A wireless heterogeneous network service caching and task offloading method assisted by aggregation nodes

Abstract

The present invention belongs to the technical field of heterogeneous network resource allocation, and discloses a method for service caching and task offloading of wireless heterogeneous networks assisted by aggregation nodes, including: establishing multiple mathematical models involving service caching and task offloading for 5G-WiFi heterogeneous network scenarios assisted by aggregation nodes; constructing an integer nonlinear programming optimization problem for minimizing task completion delay and cost consumption; and solving the integer nonlinear programming joint optimization problem. The present invention adopts the above-mentioned method for service caching and task offloading of wireless heterogeneous networks assisted by aggregation nodes, formulates a joint optimization problem of service caching and task offloading, uses an equivalent conversion method to process the complex 0-1 discrete variable form; based on the iterative solution of the alternating optimization algorithm, effectively solves the integer nonlinear programming problem, minimizes the task completion delay and cost consumption, and effectively solves the task offloading and resource management problems in the 5G-WiFi edge computing network.

Description

Technical Field

[0001] The present invention relates to the technical field of heterogeneous network resource allocation, and in particular to an aggregation node-assisted wireless heterogeneous network service caching and task offloading method. Background Art

[0002] The explosive growth of data traffic is an inevitable problem. Although Multi-access Edge Computing (MEC) can be used as a good solution to deal with the above problems, when a large number of users simultaneously offload computing tasks through the 5G cellular network, the MEC system will suffer from severe network congestion, resulting in a decrease in the QoS of the 5G cellular network. In addition, 5G works in the operator's authorized spectrum, and its security and service quality have advantages. However, due to its commercial nature, 5G usually has a higher cost per bit of data transmission. Compared with 5G base stations, the deployment of WiFi access points (APs) is more flexible, and the cost per bit of WiFi network communication is lower and more energy-efficient, which is more important for battery-powered wireless devices. Although the network conditions of 5G and WiFi are very different, both can provide good service quality to mobile devices. Therefore, 5G-WiFi coexistence heterogeneous networks is a suitable solution. Offloading 5G cellular traffic to WiFi networks plays an important role in alleviating the increasing burden on 5G cellular networks.

[0003] However, since WiFi networks are based on the CSMA / CA competitive channel access mechanism, a large number of user accesses and excessive traffic offloading will cause serious access conflicts to the WiFi network, and the coverage of WiFi APs can usually only achieve partial coverage, and there will still be terminal devices (TDs) that are not within the coverage of the WiFi network. TDs that are not covered by APs can only use the computing resources of the 5G cellular network, and cannot fully utilize the resources in the heterogeneous network, resulting in waste.

[0004] In order to solve this problem, it is considered to introduce an aggregation node (AN) that can work in both 5G cellular network and WiFi network in 5G-WiFi heterogeneous network, so that TD outside the coverage of AP can also use the internal resources of WiFi network through relay, which has become an effective technical means. In order to further reduce the computing processing delay and data transmission delay, it is critical to make good use of the cache capacity of edge servers and the edge computing capacity of nodes in various parts of the network. Existing research has proposed task caching for repeatable task offloading process. The basic concept of task caching is to cache reusable task data or hotspot content at the edge, which can effectively reduce the transmission of repeated input data, reduce task processing delay and equipment energy consumption. However, in actual work, TD rarely has repeated task offloading work, and the tasks generated by TD are often different, and the corresponding task parameters are also different. The input data and calculation results of the application are difficult to use for different tasks, and executing repeated task program code will cause additional overhead. Since the service components responsible for task processing, such as programs, tool libraries and databases, can be shared and reused by TD tasks. Therefore, caching the relevant database and program code of the application on the edge server has become an effective solution, which is called service caching.

[0005] Therefore, task caching can only meet TD's needs for the same task processing, while service caching can provide service support for TD's tasks with the same target but different parameters. Since the service cache capacity of the edge server is limited, it can only cache a limited set of services. Therefore, it is crucial for the edge server to know which TD services to cache and where the edge server caches TD services to optimize the overall network utility of the system. Summary of the invention

[0006] The purpose of the present invention is to provide an aggregation node-assisted wireless heterogeneous network service caching and task offloading method, which effectively solves the problem of task offloading and resource management in 5G-WiFi edge computing network.

[0007] To achieve the above object, the present invention provides an aggregation node-assisted wireless heterogeneous network service caching and task offloading method, comprising the following steps:

[0008] S1. For the 5G-WiFi heterogeneous network scenario assisted by aggregation nodes, multiple mathematical models involving service caching and task offloading are established, including terminal device TD communication model, aggregation node AN communication model, task completion delay model, energy consumption and commercial cost model, and service caching model;

[0009] S2. Based on the multiple mathematical models established in step S1, construct an integer nonlinear programming optimization problem for minimizing the task completion delay and cost consumption;

[0010] S3. Use the AO algorithm to solve the integer nonlinear programming joint optimization problem constructed in step S2.

[0011] Preferably, in step S1, establishing a TD communication model of a terminal device includes the following steps:

[0012] S111, 5G cellular network access system is based on OFDMA, where n is the label of each TD, TD n The uplink transmission rate of the offload task through the 5G cellular network is as follows:

[0013] ;

[0014] in, The base station BS is assigned to the TD n bandwidth; is the background noise power in the 5G cellular network; Indicates TD n The transmission power; Indicates TD n The channel gain between the BS and the base station; Indicates TD n The distance to the base station BS; ;

[0015] S112. TDs covered by both 5G cellular network and WiFi network are represented as Class A TDs and recorded as a set , Class A TD n The uplink transmission rate of the task offloaded through the WiFi network is as follows:

[0016] ;

[0017] in, Indicates Class A TD n The transmission rate that can be achieved after successfully seizing the WiFi channel; Respectively represent Class A TD n and the weight of the aggregation node AN in the channel competition;

[0018] S113. TDs covered only by 5G cellular networks are denoted as Class B TDs and are recorded as a set , Class B TD n The uplink transmission rate of the task offloaded through the 5G cellular network to the aggregation node AN is as follows:

[0019] ;

[0020] in, Indicates Class B TD nThe channel gain to the aggregation node AN, Indicates Class B TD n The distance between AN and the

[0021] Preferably, in step S1, establishing an aggregation node AN communication model includes the following steps:

[0022] After the aggregation node AN receives the task data transmitted by the Class B TD through the 5G cellular network, if there is no service cache for the corresponding task, it will be unloaded to the edge server at the WiFi access point AP for processing, following the CSMA / CA wireless communication model for task unloading, as shown below:

[0023] ;

[0024] in, It indicates the transmission rate that can be achieved after the aggregation node AN successfully seizes the WiFi channel.

[0025] Preferably, in step S1, establishing a task completion delay model includes the following steps:

[0026] S131. The two types of TD have common task completion delays, namely local calculation delay, base station BS transmission delay and edge server calculation delay, as shown below:

[0027] ;

[0028] in, For TD n The local computing latency, Indicates TD n The total computational effort of the task, Represented as TD n The CPU cycle frequency;

[0029] ;

[0030] in, For TD n Uplink transmission delay to the base station BS offloading task, Indicates TD n The total amount of data for the task;

[0031] In heterogeneous networks, TD n The task is calculated on three edge servers with different computing capabilities, and the latency is as follows:

[0032] ;

[0033] in, They represent the CPU cycle frequencies of edge servers near WiFi access points AP, base stations BS, and aggregation nodes AN, respectively. The edge servers at three locations calculate the latency respectively;

[0034] S132, since the task offloading paths of the class A TD and the class B TD are different, they each have their own unique task completion delay;

[0035] A-type TD completes task offloading by using WiFi communication in the WiFi network. n The uplink transmission delay using the WiFi network is as follows:

[0036] ;

[0037] Class B TD first uses the 5G cellular network to transmit to the aggregation node AN, and then the AN transmits it to the edge server near the WiFi access point AP through the WiFi network to complete the task offloading. n The task completion delay is as follows:

[0038] .

[0039] Preferably, in step S1, establishing an energy consumption and business cost model comprises the following steps:

[0040] S141, TD n The local computing energy consumption is as follows:

[0041] ;

[0042] in, A constant representing the effective switched capacitance, f n l Represents the local server CPU cycle frequency;

[0043] S142. Since the transmission delay and transmission power of TD to different ends are different, TD n The wireless transmission energy consumption for uplink transmission to the base station BS is as follows:

[0044] ;

[0045] Class A TD n The transmission energy consumption of offloading tasks to the edge server near the WiFi access point AP is as follows:

[0046] ;

[0047] in, A-type TDn Transmitting power when working on WiFi network;

[0048] Class B TD n Energy consumption is generated when transmitting to the aggregation node AN through the 5G cellular network. The energy consumption is as follows:

[0049] ;

[0050] S143, cost is TD n The traffic transmission cost of TD is as follows:

[0051] ;

[0052] in, Expressed as the unit traffic cost of 5G cellular network.

[0053] Preferably, in step S1, establishing a service cache model includes the following steps:

[0054] S151. For Class A TD, use Indicates Class A TD n Whether the service required by the task is cached in the edge server near the WiFi access point AP, if the A-type TD n The edge server near the AP has a service cache otherwise ;

[0055] For Class B TD, use Indicates Class B TD n Whether the service required by the task is cached in the edge server near the aggregation node AN, if the B type TD n Edge servers near AN have service caches otherwise ;

[0056] Both Class A TD and Class B TD can access the edge server near the base station BS, so use Indicates TD n Are the services required by the task cached in the edge server near the 5G BS? n There is a service cache on the edge server near the BS otherwise ;

[0057] S152. After the service cache exists, the task completion delay, energy consumption and cost of TD under different decisions are as follows:

[0058] (1) Category A TD service cache status:

[0059] 1) When When A-type TD n The task is cached in the edge server near the base station BS or WiFi access point AP, Class A TD n The task completion delay, energy consumption and cost are as follows:

[0060] ;

[0061] Among them, the decision variables Indicates TD n Whether to choose to access the edge server in the WiFi network, if TD n Select the WiFi network to connect to. otherwise ; Using decision variables Indicates TD n Whether to choose to access the edge server at the base station BS; if TD n Choose to access 5G cellular network otherwise ; Indicates the local server;

[0062] 2) When When A-type TD n The task is only cached in the edge server near the WiFi access point AP, Class A TD n The task completion delay, energy consumption and cost are as follows:

[0063] ;

[0064] 3) When When A-type TD n The task is only cached in the edge server near the base station BS, Class A TD n The task completion delay, energy consumption and cost are as follows:

[0065] ;

[0066] 4) When When A-type TD n If there is no cache on the edge server near the base station BS or WiFi access point AP, the task offloading must transmit the corresponding data. n The task completion delay, energy consumption and cost are as follows:

[0067] ;

[0068] (2) Class B TD service cache status:

[0069] 1) When When B-type TD n The tasks are cached in the edge servers near the base station BS or the aggregation node AN. Class B TD n The task completion delay, energy consumption and cost are shown as follows:

[0070] ;

[0071] 2) When When B-type TD n The task is only cached in the edge server near the aggregation node AN, and the B-type TD n The task completion delay, energy consumption and cost are as follows:

[0072] ;

[0073] 3) When When B-type TD n The task is only cached in the edge server near the base station BS, Class B TD n The task completion delay, energy consumption and cost are as follows:

[0074] ;

[0075] 4) When When B-type TD n If there is no cache in the edge server near the base station BS or the aggregation node AN, the corresponding data must be transmitted when the task is offloaded. n The task completion delay, energy consumption and cost are as follows:

[0076] .

[0077] Preferably, in step S2, based on the multiple mathematical models established in step S1, an integer nonlinear programming optimization problem for minimizing the task completion delay and cost consumption is constructed, and the specific process is as follows:

[0078] S21. Based on the above discussion of various models of TD in 5G-WiFi heterogeneous network MEC task offloading, the completion delay, energy consumption and cost model of Class A TD task after service cache decision are summarized as follows:

[0079] ;

[0080] S22. Similarly, the completion delay, energy consumption and cost model of Class B TD tasks are summarized as follows:

[0081] ;

[0082] S23. The optimization goal is to minimize the network utility consisting of the weighted sum of task completion delay and cost, and introduce the weight factor and By adjusting the relationship between task completion delay and cost, the joint optimization problem of 5G-WiFi heterogeneous network based on service cache and aggregation node AN assistance is as follows:

[0083] ;

[0084] Among them, constraints C1, C2 and C3 represent that TD's task offloading scheduling decision and service caching decision are all 0-1 discrete variables; constraint C4 represents TD n The choice constraint of the task offloading decision variable ensures that each TD n Only one network node will be selected for task offloading; in constraint C5 For TD n The energy consumption limit of TD n Energy consumption constraints; Constrain the parameters in C6, C7 and C8 They are the upper limits of the service cache capacity of the edge server near the WiFi access point AP, the base station BS, and the aggregation node AN, respectively, indicating the service cache capacity constraints of the edge server.

[0085] Preferably, in step S3, the integer nonlinear programming joint optimization problem constructed in step S2 is solved, and the specific process is as follows:

[0086] S31. The optimization problem in the above step S2 includes service caching and task offloading. Based on the AO algorithm, the high coupling in the optimization problem is solved, and the optimization problem is decomposed into two sub-problems for solution. First, the task offloading sub-problem after fixing the service caching decision variable is as follows:

[0087] ;

[0088] S32. Introducing penalty parameters And the corresponding penalty term, restricting the solution to a binary form, the above sub-problem is transformed into:

[0089] ;

[0090] Among them, the fmincon function in the Matlab toolbox is used to solve the problem of the minimum value of nonlinear multivariate functions;

[0091] S33, obtain fixed variables through the above solution process Post-variable The optimal solution of The 0-1 integer linear programming problem is as follows:

[0092] .

[0093] Therefore, the present invention adopts the above-mentioned aggregation node-assisted wireless heterogeneous network service caching and task offloading method, and formulates a joint task scheduling optimization problem of service caching and task offloading based on the service caching and AN-assisted 5G-WiFi coexistence heterogeneous network, minimizing the system total task completion delay and cost under the energy consumption constraint of TD and the service cache capacity constraint of the edge server; by using the equivalent conversion method, the complex 0-1 discrete variable form in the problem is processed; the iterative optimization algorithm based on the AO algorithm effectively solves the integer nonlinear programming problem, minimizes the task completion delay and cost consumption, and at the same time ensures the task offloading scheduling strategy under the energy consumption constraint and service cache capacity constraint, effectively solving the task offloading and resource management problems in the 5G-WiFi edge computing network.

[0094] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] Figure 1 It is a flow chart of a method for caching and offloading wireless heterogeneous network services assisted by an aggregation node of the present invention;

[0096] Figure 2 This is a 5G-WiFi heterogeneous network scenario diagram based on service caching and AN assistance of the present invention. DETAILED DESCRIPTION

[0097] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0098] like Figure 1 As shown, the present invention provides an aggregation node-assisted wireless heterogeneous network service caching and task offloading method, comprising the following steps:

[0099] S1. For the 5G-WiFi heterogeneous network scenario assisted by aggregation nodes, multiple mathematical models involving service caching and task offloading are established, including terminal device TD communication model, aggregation node AN communication model, task completion delay model, energy consumption and commercial cost model, and service caching model;

[0100] S2. Based on the multiple mathematical models established in step S1, construct an integer nonlinear programming (INLP) optimization problem for minimizing the task completion delay and cost consumption;

[0101] S3. Use the AO algorithm to solve the integer nonlinear programming joint optimization problem constructed in step S2.

[0102] Example

[0103] S1. For the 5G-WiFi heterogeneous network scenario assisted by aggregation nodes, multiple mathematical models involving service caching and task offloading are established.

[0104] like Figure 2 As shown in the figure, a 5G-WiFi heterogeneous network MEC task offloading scenario with AN assistance is considered. There is one 5G base station (5G BS), one WiFi access point (WiFi AP) and one aggregation node (AN) in the WiFi network. There is no cross-layer interference between the WiFi network and the 5G cellular network. The AN has dual connection functions for the 5G cellular network and the WiFi network. Edge servers with different computing capabilities and capable of supporting service caching are deployed near the BS, AP and AN. The TD covered by both the 5G cellular network and the WiFi network is represented as Class A TD and recorded as the set , TDs that are only covered by 5G cellular networks are represented as Class B TDs and recorded as a set For the convenience of description, all TD sets are recorded as , each TD label is denoted as n. The task profile of TD is composed of Indicates that Indicates TD n The total amount of data for the task, Indicates TD n The total computational effort of the task.

[0105] Using decision variables Indicates TD n Whether to choose to access the edge server in the WiFi network, that is, the edge server at the AP or the edge server at the AN; if TD n Select the WiFi network to connect to. otherwise Using decision variables Indicates TD n Whether to choose to access the edge server at BS; if TD n Choose to access 5G cellular network otherwise TD n In addition to choosing to offload to the BS or within the WiFi network, you can also choose to process the task locally, which is represented by , so there is a constraint: .

[0106] S11. Establish a TD communication model.

[0107] S111, 5G cellular network access system is based on OFDMA, TD n The uplink transmission rate of the offload task through the 5G cellular network is as follows:

[0108] ;

[0109] in, BS is assigned to TD n bandwidth, is the background noise power in the 5G cellular network, Indicates TD n The transmission power; Indicates TD n The channel gain between the BS and Indicates TD n The distance to the BS and set Perform subsequent simulations.

[0110] S112, assume that all TDs accessing the WiFi network share a common channel and compete for channel access rights through the CSMA / CA access mechanism. In the scenario discussed, there is only one AN, and Class A TD n The uplink transmission rate of the task offloaded through the WiFi network is as follows:

[0111] ;

[0112] in, Indicates Class A TD n The transmission rate that can be achieved after successfully seizing the WiFi channel; Respectively represent Class A TD n The weight of AN in channel competition. The higher the weight, the more advantage it has in channel competition.

[0113] S113, Class B TD is transmitted to AN through 5G cellular network, using the edge computing capability at AP or AN, so Class B TD n The uplink transmission rate of offloading tasks to AN through the 5G cellular network is as follows:

[0114] ;

[0115] in, Indicates Class B TD n Channel gain to AN, Indicates Class B TD n The distance between AN and the

[0116] S12. Establish an AN communication model.

[0117] After AN receives task data transmitted by Class B TD through 5G cellular network, if there is no service cache for the corresponding task, it needs to be offloaded to the edge server at AP for processing. The task is also offloaded according to the CSMA / CA wireless communication model, as shown below:

[0118] ;

[0119] in, Indicates the transmission rate that can be achieved after AN successfully seizes the WiFi channel.

[0120] S13. Establish a task completion delay model.

[0121] S131, task completion delay Ignoring the transmission delay of the task result return, there is a common task completion delay for the two types of TDs, namely local calculation delay, BS transmission delay and edge server calculation delay, as shown below:

[0122] ;

[0123] in, For TD n The local computing latency, Represented as TD n The CPU cycle frequency.

[0124] ;

[0125] in, For TD n Uplink transmission delay to BS offloading task.

[0126] In a heterogeneous network, there are three edge servers with different computing capabilities, so TD n The tasks of calculating the latency on different edge servers are as follows:

[0127] ;

[0128] in, They represent the CPU cycle frequencies of edge servers near the AP, BS, and AN, respectively. The latency is calculated for the three edge servers respectively.

[0129] S132. Since the task offloading paths of Class A TD and Class B TD are different, they have their own unique task completion delay. In addition to offloading to BS, Class A TD can also choose to offload directly to AP. Class A TD completes task offloading in the WiFi network by using WiFi communication. n The uplink transmission delay using the WiFi network is as follows:

[0130] ;

[0131] Unlike Class A TD, if the task of Class B TD needs to be offloaded to AP, it needs to be relayed. It is first transmitted to AN using 5G cellular network and then transmitted to edge server near AP by AN via WiFi network. n This part of the delay is as follows:

[0132] ;

[0133] S14. Establish energy consumption and business cost model.

[0134] S141. For energy consumption, the focus is on the energy consumption of TD itself, including local computing energy consumption and wireless transmission energy consumption. Because it has no direct impact on optimizing the energy consumption of TD itself, the static energy consumption and the energy consumption of edge computing terminals are ignored. n The local computing energy consumption is as follows:

[0135] ;

[0136] in, A constant representing the effective switched capacitance.

[0137] S142. Since the transmission delay and transmission power of TD to different ends are different, the wireless transmission energy consumption is also different. n The wireless transmission energy consumption for uplink transmission to the BS is as follows:

[0138] ;

[0139] For Class A TD n For example, the transmission energy consumption of offloading tasks to edge servers near the AP is as follows:

[0140] ;

[0141] in, A-type TD n The transmit power at which the WiFi network operates.

[0142] For Class B TD n For example, when transmitting to AN through 5G cellular network, energy consumption will be generated, which is as follows:

[0143] ;

[0144] S143, the cost is mainly TD nThe traffic transmission cost, that is, using 5G cellular network for task offloading will incur costs, and using WiFi network transmission will not incur additional costs. The transmission cost of TD is as follows:

[0145]

[0146] in, Expressed as the unit traffic cost of 5G cellular network.

[0147] S15. Establish a service cache model.

[0148] S151. For Class A TD, use Indicates Class A TD n Whether the service required by the task is cached in the edge server near the AP, if the A-type TD n The edge server near the AP has a service cache otherwise .

[0149] For Class B TD, use Indicates Class B TD n Whether the service required by the task is cached in the edge server near the AN, if the B TD n Edge servers near AN have service caches otherwise .

[0150] Both Class A TD and Class B TD can access the edge server near the BS, so Indicates TD n Whether the service required by the task is cached in the edge server near the 5G BS. n There is a service cache on the edge server near the BS otherwise .

[0151] S152: If the service required by TD to complete the task is cached in the corresponding edge server, then there is no need to transmit the input data of the task during the task offloading process. Therefore, after the service cache exists, the task completion delay, energy consumption and cost of TD under different decisions are as follows:

[0152] (1) Category A TD service cache status:

[0153] 1) When When A-type TD n The task is cached in the edge server near the BS or AP. In this case, the A-type TD n If you choose to offload tasks to BS or AP, you do not need to transmit the corresponding data. nThe task completion delay, energy consumption and cost are as follows:

[0154] ;

[0155] 2) When When A-type TD n The task is only cached in the edge server near the AP. In this case, the A-type TD n If you choose to offload tasks to the AP, you do not need to transmit the corresponding data. n The task completion delay, energy consumption and cost are as follows:

[0156] ;

[0157] 3) When When A-type TD n The task is only cached in the edge server near the BS. In this case, the A-type TD n If you choose to offload tasks to the BS, you do not need to transmit the corresponding data. n The task completion delay, energy consumption and cost are as follows:

[0158] ;

[0159] 4) When When A-type TD n The task is not cached by the edge server near the BS or AP. In this case, the A-type TD n If task offloading is selected, the corresponding data must be transmitted. n The task completion delay, energy consumption and cost are as follows:

[0160] ;

[0161] (2) Class B TD service cache status:

[0162] The service cache situation of Class B TD is similar to that of Class A TD. Considering the limited cache capacity of the edge server, in this scenario, the AP will not cache the services required by Class B TD. It mainly provides service cache for Class A TD. Moreover, the edge server at AN will only execute tasks with service cache due to load issues. Therefore, Class B TD utilizes the cache resources near AN and BS and the computing resources near AP and BS.

[0163] 1) When When B-type TD n The task is cached in the edge server near the BS or AN. In this case, the B TD n If you choose to offload tasks to BS or AN, you do not need to transmit the corresponding data.n The task completion delay, energy consumption and cost are shown as follows:

[0164] ;

[0165] 2) When When B-type TD n The task is only cached in the edge server near AN. In this case, the B TD n If AN is selected for task offloading, there is no need to transmit the corresponding data. n The task completion delay, energy consumption and cost are as follows:

[0166] ;

[0167] 3) When When B-type TD n The task is only cached in the edge server near the BS. In this case, the B-type TD n If BS is selected for task offloading, there is no need to transmit the corresponding data. n The task completion delay, energy consumption and cost are as follows:

[0168] ;

[0169] 4) When When B-type TD n The task is not cached by edge servers near BS or AN. In this case, the B-type TD n If task offloading is selected, the corresponding data must be transmitted. n The task completion delay, energy consumption and cost are as follows:

[0170] ;

[0171] S2. Based on the multiple mathematical models established in step S1, an integer nonlinear programming (INLP) optimization problem for minimizing the task completion delay and cost consumption is constructed.

[0172] S21. Based on the above discussion of various models of TD in 5G-WiFi heterogeneous network MEC task offloading, the completion delay, energy consumption and cost model of Class A TD task after service cache decision are summarized as follows:

[0173] ;

[0174] S22. Similarly, the completion delay, energy consumption and cost model of Class B TD tasks are summarized as follows:

[0175] ;

[0176] S23. The optimization goal is to minimize the network utility consisting of the weighted sum of task completion delay and cost. The joint optimization problem of 5G-WiFi heterogeneous network based on service cache and AN assistance is as follows:

[0177] ;

[0178] As shown above, we consider the joint optimization of task offloading decisions and edge server service cache strategies, while ensuring that the energy consumption limit of each TD and the service cache capacity limit of each edge server are met. The importance of balancing task completion delays and cost Used to balance different scales between the delay in completing the expense task and the expense cost.

[0179] Among them, constraints C1, C2 and C3 indicate that TD's task offloading scheduling decision and service caching decision are both 0-1 discrete variables. Constraint C4 indicates that TD n The alternative constraint of the task offloading decision variable ensures that each TD n Only one network node will be selected for task offloading. For TD n The energy consumption limit of TD n Constrain the parameters in C6, C7 and C8 are the upper limits of the service cache capacity of the edge servers near AP, BS, and AN, respectively, and represent the service cache capacity constraints of the edge servers. Since there are serious variable couplings in both the objective function and the constraints and all variables are 0-1 discrete variables, the optimization problem is an INLP problem.

[0180] S3. Solve the INLP joint optimization problem constructed in step S2.

[0181] S31. The optimization problem in the above step S2 involves two aspects: service cache and task offloading. Considering the need for fast solution of optimization problems in practical applications, the AO algorithm is used to solve the high coupling in the original problem and decompose the optimization problem into two sub-problems. First, the task offloading sub-problem after fixing the service cache decision variable is as follows:

[0182] ;

[0183] S32: Although the variables are separated, there are still coupled discrete binary variables in this subproblem. In order to obtain a suitable solution, we first consider relaxing the 0-1 discrete variables in the target and introducing the penalty parameter And the corresponding penalty items, appropriate The value can give a higher penalty to solutions that are not binary. When the values ​​in the variables are all binary, the penalty part in the objective function is 0, otherwise the penalty value will be very high. Therefore, this penalty mechanism can limit the solution to binary form as much as possible. The above sub-problem is converted into the following form:

[0184] ;

[0185] Due to the continuity of 0-1 discrete variables, the problem after conversion is a classic nonlinear optimization problem. The fmincon function in the Matlab toolbox can well solve the problem of minimum value of nonlinear multivariate function.

[0186] S33, by solving the above problems, fixed variables can be obtained Post-variable The optimal solution of , after substituting the optimal solution into , we can get The sub-problems are as follows:

[0187] ;

[0188] It can be seen that After fixing, this sub-question is about There are many mature solutions for 0-1 integer linear programming problems, such as cutting plane method and branch and bound method. In order to solve the problem efficiently, consider using CPLEX to solve this subproblem. CPLEX can solve 0-1 integer linear programming problems efficiently and reliably.

[0189] Based on the above process, by alternately solving the two decoupled sub-problems, the AO algorithm can quickly converge to the stable point of the original problem.

[0190] Therefore, the present invention adopts the above-mentioned aggregation node-assisted wireless heterogeneous network service caching and task offloading method, and formulates a joint task scheduling optimization problem of service caching and task offloading based on the service caching and AN-assisted 5G-WiFi coexistence heterogeneous network, minimizing the system total task completion delay and cost under the energy consumption constraint of TD and the service cache capacity constraint of the edge server; by using the equivalent conversion method, the complex 0-1 discrete variable form in the problem is processed; the iterative optimization algorithm based on the AO algorithm effectively solves the integer nonlinear programming problem, minimizes the task completion delay and cost consumption, and at the same time ensures the task offloading scheduling strategy under the energy consumption constraint and service cache capacity constraint, effectively solving the task offloading and resource management problems in the 5G-WiFi edge computing network.

[0191] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A wireless heterogeneous network service caching and task offloading method assisted by an aggregation node, characterized in that: The following steps are involved: S1. For the 5G-WiFi heterogeneous network scenario assisted by aggregation nodes, multiple mathematical models involving service caching and task offloading are established, including terminal equipment TD communication model, aggregation node AN communication model, task completion delay model, energy consumption and commercial cost model, and service caching model; S2. Based on the multiple mathematical models established in step S1, an integer nonlinear programming optimization problem is constructed to minimize the task completion delay and cost consumption, where the TDs covered by both the 5G cellular network and the WiFi network are represented as Class A TDs and recorded as a set , TDs that are only covered by 5G cellular networks are represented as Class B TDs and recorded as a set , all TD sets are recorded as , each TD label is recorded as n, and the specific process is as follows: S21. Based on various models of TD in 5G-WiFi heterogeneous network MEC task offloading, the completion delay, energy consumption and cost model of Class A TD tasks after service cache decision are summarized as follows: ； Among them, the decision variables Indicates TD n Whether to choose to access the edge server in the WiFi network, if TD n Select the WiFi network to connect to. otherwise ; Decision variables Indicates TD n Whether to choose to access the edge server at the base station BS, if TD n Choose to access 5G cellular network otherwise ; Indicates the local server; Indicates the calculation delay of the corresponding edge server near the WiFi access point AP; Indicates the computation delay of the edge server corresponding to the base station BS; Indicates Class A TD n Uplink transmission delay using WiFi network; For TD n Uplink transmission delay to the base station BS offloading task; For TD n Local computing latency; Indicates Class A TD n Whether the service required by the task is cached in the edge server near the WiFi access point AP, if the A-type TD n The edge server near the AP has a service cache otherwise ; Indicates TD n Are the services required by the task cached in the edge server near the 5G BS? n There is a service cache on the edge server near the BS otherwise ; Indicates TD n Local computing energy consumption; Indicates Class A TD n Transmission energy consumption of offloading tasks to edge servers near WiFi access points APs; Indicates TD n Wireless transmission energy consumption for uplink transmission to base station BS; It represents the transmission cost of TD; S22. Similarly, the completion delay, energy consumption and cost model of Class B TD tasks are summarized as follows: ； in, Indicates Class B TD n Whether the service required by the task is cached in the edge server near the aggregation node AN, if the B type TD n Edge servers near AN have service caches otherwise ; Indicates Class B TD n The task completion delay; represents the computation delay of edge servers near the aggregation node AN; S23. The optimization goal is to minimize the network utility consisting of the weighted sum of task completion delay and cost, and introduce the weight factor and By adjusting the relationship between task completion delay and cost, the joint optimization problem of 5G-WiFi heterogeneous network based on service cache and aggregation node AN assistance is as follows: ； Among them, constraints C1, C2 and C3 represent that TD's task offloading scheduling decision and service caching decision are all 0-1 discrete variables; constraint C4 represents TD n The choice constraint of the task offloading decision variable ensures that each TD n Only one network node will be selected for task offloading; in constraint C5 For TD n The energy consumption limit of TD n Energy consumption constraints; Constrain the parameters in C6, C7 and C8 are the upper limits of the service cache capacity of the edge servers near the WiFi access point AP, the base station BS, and the aggregation node AN, respectively, indicating the service cache capacity constraints of the edge servers; Indicates TD n The total amount of data for the task; S3. Use the AO algorithm to solve the integer nonlinear programming joint optimization problem constructed in step S2.

2. The method for wireless heterogeneous network service caching and task offloading assisted by an aggregation node according to claim 1, characterized in that: In step S1, a TD communication model of a terminal device is established, including the following steps: S111, 5G cellular network access system is based on OFDMA, where n is the label of each TD, TD n The uplink transmission rate of the offload task through the 5G cellular network is as follows: ； in, The base station BS is assigned to the TD n bandwidth; is the background noise power in the 5G cellular network; Indicates TD n The transmission power; Indicates TD n The channel gain between the BS and the base station; Indicates TD n The distance to the base station BS; ; S112. TDs covered by both 5G cellular network and WiFi network are represented as Class A TDs and recorded as a set , Class A TD n The uplink transmission rate of the task offloaded through the WiFi network is as follows: ； in, Indicates Class A TD n The transmission rate that can be achieved after successfully seizing the WiFi channel; Respectively represent Class A TD n and the weight of the aggregation node AN in the channel competition; S113. TDs covered only by 5G cellular networks are denoted as Class B TDs and are recorded as a set , Class B TD n The uplink transmission rate of the task offloaded through the 5G cellular network to the aggregation node AN is as follows: ； in, Indicates Class B TD n The channel gain to the aggregation node AN, Indicates Class B TD n The distance between AN and the 3. The method for caching and offloading wireless heterogeneous network services assisted by an aggregation node according to claim 2, characterized in that: In step S1, an aggregation node AN communication model is established, including the following steps: After the aggregation node AN receives the task data transmitted by the Class B TD through the 5G cellular network, if there is no service cache for the corresponding task, it will be unloaded to the edge server at the WiFi access point AP for processing, following the CSMA / CA wireless communication model for task unloading, as shown below: ； in, It indicates the transmission rate that can be achieved after the aggregation node AN successfully seizes the WiFi channel.

4. The method for caching and offloading wireless heterogeneous network services assisted by an aggregation node according to claim 3, characterized in that: In step S1, a task completion delay model is established, including the following steps: S131. The two types of TD have common task completion delays, namely local calculation delay, base station BS transmission delay and edge server calculation delay, as shown below: ； in, For TD n The local computing latency, Indicates TD n The total computational effort of the task, Represented as TD n The CPU cycle frequency; ； in, For TD n Uplink transmission delay to the base station BS offloading task, Indicates TD n The total amount of data for the task; In heterogeneous networks, TD n The task is calculated on three edge servers with different computing capabilities, and the latency is as follows: ； in, They represent the CPU cycle frequencies of edge servers near WiFi access points AP, base stations BS, and aggregation nodes AN, respectively. The edge servers at three locations calculate the latency respectively; S132, since the task offloading paths of the class A TD and the class B TD are different, they each have their own unique task completion delay; A-type TD completes task offloading by using WiFi communication in the WiFi network. n The uplink transmission delay using the WiFi network is as follows: ； Class B TD first uses the 5G cellular network to transmit to the aggregation node AN, and then the AN transmits it to the edge server near the WiFi access point AP through the WiFi network to complete the task offloading. n The task completion delay is as follows: 。 5. The method for caching and offloading wireless heterogeneous network services assisted by an aggregation node according to claim 4, characterized in that: In step S1, an energy consumption and business cost model is established, including the following steps: S141, TD n The local computing energy consumption is as follows: ； in, A constant representing the effective switched capacitance, Represents the local server CPU cycle frequency; S142. Since the transmission delay and transmission power of TD to different ends are different, TD n The wireless transmission energy consumption for uplink transmission to the base station BS is as follows: ； Class A TD n The transmission energy consumption of offloading tasks to the edge server near the WiFi access point AP is as follows: ； in, A-type TD n Transmitting power when working on WiFi network; Class B TD n Energy consumption is generated when transmitting to the aggregation node AN through the 5G cellular network. The energy consumption is as follows: ； S143, cost is TD n The traffic transmission cost of TD is as follows: ； in, Expressed as the unit traffic cost of 5G cellular network.

6. The method for caching and offloading wireless heterogeneous network services assisted by an aggregation node according to claim 5, characterized in that: In step S1, a service cache model is established, including the following steps: S151. For Class A TD, use Indicates Class A TD n Whether the service required by the task is cached in the edge server near the WiFi access point AP, if the A-type TD n The edge server near the AP has a service cache otherwise ; For Class B TD, use Indicates Class B TD n Whether the service required by the task is cached in the edge server near the aggregation node AN, if the B type TD n Edge servers near AN have service caches otherwise ; Both Class A TD and Class B TD can access the edge server near the base station BS, so use Indicates TD n Are the services required by the task cached in the edge server near the 5G BS? n There is a service cache on the edge server near the BS otherwise ; S152. After the service cache exists, the task completion delay, energy consumption and cost of TD under different decisions are as follows: (1) Category A TD service cache status: 1) When When A-type TD n The task is cached in the edge server near the base station BS or WiFi access point AP, Class A TD n The task completion delay, energy consumption and cost are as follows: ； Among them, the decision variables Indicates TD n Whether to choose to access the edge server in the WiFi network, if TD n Select the WiFi network to connect to. otherwise ; Using decision variables Indicates TD n Whether to choose to access the edge server at the base station BS; if TD n Choose to access 5G cellular network otherwise ; Indicates the local server; 2) When When A-type TD n The task is only cached in the edge server near the WiFi access point AP, Class A TD n The task completion delay, energy consumption and cost are as follows: ； 3) When When A-type TD n The task is only cached in the edge server near the base station BS, Class A TD n The task completion delay, energy consumption and cost are as follows: ； 4) When When A-type TD n If there is no cache on the edge server near the base station BS or WiFi access point AP, the task offloading must transmit the corresponding data. n The task completion delay, energy consumption and cost are as follows: ； (2) Class B TD service cache status: 1) When When B-type TD n The tasks are cached in the edge servers near the base station BS or the aggregation node AN. Class B TD n The task completion delay, energy consumption and cost are shown as follows: ； 2) When When B-type TD n The task is only cached in the edge server near the aggregation node AN, and the B-type TD n The task completion delay, energy consumption and cost are as follows: ； 3) When When B-type TD n The task is only cached in the edge server near the base station BS, Class B TD n The task completion delay, energy consumption and cost are as follows: ； 4) When When B-type TD n If there is no cache in the edge server near the base station BS or the aggregation node AN, the corresponding data must be transmitted when the task is offloaded. n The task completion delay, energy consumption and cost are as follows: 。 7. The method for wireless heterogeneous network service caching and task offloading assisted by an aggregation node according to claim 1, characterized in that: In step S3, the integer nonlinear programming joint optimization problem constructed in step S2 is solved, and the specific process is as follows: S31. The optimization problem in the above step S2 includes service caching and task offloading. Based on the AO algorithm, the high coupling in the optimization problem is solved, and the optimization problem is decomposed into two sub-problems for solution. First, the task offloading sub-problem after fixing the service caching decision variable is as follows: ； S32. Introducing penalty parameters And the corresponding penalty term, restricting the solution to a binary form, the above sub-problem is transformed into: ； Among them, the fmincon function in the Matlab toolbox is used to solve the problem of the minimum value of nonlinear multivariate functions; S33, obtain fixed variables through the above solution process Post-variable The optimal solution of The 0-1 integer linear programming problem is as follows: 。

Images