CN104053223A - A Time Synchronization Method for Low-power Wireless Sensor Networks - Google Patents

Abstract

The invention provides a low-power wireless sensor network time synchronization method. According to the method, synchronization information is carried by a control packet, a timestamp is printed at the tail of the periodic control packet, and a receiver extracts timestamp information and estimates synchronization delay online after receiving a synchronization packet. The receiver calibrates the time thereof according to the principle of time calibration, and realizes a synchronization mechanism for whole-network synchronization through local synchronization. During synchronization, synchronization is performed only between father and son nodes. By adopting the method of the invention, the number of nodes sending time messages in a network can be greatly reduced, the network traffic in the process of synchronization can be reduced, and the goal of reducing the power consumption of wireless sensor network time synchronization can be achieved under the premise of ensuring the time synchronization accuracy.

Description

A Time Synchronization Method for Low-power Wireless Sensor Networks

technical field

The invention belongs to the technical field of wireless sensor networks, and relates to a transmission method of synchronous messages in the wireless sensor network, in particular to a time synchronization method of the wireless sensor network with low energy consumption and high precision. the

Background technique

Wireless sensor network (WSN) is composed of a large number of micro sensor nodes deployed in the monitoring area, and forms a multi-hop self-organizing network through wireless communication. It has brought about a revolution in information perception with the characteristics of low power consumption, low cost, distribution and self-organization. It has been widely used in environmental perception and monitoring, wireless positioning and tracking, medical monitoring, smart home and other fields. prospect. the

Time synchronization is a key supporting technology of WSN. WSN is a distributed system, and the local clocks of each node are independent of each other, which causes the system time of each node to be different from each other. Many WSN applications require all nodes to work together, and the raw data collected by a single node needs to be processed in a unified manner with other node data to obtain useful information. For example, in the target tracking application scenario, it is necessary to perform Kalman filtering on the data collected by relevant nodes to extract useful location information. Such data processing and data fusion require relevant nodes to maintain time synchronization. On the other hand, time synchronization also has important applications in the design of WSN communication protocols. The design goals of the MAC (Medium Access Protocol) protocol mainly include saving energy and preventing collisions. TDMA has inherent advantages in this regard, and time synchronization is the basis of TDMA. the

The system time of the sensor node is generated by a timer, and the timing signal is generally provided by a crystal oscillator (crystal oscillator). Due to differences in the manufacturing process of crystal oscillators and changes in the operating environment, it is difficult to keep the timing frequency consistent, resulting in the time of each node being easily out of sync. Therefore, it is necessary to design a special synchronization algorithm to ensure the time synchronization of nodes in the network. the

Traditional time synchronization methods mainly include NTP (Network Time Protocol) and GPS (Global Positioning System). NTP is a standard time synchronization protocol on the Internet, which mainly provides standard UTC (Coordinated Universal Time) time for all hosts on the Internet. GPS is used for global high-precision navigation and positioning. Both NTP and GPS require specific equipment or dedicated nodes to achieve synchronization. Due to the limited hardware resources of sensor nodes and relatively high energy saving requirements, these two traditional time synchronization methods cannot be directly used in WSN. the

Currently, time synchronization methods for WSN are mainly divided into the following three types: (1) Receiver-receiver-based time synchronization methods. For example RBS (Reference Broadcast Synchronization). (2) One-way time synchronization method based on sender-receiver. Typical representatives are DMTS (Delay Measurement Time Synchronization) and FTSP (Flooding Time Synchronization Protocol). (3) Two-way time synchronization method based on sender-receiver. For example, TPSN (Timing-sync Protocol for Sensor Networks) and LTS (Lightweight Tree-based Synchronization). The following uses the LTS algorithm to synchronize the time of the wireless sensor network as an example to introduce this in detail:

The core task of the LTS lightweight spanning tree synchronization algorithm is to reduce the complexity of the synchronization algorithm, which was proposed by Jana van Greunen at the University of California, Berkeley. In some sensor network applications, the accuracy requirements for time synchronization are not very high, and the nodes that need time synchronization may not be all nodes in the entire network, so a simple lightweight time synchronization mechanism can be used to reduce the frequency of time synchronization and The number of nodes participating in the synchronization can reduce the communication and computing overhead of the nodes while meeting the synchronization accuracy requirements, and reduce the energy consumption of the network. Two LTS multi-hop time synchronization algorithms, centralized and distributed, are proposed based on the analysis of the time synchronization mechanism based on send-receive between single-hop node pairs. the

The centralized multi-hop synchronization algorithm is a simple linear extension of single-hop synchronization. Its basic idea is to construct a low-depth spanning tree, and then use the tree root as a reference node to perform step-by-step synchronization to leaf nodes, and finally achieve network-wide synchronization. The root node initiates the time synchronization process by synchronizing its neighbor child nodes; each child node then synchronizes with its child nodes; and so on, until the leaf nodes of the tree are all synchronized. The root node also initiates resynchronization when needed. In the centralized multi-hop synchronization algorithm, the root node initializes synchronization, so the nodes resynchronize at the same frequency. The running time of the algorithm is proportional to the depth of the spanning tree. The optimized spanning tree has the minimum depth, and the synchronization operation is performed in parallel along all branches . Since the depth of the spanning tree affects the synchronization time of the entire network and the accuracy error of the leaf nodes, it is necessary to transmit the depth of the tree back to the root node so that the root node can use this information when deciding to resynchronize. The communication complexity and precision of multi-hop synchronization are related to the construction method of the spanning tree and the depth of the tree, and the resynchronization frequency is related to the clock drift and single-hop synchronization precision. the

In the distributed synchronization algorithm, each node determines its own synchronization time, and the algorithm does not use spanning tree. When node i decides that it needs to resynchronize (determined by the synchronization accuracy requirement, the hop distance from the clock source node and the clock drift), it sends a synchronization request to the nearest reference node. Then, all nodes along the path from the clock source node to node i must be synchronized before i-node is synchronized. The advantage of this mechanism is that some nodes may have very little event forwarding and thus rarely need to synchronize. This saves unnecessary synchronization overhead since a node has the opportunity to determine its own synchronization. At the same time, by merging synchronous requests, the number of requests on the same path is reduced and resources are saved. the

In the LTS algorithm, nodes in the network avoid synchronizing with multiple upper nodes, but only synchronize with their direct parent nodes, which reduces the number of message exchanges and synchronization time. The purpose of this algorithm is to minimize the complexity to reduce energy consumption, and the accuracy is average. The running time of the algorithm is proportional to the depth of the tree, so the convergence time is the shortest when the spanning tree has the minimum depth, but the construction of a small spanning tree also requires certain calculation and communication overhead. the

Contents of the invention

The technical problem to be solved by the present invention is to provide a low-energy wireless sensor network time synchronization method for data collection scenarios without sacrificing synchronization accuracy, which can effectively reduce the cost of time synchronization and reduce time synchronization. purpose of energy consumption. the

For solving the problems of the technologies described above, the present invention is achieved like this:

A. To send periodic synchronization information, use the BEACON packet (control packet) to carry the synchronization information, and put the time stamp at the end of the periodic BEACON packet for time synchronization between nodes. The BEACON packet structure is shown in Figure 1. the

B. When the receiver receives the BEACON packet, it can calibrate its own time according to the timestamp information. The timestamp includes TAR and count information, indicating the current system time of the sending node. the

Assuming that nodes 1 and 2 need to implement time synchronization, node 1 is a time reference node (parent node), and node 2 is a node to be synchronized (child node). The calibration principle is as follows:

TAR ₂ =TAR ₁ +ΔT (1)

TACCR1 ₂ = TAR ₂ + (INTERVAL - TAR ₂ % INTERVAL) (2)

${\mathrm{<span\; class="notranslate">\; count</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; count</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; TAR</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; INTERVAL</span>}}<span\; class="notranslate">\; -</span>\frac{{\mathrm{<span\; class="notranslate">\; TAR</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}{\mathrm{<span\; class="notranslate">\; INTERVAL</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

Both TAR ₁ and count ₁ can be directly extracted from the synchronization packet, and INTERVAL is constant, so there is only one unknown quantity ΔT.

C. Online estimation of the synchronization packet delay ΔT between nodes in the wireless sensor network, three estimation methods can be used:

1. Fixed value correction method. the

2. Tracking correction method. the

3. Inter-cell mean correction method. the

D. A synchronization mechanism that realizes the synchronization of the whole network with local synchronization. The steps include:

In the data acquisition tree, all data exchange is carried out in the parent node-child node pair, each node is only synchronized with its parent node, and the whole network synchronization is realized step by step through the tree topology. the

1. When the network is initialized, a data collection tree with the SINK node as the root node will be established. the

2. SINK nodes periodically send BEACON packets containing timestamps. 3. After receiving the BEACON packet, the child node parses out the time stamp, and calibrates its own clock according to the formula (1-3). the

4. After each node completes synchronization (that is, network access), it will also broadcast BEACON packets in the way of SINK nodes to facilitate the addition of new nodes. the

5. When a new node joins, it will first listen for a period of time. After collecting the BEACON of the neighbor node, it will select a parent node and synchronize with it to complete the network access. the

6. When the topology changes, the disconnected node will first listen for a period of time, reselect the parent node, and complete the synchronous access to the network. the

As can be seen from the above method, the core idea of the present invention is: put the time stamp at the end of the periodic control packet, dynamically estimate the time synchronization delay of the parent and child nodes, and use the estimated synchronization delay to correct the node time, In this way, the synchronization between the parent and child nodes is realized, and the purpose of improving the time synchronization accuracy of the wireless sensor network nodes is achieved and the network power consumption is reduced. the

The present invention has the following advantages:

1. Use periodic control packets to carry synchronization information. By completely embedding the synchronization information into the control packets generated by the upper layer protocol stack, and putting the timestamp at the end of the periodic control packets, the synchronization packets only need to be in the control packets of the MAC layer. By adding 9 bytes to the end of the packet, functions such as synchronization, routing, and network maintenance can be realized uniformly. . In this way, time synchronization will not bring any additional data packet overhead, but only increase the existing packet length by several bytes, which can effectively save energy consumption for time synchronization. the

2. By analyzing the error distribution characteristics and time-domain variation characteristics of synchronization delay, combined with the principle of time synchronization calibration, three synchronization delay correction methods are proposed: fixed value correction method, tracking correction method and inter-cell average correction method. The mean value and variance of the synchronization error of the three synchronization delay correction methods have been improved to a certain extent, and the synchronization accuracy of the microsecond level (<100 microseconds) can be achieved, and the synchronization accuracy is equivalent to the classic algorithm TPSN and RBS. the

3. A data acquisition tree is proposed for data application scenarios. All data exchanges are performed in parent node-child node pairs, and each node is only synchronized with its parent node. Using tree topology and local parent-child node synchronization can realize the synchronization of nodes in the entire network, which greatly reduces the number of synchronization packets sent in the network. On the premise of ensuring the accuracy of time synchronization, it can effectively reduce the energy of nodes in the network consumption purpose. the

Description of drawings

Figure 1: Format description of the control packet. the

Figure 2: Schematic illustrating the linear dependence of synchronization delay on packet length. the

Figure 3: Schematic diagram of the time synchronization error distribution using the fixed value correction method. the

Figure 4: Schematic diagram of the timing of the time synchronization error using the fixed value correction method. the

Figure 5: Schematic diagram of time synchronization error distribution using tracking correction method. the

Figure 6: Schematic diagram of the timing of time synchronization errors using the tracking correction method. the

Figure 7: Schematic diagram of the process of time synchronization between two nodes. the

Figure 8: Schematic diagram of the basic principle of inter-cell mean correction method. the

Figure 9: Schematic diagram of time synchronization error distribution using inter-cell mean correction method. the

Figure 10: Schematic diagram of the time series of time synchronization errors using the inter-cell mean correction method. the

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the implementation of the present invention will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. the

The core idea of the present invention is: put the time stamp at the end of the periodic control packet, dynamically estimate the time synchronization delay of the parent and child nodes, and use the estimated synchronization delay to correct the node time, thereby realizing the synchronization between the parent and child nodes , to achieve the purpose of improving the time synchronization accuracy of wireless sensor network nodes. the

First introduce the realization method of the present invention below, optional one comprises 8MHz MSP430 microprocessor, 10KB RAM register, 48KB flash gravel register, the sensor network hardware node of 250kbps radio frequency transceiver, node time is produced by external clock source (crystal oscillator), It is uniquely determined by three (register) variables TAR, TACCR1 and count. the

1. TAR is a 16-bit hardware register. Counting according to the external clock source, the time in tick (1tick= ^2-15 seconds) can be obtained, and the timing accuracy reaches the microsecond level.

2. TACCR1 is a capture register. When TAR=TACCR1, a TIMER interrupt can be generated. By setting the increment step of TACCR1 to a fixed value INTERVAL, the TIMER interrupt can be counted to achieve a wider range of timing. the

3. count is a 32-bit software counter that records the number of TIMER interrupts to obtain millisecond-level timing. the

Figure 1 shows the time synchronization packet structure of the low-power time synchronization method. The so-called synchronization packet is a BEACON packet with a time stamp at the end of the packet. TYPE is a flag, indicating whether synchronization is required; SenderAddr is the address of the sender, based on which the receiver judges whether it is its parent node, and thus decides whether to synchronize with it; TAR and count are timestamps, indicating the current system time of the sending node. Since the node time is uniquely determined by TAR, TACCR1 and count, if the three pairs of parameters of the two nodes can be guaranteed to be consistent, time synchronization can be achieved. the

It is assumed that nodes 1 and 2 need to realize time synchronization, node 1 is a time reference node (parent node), and node 2 is a node to be synchronized (child node). The calibration principle is as follows:

TAR ₂ =TAR ₁ +ΔT (1)

TACCR1 ₂ = TAR ₂ + (INTERVAL - TAR ₂ % INTERVAL) (2)

${\mathrm{<span\; class="notranslate">\; count</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; count</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; TAR</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; INTERVAL</span>}}<span\; class="notranslate">\; -</span>\frac{{\mathrm{<span\; class="notranslate">\; TAR</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}{\mathrm{<span\; class="notranslate">\; INTERVAL</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

Both TAR ₁ and count ₁ can be directly extracted from the synchronization packet. INTERVAL is a constant, so there is only one unknown quantity ΔT, which is the time delay of the synchronization packet. The accuracy of ΔT estimation directly affects the synchronization accuracy.

The present invention puts the sending and receiving of the synchronization packet in the driver program of the physical layer, which is the lowest layer that can be controlled by software, and uses the control packet to carry the time stamp and the mechanism of periodically repeating the transmission of the synchronization packet to realize that the time delay of the synchronization packet introduced mainly includes Transmission delay, which is directly related to packet length. As shown in Figure 2, Figure 2 shows that the synchronization delay ΔT is linearly correlated with the packet length, and the following empirical formula can be obtained through linear fitting:

ΔT(tick)＝68.7+1.6*L(byte) (4)

According to the above formula, ΔT can be obtained for a given packet length L, and then time calibration can be performed according to formula (1-3). On this basis, the present invention proposes three time correction methods. the

1. Fixed value correction method

The calibration formula of the fixed value correction method is shown in formula (5):

t ₂ =t ₁ +ΔT (4)

Among them, t ₁ is the time of the reference node; ΔT is the synchronization delay correction parameter, in this method, ΔT is obtained according to formula (4); t ₂ is the time after correction.

The specific method of the present invention is: the child node receives the control packet sent by the parent node, analyzes the control packet, extracts the TAR ₁ information, then estimates ΔT according to the packet length L and formula (4), and calculates its current time to achieve synchronization between parent and child nodes.

Figure 3 is a schematic diagram of the time synchronization error distribution of the fixed-value correction method. The probability and statistical results of the data are shown in Table 1. It can be seen that the average value of the synchronization error is 0.968 ticks, or 29.54 microseconds, and the mean square error is 1.2054 ticks, or 36.79 microseconds . Figure 4 is a timing diagram of its synchronization error,

Table 1 Synchronization error probability statistics of three synchronization correction methods

2. Tracking correction method

As shown in Figure 4, Figure 4 shows that the synchronization error is relatively stable in the short-term small interval, but the synchronization error changes periodically in the long-term large interval. According to this characteristic, we further propose a tracking correction method to synchronize time between nodes. the

The correction formula of the tracking correction method is shown in formula (6):

t _{2, n} = t _{1, n} + (t′ _{2, n-1} -t _{1, n-1} ) (5)

Among them, the subscripts n and n-1 indicate the sampling time, and the following tables 1 and 2 indicate the node numbers. t _{1, n} , t _{1, n-1} represent the time of the reference node at the current moment and the previous moment respectively, t′ _2,n-1 is the time before the synchronous node is corrected at the previous moment, t _{2, n} is the synchronization The time the node was after the correction.

The specific method of the present invention is: the sub-node records the uncorrected time t′ _{2, n-1} at the previous moment and the time t _{1, n-1} of receiving the reference node, and then according to the sub-node according to the currently received control packet The information is analyzed to obtain the current time t _1,n of the reference node, and the current time t _2,n of the child node itself is calculated and estimated according to the formula (6), so as to realize the time synchronization between the parent and child nodes.

The distribution of synchronization errors using the tracking correction method is shown in Figure 5, and the probability and statistical results of the data are shown in Table 1. The average value of the synchronization errors is 0.4114 ticks, or 12.55 microseconds, and the mean square error is 0.7003 ticks, or 21.37 microseconds. The timing diagram of the synchronization error obtained by analyzing the timing of the error is shown in FIG. 6 . the

Compared with the fixed value correction method, the tracking correction method has higher synchronization accuracy and better performance. Its synchronization error has a mean reduction of 57.5% and a variance reduction of 41.9%. the

3. Inter-cell mean correction method

Figure 6 can reflect the time-domain variation trend of the synchronization delay ΔT, that is, ΔT is relatively stable in a short-term small interval, but changes periodically in a long-term large interval. Therefore, the present invention uses the average value within the cell to estimate the current ΔT. the

The basic principle of inter-cell mean value correction method can be described with FIG. 7 and FIG. 8 . As shown in Figure 7, Figure 7 describes the synchronization process between nodes A and B. Node A receives this control packet at time T _{n, A} , and node B receives this control packet at time T _{n, B} ', and after correction, node B Adjust your own time to T _n,B . Previously, the time difference between node A and node B looked like this:

T _{n, B} ′-T _{n, A} = ΔT _n + T _drift (7)

Among them, ΔT _n is the synchronization delay estimated by formula (4), and T _drift is the deviation caused by the frequency offset of the crystal oscillator.

Because the synchronization delay ΔT is relatively stable in a short time between cells, the present invention estimates the synchronization delay more accurately by tracking historical values. The tracking process is as follows:

ΔT _n -ΔT _n -1 = (T _{n, B} '-T _{n, A} ) - (T _{n-1, B} '-T _{n-1, A} ) (8)

The present invention estimates the current delay ΔT by tracking the historical values of three times, as shown in FIG. 8 , the interval taken in FIG. 8 is 3, and the average value in this interval is used to replace ΔT. Through experiments, it can be verified that the inter-cell mean value correction method can effectively reduce the synchronization delay fluctuation between cells. the

The distribution of synchronization delay error using the inter-cell mean value correction method is shown in Figure 9, and the probability statistics results of the data are shown in Table 1. It can be seen that the synchronization error is 0.6280 ticks, or 12.55 microseconds, and the mean square error is 0.7629 ticks, or 23.28 microseconds. Second. The timing diagram of the synchronization error obtained by analyzing the timing of the error is shown in FIG. 10 . the

Compared with the fixed value correction method, the inter-cell mean value correction method has higher synchronization accuracy and better performance. Its synchronization error has a mean reduction of 35.1% and a reduction of variance of 36.7%. the

In order to illustrate the improvement of the present invention in terms of energy consumption, the present invention is compared with the classic RBS and TPSN algorithms. The synchronization errors of RBS and TPSN are 29.13 microseconds and 16.9 microseconds respectively, which are equivalent to the synchronization accuracy of the present invention, but the time synchronization method proposed by the present invention has more advantages in terms of energy consumption. the

The energy consumption of WSN nodes is mainly generated by the data transmission and reception of the radio frequency module, so the energy consumption of different synchronization algorithms can be compared by the number of synchronization packet transmission and reception. Table 2 shows the synchronization overhead of each algorithm when considering a reference node and n receiving nodes in a broadcast domain. Among them, K is the time recording times of RBS algorithm, and the larger K is, the higher the synchronization accuracy will be. It can be seen from Table 2 that compared with RBS and TPSN, the time synchronization method of the present invention can greatly reduce the number of sending and receiving of synchronization packets, thus reducing energy consumption. the

Table 2 Comparison of synchronization overhead of different algorithms

The specific embodiments mentioned above further illustrate the purpose, technical solutions and beneficial results of the present invention. It should be understood that the above descriptions are only specific embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention. the

Claims (4)

1. A wireless sensor time synchronization method with low power consumption, characterized in that: (1) Periodic synchronization information transmission, using control packets to carry synchronization information; (2) After receiving the synchronization packet, the receiver calibrates its time according to the time stamp information according to the time calibration principle; (3) Three time correction methods estimate and correct the synchronization packet delay ΔT between nodes in the wireless sensor network online; (4) The synchronization mechanism that realizes the synchronization of the whole network with local synchronization, only needs the parent and child nodes to synchronize. 2. The time synchronization method for low-power wireless sensor networks according to claim 1, characterized in that: the packet tail of the control packet is marked with a time stamp, which can save energy consumption for time synchronization, and the synchronization packet only needs to be in the MAC layer. Add 9 bytes to the end of the packet, including: TYPE is a flag indicating whether synchronization is required; SenderAddr is the address of the sender, and the receiver judges whether it is its parent node based on this, so as to determine whether it needs to synchronize with it; TAR and count are timestamps , indicating the current system time of the sending node. 3. The time synchronization method for low-power wireless sensor networks according to claim 1, characterized in that: after receiving the synchronization packet, the receiver can calibrate its own time according to the timestamp information, and the node time is unique by TAR, TACCR1 and count It is determined that time synchronization can be achieved by ensuring that the three pairs of parameters of the two nodes are consistent. The calibration principle is as follows: Assuming that nodes 1 and 2 need to achieve time synchronization, node 1 is the time reference node (parent node), and node 2 is the node to be synchronized (child node). The calibration principle is as follows: TAR ₂ =TAR ₁ +ΔT 3-1 TACCR1 ₂ =TAR ₂ +(INTERVAL-TAR ₂ % INTERVAL) 3-2 ${\mathrm{<span\; class="notranslate">\; count</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; count</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; TAR</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; INTERVAL</span>}}<span\; class="notranslate">\; -</span>\frac{{\mathrm{<span\; class="notranslate">\; TAR</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}{\mathrm{<span\; class="notranslate">\; INTERVAL</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 3</span>}$ Both TAR ₁ and count ₁ can be directly extracted from the synchronization packet, and INTERVAL is constant, so there is only one unknown quantity ΔT. 4. The time synchronization method for low-power wireless sensor networks according to claim 1, characterized in that: three estimation methods are proposed to estimate online the synchronization packet time delay ΔT between nodes in the wireless sensor network: (1) Fixed value correction method The calibration formula of the fixed value correction method is shown in formula (3-4): t ₂ =t ₁ +ΔT 3-4 Among them, _t1 is the time of the reference node, ΔT is the synchronization delay correction parameter, and _t2 is the time after correction; (2) tracking correction method The correction formula of the tracking correction method is shown in formula (3-5): t _{2, n} = t _{1, n} + (t' _{2, n-1} -t _{1, n-1} ) 3-5 Among them, the subscripts n and n-1 indicate the sampling time, the following tables 1 and 2 indicate the node number, t _{1, n} , t _{1, n-1} respectively indicate the time of the reference node at the current moment and the previous moment, and t′ _{2 , n-1} is the time before the synchronization node is corrected at the previous moment, t _{2, n} is the time after the synchronization node is corrected; (3) Inter-cell mean correction method ΔT is relatively stable in short-term small intervals, but changes periodically in long-term large intervals. Therefore, the current ΔT can be estimated by using the average value in the cell, and the inter-cell average value correction method can effectively reduce the synchronization delay fluctuation in the cell. 5. The time synchronization method for low-power wireless sensor networks according to claim 1, characterized in that: in the data acquisition tree, all data exchanges are carried out in the parent node-child node pairs, and each node only Synchronize with its parent node, and realize the synchronization of the whole network through partial synchronization through the tree topology. The synchronization method is handled as follows: (1) When the network is initialized, a data collection tree with the SINK node as the root node will be established; (2) SINK nodes periodically send BEACON packets containing timestamps; (3) After receiving the BEACON packet, the child node parses out the time stamp, and calibrates its own clock according to the formula (3-1～3-3); (4) After each node completes synchronization (that is, network access), it will also broadcast BEACON packets in the way of SINK nodes to facilitate the addition of new nodes; (5) When a new node joins, it will first listen for a period of time, and after collecting the BEACON of the neighbor node, it will select a parent node and synchronize with it to complete the network connection; (6) When the topology changes, the disconnected node will first listen for a period of time, reselect the parent node, and complete the synchronous access to the network.