A low-overhead robust time synchronization method for industrial wireless sensor networks

Abstract

The present invention discloses a low-overhead robust time synchronization method for industrial wireless sensor networks, belonging to the field of industrial wireless communication networks. This method fully utilizes the heterogeneity of industrial network nodes to build a mesh-plus-star network topology. The upper-layer mesh network is composed of a small number of routing nodes with rich computing, communication, and storage capabilities, achieving wide-area coverage of the industrial network. Local logical clocks are updated based on time information periodically broadcast by neighboring routing nodes, implementing a distributed robust time synchronization method. The lower-layer star network is composed of a large number of resource-constrained, low-power network devices that implement a passive monitoring mechanism and synchronize with edge routing nodes in a single-hop manner, reducing the communication overhead of nodes within the star network. The present invention achieves robust time synchronization with low communication overhead and relatively fast convergence speed in a large-scale industrial wireless sensor network environment.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial wireless networks, and in particular to a low-overhead robust time synchronization method for an industrial wireless sensor network. Background Art

[0002] Wireless sensor networks (WSNs), characterized by excellent mobility and strong self-organizing capabilities, are a network technology that monitors target areas in a distributed manner. Building new industrial networks based on WSNs can effectively reduce the deployment costs of traditional wired networks like industrial Ethernet, meeting the growing demand for large-scale, ubiquitous production process monitoring in industrial networks, and has garnered widespread attention from both industry and academia. Time synchronization, which aims to ensure that nodes in a network maintain time consistency through specific technologies, supports fundamental WSN applications such as low-power sleep technology, multivariate data fusion, and time-slot transmission scheduling. It also underpins new industrial applications such as equipment fault tracing and multi-robot collaborative operation. Furthermore, the customizable and flexible manufacturing of Industry 4.0 requires plug-and-play industrial network equipment, meaning that network node failure detection and recovery must be as fast as possible. This, in turn, places higher demands on the convergence speed and robustness of time synchronization. Furthermore, industrial WSNs are often composed of resource-constrained nodes, making low overhead a crucial factor in the design of time synchronization.

[0003] The core concept of existing distributed time synchronization methods is that nodes adjust their local clocks based on neighbor information to achieve relative network synchronization. These methods offer robustness, such as the ATS (Average Time Synchronization) method based on average consistency theory, the MTS (Maximum Time Synchronization) method based on maximum consistency theory, and the GTSP (Gradient Time Synchronization Protocol) method. However, ATS and GTSP require periodic iterations, resulting in slow convergence and low efficiency. The MTS method is significantly affected by link latency, and network synchronization errors are prone to divergence. While the CCTS (Clustered Consensus Time Synchronization) method, based on network clustering, improves synchronization convergence speed, intra-cluster synchronization precedes inter-cluster synchronization, and the accuracy of inter-cluster synchronization is affected by intra-cluster synchronization accuracy. Virtual link-based methods utilize multi-hop neighbor information to improve network algebraic connectivity. While these methods can accelerate convergence, they also introduce information redundancy and increase network communication overhead.

[0004] WIA-PA (Wireless Networks for Industrial Automation-Process Automation) is an industrial wireless network protocol with independent intellectual property rights launched by the China Industrial Wireless Alliance. Unlike traditional clustered network topologies, WIA-PA is based on a hybrid mesh + star topology. The upper-layer mesh topology is flexible in layout and has strong anti-interference capabilities; the lower-layer star network topology can quickly broadcast upper-layer information to local network nodes in just one hop, overcoming the uncertainty of multi-hop transmission paths. Therefore, the present invention integrates network topology construction and distributed consistency time synchronization method design to design a low-overhead, robust, and fast-converging industrial wireless sensor network time synchronization method. Summary of the Invention

[0005] To overcome the shortcomings of the aforementioned background technologies, the present invention provides a distributed time synchronization method based on a mesh-star hybrid topology. This method initiates the time synchronization process by upper-layer mesh routing nodes (including edge routing nodes) and adjusts local clocks based on information from neighboring routing nodes, improving the robustness of time synchronization. Nodes within the local star network only need to listen to synchronization messages broadcast by edge routing access nodes to achieve time synchronization, reducing communication overhead and accelerating synchronization convergence.

[0006] The technical solution adopted by the low-overhead robust time synchronization method for industrial wireless sensor networks described in the present invention is:

[0007] Step 1: Build a mesh + star hybrid network topology;

[0008] Step 2: Initialize network node clock parameters;

[0009] Step 3: The mesh network routing node periodically broadcasts local time information and adjusts the local clock based on the time information received from neighboring routing nodes;

[0010] Step 4: The edge routing access node broadcasts the updated clock information in the star network, and the star network nodes adjust their local clocks based on the time information broadcast by the edge routing access node.

[0011] The hybrid network topology structure in step 1 consists of two layers, wherein the upper network topology structure is a mesh structure composed of routing devices; the lower network topology structure is a star structure composed of industrial field sensing network devices, and is connected to the upper mesh network through edge routing access devices, ultimately achieving full coverage of the industrial field network.

[0012] The network node clock parameter initialization in step 2 includes the logical clock slope compensation value of any node i 1, logical clock offset compensation value is 0, the relative clock slope a between the node and its neighbor nodes ij If it is 1, the period of time that the routing node broadcasts time information is T.

[0013] In step 3, the routing nodes, including edge routing nodes, are all based on the local hardware clock, i.e., H i (t)=kT,k∈N + , periodically broadcast time information, including local hardware time H i (t), logical clock slope compensation value Logical clock offset compensation value

[0014] When routing node i receives the time information broadcast by neighbor routing node j, it records the local time information and calculates the relative clock slope a between nodes i and j based on the time information of node j received at the last moment. ij (t):

[0015]

[0016] Among them, H i (t) is the hardware time information H received by node i from node j j (t) time, H i (t-1) is the hardware time information H received by node i from node j at the last moment j At time (t-1), node i updates the slope compensation value of the local logical clock

[0017]

[0018] in, is the logical clock slope compensation value of node i at the previous moment, is the logical clock slope compensation value sent by node j, η∈(0,1) is the set iteration weight parameter, which usually takes a value of 0.5.

[0019] Node i updates the logical clock offset compensation value

[0020]

[0021] in, is the logical clock offset compensation value of node i at the last moment, L i (t) is the logical time of node i, that is, ν∈(0,1) is the set iteration weight parameter, which usually takes a value of 0.5.

[0022] In step 4, after node l in the star network receives the time information update value broadcast by edge routing access node i, it calculates the relative clock slope a according to formula (1) based on the time information broadcast by the routing access node twice received in succession. li , and update the local logic clock slope and offset compensation value and as follows:

[0023]

[0024]

[0025] in, is the logical clock offset compensation value of node i at the last moment, L l (t) is the logical time of node l.

[0026] Compared with the prior art, the advantages of the above technical solution adopted in the present invention are:

[0027] 1. The present invention is based on a mesh-plus-star hybrid industrial wireless sensor network topology, which fully considers the computing, storage, and communication capabilities of routing nodes. Routing nodes are responsible for complex algorithm calculations and frequent communication iterations. Star network nodes only need to monitor and adjust their local clocks to achieve synchronization with edge routing access nodes, effectively reducing the communication and computing overhead of time synchronization methods.

[0028] 2. The present invention adopts a hierarchical network structure for time synchronization. The upper mesh topology adopts a distributed method to achieve time synchronization, and the lower star network nodes only passively listen to time information for synchronization. Therefore, the failure and movement of the star network nodes will not interrupt the network time synchronization. The above mechanism ensures the robustness of the time synchronization method.

[0029] 3. The present invention constructs a mesh network by directly connecting routing nodes, which solves the problem of loss of coverage nodes between adjacent clusters due to energy consumption, thus causing failure of inter-cluster time synchronization in the time synchronization method based on the coverage cluster network topology. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The industrial wireless sensor network provided by the embodiment of the present invention has a mesh-plus-star hybrid topology;

[0031] Figure 2 The time synchronization process provided by the embodiment of the present invention;

[0032] Figure 3 A schematic diagram of the logical clock slope synchronization error provided by an embodiment of the present invention;

[0033] Figure 4A schematic diagram of the logical clock offset synchronization error provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments, but the present invention is not limited to these embodiments.

[0035] The present invention includes the following contents: constructing a mesh + star hybrid network topology structure; initializing network node clock parameters; mesh routing nodes periodically broadcasting local time information and adjusting local clocks based on time information received from neighbor routing nodes; edge routing access nodes broadcasting updated clock information, and star network nodes adjusting local clocks based on the time information broadcast by the routing access nodes.

[0036] like Figure 1 As shown in the figure, the hybrid network topology consists of two layers. The upper network topology is a mesh, which is composed of routing devices with rich computing, communication, storage and other capabilities. It is responsible for wide area network coverage and data packet forwarding, and the edge routing device is responsible for local star network device management and access to the mesh network; the lower network topology is a star, which is composed of low-power industrial field sensing network devices. It is directly connected to the production process through sensors or actuators, responsible for collecting production data such as temperature, humidity, concentration, and pressure, and executing control commands issued by the upper computer; industrial field equipment is connected to the upper mesh network through edge routing access devices, ultimately achieving full coverage of the industrial field network.

[0037] Any node i in the industrial wireless sensor network is equipped with a timer based on local crystal oscillator counting to obtain time, and its local time is described by a first-order linear model, which is called hardware time H. i (t) = α i t+β i , where α i is the hardware clock slope, ideally the value is 1; β i The hardware clock offset is 0 in ideal case. However, due to factors such as manufacturing process, external environment interference, hardware aging, etc., the hardware clock parameter α i and β i It is not an ideal value. In addition, since the absolute time t is unknown and unmeasurable, although the hardware time of the node can be obtained, the hardware clock parameter α i and β i The value of H cannot be directly measured, and the hardware clock H cannot be directly adjusted i (t). Therefore, the logical clock is defined as in, is the logic clock slope compensation value, is the logical clock offset compensation value, is the logic clock slope, The purpose of the present invention is to adjust the logical clock compensation value and The network nodes are enabled to have globally consistent logical clock slope and logical clock offset.

[0038] During the network initialization phase, the slope compensation value of the logical clock of any node i is 1, logical clock offset compensation value is 0, the relative clock slope a between the node and its neighbor nodes ij If it is 1, the period of time that the routing node broadcasts time information is T.

[0039] like Figure 2 As shown, the time synchronization process is initiated by the routing device i (including edge routing devices) of the upper mesh network and is based on the local hardware clock with a time interval of T, that is, H i (t)=kT,k∈N + , periodically broadcast time information, including local hardware time H i (t), logical clock slope compensation value Logical clock offset compensation value

[0040] When routing node i receives the time information broadcast by neighbor routing node j, it records the local time information and calculates the relative clock slope a between nodes i and j based on the time information of node j received at the last moment. ij :

[0041]

[0042] Among them, H i (t) is the hardware time information H received by node i from node j j (t) time, H i (t-1) is the hardware time information H received by node i from node j at the last moment j At time (t-1), node i updates the slope compensation value of the local logical clock

[0043]

[0044] in, is the logical clock slope compensation value of node i at the previous moment, is the logical clock slope compensation value sent by node j, η∈(0,1) is the set iteration weight parameter, which usually takes a value of 0.5.

[0045] Node i updates the logical clock offset compensation value

[0046]

[0047] in, is the logical clock offset compensation value of node i at the last moment, L i (t) is the logical time of node i, that is, ν∈(0,1) is the set iteration weight parameter, which usually takes a value of 0.5.

[0048] When the edge routing access node updates the clock information, it broadcasts it in the local star network. After node l in the star network receives the time information broadcast by routing access node i, it calculates the relative clock slope a based on the time information broadcast by the routing access node twice in a row according to formula (1): li , and update the local logic clock slope and offset compensation value and as follows:

[0049]

[0050]

[0051] in, is the logical clock offset compensation value of node i at the last moment, L l (t) is the logical time of node l.

[0052] To verify the effectiveness of the method of the present invention (let it be FTS), the time synchronization performance (logical clock slope synchronization error and logical clock offset synchronization error) of ATS, GTSP and FTS is verified in the Matlab simulation environment. Set the initialization value of the network node clock parameter: T=1s、α i ∈[0.8,1.2], β i ∈[0,1]. The upper layer of the FTS is a mesh network consisting of eight edge routing nodes, each connected to three star network nodes. The 32 nodes in the ATS and GTSP networks are randomly distributed within a [1x1] range, with a node communication radius of 0.1. The logical clock slope synchronization error is defined as the maximum value of the logical clock slope error of each node, and the logical clock offset synchronization error is defined as the maximum value of the logical clock offset error of each node. Figure 4 The following graph shows the convergence of the logical clock slope synchronization error of the FTS method of the present invention, the ATS method, and the GTSP method after 2000 seconds of time synchronization. As can be seen from the graph, after several iterations, the logical slope error of each node in the network gradually stabilizes. The logical clock slope of FTS converges to the same value after about 500 seconds, while ATS and GTSP need about 1300 seconds to reach a stable state. Figure 4As can be seen, after the logical clock slopes of each node converge, the logical clock offsets of the nodes also converge faster. Furthermore, the FTS method converges significantly faster than the ATS and GTSP algorithms. Given a given time synchronization period, fast convergence means low communication overhead for network time synchronization. This means that the time synchronization method provided by the present invention offers fast convergence and low communication overhead.

[0053] Compared with the traditional distributed method, whose network convergence speed is constrained by the network scale and node communication range, the method of the present invention only executes the distributed time synchronization method in the upper mesh network, and uses a small number of routing devices with rich computing, communication, storage and other capabilities for networking, thereby accelerating the time synchronization convergence speed. In order to enhance the robustness of the upper mesh network, it is usually possible to configure redundant routing devices; or increase the number of neighbors of the routing devices, reduce the number of routing devices, and further reduce the number of iterations of network synchronization convergence. Industrial field equipment collects time information in a passive listening manner. When a device node fails or moves to connect to other star networks, it will not interfere with the current synchronization process. Therefore, the present invention effectively reduces the communication overhead in the time synchronization process and improves the robustness and convergence speed of time synchronization.

[0054] The above descriptions are merely a few embodiments of the present application and do not constitute any form of limitation to the present application. Although the present application discloses the preferred embodiments as above, they are not intended to limit the present application. Any technical personnel familiar with the present profession, without departing from the scope of the technical solution of the present application, using the technical content disclosed above to make slight changes or modifications are equivalent to equivalent implementation cases and fall within the scope of the technical solution.

Claims

1. A low-overhead robust time synchronization method for industrial wireless sensor networks, characterized in that: The following steps are involved: Step 1: Build a mesh + star hybrid network topology; Step 2: Initialize network node clock parameters; Step 3: The mesh network routing node periodically broadcasts local time information and adjusts the local clock based on the time information received from neighboring routing nodes; Step 4: The edge routing access node broadcasts the updated clock information in the star network, and the star network nodes adjust their local clocks based on the time information broadcast by the edge routing access node; In step 1, the hybrid network topology consists of two layers, wherein the upper network topology is a mesh structure composed of routing devices; the lower network topology is a star structure composed of industrial field sensing network devices, and is connected to the upper mesh network through edge routing access devices, ultimately achieving full coverage of the industrial field network; In step 2, the network node clock parameter initialization includes the logical clock slope compensation value of any node i. 1, logical clock offset compensation value is 0, the relative clock slope a between the node and its neighbor nodes ij If it is 1, the period of the routing node broadcasting time information is T; In step 3, routing nodes, including edge routing nodes, are based on local hardware clocks, i.e., H i (t)=kT,k∈N + , periodically broadcast time information, including local hardware time H i (t), logical clock slope compensation value Logical clock offset compensation value When routing node i receives the time information broadcast by neighbor routing node j, it records the local time information and calculates the relative clock slope a between nodes i and j based on the time information of node j received at the last moment. ij (t): Among them, H i (t) is the hardware time information H received by node i from node j j (t) time, H i (t-1) is the hardware time information H received by node i from node j at the last moment j (t-1) time; node i updates the local logical clock slope compensation value in, is the logical clock slope compensation value of node i at the previous moment, is the logical clock slope compensation value sent by node j, η∈(0,1) is the set iteration weight parameter, and its value is 0.5; Node i updates the logical clock offset compensation value in, is the logical clock offset compensation value of node i at the last moment, L i (t) is the logical time of node i, that is, It is the set iteration weight parameter, which is set to 0.5; In step 4, after node l in the star network receives the updated value of the time information broadcast by the routing access node i, it calculates the relative clock slope a according to formula (1) based on the time information broadcast by the routing access node twice received in succession. li , and update the local logic clock slope and offset compensation value and as follows: in, is the logical clock offset compensation value of node i at the last moment, L l (t) is the logical time of node l.

Images