Multi-hop task offloading method in Internet of Things system

Abstract

The present application discloses a multi-hop task offloading method applied to a large-scale heterogeneous Internet of Things system under incomplete information conditions. In order to effectively deal with the problem of obtaining incomplete information and give corresponding joint offloading decisions and routing strategies, the present application proposes a two-stage game theory method with incomplete information: based on the locally stored historical data and the hierarchical minority game method, the Internet of Things device estimates the offloading cost, and makes an offloading decision based on the estimated offloading cost and the local cost; the base station schedules the transmission path for offloading the Internet of Things device based on the received offloading request using the routing decision based on game theory, and sends the transmission path to the Internet of Things device; according to the transmission path, the Internet of Things device offloads the task to the target base station. From the perspective of two-stage game theory, the present invention solves the offloading decision and routing problems of Internet of Things devices, improves offloading performance, and reduces offloading costs.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Things, and in particular to a multi-hop task offloading method applied to a large-scale heterogeneous Internet of Things system under the condition of incomplete information. Background Art

[0002] In recent years, with the rapid development of embedded technology, sensor technology, wireless communication and other technologies, as well as the development and improvement of various devices and technologies such as information sensors, radio frequency identification technology, global positioning system, infrared sensors, laser scanners, etc., the Internet of Things has gradually moved from concept to reality and has become a hot topic in scientific research and industry. The Internet of Things integrates the sensing, computing and communication capabilities of embedded devices and is widely used in various scenarios such as autonomous driving, smart grids, industrial augmented reality, atmospheric environment monitoring, smart cities, etc. However, due to the explosive growth in the number of IoT devices and the limited computing power of most of them, most IoT devices cannot meet the requirements of some computing-intensive or layer-sensitive applications.

[0003] Multi-access edge computing (MEC) is an emerging technology that provides an IT service environment and cloud computing capabilities at the edge of mobile networks. With the help of MEC, IoT devices can offload computing tasks to edge servers with sufficient computing resources, which is a solution that can significantly reduce the latency and energy consumption of IoT devices. However, the main assumption of most existing approaches to MEC requires that all IoT devices can be directly connected to base stations equipped with edge servers, which is called single-hop task offloading. However, when the scale of IoT systems increases, the communication range of base stations becomes very limited due to obstacles in communication and congestion in information communication channels, and IoT devices may lose stable connection with the base stations.

[0004] To solve this problem, researchers consider increasing the coverage of edge services through multi-hop transmission, thereby reducing the cost of server deployment. In these methods, IoT devices cooperate with each other to forward task data to the base station, which is called multi-hop task offloading. Although some results have been achieved, they have not been fully applied to large-scale heterogeneous IoT systems because they fail to effectively deal with the situation of incomplete information and give corresponding joint offloading decisions and routing strategies.

[0005] Based on this, in order to solve the incompleteness of device information in a heterogeneous Internet of Things system, the present invention proposes an Internet of Things device decision-making problem and routing selection problem under incomplete information based on a two-layer game theory method of incomplete information. Summary of the invention

[0006] In order to solve or partially solve some or all of the above technical problems, the present invention is achieved through the following technical methods:

[0007] A multi-hop task offloading method in an Internet of Things system comprises the following steps: an Internet of Things device estimates an offloading cost based on locally stored historical data and a hierarchical minority game method, and makes an offloading decision based on the estimated offloading cost and the local cost; a base station schedules a transmission path for offloading the Internet of Things device using a routing method based on game theory according to a received offloading request, and sends the transmission path to the Internet of Things device; according to the transmission path, the Internet of Things device offloads the task to a target base station; after the target base station completes the task, it returns the execution of the offloading task and transmission data to the Internet of Things device.

[0008] In some embodiments, the hierarchical minority game method includes the following steps: the IoT devices are divided into multiple hierarchies according to the number of transmission hops from the base station; the transmission cost and computing cost of each level are estimated by the hierarchical estimation method; and the offloading decision is made by comparing the estimated offloading cost and the local cost.

[0009] In some embodiments, the layer estimation method includes the following steps: dividing the layers according to the number of transmission hops from the Internet of Things device to the base station; for the transmission of data from level j(h) to level j(h-1) in the Internet of Things device, it is divided into the following two parts: the first part is: estimating the data transmission cost from the node of level j(h) to other nodes of level j(h) or other levels through historical data; the second part is: estimating the data transmission cost from the node of level j(h) to the node of level j(h-1) through the parallel transmission method; wherein j(h) is the hth layer in base station j, and j(h-1) is the h-1th layer in base station j.

[0010] In some embodiments, the parallel transmission method includes the following steps: setting parameters and And iterate repeatedly until the trend no longer changes; among them, the parameter is the value of the number of IoT devices that base station j transmits data from level j(h) to level j(h-1); parameter Yes h,h-1 A subset of , and l h,h-1 is the set of transmission links connecting nodes in level j(h-1) to nodes in level j(h).

[0011] In some embodiments, if l h,h-1 The number of data transmission links in From l h,h-1 Select the link with the highest transmission rate as otherwise, Equal to l h,h-1 .

[0012] In some embodiments, when the IoT device decides to offload a task and the offloading cost of the task is less than the local cost, the IoT device issues the offloading request.

[0013] In some embodiments, the routing method based on game theory is a static game in which each IoT device competes for resources of the transmission link. Strategies to reach Nash equilibrium And for any i∈Λ j , R i ∈R, satisfying Among them, Λ j is the set of IoT devices that decide to offload their tasks to base station j; R represents a set of feasible routing strategies, YesR i The cost function of .

[0014] In some embodiments, the IoT device searches for the optimal transmission path and compares whether the potential function of the original transmission path is greater than the potential function of the new transmission path; if so, the IoT device updates the routing strategy; if the static game converges to a Nash equilibrium or the number of iterations is greater than the maximum number of iterations, the iterative process will terminate and the transmission path planning will end.

[0015] In some embodiments, if the running time of the game theory routing method exceeds a predetermined time, the Dijkstra method is used instead, and the result obtained by the Dijkstra method is used to schedule the transmission path of the unloaded IoT device.

[0016] In some embodiments, iteration parameters of the level estimation method are updated based on the received results of the task execution and transmission.

[0017] The various embodiments disclosed in the present invention respectively have one or more of the following beneficial technical effects, and more beneficial effects will be introduced in the specific embodiments:

[0018] (1) In large-scale heterogeneous IoT systems, due to the privacy requirements of user devices, the original uninstallation method is difficult to obtain complete information, which in turn has a certain impact on the uninstallation of related IoT devices. On this basis, the present invention proposes a new two-stage game theory method based on incomplete information to solve the uninstallation decision and routing problem of IoT devices in this application scenario.

[0019] (2) A hierarchical minority game (HMG) is proposed to solve the offloading decision problem, and the game theory routing method is used to schedule the transmission path for offloading IoT devices.

[0020] (3) In order to avoid convergence failure in dynamic scenarios, an adaptive mechanism to reduce the convergence time is proposed.

[0021] (4) A large number of simulation experiments were conducted to study the two-stage game theory method with incomplete information, and the results of comparison with three basic methods showed that the present invention improved the performance by at least 194.17% and reduced the overall offloading cost by at least 61.6% compared with the traditional greedy method and single-hop offloading method, and the message cost of this method was only 1% of that of the complete information method.

[0022] The above technical methods / features are intended to summarize the technical methods and technical features described in the specific implementation section, so the scope of the records may not be exactly the same. However, these new technical methods and technical features disclosed in this section, together with the technical features disclosed in the subsequent specific implementation section, disclose more technical methods in a reasonable combination with each other.

[0023] The technical method formed by combining all the technical features disclosed at any position of the present invention is used for summarizing the technical method, modifying the patent document, and disclosing the technical method. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a time sequence diagram of the two-stage game theory method with incomplete information;

[0025] Figure 2 This is a diagram of a large-scale edge-based IoT network model;

[0026] Figure 3 It is a flowchart of the hierarchical minority game method;

[0027] Figure 4 It is a flow chart of routing method. DETAILED DESCRIPTION

[0028] In conjunction with the various method steps or modules described in the embodiments disclosed herein, they can be implemented in hardware, software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the steps and components of each embodiment have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application or design constraints of the technical method. A person of ordinary skill in the art may use different methods to implement the described functions for each specific application, but such implementation should not be considered outside the scope of protection claimed by the present invention.

[0029] The serial numbers such as "first" and "second" in any position of the present invention are merely distinguishing marks for description and do not imply an absolute order in time or space, nor do they imply that the terms prefixed with such serial numbers necessarily have different references from the same terms prefixed with other attributes.

[0030] Since it is impossible to describe all the alternative methods, the following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the key points of the technical methods in the embodiments of the present invention. For other technical methods and details not disclosed in detail below, they generally belong to technical goals or technical features that can be achieved by conventional means in the art. Due to space limitations, the present invention will not introduce them in detail.

[0031] The present invention will describe various key points for combining into various specific embodiments, and these key points will be combined into various methods and products. In the present invention, even if the key points are only described when introducing the method / product method, it means that the corresponding product / method also explicitly includes the technical features.

[0032] When describing the existence or inclusion of a certain step, module, or feature at any position in the present invention, it does not imply that such existence is exclusive and unique. Those skilled in the art can obtain other embodiments based on the technical method disclosed in the present invention with the assistance of other technical means. Based on the key points described in the specific embodiments of the present invention, those skilled in the art can completely replace, delete, add, combine, and change the order of certain technical features to obtain a technical method that still follows the concept of the present invention. These methods that do not deviate from the technical concept of the present invention are also within the scope of protection of the present invention.

[0033] Terminology explanation:

[0034] 1. Multi-access edge computing (MEC) technology: MEC is a technology that is used on mobile communication systems and edge nodes to undertake a large number of computing tasks, providing IT service environment and cloud computing capabilities at the edge of the mobile network. With the help of MEC, IoT devices can transfer computing tasks to edge servers with sufficient computing resources, which is a solution that can significantly reduce the latency and energy consumption of IoT devices.

[0035] 2. Multi-hop transmission technology: Multi-hop transmission technology is a technology that transfers data information to multiple nodes on the communication link. Multi-hop transmission technology solves the limitation of base station coverage in single-hop MEC technology and enables longer-distance communication between senders and receivers through relay forwarding nodes. While increasing the communication range, multi-hop transmission technology also has a certain effect on improving the stability of information transmission.

[0036] Figure 1 The actual is a timing diagram of the unloading method of the two-stage game theory method with incomplete information (TGAII) proposed by the present invention. In general, the method is mainly divided into two stages:

[0037] (1) In the first stage, IoT devices are divided into multiple levels according to the number of multi-hop relays required for transmission. On this basis, a hierarchical minority game method is proposed to estimate transmission delay and energy consumption. Each IoT device makes a corresponding offloading decision by comparing the estimated offloading cost with the local cost.

[0038] (2) In the second stage, a game theory routing method is proposed to schedule the transmission paths of offloaded IoT devices, and the existence of Nash equilibrium is proved by combining potential game theory.

[0039] In other words, a multi-hop task offloading method in an Internet of Things system includes the following steps:

[0040] According to the historical data stored locally and the hierarchical minority game method, the IoT device estimates the offloading cost and makes an offloading decision based on the estimated offloading cost and the local cost; the base station schedules the transmission path for offloading the IoT device based on the received offloading request using a routing method based on game theory and sends the transmission path to the IoT device; according to the transmission path, the IoT device offloads the task to the target base station; after the target base station completes the task, it returns the execution of the offloading task and transmits the data to the IoT device. Furthermore, more steps of the offloading method are described below.

[0041] The Internet of Things network method of the present invention is as follows Figure 2 As shown in the figure, it is a large-scale Internet of Things based on edge computing, which involves some heterogeneous IoT devices D = {1,2,…,N} and base stations E = {1,2,…,M} distributed throughout the network space. Each base station is equipped with an edge server with limited computing resources and storage capacity. The base station establishes a cellular connection communication range with other IoT devices in the network through a self-organizing network connection. In heterogeneous scenarios, due to privacy requirements, IoT devices are usually reluctant to share their private information about offloading decisions and local computing capabilities with others. And in actual situations, due to the limited communication range of base station s or congestion caused by other obstacles, some IoT devices may not be able to connect directly to base station s. Multi-hop task offloading technology can transmit tasks to the target base station through a multi-hop cooperative forwarding path. In order to obtain the optimal forwarding path / data transmission path, the sequence parameter is defined. As a path. Path R i The first element is the IoT device i, and the path R i The last element of R is defined as i [-1]. If IoT device i offloads its tasks to base station j, R i [0] = i and R i [-1] = j. Path R i The discrete segment i of When IoT device i performs its task locally, R i =Φ. In addition, there is a binary indicator variable Π i Defined as indicating whether to execute the task locally on the base station or remotely. If the IoT device decides to execute its task locally, then Π i = 0. When it is necessary to offload to one of the base stations, Π i =1.

[0042] The task method requires that each IoT device has a computation task in each available edge service time slot τ∈{1,2,3,…}. The task configuration method is defined as i=(S i ,C i ), where S i is the data Φ i The number of bits, C i is the number of CPU cycles required. For IoT device i, its offloading strategy can be represented by a tuple s i =(θ i ,R i ). To decide where to perform a task, each IoT device has two options: compute the task locally or offload the task to a suitable base station in an edge computing manner. The offloading decision of IoT device i is defined as θ i ∈{0,1,…,M},θ i = j means that IoT device i decides to offload its task to base station j. If IoT device i performs the task locally, θ i = 0. In addition, the computing capacity of the edge server in base station j is defined as The computing power of IoT device i is defined as

[0043] The communication method uses the orthogonal frequency division multiple access (OFDMA) method to allocate channel resources. Each IoT device is evenly allocated to a grid-shaped orthogonal channel. In the network link, the transmission rate from IoT device i to base station j can be calculated as follows:

[0044]

[0045] Among them B u is the bandwidth of the network connection, P i is the communication power of IoT device i, h i,j is the channel gain of the wireless link (i, j) in the communication network, σ 2 is the variance variable of the noise power.

[0046] Similar to the above mesh structure, the transmission power between IoT devices i and m can be expressed as:

[0047]

[0048] Same as above, where B d is the bandwidth of IoT device connections, P i is the communication power of IoT device i, h i,m is the channel gain of the communication link (i, m) between two IoT devices in the communication network, σ 2 is the variance variable of the noise power. For the sake of convenience, the above two formulas are unified as follows:

[0049]

[0050] Each IoT device must decide whether to perform the computing task locally or offload it to the base station. The task execution cost method for each IoT device includes two computing scenarios: 1) local computing; 2) edge computing. For each computing scenario, the quality of service (QoS) requirements are met.

[0051] For any IoT device, the total cost in the task execution method includes two parts: execution time and execution energy.

[0052] Local computing: If IoT device i performs task Φ locally i , the execution time is mainly determined by the number of CPU cycles of the IoT device. Execution time It can be calculated by the following formula:

[0053]

[0054] The energy consumption of local computing can be calculated by the following formula:

[0055]

[0056] where β i is the energy consumption of the CPU per cycle. Therefore, the total energy consumption is:

[0057]

[0058] in and is a coefficient related to time delay and energy consumption,

[0059] Edge offloading: For edge computing, IoT devices must offload tasks to edge servers equipped by base station devices through multi-hop transmission paths. Unlike local computing, the total cost of edge computing includes computing cost and transmission cost.

[0060] The time consumption at base station (BS) j can be recorded as:

[0061]

[0062] in is the number of computational tasks offloaded to base station j. i Path offloading taskΦ i The time cost to BSj can be written as:

[0063]

[0064] in is the number of transmission tasks in link (k[0],k[1]), and YesR i and R i [-1] = j, IoT device i only bears the section directly connected to it Energy consumption, χ i is the transmission power of IoT device i, task Φ i The energy consumption unloaded to BSj is:

[0065]

[0066] According to the above equation, the total energy consumption can be written as:

[0067]

[0068] Next, we will introduce how the hierarchical minority game model in the first stage of the two-stage game model with incomplete information solves the unloading decision problem.

[0069] IoT devices are divided into multiple levels according to the number of transmission hops from base station s. IoT devices estimate the transmission cost and computation cost of each level through the level estimation method. By comparing the estimated offloading cost and local cost, an offloading decision is made. After base station s completes the task, the results of task execution and transmission will be returned from base station s to update the iteration parameters of the level estimation method.

[0070] Layer estimation method: First, the layers are divided according to the number of transmission hops from IoT devices to base station s. is defined as the number of layers transmitted by IoT device i in BSj, that is, the device i passes at least Jump to base station j to transmit data. max is the maximum number of hops from any IoT device to base station s, j(h) represents the hth layer in base station j, is a set with the number of layers equal to j(h),

[0071] The process of data transmission from level j(h) to level j(h-1) in IoT devices is divided into two parts: 1) transmission from nodes at level j(h) to other nodes at level j(h) or other levels (called part 1); 2) transmission from nodes at level j(h) to nodes at level j(h-1) (called part 2). The transmission costs of part 1 and part 2 are estimated using historical data and parallel transmission methods, respectively.

[0072] Using historical data for estimation, first in Section 1 the historical transmission data is defined as represent The line, in the initialization phase, The historical transmission path is the result of the last moment of the path method, which is described in detail later. Therefore, the first part of the transmission delay of the j(h)th level is:

[0073]

[0074] in is the number of task transfers on link k at the last moment.

[0075] The parallel transmission method is used to estimate the transmission cost in Part 2. In this method, there are two parameters and needs to be set and iterated repeatedly according to the relevant method until the trend no longer changes, where is defined as the number of IoT devices that transfer data from level j(h) to level j(h-1), l h,h-1 is the set of transmission links connecting nodes in j(h-1) to nodes in j(h), Yes n,h-1 A subset of . If l h,h-1 The number of transmission links in From l h,h-1 Select the link with the highest transmission rate as otherwise, Equal to l h,h-1 The estimated transmission rate from j(h) to j(h-1) in the first stage is It can be expressed as:

[0076]

[0077] In summary, the expected transmission time for IoT device i to transmit its task to BSj can be expressed as:

[0078]

[0079] in is the average deviation between the estimated transmission time from j(h) to j(h-1) and the actual value of the transmission time. After the task is completed, the base station s will return the true value of the transmission time to the IoT device. The estimated transmission energy consumption of IoT device i to transfer its task to BSj can be calculated:

[0080]

[0081] in It is an estimate Compared with the actual value (e i,j E,trans ) The average energy deviation between . Estimated computation time Therefore, in the edge computing where IoT device i transfers its task to BSj, the estimated total cost is It can be expressed as:

[0082]

[0083] For IoT device i, the premise of offloading the task to base station j is that the estimated offloading cost is less than the local cost Uninstall threshold Defined as the offloading task Φ i The boundary value to base station j can be calculated by the following formula:

[0084]

[0085] Cutoff threshold: Set a critical value Defined as the threshold of the offloading success rate at level j(h). For an IoT device, successful offloading requires two conditions: 1) the IoT device decides to offload the task; 2) the offloading cost of the task is less than its local cost. The offloading success rate of j(h) in time τ pass

[0086]

[0087] in represents the number of successful uninstallation of IoT devices at level j(h) in time τ, is the number of all j(h)-level IoT devices at time τ. In layman's terms, IoT devices need to compete for the transmission resources of the wireless link and the computing resources of the base station s. If the offloading success rate is greater than the threshold Then the base station has extra resources to accept more offload tasks. Otherwise, fewer tasks should be offloaded. is defined as and The incremental value of .

[0088] Hierarchical Minority Game Method: Each IoT device estimates the total offloading cost through a hierarchical estimation method. By comparing the local cost and the estimated offloading cost, each IoT device decides where to perform the task to minimize the execution cost. Then, the base station schedules the transmission path of the IoT devices that need to be offloaded. After the task is completed, the base station will return the data of task execution and transmission of each IoT device. Based on the received data, the IoT device updates the parameters. The hierarchical estimation method is as follows: Figure 3 shown.

[0089] Uninstall Decision Threshold As the number of CPU cycles C is calculated i increases with the increase of the transmitted data bit S i This indicates that IoT devices with computationally intensive and transmission-light tasks in the hierarchical minority game have higher priority in offloading tasks to base station s.

[0090] Uninstall Decision Threshold With computing power and as the number of IoT devices increases This decreases with the increase of , which transfers their tasks to gateway j. This shows that IoT devices are more likely to offload when the base station has more powerful computing power. As more and more devices decide to offload tasks and compete with the transmission and computing resources of the link and server, the IoT devices are less likely to offload tasks.

[0091] This section introduces how to solve the routing problem based on the sub-game theory routing method in the first stage of the two-stage game method with incomplete information.

[0092] Based on the hierarchical minority game derived in the first phase, each base station will receive requests from offloaded IoT devices to perform their tasks. The base station needs to schedule the transmission path with the lowest transmission cost for offloading IoT devices. To solve this problem, a game theory-based routing method for task offloading of IoT devices is proposed.

[0093] First, the game is defined and analyzed: Each base station receives a request and generates a set of offloaded IoT devices Λ. j contains IoT devices that decide to offload their tasks to base station j. Each IoT device competes for the resources of the transmission link, which can be regarded as a resource competition game. The routing problem can be expressed as a static game in YesR i The cost function is the same as in the previous article. And if the game To reach the potential Nash equilibrium, then the strategy situation Should satisfy Where R represents a feasible routing strategy set. is a potential weight game. Global situation function F(R i ,R -i ) changes when each IoT device updates its path strategy. The potential function is described as follows:

[0094]

[0095] in

[0096] For potential gaming Each vector Can get

[0097]

[0098] According to the feasible routing strategy set R, the improved path sequence is defined as {R i (0),R i (1),R i (2),…}. We can get F(R i (0))>F(R i (1))>F(R i (2))>···. Since the set R is finite, the game After a finite number of iterations, a Nash equilibrium (NE) will be reached.

[0099] Then, a routing method is proposed to offload tasks from IoT devices to find the optimal path. m Represents the maximum number of iterations. The specific process of the routing method is described in the previous article. Each IoT device can find the transmission path R′ with the minimum delay and energy cost i By comparing R′ i and R i The IoT device decides whether to modify the routing strategy. If the game converges to the Nash equilibrium or the number of iterations is greater than Iter m , the iteration process will terminate. The specific method process is as follows Figure 4 shown.

[0100] In practical scenarios, environmental noise and network topology are usually time-varying. In order to improve the robustness of TGAII, an adaptive mechanism is proposed to adapt to environmental changes.

[0101] The adaptive mechanism includes two adjustments: 1) estimation error elimination; 2) fast routing. When the noise changes, the IoT device will notify the nearby base stations, so that the base stations can sense the occurrence of noise and network topology changes, assuming that the frequency of changes is lower than

[0102] First, in the estimation error elimination, the error parameters in the hierarchical estimation method Every In a dynamic environment, time-dependent parameters may change during the execution of the method, leading to convergence failure.

[0103] Then, in order to balance optimality and convergence, the present invention also proposes a fast routing method to solve the problem of accelerating convergence. The process of the method is that when the running time of the routing method based on game theory is greater than When the optimal solution is obtained, the Dijkstra method is used to replace it and the results are used to schedule the transmission path of the offloaded IoT devices. In this process, although the probability of the Dijkstra method generating the optimal solution may be lower, the running time is shorter. This method can quickly reach a stable point and avoid convergence failure.

[0104] In summary, in order to study the complexity and performance deviation of the optimal solution of TGAII, the present invention has conducted a rigorous analysis. In order to improve the robustness of TGAII in dynamic scenarios, an extended adaptive mechanism is proposed to reduce the convergence time of the method and avoid convergence failure.

[0105] Although the present invention has been described with reference to specific features and embodiments of the present invention, various modifications, combinations, and substitutions may be made thereto without departing from the present invention. The scope of protection of the present invention is not intended to be limited to the specific embodiments of the processes, machines, manufactures, compositions of matter, devices, methods, and steps described in the specification, and these methods and modules may also be implemented in one or more products and methods that are associated, interdependent, mutually coordinated, and preceding / following stages.

[0106] Some of the technical features mentioned in the attached claims may have alternative technical features, or the order of certain technical processes and the order of material organization may be reorganized. After knowing the present invention, ordinary technicians in this field can easily think of these replacement means, or change the order of technical processes and the order of material organization, and then use basically the same means to solve basically the same technical problems and achieve basically the same technical effects. Therefore, even if the above-mentioned means and / or order are clearly defined in the claims, these modifications, changes, and substitutions should fall within the scope of protection of the claims based on the principle of equivalents.

Claims

1. A multi-hop task offloading method in an Internet of Things system, characterized in that: The method comprises the following steps: Based on the historical data stored locally and the hierarchical minority game method, the IoT device estimates the offloading cost and makes an offloading decision based on the estimated offloading cost and the local cost; The base station schedules the transmission path for the IoT device to be unloaded according to the received unloading request using a routing method based on game theory, and sends the transmission path to the IoT device; According to the transmission path, the IoT device offloads the task to a target base station; After the target base station completes the task, it returns to the execution of the offload task and transmits data to the IoT device; The hierarchical minority game method comprises the following steps: IoT devices are divided into multiple tiers based on the number of transmission hops from the base station; Estimate the transmission cost and computation cost of each level through the hierarchical estimation method; Make unloading decisions by comparing the estimated unloading costs with the local consumption costs; The hierarchical estimation method comprises the following steps: Divide the levels according to the number of transmission hops from IoT devices to base stations; For IoT devices, data transmission from level j(h) to level j(h-1) is divided into the following two parts: The first part is: estimating the transmission cost from a node at level j(h) to other nodes at level j(h) or other levels through historical data; The second part is: estimating the transmission cost from the node at level j(h) to the node at level j(h-1) by parallel transmission method; Among them, j(h) is the h-th layer in base station j, and j(h-1) is the h-1-th layer in base station j.

2. The multi-hop task offloading method in the Internet of Things system according to claim 1, characterized in that: The parallel transmission method comprises the following steps: Setting parameters and , and repeat the process until the trend stops changing; Among them, the parameters is the value of the number of IoT devices that base station j transmits data from level j(h) to level j(h-1); parameter yes A subset of is the set of transmission links connecting nodes in level j(h-1) to nodes in level j(h).

3. The multi-hop task offloading method in the Internet of Things system according to claim 2, characterized in that: if The number of data transmission links in ,from Select the link with the highest transmission rate as ; otherwise, equal .

4. The multi-hop task offloading method in the Internet of Things system according to claim 1, characterized in that: When the IoT device decides to offload a task and the offloading cost of the task is less than the local cost, the IoT device issues the offloading request.

5. The multi-hop task offloading method in the Internet of Things system according to claim 1, characterized in that: The routing method based on game theory is a static game for each IoT device to compete for the resources of the transmission link. , the strategy to achieve Nash equilibrium , and for any , ,satisfy ; in, is the set of IoT devices that decide to offload their tasks to base station j; R represents a set of feasible routing strategies, yes The cost function of .

6. The multi-hop task offloading method in the Internet of Things system according to claim 5, characterized in that: The IoT device searches for the optimal transmission path and compares whether the potential function of the original transmission path is greater than the potential function of the new transmission path; if so, the IoT device updates the routing strategy; If the static game converges to a Nash equilibrium or the number of iterations is greater than the maximum number of iterations, the iteration process will terminate and the transmission path planning will end.

7. The multi-hop task offloading method in the Internet of Things system according to claim 1, characterized in that: If the running time of the game theory routing method exceeds the predetermined time, the Dijkstra method is used to replace it, and the result obtained by the Dijkstra method is used to schedule the transmission path of the unloaded Internet of Things device.

8. The multi-hop task offloading method in the Internet of Things system according to claim 1, characterized in that: Based on the received task execution and transmitted results, the iteration parameters of the level estimation method are updated.

Images