Time synchronization method for wireless load testing system

Abstract

A time synchronization method for a wireless load testing system, comprising two stages: a wireless sensor network system building stage and a network-wide time synchronization stage for suppressing communication delay and realizing accuracy. The influence caused by time delay in wireless communication is suppressed by means of a relative skewness estimator based on the Bayesian estimation theory, and the estimation precision of the relative skewness estimator is improved while the memory requirement is reduced; by using the average consensus protocol, high-precision network-wide time synchronization is realized; and the overall method enhances the robustness for time delay and realizes high precision. The problems of reduction of time synchronization precision caused by the time delay and a time synchronization failure in some child nodes in a complex spatial layout of a traditional time synchronization algorithm are solved.

Description

A time synchronization method for wireless load testing system Technical Field

[0001] The present invention relates to the technical field of wireless sensor networks for measuring aircraft loads, and in particular to a time synchronization method for a wireless load testing system. Background Art

[0002] Many distributed time synchronization algorithms have been proposed for traditional wireless sensor networks. These algorithms are robust to unreliable communication links and adaptable to dynamic network topologies. However, these algorithms often overlook the impact of communication delay on time information. In practical wireless sensor networks, communication delay is a significant limitation, significantly impacting both the accuracy and efficiency of time synchronization. Traditional time synchronization algorithms suffer from reduced accuracy and failure to synchronize some subnodes due to time delay in complex spatial layouts.

[0003] Summary of the Invention

[0004] The present disclosure provides a time synchronization algorithm for a wireless load testing system, which can accurately achieve time synchronization for the entire network. It solves the problems of reduced time synchronization accuracy and failure of time synchronization of some sub-nodes caused by time delays in traditional time synchronization algorithms in complex spatial layouts. It is suitable for point-to-multipoint and wireless networks where multiple points do not communicate with each other.

[0005] The time synchronization method for a wireless load testing system provided in the present disclosure includes two stages:

[0006] S1: wireless sensor system construction stage; S2: suppressing communication delay to achieve accurate full network time synchronization stage. The specific process is as follows:

[0007] S1: Wireless sensor system construction phase, including: S1.1 Providing a network model through a consensus algorithm, S1.2 Building a clock model, S1.3 Updating the compensation parameters of the node virtual logical clock, and S1.4 Proposing a timestamp exchange mechanism for a pseudo-periodic communication scheme;

[0008] S2: Suppressing communication delays to achieve accurate network-wide time synchronization, mainly including:

[0009] S2.1 proposes a relative skewness estimation method based on Bayesian theory to perform relative skewness estimation;

[0010] S2.2 Bias and offset compensation.

[0011] Compared with the prior art, the beneficial effects of the present invention are: (1) the influence of time delay in wireless communication is suppressed by a relative skew estimator based on Bayesian estimation theory, thereby improving the estimation accuracy of the relative tilt estimator while reducing memory requirements; (2) high-precision full-network time synchronization is achieved by applying the average consistency protocol; (3) its overall method enhances the robustness to time delay and has higher accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The above and other objects, features and advantages of the present disclosure will become more apparent through a more detailed description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, wherein like reference numerals generally represent like components throughout the exemplary embodiments of the present disclosure.

[0013] FIG1 is a schematic diagram of a flow chart of an exemplary embodiment according to the present disclosure;

[0014] FIG2 is a diagram illustrating an exemplary wireless network node distribution;

[0015] FIG3 is a block diagram of the time synchronization algorithm of the present disclosure, including (a) a block diagram of the wireless sensor network system construction stage, and (b) a block diagram of the stage of suppressing communication delay to achieve accurate full-network time synchronization. DETAILED DESCRIPTION

[0016] The preferred embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although preferred embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0017] The present disclosure provides a time synchronization method for a wireless load testing system. An exemplary embodiment flow chart is shown in FIG1 . The exemplary wireless load testing system includes a central node and 12 subnodes, which transmit load data to a host computer via wireless communication, as shown in FIG2 .

[0018] The method disclosed in this disclosure mainly includes two stages: S1: wireless sensor system construction stage, S2: suppressing communication delay to achieve accurate full network time synchronization stage, and the stage-by-stage flow chart is shown in Figure 3.

[0019] S1: The wireless sensor system construction phase includes: S1.1 providing a network model through a consensus algorithm, S1.2 building a clock model, S1.3 updating the compensation parameters of the node virtual logical clock, and S1.4 proposing a timestamp exchange mechanism for a pseudo-periodic communication scheme.

[0020] The steps of the wireless sensor system construction phase are as follows:

[0021] S1.1 Network model given by consensus algorithm

[0022] Assume that the communication topology is an undirected graph G = (N, ε) in a wireless network, where N = {1, 2..., N} represents the set of sensor nodes. Represents the set of valid communication links. a ij Represents a non-empty adjacent element. If and only if there is a directed link in G, the set of directly adjacent nodes of child node i can be represented as N i ={j|(i,j)∈ε}, where (i,j)∈ε indicates that node i and node j can communicate in one hop. For the state of a child node i, the consensus algorithm can give the following linear model:

[0023] Where k is the number of iterations. Applying the linear model of the consensus algorithm to the clock update equation can drive the node clock to a common clock.

[0024] S1.2 Construction of clock model

[0025] Each sensor node has a local hardware clock, whose first-order dynamic function is expressed as: i (t) = α i t+β i

[0026] Among them, c i (t) is the hardware clock reading of the absolute reference time t, α i and β i The local hardware clock deviation and offset are respectively. Using the hardware clock data of any two nodes to obtain their indirect information, the clock model is established as follows:

[0027] in,

[0028] Since the local hardware clock cannot be manually modified, it is proposed that the node maintain a virtual logical clock to represent the synchronized time. The model is as follows:

[0029] in, and Represents deviation compensation and offset compensation respectively, and and Defined as virtual logical clock deviation and offset respectively.

[0030] S1.3 Update the compensation parameters of the node virtual logical clock

[0031] By updating the compensation parameters of the virtual logical clock model in a periodic and iterative manner, the logical clock of each sensor node is close to a common clock.

[0032] Update the compensation parameters of the node virtual logical clock so that satisfy:

[0033] S1.4 A timestamp exchange mechanism for pseudo-periodic communication scheme is proposed

[0034] Each node broadcasts time information with MAC layer timestamp to neighboring nodes in each synchronization period T. The neighboring nodes immediately record the local clock after receiving the information. Assume that the kth broadcast from node i to neighboring node j with communication delay is represents the kth transmission time of node i, where Expressed as the corresponding absolute moment of transmission, we continue to assume that node j is in its local time Get the kth message, where represents the instantaneous time when node j receives the message in absolute time, so Right now

[0035] in, Indicates the communication delay, which is caused by random delay and fixed delay composition.

[0036] According to the communication delay hypothesis, for any communication link between any two nodes, the delay exists as a positive bounded independent random variable. Among them, M d is a positive real constant. In addition, we continue to assume that each node neither updates its virtual logical clock nor obtains the time information sent by its adjacent node until the time length is greater than M d .

[0037] S2: Suppressing communication delays to achieve accurate network-wide time synchronization

[0038] Two steps are proposed at this stage: S2.1 relative skew estimation, S2.2 tilt and offset compensation

[0039] The steps for suppressing communication delay and achieving accurate network-wide time synchronization are as follows:

[0040] S2.1 Relative skewness estimation

[0041] In order to offset the influence of time delay, a relative skew estimation method based on Bayesian theory is proposed.

[0042] Assume that each node i periodically broadcasts time information to its neighboring node j, including node ID, information transmission sequence number, current hardware clock value, logic deviation, and offset compensation value. Considering communication delay, including random delay and fixed delay, the time information exchanged between node i and node j is reconstructed as

[0043] The influence of other unknown parameters of relative skew is analyzed by subtracting the time information, i.e.

[0044] In the actual time synchronization process, when the message of node i is not received by the adjacent node j within the time length Md, the synchronization message will be regarded as discarded.

[0045] According to Bayesian estimation theory, the parameter to be estimated α ij The error function is expressed as

[0046] Then the quadratic function is defined as:

[0047] Estimating relative skewness using Bayesian mean squared error

[0048] Applied to the probability density function p(Δc j (n),α ij ), according to the Bayesian principle, we can get:

[0049] p(Δc j (n),α ij )=p(α ij |Δc j (n)pΔc j (n))

[0050] Therefore

[0051] Since p(Δc j (n))≥0 for all Δc j (n) are satisfied if the integral in the brackets is j (n) can be minimized, then Bmse will be minimized, and the estimator that minimizes Bmse is The mean of the posterior joint conditions of .

[0052] The equivalent recursive solution of the above formula is as follows:

[0053] In the case of bounded delay, the bounded convergence of the time synchronization algorithm depends on the convergence rate and decay rate of the relative skew estimation error. The convergence and convergence rate of the above equation are analyzed to determine the validity of the relative skew estimation. The analysis method is as follows:

[0054] The hardware local time when node j receives the kth data packet Substituting into the above formula we get:

[0055] when

[0056] In summary,

[0057] It follows that for any n∈N + , That is ε ij The convergence rate of (n) is equal to O(1 / (n-1)), so The effectiveness of the relative skewness estimator is demonstrated.

[0058] S2.2 Tilt and offset compensation

[0059] When node j gets the nth synchronization packet from node i, it will The nth tilt and skew compensation update is performed using the following update rules.

[0060] Among them, ρ a ∈(0,1),ρ o ∈(0,1) is the tuning parameter, a(n)=(n-1) -u is the attenuation factor, where u∈(0,1), To compensate for the skewness of node i at t, the initial condition of any node j∈N is set to The introduction of a(n) in the above formula is to further weaken the adverse effects of relative skew estimation errors and ensure the consistency of logical skew under communication delays.

[0061] The above technical solutions are only exemplary embodiments of the present invention. For those skilled in the art, it is easy to make various types of improvements or modifications based on the application methods and principles disclosed in the present invention, and are not limited to the methods described in the above specific embodiments of the present invention. Therefore, the methods described above are only preferred and do not have a restrictive meaning.

Claims

1. A time synchronization method for a wireless load test system, The following steps are involved: S1, establish a network model based on the consensus algorithm for updating the node virtual logic clock; S2, establishes a node virtual logic clock update model based on the timestamp exchange mechanism of the pseudo-periodic communication scheme; S3, build a relative skew estimator based on Bayesian theory to estimate the undetermined parameters in the node virtual logical clock update, so that the logical clock of each sensor node is close to a common clock; S4, updating the logical deviation compensation and offset compensation of the node.

2. The method according to claim 1, It is characterized in that The network model based on the consensus algorithm in step S1 includes: Assume that the communication topology is an undirected graph G = (N, ε) in a wireless network, where N = {1, 2, ..., N} represents the set of sensor nodes. represents the set of valid communication links, a ij represents a non-empty adjacent element. If and only if there is a directed link in G, the set of direct adjacent nodes of child node i can be represented as N i ={j|(i,j)}∈ε, where (i,j)∈ε means that node i and node j can communicate in one hop. For the state of a child node i, the consensus algorithm can give the following linear model: Where k is the number of iterations, x i (k) represents the linear model of node i after k iterations. Applying the linear model of the consensus algorithm to the clock update equation can drive the node's clock to a common clock.

3. The method according to claim 1 or 2, It is characterized in that The step S2 specifically includes: S21, construction of clock model: Each sensor node has a local hardware clock, whose first-order dynamic function is expressed as: c i (t)=a i t+b i Among them, c i (t) is the hardware clock reading of the absolute reference time t, α i and β i They are the local hardware clock bias and offset respectively; Using the hardware clock data of any two nodes to obtain their indirect information, the clock model is established as follows: in, Since the local hardware clock cannot be modified manually, the node maintains a virtual logical clock to represent the synchronization time. The model is as follows: Among them, α i and β i Respectively represent deviation compensation and offset compensation, and α i =α i α i and β i =α i β i +β i They are defined as virtual logical clock bias and offset respectively; S22, update the compensation parameters of the node virtual logical clock: By updating the compensation parameters of the virtual logical clock model in a periodic and iterative manner, the logical clock of each sensor node is made close to a common clock. That is, update the compensation parameters of the node virtual logical clock so that (α i ,β i )satisfy: in, Represented as a common clock, and Indicates the compensation parameters for updating the node virtual logical clock; S23, proposes a timestamp exchange mechanism for a pseudo-periodic communication scheme: Each node broadcasts time information with MAC layer timestamp to adjacent nodes in each synchronization period T. Adjacent nodes immediately record local clocks after receiving the information. Assuming that there is communication delay, the kth broadcast from node i to adjacent node j is represents the kth transmission time of node i, where Expressed as the transmission instant of the corresponding absolute time, let node j be in its local time Get the kth message, where represents the instantaneous time when node j receives the message in absolute time, so Right now in, Represents the communication delay, which is composed of random delay and fixed delay composition; Based on the communication delay, we propose the following hypothesis: for any communication link between any two nodes, the delay has a positive bounded independent random variable, namely Among them, M d is a positive real constant. In addition, it is assumed that each node neither updates its virtual logical clock nor obtains the time information sent by its neighboring nodes until the time length is greater than M d .

4. The method according to claim 3, It is characterized in that The step S3 specifically includes: Assume that each node i periodically broadcasts time information to adjacent node j, including node ID, information transmission sequence number, current hardware clock value, logic deviation and offset compensation value; consider communication delay, including random delay and fixed delay, and reconstruct the time information exchanged between node i and node j as The influence of other unknown parameters of relative skewness is analyzed by subtracting the time information, i.e. In the actual time synchronization process, when the message of node i is within the time length M d If the synchronization message is not received by the neighboring node j within , the synchronization message will be considered discarded; According to Bayesian estimation theory, the parameter to be estimated α ij The error function is expressed as e＝α ij -α ij Then the quadratic function is defined as: C(e)=e 2 =(a ij -a ij ) 2 Estimating relative skewness using Bayesian mean squared error R=E(C(e))=E((α ij -α ij ) 2 ) Applied to the probability density function p(Δc j (n),α ij ), according to the Bayesian principle, we can get: p(Δc j (n),α ij )=p(α ij ||Δc j (n)pΔc j (n)) Therefore B mse (α ij )=∫(∫(α ij -α ij ) 2 p(α ij |Δc j (n)d(α ij ))×p(Δc j (n))d((Δc j (n)); Since p(Δc j (n))≥0 for all Δc j (n) are satisfied if the integral in the brackets is j (n) can be minimized, then Bmse will be minimized, and the estimator that minimizes Bmse is The posterior joint conditional mean of : The equivalent recursive solution of the above formula is as follows:

5. The method according to claim 4, It is characterized in that The step S4 specifically includes: When node j receives the nth synchronization packet from node i, it will The nth skew and offset compensation update is performed using the following update rule: Among them, ρ a ∈(0,1),ρ o ∈(0,1) is the tuning parameter, a(n)=(n-1) -u is the attenuation factor, where u∈(0,1), α i (t) is the skewness compensation of node i at t; The initial condition of any node j∈N is set to β j (0)＝1。

Images