Industrial Internet of Things low-overhead time synchronization method based on event triggering

Abstract

The invention discloses an industrial Internet of Things low-overhead time synchronization method based on event triggering, and belongs to the field of industrial Internet of Things. According to the method, nodes regularly update logic clock drift and offset compensation parameters according to a set update period, an estimator is used for estimating the real-time clock state of neighbor nodes to construct a dynamic threshold value, and a state error is increased along with accumulation of absolute time until the state error exceeds the dynamic threshold value, so that broadcast is triggered. In addition, a simple trigger condition is added to improve the early convergence speed. And the global consistency of the network logic clock is achieved through continuous updating iteration. According to the method, the problems that the time-triggered communication overhead is too high in the industrial Internet of Things random topology environment and the nodes in the event triggering mechanism cannot be triggered stably and are not triggered uniformly are considered, the synchronization precision is improved under the condition that the additional communication overhead is not increased, and the blockage probability of the shared network is reduced.

Description

Technical Field

[0001] The present invention relates to the field of industrial Internet of Things, and in particular to an event-triggered low-overhead time synchronization method for industrial Internet of Things. Background Art

[0002] Time synchronization technology, which provides a unified time reference for network nodes, plays a vital role in the Industrial Internet of Things. Promoting the efficient and in-depth application of the Industrial Internet of Things is inseparable from the timely and accurate data transmission and collaborative operation between various industrial devices. In actual scenarios, a large number of wireless sensor nodes used in the perception layer of the Industrial Internet of Things will show slight differences in clock rates due to the manufacturing process, resulting in time asynchrony. Therefore, it is inevitable that clock information needs to be exchanged between network nodes to ensure their consistency in time. The fully distributed time synchronization method is more suitable for large-scale Industrial Internet of Things due to its high flexibility and strong robustness. However, it is worth noting that the distributed synchronization method requires frequent communication iterations between nodes, which violates the energy-saving design concept of the relevant algorithm in the resource-constrained Industrial Internet of Things environment, and greatly limits the application of the fully distributed time synchronization method that broadcasts periodically according to time.

[0003] Different from the time-triggered control mechanism, the event-triggered control mechanism triggers broadcast communication only when certain conditions are met, which can effectively save unnecessary communication overhead. Therefore, the design of a time synchronization method integrating the event-triggered mechanism provides a new idea for reducing the communication overhead of the distributed time synchronization method in a large-scale industrial Internet of Things environment, thereby extending the service life of the network.

[0004] In the event-triggered time synchronization method, the node only broadcasts the message when the state has changed sufficiently since the previous transmission, rather than the traditional frequent periodic broadcast based on time, which significantly reduces the communication energy consumption of the industrial Internet of Things. However, the existing event-triggered time synchronization method has problems such as unstable and uneven node triggering. That is, in the later stage of the time synchronization stage, the triggering of the node is prone to stagnation, which is not conducive to improving the accuracy of time synchronization. In addition, the events triggered by the node will appear in an uneven state with different densities, such as frequent triggering of the node in the early stage and sparse triggering in the later stage, which will increase the number of communication broadcasts and increase the probability of congestion of the shared network. Therefore, there is an urgent need for an event-triggered time synchronization method that ensures stable and uniform triggering of events without increasing communication energy consumption, and realizes clock synchronization of the entire network. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is to provide a fully distributed time synchronization method based on event triggering applied in the industrial Internet of Things environment, to deal with the problems of frequent communication in the time triggering mechanism and the inability of nodes to be triggered stably and unevenly in the event triggering mechanism; to provide the system with an event-triggered node interaction strategy suitable for large-scale industrial Internet of Things scenarios, which can achieve reasonable time synchronization accuracy with lower communication overhead, and all nodes can stably trigger broadcasts, and at the same time, the broadcast intervals are uniform and the broadcast tasks are similar, thereby reducing the probability of network congestion.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A low-overhead time synchronization method for industrial Internet of Things based on event triggering is proposed. Aiming at the problem of unstable and uneven node triggering in the event-triggered time synchronization method, a method based on joint estimation of neighbor clock state errors to construct a dynamic threshold is proposed. The state error of the node will increase with the accumulation of absolute time until it exceeds the set dynamic threshold, that is, stable and sparse trigger communication and timely synchronization error correction are achieved, thereby achieving global clock consistency. In addition, an estimator is designed to estimate the real-time clock state of neighbors to avoid additional communication.

[0008] The technical solution adopted by the event-triggered low-overhead time synchronization method of the industrial Internet of Things described in the present invention is:

[0009] Connect sensor nodes with similar hardware parameters through a wireless shared network to generate a random topology;

[0010] The local node initializes the clock parameters and sets the update cycle. All nodes update the logical clock drift and offset compensation parameters according to the cycle.

[0011] Based on the designed estimator, the local node uses its own existing information to estimate the real-time status information of its neighbors and construct a dynamic threshold. At the same time, in order to speed up the convergence speed in the early stage, the constraint of the logical clock offset compensation parameter is added on the basis of the dynamic threshold, that is, a trigger condition of a static threshold is added;

[0012] The local node determines whether the trigger condition is met, and saves and broadcasts its own clock status information if the trigger threshold is exceeded;

[0013] The local node adjusts the local clock until the global clock is synchronized.

[0014] The random topology structure means that sensor nodes with the same function are randomly distributed, and their neighbor nodes are determined by communication radius and two-way communication is performed to form a random network topology structure, but it is ensured that each local node has at least one neighbor node.

[0015] Initializing the clock parameters and setting the update cycle means: setting the node update cycle to T, the initial update number to k=1, and setting the initial logical clock drift and offset compensation parameters of all nodes to And the relative clock slope η is set to 1, t k Indicates the physical time when the node is updated for the kth time.

[0016] The node updates the logical clock drift and offset compensation parameters, which means that the local node updates the local logical clock parameters using the collected clock information according to the set update cycle. The specific process is as follows:

[0017] For any sensor node i, the kth update time of its hardware clock can be expressed as When the number of updates k is greater than 2 and satisfies When the logic clock drift compensation parameters and offset compensation parameters are updated respectively:

[0018]

[0019] in, and They represent the logical clock drift compensation parameter and offset compensation parameter of node i respectively; η ij represents the relative clock slope of node i and node j, that is, Among them, α i and α j denote the hardware clock drift of nodes i and j respectively, The hardware clock value received by node i from node j is The hardware clock value recorded when is the kth broadcast time of node j, and node i records the logical clock drift compensation parameter broadcast by neighbor node j as This means that in the interval between two broadcasts of node j, the logical clock drift compensation parameter of node j recorded by node i remains unchanged; |N i | represents the degree of node i, N i is the set of neighbor nodes of node i; Indicates update time The instant before, Indicates the logical clock drift compensation parameter of node j recorded by node i before updating. Indicates the logical clock drift compensation parameter of the last broadcast of node i. In addition, Indicates update time The estimated value of the logical clock of node j recorded by node i at time Indicates update time The previous logical clock value of node i.

[0020] The estimator means that the node does not broadcast the local clock information at all update times, so the real-time hardware clock value and logical clock value of the neighboring node are unknown. In order to improve the accuracy of the update parameters without increasing additional communication, the present invention designs an estimator here to estimate the real-time hardware clock and logical clock value of node j:

[0021]

[0022] in, and They represent the estimated values ​​of the real-time hardware clock and logical clock of node j respectively. represents the last transmission time of node j, Indicates that node i receives the clock information of node j last transmitted The local hardware clock value recorded at and They represent the drift and offset compensation parameters of the logical clock of node j recorded by node i before the update. Considering the ideal fixed delay, the estimated value of the real-time hardware clock is consistent with the actual value, while there is an acceptable small error between the estimated value of the real-time logical clock and the actual value.

[0023] The dynamic threshold refers to a threshold constructed based on the estimated value of the neighbor's real-time status information. In order to avoid frequent communication, the present invention uses the neighbor's real-time status information in the estimator. At the same time, the measurement error is constructed to be compared with the dynamic threshold. The specific process is as follows:

[0024] Use e i represents the measurement error of node i, in, It represents the estimated value of the logical clock formed by node i using the clock information broadcast to neighboring nodes for the last time, that is, and They represent the logical clock drift and offset compensation parameters of the most recent broadcast of node i; z i It represents the combined measurement state of node i, Indicates update time The actual logical clock value of node i and the dynamic threshold are expressed as

[0025] After node i is updated, it is necessary to determine whether the following formula is satisfied:

[0026]

[0027] When this condition is met, node i broadcasts a clock information packet after updating the clock, where σ is a trade-off parameter, satisfying 0<σ<1, and the trade-off parameters of different nodes may be different.

[0028] The static threshold refers to: setting the currently updated logical clock offset compensation parameter and the value before update Subtract and compare with the set static threshold:

[0029]

[0030] Wherein, c is a set constant. When this condition is met, node i broadcasts the clock information packet after updating the clock. The purpose of setting this static threshold is to improve the early convergence speed. That is, the threshold set in the present invention is a combination of the dynamic threshold and the static threshold:

[0031] or

[0032] The storing and broadcasting of the clock status information of the node means that when the set dynamic or static threshold is exceeded, the node stores the updated logical clock drift and offset compensation parameters, and broadcasts them together with the hardware clock value of the node to neighboring nodes.

[0033] The local node adjusts the local clock until the global clock is synchronized, which means that the local node i adjusts the local logical clock based on the updated logical clock compensation parameter:

[0034] The advantages of the present invention using the above technical solution are:

[0035] 1. The low-overhead time synchronization method of industrial Internet of Things based on event triggering of the present invention fully considers the characteristics and application requirements of the industrial Internet of Things, designs a new threshold event triggering method based on an undirected network with random topology, and solves the problem of unstable and uneven triggering of wireless sensor nodes in the event-triggered time synchronization method without adding additional communication.

[0036] 2. The algorithm proposed in the present invention is completely distributed, that is, it does not require any global information. Each node completes the update by collecting the broadcast information of neighboring nodes, which significantly reduces the frequent communication of wireless sensor nodes in the industrial Internet of Things.

[0037] 3. The present invention proposes a new state-dependent dynamic threshold, in which the dynamic threshold condition is determined by the clock state error between each node and its neighbors, and the trigger frequency and synchronization accuracy are controlled by adjusting the trade-off parameters.

[0038] 4. The present invention uses an estimator to estimate the neighbor real-time clock, which overcomes the problem that the old clock information of the neighbor node used when the node is updated has become inaccurate over time, and improves the time synchronization accuracy of the node without increasing the communication energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic diagram of a random topology structure provided by an embodiment of the present invention;

[0040] Figure 2 This is a flow chart of event-triggered time synchronization used in the present invention;

[0041] Figure 3 A node-triggered communication event graph generated by the present invention using simulation means;

[0042] Figure 4 This is a global logic clock synchronization effect diagram generated by simulation means in the present invention. DETAILED DESCRIPTION

[0043] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments, but the present invention is not limited to these embodiments, and the various contents mentioned in the present invention can also be well applied in different fields and occasions.

[0044] The present invention includes the following contents: connecting sensors through a wireless shared network and forming a random topology structure, the node uses an estimator to estimate the real-time clock information of neighboring nodes, combines the information collected in the update interval, and regularly updates the local clock according to the update cycle. In addition, the system status around the node is used to set dynamic and static thresholds, and the clock information is broadcast when the comparison value of the relevant clock information exceeds the threshold, and all nodes adjust the local clock until the global clock is synchronized.

[0045] like Figure 1 As shown, it is an example of random topology of distributed wireless network considered in the present invention, which is an industrial Internet of Things topology composed of 10 sensor nodes randomly. Its topological structure can be described by a graph G∈(V,E), where V={1,2,...,n} represents the set of nodes in the Internet of Things; Represents the edge set between nodes. If (i,j)∈E, it means that nodes i and j are adjacent, node i can obtain information from node j, and N i ={j|(i,j)∈E} is the set of neighbor nodes of node i. In the present invention, since the channels between nodes are bidirectional, that is, (i,j)∈E means (j,i)∈E, the graph G is an undirected graph.

[0046] Each wireless sensor node in the industrial Internet of Things is equipped with a crystal oscillator, whose output is used to trigger a timing counter. For node i, the hardware clock τi (t) can be expressed as:

[0047] τ i (t) = α i t+β i

[0048] Among them, t represents absolute time, α i and β i Represent the hardware clock drift and offset of node i respectively. However, since the node cannot access the absolute time t, it is not possible to directly obtain and modify the hardware clock parameters. To ensure smooth time synchronization, each node is equipped with a logical clock based on the hardware clock. The logical clock L of node i i (t) is defined as:

[0049]

[0050] in, and They represent the logical clock drift and offset compensation parameters respectively; and represent the logical clock drift and offset respectively. Therefore, the time synchronization goal of the present invention is to make the logical clock drift x i and offset y i Tend to be globally consistent, and then achieve consistent time synchronization of the entire network, that is:

[0051]

[0052] Before the synchronization process begins, the parameters are initialized. In this embodiment, the number of node updates is set to k = 1, and the hardware clock drift α and offset β are randomly selected from the intervals [0.9999, 1.0001] and [0, 0.0002], respectively. The initial logical clock drift and offset compensation parameters are set to The update cycle is set to T = 1s, and the actual update time of the node is t k =(kT-β) / α. Since there are slight differences in α and β between nodes, the updates of each node in the network are not strictly synchronized.

[0053] When the update times k = 1 and k = 2, the node cannot update the clock information due to incomplete information collected. At the first two moments, the node only broadcasts the local clock information. For node i, the kth update time of its hardware clock can be expressed as When the number of updates k is greater than 2 and satisfies When , the node will use the estimator to estimate the real-time clock of neighbor node j and update the logical clock compensation parameters as follows:

[0054] The real-time hardware clock estimate is:

[0055] The estimated real-time logical clock value is: in, and Represents the real-time hardware clock and logical clock of node j estimated by node i. Represents the hardware clock of the latest transmission of node j, received The hardware clock of node i is recorded as Indicates update time The instant before, and They represent the logical clock drift and offset compensation parameters of node j recorded by node i before updating, respectively, ij represents the relative clock slope of node i and node j, that is, Among them, α i and α j denote the hardware clock drift of nodes i and j respectively, and They represent that node i receives the hardware clock from node j and The node then updates its own hardware clock according to the following formula:

[0056]

[0057] Among them, |N i | represents the degree of node i, Indicates the logical clock drift compensation parameter of node j recorded by node i before updating. Indicates the logical clock drift compensation parameter of the most recent broadcast by node i, is the estimated value of the logical clock of node j, Indicates the logical clock value of node i before the update.

[0058] After the update is completed, the measurement error is calculated Right now in It represents the estimated value of the logical clock formed by the node using the clock information broadcast to the neighboring nodes for the last time, that is, and They represent the logical clock drift and offset compensation parameters of the most recent broadcast of node i; z i represents the combined measurement state of node i, Indicates update time The actual logical clock value of node i will measure the error Compare with the dynamic threshold; also update the current logic clock drift compensation parameters and the value before update Subtract and compare with the set static threshold:

[0059] or in, Represents the dynamic threshold, and the trade-off parameter σ is a constant. Its value is the key factor in balancing synchronization accuracy and energy consumption. The smaller the value of σ, the stricter the dynamic threshold trigger condition will be, resulting in more trigger times. Nodes can replace old information with new broadcast information more frequently to update compensation parameters and obtain higher synchronization accuracy, but the communication overhead will also increase accordingly. In this embodiment, σ is taken as 0.05; the trigger formula of the static threshold can solve the problem of slow convergence speed in the early stage, and selecting a constant as the static threshold will not increase the number of broadcasts triggered in the subsequent stage. Considering the hardware parameters of the sensor node, the constant c is selected as 0.000002 in this embodiment.

[0060] When the set dynamic threshold or static threshold is exceeded, the node stores the updated logical clock drift and offset compensation parameters and broadcasts them to neighboring nodes together with its own hardware clock value.

[0061] Finally, the local node i adjusts the local clock based on the updated logical clock compensation parameter until the global clock is synchronized:

[0062] like Figure 3 As shown in the figure, it is a triggering time diagram of the sensor node. Each point in the figure is mapped to a communication broadcast of the corresponding node, and the density of the points reflects the level of communication energy consumption. As can be seen from the figure, the method of the present invention can ensure the uniform sparsity of the broadcast while stably triggering the broadcast event. Specifically, the number of communication interactions of all nodes is small, the broadcast interval of the same node is uniform, and the amount of broadcast tasks undertaken by different nodes is similar.

[0063] like Figure 4 As shown in FIG. 1 , it is a graph showing the maximum error variation of the global logical clock. The smaller the maximum error, the higher the time synchronization accuracy. As shown in the figure, the logical clock using the method of the present invention gradually converges after about 10 seconds. As the number of updates increases, the logical clock can achieve higher convergence accuracy.

[0064] It should be noted that the above are only embodiments of the present application to explain the technical solution and not to limit it. Although the present invention has been explained in detail with preferred embodiments, any technician familiar with this profession needs to be clear that without departing from the scope of the technical solution of the present application, making slight changes or modifications using the technical contents disclosed above are equivalent to equivalent implementation cases and are within the scope of the technical solution.

Claims

1. A low-overhead time synchronization method for industrial Internet of Things based on event triggering, characterized in that: The following steps are involved: Connect sensor nodes with similar hardware parameters through a wireless shared network to generate a random topology; The local node initializes the clock parameters and sets the update cycle. All nodes update the logical clock drift and offset compensation parameters according to the cycle. Based on the designed estimator, the local node uses its own existing information to estimate the real-time status information of its neighbors and construct a dynamic threshold. At the same time, in order to speed up the convergence speed in the early stage, the constraint of the logical clock offset compensation parameter is added on the basis of the dynamic threshold, that is, a trigger condition of a static threshold is added; The local node determines whether the trigger condition is met, and saves and broadcasts its own clock status information if the trigger threshold is exceeded; The local node adjusts the local clock until the global clock is synchronized.

2. The method according to claim 1, characterized in that Sensor nodes with similar hardware parameters are connected together through a wireless shared network to generate a random topology, including: The sensor nodes with the same functions are randomly distributed, and use the communication radius to determine their neighbor nodes and conduct two-way communication to form a random network topology, but it is necessary to ensure that each local node has at least one neighbor node.

3. The method according to claim 1, characterized in that The local node initializes the clock parameters and sets the update period. All nodes update the logical clock drift and offset compensation parameters according to the period, including: Set the node update cycle to T, the initial update number to k = 1; set the initial logical clock drift and offset compensation parameters of all nodes to And the relative clock slope η is set to 1, t k Indicates the physical time of the kth update of the node; The local node updates the local logical clock parameters using the collected clock information according to the set update cycle.

4. The method according to claim 3, characterized in that The local node updates the local logical clock parameters using the collected clock information according to the set update cycle, including: For any sensor node i, the kth update time of its hardware clock can be expressed as When the number of updates k is greater than 2 and satisfies When the logic clock drift compensation parameters and offset compensation parameters are updated respectively: in, and They represent the logical clock drift compensation parameter and offset compensation parameter of node i respectively; η ij represents the relative clock slope of node i and node j, that is, Among them, α i and α j denote the hardware clock drift of nodes i and j respectively, The hardware clock value received by node i from node j is The hardware clock value recorded when is the kth broadcast time of node j, and node i records the logical clock drift compensation parameter broadcast by neighbor node j as This means that in the interval between two broadcasts of node j, the logical clock drift compensation parameter of node j recorded by node i remains unchanged; |N i | represents the degree of node i, N i is the set of neighbor nodes of node i; Indicates update time The instant before, Indicates the logical clock drift compensation parameter of node j recorded by node i before updating. Indicates the logical clock drift compensation parameter of the last broadcast of node i. In addition, Indicates update time The estimated value of the logical clock of node j recorded by node i at time Indicates update time The previous logical clock value of node i.

5. The method according to claim 1, characterized in that Based on the designed estimator, the real-time status information of the neighbors is estimated using the existing information, including: Nodes do not broadcast local clock information at all update times, so they do not know the real-time hardware clock value and logical clock value of neighboring nodes. In order to improve the accuracy of updating parameters without increasing additional communication, the present invention designs an estimator to estimate the real-time hardware clock and logical clock value of node j: in, and Represent the estimated values ​​of the real-time hardware clock and logical clock of node j, respectively. represents the last transmission time of node j, Indicates that node i receives the clock information of node j last transmitted The local hardware clock value recorded at and They represent the drift and offset compensation parameters of the logical clock of node j recorded by node i before the update. Considering the ideal fixed delay, the estimated value of the real-time hardware clock is consistent with the actual value, while there is an acceptable small error between the estimated value of the real-time logical clock and the actual value.

6. The method according to claim 1, characterized in that Build dynamic thresholds, including: The threshold is constructed based on the estimated value of the neighbor's real-time status information. In order to avoid frequent communication, the present invention uses the neighbor's real-time status information in the estimator, and at the same time, constructs the measurement error to compare with the dynamic threshold; the specific process is as follows: Use e i represents the measurement error of node i, in, It represents the estimated value of the logical clock formed by node i using the clock information broadcast to neighboring nodes for the last time, that is, and They represent the logical clock drift and offset compensation parameters of the most recent broadcast of node i; z i It represents the combined measurement state of node i, Indicates update time The actual logical clock value of node i and the dynamic threshold are expressed as After node i is updated, it is necessary to determine whether the following formula is satisfied: When this condition is met, node i broadcasts a clock information packet after updating the clock, where σ is a trade-off parameter, satisfying 0<σ<1, and the trade-off parameters of different nodes may be different.

7. The method according to claim 1, characterized in that Add constraints on the logical clock offset compensation parameters on the basis of the dynamic threshold, that is, add a trigger condition for the static threshold, including: Update the current logical clock offset compensation parameter and the value before update Subtract and compare with the set static threshold: Wherein, c is a set constant. When this condition is met, node i broadcasts the clock information packet after updating the clock. The purpose of setting this static threshold is to improve the early convergence speed. That is, the threshold set in the present invention is a combination of the dynamic threshold and the static threshold: or 8. The method according to claim 1, characterized in that The local node determines whether the trigger condition is met. If the trigger threshold is exceeded, the local node saves and broadcasts its own clock status information, including: When the set dynamic or static threshold is exceeded, the node saves the updated logical clock drift and offset compensation parameters and broadcasts them to neighboring nodes together with its own hardware clock value.

9. The method according to claim 1, characterized in that: The local node adjusts the local clock until the global clock is synchronized, including: Local node i adjusts the local logical clock based on the updated logical clock compensation parameter:

Images