Distributed edge computing node network lightweight middleware migration method
By introducing the collaborative work of main components, interactive components and sub-components into the distributed edge computing node network, the Euclidean distance and information threshold are used to solve the efficiency of middleware migration, and efficient and low-energy middleware application migration is achieved, and the robustness and fault tolerance of the network are enhanced.
Patent Information
- Application Number
- CN202510048228.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-13
- Publication Date
- 2025-04-25
AI Technical Summary
In the prior art, the collaboration and migration of middleware in distributed edge computing node network lacks effective strategies, resulting in the need to recycle nodes when upgrading and modifying the application, resulting in inconvenience.
A lightweight middleware migration method for distributed edge computing node network is adopted. Through the collaborative work of the main component, interactive component and sub-component, the Euclidean distance and information threshold are used to achieve efficient migration and collaboration of middleware.
Reduce network data traffic, reduce energy consumption, improve the efficiency and fault tolerance of the migration process, support asynchronous and autonomous lightweight middleware applications, and enhance network robustness.
Smart Images
Figure CN119892909A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of distributed edge computing node networks, and in particular to a distributed edge computing node network lightweight middleware migration method. Background Art
[0002] Distributed edge computing node network is the product of combining sensor system, computing system and communication system. With the rapid development of wireless communication technology and electronic device technology, the application field of low-cost, low-power and multi-functional distributed computing nodes has become extremely extensive, which also promotes the application of distributed edge computing node network composed of hundreds of micro sensor nodes, mainly in military, environmental monitoring, medical, smart buildings and other commercial fields.
[0003] However, the current distributed node network is generally built for specific applications, which need to be pre-installed on the sensor nodes after compilation, and then deployed in a specific physical environment. This results in the need to recycle some nodes when upgrading and modifying applications, which will terminate some other applications in progress on the nodes, thus causing great inconvenience.
[0004] Existing research has demonstrated the effectiveness of using middleware to use instant applications in distributed edge computing node networks. However, there are two problems involving middleware collaboration and migration that make the application of lightweight middleware applications lack effective strategies. The former is the lack of a suitable system that allows agents to collaborate and interact; the latter is the lack of an effective way to guide migration. Summary of the invention
[0005] The technical problem to be solved by the present invention is to overcome the defects of the prior art and provide a lightweight middleware migration method for a distributed edge computing node network, which is used to solve the current problems of middleware application migration on a distributed edge computing node network, thereby efficiently migrating middleware applications and serving the distributed network.
[0006] In order to solve the above technical problems, the technical solution of the present invention is: a distributed edge computing node network lightweight middleware migration method, the method comprising:
[0007] Step S1, the distributed nodes monitor around the target entity. The first node that detects the target entity information and receives relevant information becomes the cluster head. The main component migrates to the cluster head. The cluster head is the 0th node, and i is defined as 0.
[0008] Step S2, obtaining t=T i The position of the target entity at time X(t = T i ), based on t = T iThe Euclidean distance calculation from the jth node to the target entity at time t = T i The amount of information I obtained by the jth node at time j (t=T i );
[0009] Step S3: The main component generates a new interactive component and converts the information volume I j (t=T i ), the position of the target entity X(t=T i ) and the information threshold Threshold are passed to the interactive component, and the interactive component migrates to the first node that detects the information. The time when the interactive component migrates to the jth node at time i is recorded as t = T i , let the information threshold variable Threshold_UP = f(I j (t=T i ), φ), if the information amount threshold variable Threshold_UP does not exceed the set threshold Threshold, set i=i+1, and return to step S2; wherein φ is a specific variable, and f() is a function that is positively or negatively correlated with the information amount and the specific variable;
[0010] Step S4: The interaction component stores the node information of the path of the interaction component to form a sequence group {I0(t=T0), I1(t=T1), I2(t=T2)…I i (t=T i )} and {X(t=T0),X(t=T1),X(t=T2),…,X(t=T i )}, sent to the main component;
[0011] Step S5, the main component sends a completion instruction to the middleware, and after receiving the instruction, the middleware recycles the original interactive components and subcomponents distributed on each node;
[0012] Step S6, the main component migrates once or multiple times according to the sequence to the last hop node of the sequence group (selected according to the node signal transmission power) according to the node information, and the destination node becomes the new cluster head.
[0013] Further, in step S2, t=T is obtained. i The position of the target entity at time X(t = T i ); specifically including:
[0014] Node according to t = T i The received signal strength indication RSSI at the moment selects three nodes around the target entity, and based on the position coordinates of these three nodes and the signal distances from the target entity to these three nodes, calculates t = T i The position of the target entity at time X(t = T i ).
[0015] Furthermore, the calculation formula for the position of the target entity is:
[0016]
[0017] The position coordinates of the three nodes around the selected target entity are D1(x1, y1, z1), D2(x2, y2, z2), and D3(x3, y3, z3), and the signal distances from the target entity to these three nodes are r1, r2, r3, x 物 ,y 物 ,z 物 The current three-dimensional position vector of the target entity.
[0018] Furthermore, in step S2, the amount of information I obtained by the i-th node i (t=T i ) is calculated as:
[0019]
[0020] where ||x(t)-x i || is the Euclidean distance from the i-th node to the target entity at time t; π is the pi constant; σ is the standard deviation of the normal distribution, which describes the degree of dispersion of the normal distribution data; е is the natural logarithm.
[0021] Furthermore, the target entity is an animal in a zoo, the target entity information is animal environment update information, and the set threshold Threshold is a boundary vector of the animal area.
[0022] Furthermore, the target entity is an animal in a zoo, the target entity information is an animal identifier, and the set threshold Threshold is a Boolean value used to test whether the animal identifiers are consistent.
[0023] After adopting the above technical solution, the distributed edge computing node network lightweight middleware migration strategy in the present invention has multiple advantages. First, it can greatly reduce the original data traffic on the network; second, for some important real-time systems, it can overcome network risks; third, lightweight middleware applications can be asynchronous and autonomous. Fourth, lightweight middleware applications have firmness and strong fault tolerance.
[0024] In addition, the multiple sub-components of the present invention are pre-distributed in each node, and preliminary calculations can be performed. The interactive components only store and compare. After the migration sequence is formed, the main component is migrated according to the sequence. On the one hand, it can effectively reduce the time for lightweight middleware applications to operate in the entire network. On the other hand, it can also reduce the energy consumption caused by blind migration.
[0025] In summary, the architecture of this solution is simple and clear, and the three different types of components have clear division of labor, which lays a good foundation for the feasibility and robustness of the solution. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 It is a flow chart of the distributed edge computing node network lightweight middleware migration method of the present invention;
[0027] Figure 2 It is a structural diagram of the software entity of the present invention. DETAILED DESCRIPTION
[0028] In order to make the contents of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments in conjunction with the accompanying drawings.
[0029] This embodiment is based on a distributed edge computing node network and middleware application technology, and first gives the following definitions:
[0030] Master application: A middleware application deployed in a distributed node network to perform a specific process. The master application is related to the sensory data, needs to evaluate the sensory data, and needs to cooperate and interact according to the program needs, and needs to have the ability to redeploy itself.
[0031] Master component: The master component manages the migration information of all components and the activity cycle of other components. The master component processes the perception data provided by other agents according to the needs of the main application, and collects and exchanges information with subcomponents. Each main application has only one master component.
[0032] Subcomponent: Subcomponents are used to obtain distributed node data and report it to the main component. Multiple subcomponents can coexist in one main application. Each main application can contain multiple subcomponents.
[0033] Interaction components: Interaction components are used to exchange distributed node network entity information, and need to be transferred to collect and exchange information. Each main application can have one or more interaction components.
[0034] Middleware software: a software entity that shields the physical layer and the operating system layer and is used to run the mobile code mechanism, such as Figure 2 shown.
[0035] Among them, a main component, an interactive component and several sub-components distributed on each distributed edge computing node together constitute the main application running in the distributed node network.
[0036] like Figure 1As shown, a distributed edge computing node network lightweight middleware migration method, the method includes:
[0037] Step S1, the distributed nodes monitor around the target entity. The first node that detects the target entity information and receives relevant information becomes the cluster head. The main component migrates to the cluster head. The cluster head is the 0th node, and i is defined as 0.
[0038] Step S2, obtaining t=T i The position of the target entity at time X(t = T i ), based on t = T i The Euclidean distance calculation from the jth node to the target entity at time t = T i The amount of information I obtained by the jth node at time j (t=T i );
[0039] Step S3: The main component generates a new interactive component and converts the information volume I j (t=T i ), the position of the target entity X(t=T i ) and the information threshold Threshold are passed to the interactive component, and the interactive component migrates to the first node that detects the information. The time when the interactive component migrates to the jth node at time i is recorded as t = T i , let the information threshold variable Threshold_UP = f(I j (t=T i ), φ), if the information amount threshold variable Threshold_UP does not exceed the set threshold Threshold, set i=i+1, and return to step S2; wherein φ is a specific variable, and f() is a function that is positively or negatively correlated with the information amount and the specific variable;
[0040] Step S4: The interaction component stores the node information of the path of the interaction component to form a sequence group {I0(t=T0), I1(t=T1), I2(t=T2)…I i (t=T i )} and {X(t=T0),X(t=T1),X(t=T2),…,X(t=T i )}, sent to the main component;
[0041] Step S5, the main component sends a completion instruction to the middleware, and after receiving the instruction, the middleware recycles the original interactive components and subcomponents distributed on each node;
[0042] Step S6, the main component migrates once or multiple times according to the sequence to the last hop node of the sequence group (selected according to the node signal transmission power) according to the node information, and the destination node becomes the new cluster head.
[0043] In one embodiment, in step S2, t=T is obtained. i The position of the target entity at time X(t = T i ); specifically including:
[0044] Node according to t = T i The received signal strength indication RSSI at the moment selects three nodes around the target entity, and based on the position coordinates of these three nodes and the signal distances from the target entity to these three nodes, calculates t = T i The position of the target entity at time X(t = T i ).
[0045] In one embodiment, the calculation formula for the position of the target entity is:
[0046]
[0047] The position coordinates of the three nodes around the selected target entity are D1(x1, y1, z1), D2(x2, y2, z2), and D3(x3, y3, z3), and the signal distances from the target entity to these three nodes are r1, r2, r3, x 物 ,y 物 ,z 物 The current three-dimensional position vector of the target entity.
[0048] In one embodiment, in step S2, the amount of information I obtained by the i-th node is i (t=T i ) is calculated as:
[0049]
[0050] where ||x(t)-x i || is the Euclidean distance from the i-th node to the target entity at time t; π is the pi constant; σ is the standard deviation of the normal distribution, which describes the degree of dispersion of the normal distribution data; е is the natural logarithm.
[0051] The solutions involved in the above embodiments are described in detail below in conjunction with specific embodiments.
[0052] Embodiment 1
[0053] Take the monitoring of rare animals in a zoo as an example to illustrate the implementation of the lightweight middleware migration strategy. Assume that the zoo is covered with edge computing nodes that can detect the status of such animals. Here, four parameters of the animal environment are defined: the temperature around the animal, the animal tag (animal type), and the location. The zoo has a central management system to coordinate the nodes, and the central management system has a base station function.
[0054] When the zoo needs to raise the temperature of a certain animal's activity environment from 15 degrees Celsius to 16 degrees Celsius, a heating command is set through the central management system. The first node in the surrounding area that detects the increase in the temperature of the animal's environment queries the central information system, and after receiving the return of the temperature increase, it needs to migrate the updated information of the animal's environment to the surrounding nodes. The specific migration steps are as follows:
[0055] Step 1: Before receiving the heating instruction, the edge computing nodes in the zoo run various initial middleware applications: a main component M1, an interactive component C1 and several subcomponents S. The three types of components constitute a local centralized environment detection program running in the distributed network, which is named APP_TempCtrl here;
[0056] Step 2: After the first node that detects the temperature rise around the animal receives the confirmation information fed back by the central information system, the main component migrates to this node, which is recorded as the 0th node;
[0057] Step 3: Select the location coordinates D1(x1,y1), D2(x2,y2), D3(x3,y3) of the three nodes around the target entity and the signal distances r1, r2, r3 from the target entity to the three nodes according to the received signal strength indication RSSI.
[0058]
[0059] Calculate the position X(t=T0) of the animal at this moment, where a, b, and c are the X, Y, and Z coordinates respectively. Calculate the animal position information X(t=T0) at this moment (set as T0), X(t=T0)=(a0, b0, c0);
[0060] Step 4, calculate the current information amount I0(t=T0) according to the information amount collection formula, I0(t=T0)=Θ0, and initialize the information amount threshold variable Threshold_UP to f(I0(t=T0), φ)=f(Θ0, a0, b0, c0), where φ is the position vector of the animal in this example. Here, the f function is used to determine whether the node can perceive the entity information of the animal.
[0061] Step 5: The main component M1 generates a new interactive component C2, and passes I0(t=T0), X(t=T0), Threshold_UP and the set threshold Threshold to the interactive component, and the interactive component migrates to the first node that detects the information.
[0062] Step 6: When the interactive component C2 reaches the first node at the second time T1, it also uses the target positioning algorithm to calculate X(t=T1)=(a1,b1,c1), uses the information collection formula to calculate I1(t=T1)=Θ1, and updates Threshold_UP. The interactive component stores the node information of the path; if Threshold_UP=f(Θ1,a1,b1,c1) is greater than the set threshold Threshold (in this example, the set threshold Threshold can be the limit value of the perceived animal information), then jump to step 7, otherwise record the third time when the interactive component C2 reaches the second node. T2, the target positioning algorithm is also used to calculate X(t=T2)=(a2,b2,c2), and the information collection formula is used to calculate I2(t=T2)=Θ2, and Threshold_UP is updated. The interactive component stores the node information of the path again. If Threshold_UP is greater than the set threshold Threshold at this time, jump to step 7, otherwise record the fourth time T3 when the interactive component C2 arrives at the third node, and so on. At this time, the local centralized application composed of all nodes is named APP_TempCtrl1; among them, the last node in the path stored by the interactive component is recorded as the Kth node.
[0063] Step 7: The main component M1 exchanges information with the interactive component and converts the target entity position information sequence {(a0, b0, c0)...(a K ,b K ,c K )} and the information sequence {I0(t=T0),I1(t=T1),I2(t=T2)…I K (t=T K )}delivered to the main component.
[0064] Step 8: The main component M1 then informs the middleware software to recycle the interactive component C1 and subcomponents previously distributed on each node.
[0065] Step 9, the main component M1 calculates the position of each node based on the position information sequence and the information volume sequence, and then migrates to the end of the sequence in one or more hops, at which time a new local centralized application APP_TempCtrl2 is formed, and then jumps to step 1 to continue monitoring.
[0066] Embodiment 2
[0067] When a zoo needs to add a new type of animal, the distributed nodes also need to update information in order to deploy related functions for monitoring the new animal. Each node dynamically updates information along with the animal's activity trajectory. The central information system sends the new animal ID to the edge computing node where the first main component that recognizes the animal is located. The main component interacts with the interactive component and sends the animal ID to the interactive component. The interactive component migrates to the node of the subcomponent that detects the animal ID and interacts with the subcomponent to obtain the animal's information. Then the interactive component will generate a set of sequences through a series of migration steps and send the information to the new subcomponent. The whole process is:
[0068] Step 1: A main component M1, an interactive component C1, and several subcomponents S distributed on each node constitute an application running in a distributed edge computing network, which is named APP_Detect here;
[0069] Step 2: The central information system sends the new animal ID to the edge computing node where the main component that first identifies the animal is located.
[0070] Step 3: Calculate the position of the animal at this moment X(t=T0) according to the previous similar algorithm, X(t=T0)=(a0,b0,c0)
[0071] Step 4, calculate I0(t=T0)=Θ0 at this time according to the information collection formula. In addition, the main component M1 sets an identification mark for confirming the animal model, and initializes Threshold_UP to f(I0(t=T0), φ)=f(Θ0,mark). At this time, the f function is used to determine whether the node has the animal information, and generate Array={(a0,b0,c0,mark)}.
[0072] Step 5: The main component M1 generates a new interactive component C2, and passes I0(t=T0), X(t=T0), Threshold_UP and the set threshold Threshold to the interactive component.
[0073] Step 6: When the interactive component C2 moves to the node paired with the first mark at the second time T1, the target entity position X(t=T1) and I1(t=T1)=Θ1, X(t=T1)=(a1,b1,c1) at time T1 are calculated and stored according to the above method, and Threshold_UP is set to f(I1(t=T1),φ)=f(Θ1,mark), and Array={(a1,b1,c1),mark)} is generated; the threshold Threshold is set as a Boolean value to test whether the mark is consistent. If not, jump to step 7, otherwise record the third time T2 when the interactive component C2 arrives at the second node , the target positioning algorithm is also used to calculate X(t=T2)=(a2,b2,c2), the information collection formula is used to calculate I2(t=T2)=Θ2, and Threshold_UP is updated to f(I2(t=T2),φ)=f(Θ2,mark). The interactive component stores the node information of the path again. If Threshold_UP is greater than the specific threshold at this time, jump to step 7, otherwise record the fourth time T3 when the interactive component C2 arrives at the third node, ... and so on. At this time, the local centralized application composed of all nodes is named APP_TempCtrl1; among them, the last node in the path stored by the interactive component is recorded as the Kth node.
[0074] Step 7: The main component agent M1 exchanges information with the interactive component C2, and converts the node information sequence Array = {(a0, b0, c0, mark)...(a K ,b K ,c K ,mark)} is delivered to the main component.
[0075] Step 8, the main component agent M1 then informs the middleware software to recycle the interactive component C2 and subcomponents previously distributed on each node.
[0076] In step 9, the main component M1 migrates to the end of the sequence, at which time a new local centralized application APP_APP_Detect2 is formed, and then jumps to step 1 to continue monitoring.
[0077] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.
Claims
1. A distributed edge computing node network lightweight middleware migration method, characterized in that: Methods include: Step S1, the distributed nodes monitor around the target entity. The first node that detects the target entity information and receives relevant information becomes the cluster head. The main component migrates to the cluster head. The cluster head is the 0th node, and i is defined as 0. Step S2, obtaining t=T i The position of the target entity at time X(t = T i ), based on t = T i The Euclidean distance calculation from the jth node to the target entity at time t = T i The amount of information I obtained by the jth node at time j (t=T i ); Step S3: The main component generates a new interactive component and converts the information volume I j (t=T i ), the position of the target entity X(t=T i ) and the information threshold Threshold are passed to the interactive component, and the interactive component migrates to the first node that detects the information. The time when the interactive component migrates to the jth node at time i is recorded as t = T i , let the information threshold variable Threshold_UP = f(I j (t=T i ), φ), if the information amount threshold variable Threshold_UP does not exceed the set threshold Threshold, set i=i+1, and return to step S2; wherein φ is a specific variable, and f() is a function that is positively or negatively correlated with the information amount and the specific variable; Step S4: The interaction component stores the node information of the path of the interaction component to form a sequence group {I0(t=T0), I1(t=T1), I2(t=T2)…I i (t=T i )} and {X(t=T0),X(t=T1),X(t=T2),…,X(t=T i )}, sent to the main component; Step S5, the main component sends a completion instruction to the middleware, and after receiving the instruction, the middleware recycles the original interactive components and subcomponents distributed on each node; Step S6: The main component migrates once or multiple times according to the sequence to the last hop node of the sequence group according to the node information, and the destination node becomes the new cluster head.
2. The distributed edge computing node network lightweight middleware migration method according to claim 1, characterized in that: In step S2, obtain t=T i The position of the target entity at time X(t = T i ); specifically including: Node according to t = T i The received signal strength indication RSSI at the moment selects three nodes around the target entity, and based on the position coordinates of these three nodes and the signal distances from the target entity to these three nodes, calculates t = T i The position of the target entity at time X(t = T i ).
3. The distributed edge computing node network lightweight middleware migration method according to claim 2, characterized in that: The position of the target entity is calculated as: The position coordinates of the three nodes around the selected target entity are D1(x1, y1, z1), D2(x2, y2, z2), and D3(x3, y3, z3), and the signal distances from the target entity to these three nodes are r1, r2, r3, x 物 ,y 物 ,z 物 The current three-dimensional position vector of the target entity.
4. The distributed edge computing node network lightweight middleware migration method according to claim 1, characterized in that: In step S2, the amount of information obtained by the i-th node is I i (t=T i ) is calculated as: where ||x(t)-x i || is the Euclidean distance from the i-th node to the target entity at time t; π is the pi constant; σ is the standard deviation of the normal distribution, which describes the degree of dispersion of the normal distribution data; е is the natural logarithm.
5. The method for migrating lightweight middleware of a distributed edge computing node network according to claim 1, characterized in that: The target entity is an animal in a zoo, the target entity information is animal environment update information, and the set threshold Threshold is a boundary vector of an animal area.
6. The distributed edge computing node network lightweight middleware migration method according to claim 1, characterized in that: The target entity is an animal in a zoo, the target entity information is an animal identifier, and the set threshold Threshold is a Boolean value used to test whether the animal identifiers are consistent.