λ on the device, where n k,λ q represents the quantity of resource λ sold at time k. τ,λ Let λ represent the price of resource λ at time τ. In this embodiment, λ is used. Let λ represent the quantity of resources sold up to time τ. Therefore, the expected revenue of the equipment can be expressed as follows:

[0150]

[0151] In this embodiment, the resource pricing problem is transformed into the problem of maximizing the expected revenue of the equipment:

[0152]

[0153] Subsequently, the demand for resource λ is modeled as a Poisson distribution with intensity Φ. λ A stochastic process. Using the classic demand function:

[0154]

[0155] Where k and α are both constants and have no effect on the price trend with time and the amount of available resources. For ease of representation, in this embodiment, α = 1 is set. Then, the optimal revenue of the equipment can be expressed as follows:

[0156]

[0157] The optimal price for a resource can be calculated using the following formula:

[0158] q τ,λ =V * (λ,t,n)-V * (λ,t,n-1)+1

[0159] It is known that less remaining resources or more remaining time to sell will result in higher resource prices.

[0160] S4. Using the multi-level offloading network task offloading algorithm MOS, a multi-level task offloading network is constructed with the user as the center and the maximum task completion delay as the constraint. The user decides on the final task offloading scheme through a layer-by-layer bidding process.

[0161] like Figure 3 As shown, in this embodiment, the task offloading algorithm MOS of the multi-level offloading network further includes the following specific steps:

[0162] S41. Build the unloading network;

[0163] In this embodiment, since the task set has a maximum tolerable latency, the auxiliary computing devices suitable for participating in the task offloading must be carefully selected. Furthermore, to ensure the smooth and efficient execution of the offloading task, the predecessor and successor devices of each device in the offloading network must be identified, i.e., the direction of task transmission must be clearly defined. This is crucial for ensuring the coordination and orderliness of the offloading process. To maintain the accuracy of the bidding table received by each device from its successor devices, it must be ensured that if a device has multiple predecessor devices, it only selects one to communicate with for task transmission. This approach helps prevent conflicts and inaccuracies in the bidding process and ensures the effectiveness of the offloading strategy.

[0164] In this embodiment, when a user generates a set of tasks to be unloaded, they broadcast it to connected devices that meet the latency requirements for completing all tasks. The broadcast includes the number of tasks, the tolerable latency, and the number of hops. The receiving device calculates which connected devices meet the requirements—whether the time to complete all tasks plus the inter-task transmission time satisfies the experimental constraints. Since the unloading decision is not yet determined, C is used to represent the task size, i.e.:

[0165] t e +t i,j ≤τ j

[0166]

[0167] Among them, t e It is the total execution time of the task, t i,j It is the transmission time, τ j Let t be the latency constraint received by device j for the task, where I and O are the input and output sizes of a single task, respectively, and r is the transmission rate. If j satisfies the latency constraint, then device i will subtract t from the latency constraint in the message. i,j After incrementing the hop count by 1, the message is sent to device j, which then makes the same choice. If none of the successor devices of device i can meet the latency constraint, then device i is the last node to offload the network.

[0168] S42. A user benefit function was defined to represent the revenue of each intermediate auxiliary computing device (not the last layer of the network), and the price that the device should pay to purchase resources from its successor devices was determined. The goal of maximizing user revenue was transformed into each device maximizing its own revenue, thereby minimizing its own cost.

[0169] In this embodiment, the focus is on the user's perspective, aiming to maximize the user's efficiency in completing the task. Therefore, a user efficiency function is defined:

[0170]

[0171] in, The first two terms represent the value of resources a user needs to consume to complete the task alone, while Cost' represents the value of the user's actual resources consumed in completing the task. Therefore, the first two terms represent the value of resources saved by the user in completing the task. These saved resources may still be sold, and are referred to as the user's retained earnings from completing the task. V represents the incentive given to device j when a user offloads a task to device j, and V represents the reward the user can obtain by completing the task. The latter two terms are called the user's actual reward for completing the task. Therefore, in order to maximize their own benefits, users will allocate as many tasks as possible to devices with lower bids, pay lower rewards, and retain their higher-value resources.

[0172] In this embodiment, since the devices are all selfish, each device wants to maximize its own revenue. However, the devices are unaware of each other's resource information. If a device bids too high, it may not be able to obtain tasks from its predecessor node, resulting in zero revenue for itself. Therefore, a device's bid considers not only its own resource pricing but also the resource bids of other connected devices. Here, we need to distinguish between two concepts: pricing table and bidding table. For the last layer device offloading the network, the pricing table is a price table created by pricing the device's own resources using the pricing method described above, while the bidding table is a compilation of its own resource pricing table. Since the pricing of resources is related to the remaining selling time and the remaining amount of resources, the revenue obtained by the device is not linearly related to the number of resources sold. To minimize the bid, the device will successively search for the combination of time slices with the lowest price for executing n tasks within the total task execution time range and create a bidding table. For intermediate layer nodes (not the last layer node), the device's bidding table is a price table jointly formulated by the device based on its own compiled resource pricing table and the bidding tables of subsequent devices, and then the device sends this bidding table to the predecessor node.

[0173] Meanwhile, to avoid situations where devices simply act as relay nodes without generating revenue, each device has a coefficient, which we call the Expected Rate of Return (ERR). After receiving the bidding table from subsequent nodes, the device multiplies it by its own ERR and then combines this with its own resource pricing table to formulate its own bidding table. This ensures that even if a device doesn't sell its resources, it can still generate revenue by participating in task transmission. This also encourages devices to actively expand the offloading network and find better offloading strategies for users. This explains the selection method for the optimal predecessor device mentioned earlier: each additional hop between the device and the user adds a relay fee, reducing the device's probability of receiving tasks. The revenue of each intermediate auxiliary computing device (not the last layer of the network) can be expressed as:

[0174] W i=R i -Q i

[0175] In this embodiment, the revenue of the auxiliary computing devices at the last layer of the network is the selling price of the resources:

[0176] W′=Q′

[0177] Among them, R i Q is the reward given to device i by the predecessor node, which is determined by the predecessor node based on device i's bidding table. i Q is the price that the equipment pays to its successor equipment for purchasing resources. i It is expressed as follows:

[0178]

[0179] in, q represents the total cost of all resources used by device j to complete its assigned task. λ,τ Let λ represent the price of resource λ during time period τ, o represent the quantity of all resources, Su(i) represent the set of successor devices of device i, and Se(j) represent the set of time slices for selecting to purchase resource j.

[0180] In this embodiment, device i cannot control R. i To maximize revenue, device i will choose the offloading strategy with the lowest bid, i.e., minimizing Q. i Therefore, minimizing the offloading cost at each node transforms the goal of maximizing user benefits into maximizing the benefits of each device itself, i.e., minimizing its own costs.

[0181] minQ i

[0182] stj∈Su(i)

[0183]

[0184]

[0185] Where C is the total size of the task. It is the amount of tasks executed locally on device i, C i This is the amount of tasks received by device i.

[0186] S43. Implement a multi-level unloading strategy, sort the pricing of equipment, reflect the price of resource sales in a price list, and make task allocation decisions.

[0187] In this embodiment, solving the task allocation problem requires addressing two parts: determining the bidding table and making offloading decisions for each device. To determine the bidding table, a bottom-up approach is used, starting from the last layer of the offloading network. Since the pricing of device resources depends on remaining sales time and remaining resource quantity, the pricing does not increase linearly. Therefore, it is necessary to sort the device prices to ensure that the bidding table accurately reflects the selling price of resources, thereby enabling more effective task allocation decisions.

[0188] like Figure 4 and Figure 5 As shown, in this embodiment, the specific steps for organizing the price list include:

[0189] S431. Within the time constraints, calculate the price required to execute a task in a given time slice based on the types and quantities of resources required per time slice for a task.

[0190] S432. Select the time slice with the lowest price required to execute tasks, calculate the number of tasks m that can be processed in this time slice, and fill the required price p into the revised pricing table. This means that the price required to execute a task is p, and the prices in the subsequent min(m-1,n-1) rows are equal to the price in the previous row plus p. Finally, remove this time slice from the selection range.

[0191] S433. Repeat step S432 until the compiled price list is filled with n rows, where n is the number of tasks.

[0192] In the price list after the aforementioned processing steps, the price for task execution remains monotonically unchanged as the number of tasks increases. After processing all devices, we start calculating the pricing table from the second-to-last layer device in the unloading network.

[0193] like Figure 6 As shown, in this embodiment, the specific steps for calculating the price list include:

[0194] S431' Receive bid lists from all successor devices;

[0195] In this embodiment, the bid table is denoted as T. b ,b∈Su(i). T 0 T' represents the pricing table after the equipment has been organized, and T' represents the pricing table i of the equipment, which is initially empty.

[0196] S432' Multiply the bid table value received by the subsequent device by the device's own ERR value;

[0197] S433', Set |Su(i)|+1 pointers p k , making p k=1, k∈Su(i). Initialize variables S=0, l=1.

[0198] In this embodiment, the value of the pointer represents the amount of tasks assigned to device k. Initially, they all point to the first row of the table. S represents the total price for unloading the current amount of tasks, and l represents the task number currently being assigned.

[0199] S434': Based on the value pointed to by the current |Su(i)|+1 table row pointer, select the smallest value as y, and simultaneously record the value in table a.

[0200] For the corresponding table number, update S = S + y, T'(l) = S;

[0201] S435', Update Table T a The value;

[0202] In this embodiment, table T is updated using the following logic. a Value:

[0203] T a (m)=T a (m)-y,m∈{p a +1,p a +2,...,n}

[0204] Update table T a The pointers and task labels: p = p + 1, l = l + 1.

[0205] S436', Repeat steps S434' and S435' until l>n.

[0206] After the above steps are completed, table T' is the bidding table for device i. Once all penultimate layer devices transmit their bidding tables to their respective predecessor devices, the third-to-last layer devices in the offloading network begin the same operation until the first layer devices in the offloading network report their bidding tables to the user devices. Note that in step 5, the bidding table selected in step 4 is modified, and each subsequent selection is based on the modified value. The reason for this is to find the device with the smallest price difference when executing task k, that is, given that the allocation decisions for the first k-1 tasks have been made, select the minimum reward required for all devices to execute task k. It is easy to see that this method is a decision based on a greedy strategy, which selects the optimal solution for the next step based on the current state, and often only achieves a local optimum rather than a global optimum. However, this method can obtain a globally optimal solution in this problem, which is why the price table is organized before calculating the device bidding table. The organized price table satisfies the following conditions:

[0207] And x, y > 0. When 0 < x < y < n, there is T a (y) - T a (y - 1) ≥ T a (y) - T a (y - 1)

[0208] ≥ T a (x) - T a (x - 1).

[0209] After satisfying the foregoing conditions, for each optimal decision made in the current state of the price list, it is a step closer to the global optimal solution. Thus, the first part of this problem, that is, the determination part of the bid list, has been well solved.

[0210] As for the second part, the user offloading decision problem, it is the reverse process of the first part. For the solution of this part of the problem, starting from the user device and along the direction of deepening the offloading network layer by layer, each device performs the following operations:

[0211] Set |Su(i)| + 1 pointers p k , such that p k = 1, k ∈ Su(i). Initialize the variable l = 1, and the initial pointers all point to the first row of the table. l represents the task number currently being allocated; according to the values in the rows pointed to by the pointers in the current |Su(i)| + 1 tables, select the smallest one and denote it as |Su(i)| + 1, and at the same time record the corresponding table number a; update the value of the table T a , T a (m) = T a (m) - y, m ∈ {p a + 1, p a + 2,..., n}. Update the pointers and task labels of the table T a : p = p + 1, l = l + 1; repeat the foregoing sub-steps until l > n', where n' is the number of tasks allocated to device i by its predecessor device. For the user device, n' = n; obtain the task allocation scheme according to the pointer pointing situation of each table. For example, if p a = 3, then allocate 2 tasks to the successor device a, and calculate its own revenue according to this task allocation decision.

[0212] In this embodiment, each device must follow the steps from the user device to the last layer of the offloading network to determine the number of tasks to be allocated to it. Once the task allocation decision is made, the user and the participating devices can calculate their respective revenues respectively.

[0213] Embodiment 2

[0214] In this embodiment, video processing is used as an example to describe the user offloading process. After a user generates a task, they calculate the cost required to complete the task based on the pricing method proposed in this paper. Then, based on the task's deadline, they calculate which connected devices can complete the task on time and send a message containing the task size, deadline, and hop count to these devices via a D2D connection. These devices form the first layer of the offloading network. After the aforementioned devices return a confirmation message to the user, this process is repeated until the offloading network is fully constructed.

[0215] like Figure 7 As shown, in this embodiment, the last-layer device in the network aggregates the resource prices for the video processing time into a price list and organizes it, then sends it to the upper-layer device via a message. The upper-layer device receives the price lists of all subsequent devices, multiplies them by its own ERR (Earning Resource Rate), and combines this with its own price list to select the shortest time combination of resource prices required to process the entire task. This is then compiled into a bid list and returned to its predecessor device. This process continues layer by layer until the user collects all the bid lists from its subsequent devices, compares them with its own resource price list, formulates a final task offloading strategy, and then performs the task offloading.

[0216] In this embodiment, based on the model provided by the present invention, we set up 22 devices, randomly selected one device as a user, and simulated the task unloading process. We also set up three other comparative methods for a control experiment:

[0217] FDS: This method is also a multi-device, multi-level task offloading method, which finds the most efficient offloading solution for users through a layered offloading process. However, this method does not consider the benefits of participating offloading devices simply acting as relay devices. The method in this chapter addresses this issue.

[0218] SOSRAS: This method selects one device from among many connected devices to partially offload tasks, minimizing the overall task completion time. Users receive additional rewards when tasks are completed before the deadline, with higher rewards for shorter completion times.

[0219] TAROC: This method is a two-stage auction method. When a device directly connected to the user successfully bids for a task, it will auction the task again to earn revenue.

[0220] In this embodiment, to verify the effectiveness of the proposed method under varying task loads, we changed the task load and recorded the user's earnings.

[0221] like Figure 8As shown, in this embodiment, the user benefits obtained by the aforementioned four methods increase with the increase in the number of tasks. This is reasonable because each task brings a certain benefit to the user, and the more tasks a user has, the greater the total benefit. When the number of tasks is small, the differences between the aforementioned four methods are not significant. However, as the number of tasks increases, the performance of the method presented in this paper is significantly better than the other three methods. This is because in the pricing methods mentioned above, the difference in resource prices between different time slices is not very large, and when the number of tasks is small, the number of time slices involved is small, so the difference in user benefits is not very obvious. As the number of tasks increases, the number of time slices available for users to choose from in the entire offloading network also increases. The results show that when the difference in resource prices under different states accumulates over multiple time slices, the difference in user benefits is significant.

[0222] like Figure 9 As shown, in this embodiment, the impact of the number of devices at each layer on the method is examined. Specifically, an offload network with a network depth of 2 is created, and the number of subsequent devices is varied for each device in the network. Then, changes in user revenue are recorded, and... Figure 9 The experimental results are plotted below. As the number of subsequent devices increases, the user revenue growth rate of both the proposed method and the FDS method gradually slows down and eventually levels off. This can be attributed to the fact that as the number of subsequent devices increases, the range of options available to the user also increases. However, once the number of subsequent devices reaches a certain value, it becomes difficult to find a better allocation scheme, resulting in no significant change in user revenue. When users use the proposed method for uninstallation, the revenue is significantly better than other methods.

[0223] In summary, this invention, based on a multi-layered offloading network and incorporating a demand function to allow service pricing to adapt to changing demand, designs a D2D service offloading method. The computational offloading mode designed in this invention incentivizes more auxiliary computing devices to participate in task offloading, improves the utilization of idle devices, and optimizes the revenue for task requesters.

[0224] This invention proposes a resource pricing scheme for devices, aiming to maximize profits for users when offloading tasks in D2D communication. The invention designs a multi-level task offloading algorithm, MOS, to maximize the user's revenue from task completion.

[0225] This invention integrates market-based concepts into a service supply framework, pricing equipment resources. The interconnection between devices creates an offloading network, within which users seek the optimal resource purchasing options to maximize profits.

[0226] This invention, through the development of a service offloading algorithm, allows users to purchase resources from idle devices to complete tasks. A pricing function based on the device's own resource availability is also established, ensuring that all devices receive satisfactory returns.

[0227] The multi-level task offloading algorithm provided by this invention constructs a multi-level task offloading network with the user as the center and the maximum task completion delay as the constraint. The user decides the final task offloading scheme through a layer-by-layer bidding process.

[0228] This invention generates an offloading network through device interconnection, allowing users to complete tasks at a lower cost and improving resource utilization across a wider range of devices. The invention also measured its scalability, revealing that the algorithm exhibits good scalability as the system scales up.

[0229] This invention solves the technical problems of resource waste, low utilization rate of auxiliary computing equipment in the community, and low benefits for task requesters in the prior art.

[0230] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A service offloading method based on a multi-level offloading network, characterized in that, The method includes: S1. Construct a network model, wherein the devices in the network model include: user equipment and service provider equipment; S2. Abstract the network model by using pre-defined quadruples to describe the devices in the network model and the task sets generated by the devices: S3. Execute the pricing algorithm of the quantitative resource model, which involves obtaining and based on historical transaction information, acquiring the demand function, and calculating the request price. n The demand rate of various resources is used to fit the relationship between the price and demand of a single resource, transforming the resource pricing problem into a problem of maximizing the expected revenue of equipment, in order to find the optimal revenue of equipment and the optimal price of resources. S4. With the user equipment as the center and the maximum task completion delay as the constraint, construct a multi-level task offloading network, execute the multi-level offloading network task offloading algorithm MOS, perform layer-by-layer quotation operation, and obtain the applicable task offloading scheme. The task offloading algorithm MOS of the multi-level offloading network includes: S41. Construct the multi-level offloading network, wherein, from the devices, select an applicable auxiliary computing device; identify the predecessor and successor devices of the auxiliary computing device to determine the task transfer direction; S42. A user benefit function is defined to represent the revenue of each intermediate auxiliary computing device. The price payable by the intermediate auxiliary computing device for purchasing resources from subsequent devices is determined. The goal of maximizing user revenue is converted into the goal of maximizing the revenue of each device, so as to minimize its own cost. S43. Execute a multi-level offloading strategy, sort the pricing of the intermediate auxiliary computing devices, determine and formulate an offloading task allocation decision based on the resource selling price in the bidding table.

2. The service offloading method based on a multi-level offloading network according to claim 1, characterized in that, S2 includes: S21. Perform device definition operation, wherein the device is described using the preset quadruple (d, s, p, r), where d is the device number; s represents the connection vector between the devices; p represents the price list; and r represents the task set. S22. Perform a task set definition operation. The task set consists of no less than two sub-tasks. The task set is described using the preset quadruple (d, n, q, td), where d represents the unique identifier of the task set, n represents the size of the task set, q is the resource list, and td represents the deadline for the completion of the task set.

3. The service offloading method based on a multi-level offloading network according to claim 1, characterized in that, S3 includes: S31. Using the following logic, map the task set to the request type vector: This represents the number of resources 1, 2, ..., n required by each task in the task set S at each time step; S32. Based on the transaction history information, the demand function is obtained: In the formula, and P These represent the request arrival rate and the service price, respectively. S33. Calculate the requested price using the following logic: In the formula, Representing resources In time slice The price at that time; S34. Using the following logic, determine the... n Demand rate of the resource: In the formula, Indicates the deployment of the device Such services It is a request The arrival rate; S35. Based on the request type vector, the demand function, the request price, and the... The demand rate of a resource is used to fit the relationship between the price and demand of that single resource: S36. The resource pricing problem of the aforementioned equipment is formally expressed; S37. Determine the optimal benefit of the device using the following logic: S38. Determine the optimal price of the resource using the following logic: 。 4. The service offloading method based on a multi-level offloading network according to claim 3, characterized in that, S36 includes: S361. Express the expected revenue of the device using the following logic: ； S362. Using the following logic, the resource pricing problem is transformed into a problem of maximizing the expected revenue of the equipment: ； S363. Using the following demand function logic, the demand for resource λ is modeled as following a Poisson distribution with intensity... stochastic processes: In the formula, k and All are constants.

5. The service offloading method based on a multi-level offloading network according to claim 1, characterized in that, S41 includes: S411, Based on task size C Determine the total execution time and transmission time of the task: In the formula, This is the total execution time of the task. It is the transmission time. It is equipment Received the task's time delay constraint, and These are the input and output sizes of a single task, respectively. r It's the transmission rate. j Satisfy delay constraints; S412, Current Device Subtract the latency constraint from the message. Then increment the hop count by 1 and send the message to the next device. j ; S413, in the current device i When all of the successor devices fail to meet the delay constraint, the current device is determined to be... i This is the last layer node of the multi-level offloading network.

6. The service offloading method based on a multi-level offloading network according to claim 1, characterized in that, S42 includes: S421. Using the following logic, formulate the user benefit function: In the formula, This represents the value of resources required for a user to complete the task alone. This represents the actual value of the user's own resources consumed in completing the task. The user uninstalls the task to the device. j Time to give equipment j Incentives This indicates the reward a user can receive for completing the task; S422. Obtain the expected rate of return (ERR) for each of the intermediate auxiliary computing devices. When the intermediate auxiliary computing device receives the bidding table from the successor node, it multiplies the bidding table by the current expected rate of return (ERR) of the intermediate auxiliary computing device, and formulates the current bidding table for the intermediate auxiliary computing device based on the resource pricing table and the expected rate of return. The revenue of each intermediate auxiliary computing device is determined using the following logic: The revenue from the last-layer auxiliary computing device of the multi-level offloading network is used as the resource price: S423, Based on the predecessor node and the equipment i The aforementioned bid table determines the amount of money the predecessor node will offer to the device. i Remuneration Based on the following logic, the payable price for the intermediate auxiliary computing device to purchase resources from the subsequent device is determined. : In the formula, Indicates equipment j The sum of all resource costs required to complete the assigned task. Represents the time period resource The price For the quantity of all resources, It is equipment i The successor equipment set, Is it a choice to purchase equipment? j The set of time slices for resources; S424. Using the following logic, the goal of maximizing user revenue is transformed into the goal of maximizing the revenue of each device, in order to obtain the goal of minimizing its own cost: In the formula, C It is the total size of the task. It is equipment i The amount of tasks executed locally. It is equipment i The number of tasks received.

7. The service offloading method based on a multi-level offloading network according to claim 1, characterized in that, In step S43, the price list arrangement operation includes: S431. Within the preset time constraints, calculate the time-slice execution price of a task based on the types and quantities of resources required by a task in each time slice. S432. Based on the time slice with the lowest task execution price, select the time slice with the lowest price and calculate the number of tasks that can be processed in the lowest time slice. m The price required for the minimum time slice p Fill the price table by adding the price to the previous row. p ,get min(m-1,n-1) The price of the row; remove the current minimum time slice of the price from the preset time constraint range; S433. Repeat step S432 until the price table is filled to... n Okay, among them, n Indicates the number of tasks.

8. The service offloading method based on a multi-level offloading network according to claim 5, characterized in that, In step S43, the operation of calculating the bid table includes: S431' Receive the bid list from all subsequent devices; S432' Multiply the bid table value received by the successor device by the expected rate of return (ERR) of the intermediate auxiliary computing device; S433', Settings |Su(i)|+ 1 pointer , making Initialize variables ; S434', according to the current |Su(i)|+ The smallest value is selected from the pointers pointed to in the first row of the table. y At the same time, write down a The table number corresponding to the table is updated using the following logic: ； S435', Update the bid table Value: Update table Pointers and task labels: p=p+ 1 , l=l+ 1; S436', Repeat steps S434' and S435' until l>n.

9. A service uninstallation system based on a multi-level offloading network, used to execute the service uninstallation method based on a multi-level offloading network as described in any one of claims 1 to 8, characterized in that, The system includes: A network model construction module is used to construct a network model, wherein the devices in the network model include: user equipment and service provider equipment; The network device description module is used to abstract the network model, using pre-defined four-tuples to describe the devices in the network model and the task sets generated by the devices. The network device description module is connected to the network model construction module. The pricing algorithm module executes the pricing algorithm for the quantitative resource model. This includes acquiring and using historical transaction information to obtain the demand function and calculate the request price. n The demand rate of various resources is used to fit the relationship between the price and demand of a single resource, transforming the resource pricing problem into a problem of maximizing the expected revenue of equipment, in order to find the optimal revenue of equipment and the optimal price of resources. The task offloading scheme acquisition module is used to construct a multi-level task offloading network with the user equipment as the center and the maximum task completion delay as the constraint, to execute the multi-level offloading network task offloading algorithm MOS to perform layer-by-layer pricing operations and obtain applicable task offloading schemes. The task offloading scheme acquisition module is connected to the pricing algorithm module and the network device description module.

Images