An Edge Computing Resource Allocation Method Based on Binary Offloading

Abstract

The present invention discloses a method for edge computing resource allocation based on binary offloading, including: establishing an edge computing network, and the decision m of each wireless node i ∈{0,1}; initializing the offloading decision M 0 ; generating N candidate offloading decisions according to the offloading decision M l at the l-th iteration; obtaining the resource allocation results and the corresponding total network rates of each candidate offloading decision; taking the candidate offloading decision with the maximum total network rate as M l (j * ); determining whether the total network rate corresponding to M l (j * ) is greater than the total network rate corresponding to M l ; if so, taking M l (j * ) and its corresponding total network rate as M l+1 and the total network rate corresponding to M l+1 , returning to execute the next iteration; otherwise, taking M l and the corresponding resource allocation result as the final decision allocation result. This method can quickly determine the binary offloading decision and the corresponding resource allocation result, achieving a relatively high total network computing rate.

Description

Technical Field

[0001] The present invention belongs to the technical field of edge computing, and particularly relates to an edge computing resource allocation method based on binary offloading. Background Art

[0002] With the rapid development of the Internet of Things, many wireless devices applied in the Internet of Things cannot handle computationally intensive and latency-sensitive computing tasks well due to limited computing resources, working power, and battery capacity. Mobile Edge Computing (MEC) technology allows wireless nodes to offload computing loads to edge computing servers, thus greatly enhancing the task processing capabilities of wireless nodes.

[0003] Many existing research works on edge computing networks have adopted a partial offloading mode, that is, it is considered that the computing tasks of wireless devices can be arbitrarily divided, with part of the tasks offloaded to the edge server for computing and the other part of the tasks remaining for local computing. However, in actual applications, many tasks are often indivisible. Therefore, considering the indivisibility of tasks and higher network rates, the present application proposes an edge computing resource allocation method based on binary offloading. Summary of the Invention

[0004] The purpose of the present invention is to propose an edge computing resource allocation method based on binary offloading for the above problems, which can quickly determine the binary offloading decision and simultaneously solve the resource allocation result corresponding to the offloading decision, achieving a higher total network computing rate.

[0005] To achieve the above purpose, the technical solutions adopted by the present invention are as follows:

[0006] An edge computing resource allocation method based on binary offloading proposed by the present invention includes the following steps:

[0007] S1. Establish an edge computing network, which includes an edge server and N wireless nodes. The decision mi made by each wireless node at time slot T i ∈{0,1}, i = 1, 2,..., N. If mi i = 1, it means that the i-th wireless node chooses to completely offload the computing task to the edge server through time division multiplexing, denoted as the first type of wireless node. If mi i = 0, it means that the i-th wireless node chooses to perform complete local computing, denoted as the second type of wireless node;

[0008] S2. Initialize the offloading decision M of N wireless nodes 0 = [m 1 , m 2 ,..., m i ,..., m N , and obtain M 0The resource allocation results and the corresponding total network rate, where the resource allocation results include the offloading transmission duration and offloading transmit power consumption of the first type of wireless nodes, and the CPU computing frequency of the second type of wireless nodes;

[0009] S3. According to the offloading decision M in the l-th iteration l Generate N candidate offloading decisions {M l (1), M l (2), …, M l (j), …, M l (N)}, where Denotes binary summation, that is, changing the decision of the j-th wireless node in the l-th iteration process, l = 0, 1, 2, …;

[0010] S4. Obtain the resource allocation results and the corresponding total network rate of each of the N candidate offloading decisions in the current iteration process;

[0011] S5. Take the candidate offloading decision with the maximum total network rate in the current iteration process as the best candidate offloading decision M l (j * );

[0012] S6. Judge whether the total network rate corresponding to M l (j * ) is greater than the total network rate corresponding to M l . If so, take M l (j * ) as the offloading decision M for the next iteration l+1 , and take the total network rate corresponding to M l (j * ) as the total network rate corresponding to M l+1 , and return to step S3 to perform the next iteration. Otherwise, take M l and the corresponding resource allocation results as the final decision allocation results.

[0013] Preferably, the resource allocation results and the corresponding total network rate are calculated as follows:

[0014] 1) The offloading transmission duration and offloading transmit power consumption of the first type of wireless nodes, the formulas are as follows:

[0015]

[0016] Where Is the offloading transmit power consumption of the o-th wireless node, Is the maximum transmit power consumption of the offloading task of the o-th wireless node, Is the offloading transmission duration of the o-th wireless node, h o Is the channel gain from the o-th wireless node to the edge server, is the maximum transmit power consumption for offloading tasks of the i-th wireless node, h i is the channel gain from the i-th wireless node to the edge server, and T is the length of a time slot;

[0017] 2) The CPU computing frequency of the second type of wireless node is as follows:

[0018]

[0019] where, is the CPU computing frequency of the k-th wireless node, is the maximum CPU frequency for local computing of the k-th wireless node, is the upper limit of the maximum power consumption for local computing of the k-th wireless node, and κ is the computing energy efficiency coefficient of the wireless node;

[0020] 3) Total network rate The formula is as follows:

[0021]

[0022] where, N 0 is the noise power, φ is the number of CPU cycles required for the k-th wireless node to locally compute one bit of task, is the offloading transmission duration of the i-th wireless node, is the offloading transmit power consumption of the i-th wireless node.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows: Compared with the existing optimization methods, by using time division multiplexing (TDMA) communication mode and binary offloading mode in the edge computing network, the offloading decision-making and computing resource allocation problem is transformed from a continuous problem into a mixed integer programming problem, and there is no need to enumerate all binary offloading decisions, avoiding the problems of dimensional explosion and slow algorithm operation speed, achieving fast solution of binary offloading decisions and corresponding resource allocation results (including CPU computing frequency of wireless nodes with full local computing, offloading transmission duration of nodes with full task offloading, and offloading transmit power consumption allocation), and obtaining a relatively high total network computing rate. Description of the Drawings

[0024] Figure 1 is the flowchart of the edge computing resource allocation method based on binary offloading of the present invention;

[0025] Figure 2 is the schematic diagram of the edge computing network structure of the present invention.

[0026] Description of the reference numerals: 1. Edge server; 2. Base station. Detailed Embodiment

[0027] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0028] It should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0029] like Figure 1-2 As shown, a method for allocating edge computing resources based on binary offloading includes the following steps:

[0030] S1. Establish an edge computing network. The edge computing network includes edge servers and N wireless nodes. Each wireless node makes a decision m in time slot T. i ∈{0,1}, i=1,2,…,N, if m i =1, indicating that the i-th wireless node chooses to completely offload the computing task to the edge server through time division multiplexing, which is recorded as the first type of wireless node. i =0, indicating that the i-th wireless node chooses completely local calculation and is recorded as the second type of wireless node;

[0031] S2. Initialize the offloading decision M of N wireless nodes 0 =[m 1 ,m 2 ,…,m i ,…,m N ], get M 0 The resource allocation results and the corresponding total network rate, the resource allocation results include the unloaded transmission duration and unloaded transmission power consumption of the first type of wireless nodes, and the CPU calculation frequency of the second type of wireless nodes;

[0032] S3, according to the unloading decision M of the first iteration l Generate N candidate offloading decisions {M l (1),M l (2),…,M l (j),…,M l (N)}, where represents the binary sum, i.e., transforming the decision of the jth wireless node in the lth iteration, l = 0, 1, 2, ...;

[0033] S4. Obtain the resource allocation results and the corresponding total network rate of each of the N candidate offloading decisions in the current iteration process;

[0034] S5. Take the candidate offloading decision with the maximum total network rate in the current iteration process as the optimal candidate offloading decision M l (j * )

[0035] S6. Determine whether the total network rate corresponding to M l (j * ) is greater than the total network rate corresponding to M l . If so, take M l (j * ) as the offloading decision M for the next iteration l+1 , and take the total network rate corresponding to M l (j * ) as the total network rate corresponding to M l+1 , and return to step S3 to perform the next iteration. Otherwise, take M l and the corresponding resource allocation result as the final decision allocation result.

[0036] In one embodiment, the resource allocation result and the corresponding total network rate are calculated as follows:

[0037] 1) The offloading transmission duration and offloading transmit power consumption of the first type of wireless node, the formula is as follows:

[0038]

[0039] Among them, is the offloading transmit power consumption of the o-th wireless node, is the maximum transmit power consumption of the offloading task of the o-th wireless node, is the offloading transmission duration of the o-th wireless node, h o is the channel gain from the o-th wireless node to the edge server, is the maximum transmit power consumption of the offloading task of the i-th wireless node, h i is the channel gain from the i-th wireless node to the edge server, and T is the length of one time slot;

[0040] 2) The CPU computing frequency of the second type of wireless node, the formula is as follows:

[0041]

[0042] Among them, is the CPU computing frequency of the k-th wireless node, is the maximum CPU frequency of the local computing of the k-th wireless node, $P_{max,k}$ is the upper limit of the maximum power consumption for local computing of the $k$-th wireless node, and $\kappa$ is the computing energy efficiency coefficient of the wireless node;

[0043] 3) Total network rate The formula is as follows:

[0044]

[0045] where $N$ 0 is the noise power, $\varphi$ is the number of CPU cycles required for the $k$-th wireless node to locally compute one-bit task, $t_{off,i}$ is the offloading transmission duration of the $i$-th wireless node, $P_{off,i}$ is the offloading transmit power of the $i$-th wireless node.

[0046] Specifically, as Figure 2 shown, the edge server 1 can be equipped to the base station 2. In this embodiment, there are a total of 5 wireless nodes with computing requirements. The task computing of the wireless nodes works based on time frames. Each wireless node needs to make a decision in the time slot $T$, that is, to choose to completely offload the computing task to the edge server or to completely perform local computing. Among them, the wireless nodes that choose to completely offload the computing task to the edge server perform task offloading communication through time division multiplexing (TDMA).

[0047] Initialize the offloading decision $M$ of 5 wireless nodes 0 $=[1, 0, 1, 0, 1]$, which means that in this edge computing network, the 1st, 3rd, and 5th wireless nodes (i.e., the first type of wireless nodes) choose to offload all computing tasks to the edge server, and the 2nd and 4th wireless nodes (i.e., the second type of wireless nodes) choose to perform complete local computing. And obtain the resource allocation result of $M$ 0 and the corresponding total network rate. The resource allocation result includes the offloading transmission duration and offloading transmit power of the first type of wireless nodes, as well as the CPU computing frequency of the second type of wireless nodes.

[0048] According to the initialized offloading decision $M$ 0 generate 5 candidate offloading decisions $\{M$ 0 (1), $M$ 0 (2), $M$ 0 (3), $M$ 0 (4), $M$ 0 (5)$\}$, where represents binary summation ($ or $ $), that is, change the decision of the $j$-th wireless node in the 0th iteration process, such as $M$ 0 (1) $=[0, 0, 1, 0, 1]$.

[0049] For 5 candidate offloading decisions, the respective resource allocation results and the corresponding total network rate are calculated. That is, for the wireless nodes with full local computing, the CPU computing frequency is calculated, and for the wireless nodes with full task offloading, the offloading transmission duration and offloading transmission power consumption are calculated. And the corresponding total network rate is calculated according to the CPU computing frequency, offloading transmission duration, and offloading transmission power consumption of each wireless node in the given offloading decision.

[0050] Taking the candidate offloading decision M 0 (1) = [0, 0, 1, 0, 1] as an example, for the other candidate offloading decisions M 0 (2), M 0 (3), M 0 (4), M 0 (5), and the offloading decision M 0 Similarly, determine their resource allocation results and the corresponding total network rate, as follows:

[0051] 1) The offloading transmission duration and offloading transmission power consumption of the 3rd and 5th wireless nodes (i.e., the first type of wireless nodes) are calculated as follows:

[0052]

[0053] 2) The CPU computing frequencies of the 1st, 2nd, and 4th wireless nodes (i.e., the second type of wireless nodes) are calculated as follows:

[0054]

[0055] 3) Under the time slot T, the achievable total network rate is calculated as follows:

[0056]

[0057] Where, N 0 is the noise power, and φ is the number of CPU cycles required for a wireless node to locally compute one bit of task. Similarly, the total network rates of the offloading decision M 0 and its corresponding other 4 candidate offloading decisions can be obtained.

[0058] Compare the total network rates corresponding to the resource allocation results of all candidate offloading decisions, and select the candidate offloading decision with the maximum total network rate as the optimal offloading decision M 0 (j * ), where, If the total network rate corresponding to the optimal offloading decision M 0 (j * ) is greater than the offloading decision M 0 , it means that the resource allocation result corresponding to M 0 (j * ) is better than M 0If it is the resource allocation result corresponding to, then M 0 (j * ) is used as the offloading decision M for the next iteration 1 , and the total network rate corresponding to M 0 (j * ) is used as the offloading decision M for the next iteration 1 corresponding total network rate, and return to continue iterative update until the total network rate corresponding to all candidate offloading decisions is not greater than the offloading decision in the corresponding iterative process. Then, use the offloading decision in the corresponding iterative process and its corresponding resource allocation result as the output. If the total network rate corresponding to the optimal offloading decision M 0 (j * ) is less than or equal to the offloading decision M 0 , then directly output the offloading decision M 0 and its corresponding resource allocation result as the final offloading task allocation scheme.

[0059] Compared with the existing optimization methods, this method transforms the offloading decision and computing resource allocation problem from a continuous problem into a mixed integer programming problem by using time division multiplexing (TDMA) communication mode and binary offloading mode in the edge computing network, and there is no need to enumerate all binary offloading decisions, avoiding the problems of dimensional explosion and slow algorithm running speed, achieving fast solution of binary offloading decisions and corresponding resource allocation results (including CPU computing frequency of wireless nodes with full local computing, offloading transmission duration of nodes with full task offloading, and offloading transmit power consumption allocation), and obtaining a higher total network computing rate.

[0060] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0061] The above embodiments only express the embodiments of the present application that are described more specifically and in detail, but should not be construed as limiting the scope of the patent application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several deformations and improvements can be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. An edge computing resource allocation method based on binary offloading, characterized in that: the edge computing resource allocation method based on binary offloading includes the following steps: S1. Establish an edge computing network, where the edge computing network includes edge servers and N wireless nodes, and the decision m made by each wireless node at time slot T i ∈{0, 1}, i = 1, 2, …, N. If m i = 1, it means that the i-th wireless node selects to completely offload the computing task to the edge server through time-division multiplexing, denoted as the first type of wireless node. If m i = 0, it means that the i-th wireless node selects to perform complete local computing, denoted as the second type of wireless node; S2. Initialize the offloading decisions M of N wireless nodes 0 = [m 1 , m 2 , …, m i , …, m N , and obtain the resource allocation result of M 0 and the corresponding total network rate. The resource allocation result includes the offloading transmission duration and offloading transmit power consumption of the first type of wireless nodes, and the CPU computing frequency of the second type of wireless nodes. The resource allocation result and the corresponding total network rate are calculated as follows: 1) The offloading transmission duration and offloading transmission power consumption of the first type of wireless node are as follows: wherein, is the offloading transmission power consumption of the o-th wireless node, is the maximum transmission power consumption of the offloading task of the o-th wireless node, is the offloading transmission duration of the o-th wireless node, h o is the channel gain from the o-th wireless node to the edge server, is the maximum transmission power consumption of the offloading task of the i-th wireless node, h i is the channel gain from the i-th wireless node to the edge server, and T is the length of a time slot; 2) The CPU computing frequency of the second type of wireless node is as follows: Among them, is the CPU computing frequency of the k-th wireless node, is the maximum CPU frequency for local computing of the k-th wireless node, is the maximum power consumption upper limit for local computing of the k-th wireless node, and κ is the computing energy efficiency coefficient of the wireless node; 3) The total network rate The formula is as follows: Among them, N 0 is the noise power, φ is the number of CPU cycles required for a wireless node to locally compute one-bit task, is the offloading transmission duration of the i-th wireless node, is the offloading transmit power consumption of the i-th wireless node; S3. Generate N candidate offloading decisions {M l l (1), M l l (1), M l l (2), …, M l l (j), …, M l l (N)} according to the offloading decision M of the l-th iteration, where represents binary summation, that is, changing the decision of the j-th wireless node in the l-th iteration process, l = 0, 1, 2, …; S4. Obtain the resource allocation results and corresponding network total rates of N candidate offloading decisions in the current iteration process; S5. Take the candidate offloading decision with the maximum total network rate in the current iteration process as the best candidate offloading decision M l (j * ); S6. Judge M l (j * ) Whether the corresponding total network rate is greater than M l For the corresponding total network rate, if so, set M l (j * ) as the offloading decision M for the next iteration l+1 , and set the corresponding total network rate of M l (j * ) as the corresponding total network rate of M l+1 , return to step S3 to perform the next iteration. Otherwise, set M l and the corresponding resource allocation result as the final decision allocation result.

Images