KR100999686B1 - Real-time synchronization method for hybrid network - Google Patents

Abstract

The present invention relates to a real time synchronization method for a hybrid network.

More specifically, (a) the coordinator broadcasts a beacon message to the coordinator's network area at the beginning of a repetitive superframe, and through the beacon message, each node sends the clock of the node delayed by the frequency error to the coordinator. Calibrating to an approximate value, (b) the sequential node broadcasts to send a message to the coordinator in a time slot, and the contention access node obtains channel transmission delay 1 through the message, and (c) the (a) The coordinator broadcasts a beacon message in the next sequence of performing the superframe, and obtains a channel transmission delay 2 through the broadcast beacon message; and (d) calculates an average value by adding bidirectional values of the channel transmission delay. Implementation for a hybrid network comprising calibrating the clock of each node through the calculated mean value It relates to a method of synchronization.

The present invention has an effect of improving the efficiency of the synchronization data transmission and control between heterogeneous networks by providing a real-time synchronization method considering the wired and wireless mixed situation.

Hybrid, network, weight offset, RBS

Description

REAL-TIME SYNCHRONIZATION METHOD FOR HYBRID NETWORK}

The present invention relates to a real time synchronization method for a hybrid network, and more particularly, to a real time synchronization method for a hybrid network to which an improved RBS algorithm, a weight-based offset algorithm, and a differential BP technique are applied.

In general, the network connecting the controller and the sensor / actuator is an important element in a factory automation system. Many sensors and drivers are used in robots, conveyors, automatic painting lines, etc. Since they must operate organically, a network system for communication and control between devices is required.

Currently, the network protocol mainly used in the industrial network uses a wired fieldbus that supports real time. If wired network is used, stable communication is possible, but wiring and facility maintenance are difficult. As an alternative to these drawbacks, a control network using wireless has been proposed.

Wireless communication technology has the advantage of reducing the difficulty of connecting devices and wiring costs in the factory environment. However, due to transmission inaccuracy, the real-time is reduced and the bandwidth of the channel is small, making it difficult to apply. Recently, as the performance and reliability of the wireless communication technology is improved, and the price of the wireless devices is lowered, researches on the industrial network device using the wireless communication technology are actively conducted.

Wired and wireless gateways are a type suitable for applying wireless communication technology to existing industrial communication networks. Since the protocols used for wired and wireless are different in media access, simple connection is impossible, so the wired / wireless gateway has a function to control the communication priority and maximum delay factor according to different media access. And between wired and wireless, it plays the role of message transmission and node control of both networks.

Wired and wireless gateways connect heterogeneous networks to form a distributed system structure between wireless and wired. In such a distributed system, delays due to media access time differences between wired and wireless networks reduce real-time control. In order to guarantee the network real-time, it is necessary to control the synchronization of both networks through the gateway, and additional techniques are required for the existing algorithm for real-time synchronization of each node.

In many types of distributed system architectures, each node needs synchronization within a certain time frame. Synchronization algorithms used in the existing wired and wireless networks are similar, but have different schemes for the application of each node. In the wireless network, due to the spatial constraints of the medium, synchronization of each node is relatively difficult compared to the wire. Nodes in the wireless network are synchronized according to the message transmission interval, so the time error increases when there is no transmission for a long time. Accumulated errors affect the time of message delivery, and control messages cause network malfunction. In addition, the utilization rate of the channel is reduced by starting at different times in the competition section for channel access.

In order to solve these synchronization problems, a synchronization algorithm is used. However, when heterogeneous networks coexist, there is a problem that accurate synchronization is difficult because both applications are different.

References

[1] Morel, P., Croisier, A., “A wireless gateway for fieldbus”, IEEE International Symposium of Wireless: Merging onto the Information Superhighway, Personal, Indoor and Mobile Radio Communications, 27-29 Sept, 1995, Page: 105-109 vol. 1

[2] IEEE 1588 standard: Precision clock synchronization protocol for networked measurement and control systems

[3] J. Elson, L. Girod, D. Estrin, “Fine-grained network time synchronization using reference broadcasts”, ACM Operating Systems Review, Volume 36, Issue SI, December 2002, Pages 47-163

[4] IEEE standards for local area networks: token ring access method and physical layer specifications, 29 Sept, 1989

[5] J. Elson, K. Romer, “Wireless Sensor Networks: A new regime for time synchronization”, ACM SIGCOMM Computer Communication Review, Volume 33, Issue 1, January 2003, Pages 149-154

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems, and an object thereof is to provide a real-time synchronization method for a hybrid network for improving the efficiency of synchronization data transmission and control between heterogeneous networks.

The object of the present invention described above is that (a) the coordinator broadcasts a beacon message in the network area of the coordinator at the first part of the repetitive superframe, and through the beacon message, each node receives the clock of the node delayed by the frequency error. Calibrating to a value close to the coordinator, (b) the sequential node broadcasts to send a message to the coordinator in a time slot, and the contention access node obtains channel transmission delay 1 through the message, and (c) Acquiring a beacon message by the coordinator in a next sequence of the superframe in which step (a) is performed, and obtaining a channel transmission delay of 2 through the broadcast beacon message; Calculating a value and calibrating the clock of each node using the calculated average value. It is achieved by providing a real-time synchronization method.

The real-time synchronization method for a hybrid network according to an embodiment of the present invention, (e) calculating the average value of the channel transmission delay, (f) calculating the frequency error with the partner node using the average value and (g) further comprising maintaining an average value of the frequency errors obtained through step (f).

A real time synchronization method for a hybrid network according to an embodiment of the present invention is obtained by the following equation.

여기서, T1은 상기 코디네이터의 방송 시각, T2는 해당 노드의 방송 수신 시각, T3은 해당 노드의 방송 시각, T4는 상기 코디네이터의 방송 수신 시각이다.(Mathematical formula)

Here, T1 is a broadcast time of the coordinator, T2 is a broadcast reception time of a corresponding node, T3 is a broadcast time of a corresponding node, and T4 is a broadcast reception time of the coordinator.

In the real-time synchronization method for a hybrid network according to an embodiment of the present invention, a weight W is obtained through the following equation.

(Mathematical formula)

Where BP is any backoff value

In the real-time synchronization method for a hybrid network according to an embodiment of the present invention, the average channel transmission delay value is obtained through the following equation.

(expression)

Here, d is an average channel transmission delay value.

In the real-time synchronization method for a hybrid network according to an embodiment of the present invention, the number of frequency errors is obtained by the following equation.

(Mathematical formula)

Where α is the number of frequency errors.

As described above, the real-time synchronization method for the hybrid network of the present invention has an effect of improving the efficiency of synchronization data transmission and control between heterogeneous networks by providing a real-time synchronization method considering the wired and wireless mixed situation.

Hereinafter, a real time synchronization method for a hybrid network according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

The present invention designs a gateway for controlling a wired and wireless network having different transmission technologies in between. We present a model that guarantees real-time synchronization inside the gateway, which is an intermediate control layer, for messages passing through both networks. The protocols used for wired and wireless are CAN (ISO11898) and LR-WPAN (802.15.4).

For the network using two different media, we analyze the problems of the total reference time and the time synchronization of each node. The present invention relates to an algorithm that solves a reduction in channel usage efficiency caused by synchronization problems in a shared channel and an error in sensing time caused by different clocks of each node.

The present invention simulates the existing synchronization scheme and the proposed algorithm used for each of the wired and wireless, and compares and analyzes the performance.

The wired / wireless protocol used in the present invention has a primary hierarchical structure consisting of a master and a slave, and a gateway, which is an intermediate layer, becomes a master of both networks.

It has a non acknowledgment communication method for application in industrial environment. The delay factor is applied to the transmission delay and local clock frequency delay of the node and not to the external delay factor.

The delay due to processor processing is relatively small and is not included.

Most industrial distributed networks have been studied to solve time synchronization. There are two basic approaches: calibrating the clock through internal feedback and synchronizing through the time exchange of nodes. In the former case, the reference clock must be precise as each node synchronizes its own time in absolute time. As a result, most network synchronizations take the latter way of distributing time. Vibration elements between nodes have performance differences, but Coordinated Time (UTC) is very accurate. Representative methods of delivering such a reference clock to each node are NTP and RBS.

Network Time Protocol (NTP) is a synchronization method that performs well on multi-hop networks. It has become a reference model for many distributed system algorithms and is used as a time synchronization standard on the Internet. NTP guarantees synchronization of the computer clock time to less than 1ms and uses Coordinated Time (UTC) as the reference clock. Classify each node by layer and distribute time between server and client. It uses independent frames for synchronization and keeps various time information inside the frames. As shown in FIG. 1, the basic principle may be simply expressed by Equation 1 using time stamps T1 to T4 using time stamps T1 to T4.

It uses transmission delay (D _p ) between channels in a structure where master becomes time server and slave becomes client. It compares the time except transmission delay and synchronizes with the time of server. This approach is used for most synchronization algorithms.

NTP continuously maintains visual information for each layer to synchronize between layers and shows good performance in multiple layers.

Communication used in a wireless network basically operates through a broadcast. RBS is a technique that proposes a time message exchange method using the characteristics of such a wireless medium. Since transmission and reception by broadcasting have a meaning of sharing one channel, several nodes may refer to the information.

The broadcast reference method of the RBS reduces energy consumption and facilitates synchronization with a node that is a reference in the area, and thus is suitable as a synchronization method of a sensor network. 2 is a diagram illustrating a simple visual information distribution method of the RBS.

This is a method in which other receivers refer to time information transmitted by a sender. Frequent transmission of visual information in one area requires reference of several nodes, which puts a heavy load on the processing process.

To solve this problem, each node selects a node to refer to through the reception sensitivity of radio waves.

This approach also applies to bus-like networks that share channels over the wire. Ethernet and industrial networks use a reference scheme to construct a bus-type network and detect frame collisions on channels.

The environment proposed in the present invention is a wired / wireless sharing network. This is a network that uses a medium of different characteristics, wired and wireless, unlike a single protocol network.

Wirelines generally use CSMA / CD as a way to share one channel. CSMA / CD secures media access by checking channel preemption and detecting transmission collisions after transmission. On the other hand, since wireless communicates using a specific space as a shared medium, it is difficult to check a transmission collision like a wire. In general, CSMA / CA is used to apply a collision avoidance method through arbitrary time. These different media access characteristics affect the synchronization of each node. 3 is a network diagram illustrating a basic environment according to an embodiment of the present invention.

As shown in FIG. 3, network A is configured by wire, and network B is configured by wireless. The gateway, which acts as the middle layer, acts as a master of networks A and B, and acquires necessary sensing data or transmits data for control.

Certain messages originating from network A can affect nodes in network B and vice versa. However, due to the nature of medium access, the transmission from network A to B is indeterminate due to radio avoidance of collision by arbitrary time.

In the case of wired network A, the time for accessing the media is relatively shorter and the transmission speed is higher than that of the wireless network B. Therefore, message and control signal propagation from network A to network B is limited by the network condition of network B. As mentioned in the experimental section, each node synchronization and channel access delay in the wired network is small compared to the wireless network. Therefore, the present invention has focused on an algorithm for correcting channel access delay and visual error of each node in a wireless network.

Application of wireless communication technology to an existing wired network in an industrial communication network reduces costs due to wiring and allows a topology to be freely configured compared to a wired network.

However, the wireless network has a smaller transmission bandwidth than the wired network and the transmission at the correct time is uncertain due to the delay caused by the retransmission.

In an industrial network environment, communication between devices must occur in a time that is accurate and reliable. Therefore, in wired and wireless gateways, a technique for managing message transmission and time synchronization of the entire network is required for both wired and wireless networks.

Wireless communication techniques and industrial networks are closely related. This is because it has a hierarchical network structure for sharing channels. CAN communication uses CSMA / CD-NDBA, and LR-WPAN communicates with CSMA / CA-CCA. CSMA is a method used for communication sharing one channel as a common element. By sharing one channel, multiple nodes need synchronization because the exact time is required for channel participation. This synchronization requires analysis of delay between nodes.

4 is a diagram illustrating a transmission delay of a simple wired / wireless sharing network.

As shown in FIG. 4, wired and wireless share and use one channel in each network. The wired and wireless gateways serve as masters of both networks as the middle layer. In FIG. 4, the green line represents the transmission delay that occurs when communication occurs between the master and the slave.

There are many factors in the transmission delay, mainly due to transceiver processing, channel noise and signal collisions. The transmission delay between the above two nodes is expressed as shown in FIG.

Delay when the sensor 1 of the wired network transmits data to the gateway can be defined as D ^s _p and each D ^s _Pi . Each drive acting as a driver is represented by D ^D _Pi .

Since the sensor and the driver differ only in the role of the operation, there is no significant difference in the transmission delay, as shown in Equation 2 below.

D ^s _Pi ≒ D ^D _Pi

If both transmission delays are applied equally, the transmission delay on the wire can be expressed as D _Pi in common. Similarly, the transmission delay in the wireless network is expressed as D ^W _Pi because there is no difference in the delay between the sensor and the driver.

Another factor that reduces the efficiency of the channel is the local clock change at each node. Power elements that process the time of the node itself have frequency errors as internal and external factors.

Mainly, as shown in the following Table 1, the vibrating element has errors due to various factors.

As shown in Table 1 above, an error occurs depending on the temperature and pressure time. The unit of error is PPM (part per million), and 50ppm means that 50 errors occur when the device vibrates one million times. Based on this, the time C (t) in the vibrating element used at one node is the frequency k · w (τ) of the vibrating element at the initial value C (t ₀ ) (where k is a constant and w (τ) is the angular frequency. The sum is expressed by Equation 3 below.

Where k is a constant and w (τ) is the angular frequency.
If this is expressed by the time t and the time C by the vibrating element to express the accuracy of the time as shown in Equation 4 below.

If the change in time C is equal to the change in real time t, the accuracy can be expressed as one. ρ is an error value of accuracy caused by the error of the vibration element. Delay time due to frequency error in sensor 1 is expressed as D _D1 . Since frequency error occurs regardless of sensor and driver, the frequency error for each node can be defined as D _Di.

The above two transmission delay factors cause delays in shared channel participation of multiple nodes and problems in sensing / drive time of nodes.

6 is a process in which a master receives a message for each node. Three nodes A, B, and C transmit through the original CSMA process and a collision occurs. After collision, Node A transmits with channel access priority by applying priority which is characteristic of CAN protocol. Since there is a delay in transmission, it is recorded in the time slot of the last delay.

After the message is sent, the two nodes B and C race again, causing a delay due to the frequency error inside the node. In Fig. 6 the same size is given, which is for the application of the maximum size. And the frequency error accumulates over time, but it is not applied because it is a short time in the figure. B, which has a high priority, transmits like A, and because the transmission delay occurs, it is recorded in the delay part. It can be seen that the recorded delay factors reduce the efficiency of the channel based on the overall time width.

7 shows the channel efficiency reduction in the radio. Like a wire, it shows a beacon and a CAP in a network composed of one coordinator and three nodes. After the beacon is broadcast on the channel, competition begins after waiting for T _ack (the sum of IFS and frequency drift), but the competition starts at different times due to the transmission delay of the beacon. At the beginning of the competition section, there is a delay due to the frequency error of each node. When these small delays accumulate and lead to incorrect channel participation, a collision occurs. It can be seen that the delay portion in which the delay elements are written at the bottom of FIG. 7 reduces the efficiency of the channel.

The transmission delay is not accumulated continuously by confirming channel participation from other nodes, but the frequency error is continuously accumulated and changes in the time of the node itself.

Channel delays in wired and wireless networks reduce transmission efficiency due to continuous accumulation, so a synchronization algorithm is applied to eliminate delays. In the wired network, due to the characteristics of the existing synchronization algorithm and the protocol itself, frequent synchronization is performed, so it is difficult to accumulate delay factors. However, due to the nature of the protocol, the time synchronization of nodes is difficult due to the random latency for avoiding collision and the less frequent media access. In addition, in the case of a node with a large time interval for accessing a message, synchronization is only performed by beacons, so accurate time synchronization is difficult.

The problem in the application of network synchronization algorithms in wired and wireless networks is channel delay and frequency delay as described in the delay factor.

For each delay, the CAN network applied to the present invention calculates and synchronizes channel delay using NTP, and frequency delay is solved due to frequent transmission and reception and synchronization in channel sharing.

However, in a wireless network, a collision avoidance method is used for communication, so each node has minimal media access. In the case of a node with a large interval of medium access such as LR-WPAN applied to the present invention, it is difficult to apply a synchronization algorithm. In addition, since LR-WPAN uses node transceiver power only for transmission and reception for low power, aperiodic nodes with large transmission intervals are not synchronized. This synchronization problem is a common problem of sensor networks. The synchronization algorithm for the existing sensor network will be described, and the problems of the application of the wired / wireless sharing network will be discussed.

Various synchronization algorithms applied to sensor networks are divided into internal clock correction method, event sequencing method, and clock reference method, and have advantages and disadvantages.

In the case of TPSN, TSync, Global sync, and FTSP, the internal clock correction method corrects the error value of the internal clock of the sensor node through calculation. And it can be divided into internal synchronous and external synchronous according to the source of reference time.

TPSN has a hierarchical structure by sequential synchronization with the level setting for the nodes of the network. In the case of the wired / wireless shared network, the level setting method of TPSN is inefficient because it has only the layers of both the gateway and the wired and the wireless. TSync, which has improved this, is an algorithm that the master node broadcasts the clock error value by calculating the timestamp with the lower node, so that each node synchronizes even with small processing.

Structurally, it is a good algorithm to apply in wired / wireless sharing network, but in case of LR-WPAN, competition with other nodes is required for error value broadcasting, and nodes that do not use channel for a long time are difficult to synchronize. Global sync is a synchronization algorithm suitable for the cluster structure, and it is difficult to apply it because the delay of the middle layer is high in the wired / wireless shared network. The FTSP drifts clock information to the network through flooding of reliable reference time, so that each node is freely synchronized so that it is difficult to achieve efficiency in a wired / wireless shared network having only two layers of wireless and wired.

Event sequencing is a TS-ad hoc algorithm that infers the order and temporal relationship of events or messages in the network. This is an algorithm that does not modify the time difference with respect to the node itself. It is an algorithm that holds only the relative time difference by the master node where information is concentrated. In the wired / wireless shared network, the algorithm is difficult to apply because each node must be closely sensed / driven.

The clock reference method is an algorithm that is widely applied to the radio. The clock reference method is an algorithm that calculates and maintains a time difference at each node by receiving time information of each node for time synchronization. In this algorithm, the channel sharing of the wireless sensor network is an efficient means of exchanging visual information.

The reference method is a representative algorithm of RBS described above, and there are Tiny / mini-sync, LTS, and Adaptive sync. LTS is an algorithm that focuses on high energy efficiency, even with poor accuracy and precision. In the case of wired / wireless shared gateways to be applied to the factory environment, the network is maintained in a limited space, so time accuracy is required rather than energy efficiency. Tiny / mini-sync is an efficient synchronization algorithm for low processing speed and limited storage space of sensor nodes, which, unlike LTS, differs from node characteristics in the factory environment.

Currently, most synchronization algorithms need to be compensated for in real-time because they are not real-time efficient in wireless networks in industrial environments.

In the case of the LR-WPAN applied to the present invention, since there are two methods of competition access and GTS for channel access, an appropriate algorithm is required. In the present invention, the channel propagation delay through the RBS is solved, and the frequency error of each node is solved through an offset operation.

RBS is an algorithm suitable for applying to the sensor network. As in the case of LR-WPAN, time synchronization is applicable to CAP (competitive access) nodes where channel access is limited. CAP nodes can calculate the channel propagation delay by referring to the broadcast signal of the nearest GTS (sequential access) without accessing the channel. This is also true for nodes that linger for a long time. FIG. 8 is a diagram of an improved RBS reference method and applied to an LR-WPAN.

As shown in FIG. 8, CAP nodes are synchronized through the broadcast information reference scheme of the RBS. The synchronization information between the sequential node and the coordinator is suitable as time synchronization information for CAP nodes using the same channel. The closer the CAP node is to the reference node, the more accurate synchronization information can be obtained. An operation scenario of the above drawings will be described with reference to FIG. 9.

Referring to FIG. 9, (a) the coordinator broadcasts a beacon message to its network area at the beginning of a repetitive superframe, through which each node corrects the clock of the node delayed by the frequency error to be close to the coordinator. do.

(b) The referenced sequential node broadcasts to send a message to its coordinator in its time slot, and the contention access node obtains channel transmission delay 1 through that message.

(c) In the next superframe, the coordinator broadcasts the beacons, resulting in channel transmission delay 2.

(d) Add the bidirectional values of the channel transmission delays to get an average and correct the clock of the node itself.

The above algorithm is a simple time synchronization algorithm using RBS and has the advantage of low computational load at the sensor / driver node.

RBS is an algorithm suitable for sharing a bus-type network and can be applied to COS node or CAP node where channel access is restricted. The RBS can calculate the bidirectional channel propagation delay by referring to the broadcast signals of adjacent polling (sequential generation service) and GTS (sequential access) without accessing the channel. .

10A is a diagram illustrating a process of obtaining a bidirectional propagation delay through a reference method of RBS and synchronizing a periodic node and an aperiodic node using the same.

As shown in FIG. 10A, the non-periodic node is synchronized through the reference method of the broadcast information of the periodic node.

First, the gateway exchanges the transmission with the time of each node, and the gateway receives the transmission message of the periodic node and the neighbor node refers to the received message.

Obtaining a one-way transmission delay D _P1 to the gateway from the node via a reference to obtain the delay D in the direction _P2 transmitted from the gateway to the node, the node synchronizes with the D and D _{P1 P2.}

In the present invention, a process of selecting a periodic node for reference is required in order to use the above process.

In the case of CAN, the delay in the channel is nearly the same at all nodes, so you can simply refer to nodes at close range. Therefore, the neighbor node can be predefined in the network initialization state.

In the case of LR-WPAN, although the channel is shared, transmission errors and transmission / reception sensitivity decrease due to noise in the wireless. Therefore, a reference node with a large reception sensitivity is found and applied to synchronization at regular intervals.

In the referenced GTS, up to seven nodes are repeated every 15.36 ms.

In message transmission between networks using a gateway, the process of FIG. 8 should be connected to two networks and applied. Therefore, the message delivery process is divided into two according to directions, and the delay due to the message conversion process in the gateway should be considered.

10B shows a message transfer process from LR-WPAN to CAN.

Referring to FIG. 10B, the total transmission delay from the LR-WPAN to the CAN node may be represented by Equation 5 below.

Each delay is obtained by the difference of timestamps t1, t2 ...., t10, where D _{pi represents a} transmission delay (where i is a sequence of delay intervals) and D _C represents a conversion delay. D _C is a process of converting data related to a message into a frame in CAN inside the gateway, and most of them have a fixed time.

The sum of the delays result in time synchronization of the message, and the bidirectional transmission delay between the gateway and the CAN node can be obtained from Equation 6 below.

The above equation is a formula for obtaining a bidirectional delay between a COS connection node and a gateway.

In CAN, the channel is shared over the wire and represented by D _P2 because the delay of each node for a message sent at time t3 is nearly the same within a few μs.

The COS connection does not have a constant channel access interval, so refer to the polling connection node when there is no transmission for a long time.

The transmission delay of the message transmitted at time t8 in FIG. 10B is synchronized by the COS connection node with the delay D _P4 .

10c shows a message transfer process from CAN to LR-WPAN.

Referring to FIG. 10C, the LR-WPAN protocol allows a node to have a procedure of receiving a message in order to transmit a message in a coordinator. Therefore, the message of the CAN arriving at the gateway at time t2 in FIG. 10c is in the transmission buffer until the transmission of the next beacon. In the beacon transmission, the size and destination address of the message are recorded in the transmission address area and transmitted to the corresponding node.

If the node receiving the beacon is a CFP node, a transmission request is made in its GTS slot. In the case of a CAP node, the node participates in a competition section and preempts a channel and requests transmission. Total Delay D _W Inside Gateway After receiving the transmit request at time t7, the transceiver's delay D _T Then transfer at time t8.

As shown in FIG. 10C, in the present invention, the CAP node transmits a request to the destination node.

The delay for time synchronization of messages is obtained from Equation 7 below.

Also, the bidirectional transmission delay between the gateway and the LR-WPAN node can be obtained from Equation 8 below.

In the case of FIG. 10C, Equation 9 is applied to perform a synchronization process between the message reception request time t6 and the reception time t9.

Equation 10 is applied when there is no message reception, and is also applied when the beacon transmission delay D _P2 is synchronized with the node transmission delay.

The CFP node transmits and receives data at a minimum of 15,36ms and can be continuously synchronized. The CAP node can be synchronized through an RBS method referring to an adjacent CFP node.

Equation 11 regarding the frequency error of the node is expressed by Equation 12 below by dividing the accuracy into real time and time by a clock.

If the clock change with respect to the change in the real time t is the same, the accuracy can be expressed as 1 since the synchronization is made with respect to the real time. The error value of accuracy caused by the error of the vibration element is defined as ρ. The error value ρ can be caused by various factors. The value ρ is changed to about 20 ~ 100ppm according to the change of temperature, pressure and time for the vibrating element.

The range of the clock and the actual time can be obtained from Equation 13 below, as shown in Equation 14 below.

The time range when there is only one node is shown in Equation 15 below.

When there are two nodes 1 and 2, the time range of node 2 with respect to node 1 may be expressed as in Equation 16 below.

This can be applied to the message exchange between two nodes to find the transmission delay value d of two nodes over the actual time.

11 shows a process in which data and acknowledgment messages are exchanged between two nodes. The time changes from t1 to t4, which records the time (C (t1) to C (t4)) that changes with the clock at each node. When the transmission delay value d is obtained, Equation 17 below is used.

여기서 ρ_S는 송신단의 진동소자에 의한 오차, ρ_R은 수신단의 진동소자에 의한 오차이다.
D_S는 측정된 센더(sender)기준의 전송 지연값으로 계속해서 변화하는 전송 시간값에 대해 연산의 부하를 초래한다. 또한 오랜 시간동안 잠복기를 가지는 노드의 경우 이전에 측정된 전송 지연을 기준으로 연산을 할 수는 없다. 그러므로 두 노드간의 전송지연은 구간 t1~t2, t3~t4를 더하여 평균으로 만든 값으로 한다. 평균 전송지연을 반영한 식 18은 다음과 같다.

Where ρ _S is the error due to the oscillation element at the transmitter and ρ _R is the error due to the oscillation element at the receiver.
D _S is a measured sender reference transmission delay value, which incurs computational loads on continuously changing transmission time values. In addition, a node with a long latency may not be able to compute based on the previously measured transmission delay. Therefore, the transmission delay between two nodes is the average value of the intervals t1 to t2 and t3 to t4. Equation 18 reflecting the average transmission delay is

delete

As in the equation above, the error of the sender's internal clock still remains. This is an error caused by the frequency inside the node, and accumulates as the number of transmissions increases.

The cumulative degree of synchronization error depends on the protocol of the network and the application of the synchronization algorithm. In CAN, it is easier to apply synchronization and accumulates more errors than LR-WPAN due to frequent transmission and multiple reference nodes. Therefore, in synchronized CAN, the synchronization error has a small effect on the channel efficiency.

In LR-WPAN, the minimum message generation period is a minimum length of 15 to 36ms of a super-frame, and this is applied to a maximum of seven nodes that participate in CFP and are assigned a GTS.

Nodes participating in the CAP are not guaranteed the accuracy of the transmission. Due to the longer channel access time than the CAN, the accumulation of errors is large even after the synchronization algorithm is applied, which reduces the efficiency of the CAP and reduces the throughput of the entire network.

The following calculation process finds the efficiency reduction of the channel due to the frequency error and calculates the error rate.

In LR-WPAN, one transmission is performed in one slot, and the frequency and error rate of channels and nodes are assumed as shown in Table 2 below.

The local clock frequency for one superframe is shown in Equation 19 below.

To calculate the sum of the total frequency errors, we need to add all the maximum error values that occur once in each slot, as shown in Equation 20 below.

The maximum frequency error value generated when each node transmits in one slot is expressed by Equation 21 below.

Since the frequency error is ± 40 ppm, the maximum is 2 * 40 ppm. If the sum of each maximum frequency error value is obtained, the total max drift can be obtained as shown in Equation 22 below. Where I is the array of slots and i is the slot number.

The maximum frequency error rate of the channel is expressed as the ratio of the frequency value of the error to the frequency during the superframe as shown in Equation 23 below. Where ∑I is the sum of the slot arrays.

Here, i may be referred to as the frequency of channel access, i is 0.00068% at 16, and 0.16448% at 256.

Although very small, in LR-WPAN, the superframe repeats 65 times per second, reducing the efficiency of the channel due to continuous accumulation. In addition, various operations such as IFS, ACK, Collision, etc. occur not only in data transmission within a channel. Sensing and drive driving are also delayed due to frequency error, so an algorithm is needed to improve this.

In order to control the time of each node, the present invention proposes an offset algorithm.

Referring to FIG. 12, (e) an average channel delay is obtained using a basic timestamp synchronization algorithm for solving channel delay. (f) Find the frequency error with the other node through the average channel delay, and (g) Keep the average value of the error. This can solve the frequency error of nodes that do not use channels in the long run. FIG. 13 illustrates a frequency error in a receiver, which is a slave node, based on a sensor, which is a master node.

The frequency of the slave is represented at a constant rate with respect to the master, and thus there is a variable α. Α at a specific time is calculated as in Equation 24 and Equation 25 below.

Where α is the number of times the frequency error has occurred and ρ is the frequency error ratio over time C (t2).

It is inefficient to calculate α using the error ratio because the computational load of the node is large. In the present invention, the average channel delay E [D _Pi ] is used to obtain α as shown below.

Since the average channel delay E [D _Pi ] is not affected by the frequency error as in the timestamp synchronization algorithm, we can simply find α by taking the average of the delays. However, in the context of ever-changing channels, E [D _Pi ] changes, so an appropriate average value is needed. Simply applying Equation 27 below to find the average channel delay with the new delay value does not apply deviations from the previous channel delay.

To apply all of the previous channel delay deviation values contained in E [D _Pi ], multiply the total number of delays divided by the average to find the average. Equation 28 below calculates E [D _Pi ] for all delay values.

The new channel transmission delay D _pn and E [D _Pi ] can be added and divided by N + 1 to apply the new E [D _Pi ]. However, calculating E [D _Pi ] for all N slows the adaptation to the delay of the changing channel. Therefore, an appropriate weight value W of the coverage for the entire N is needed. Equation 29 to obtain the average using W is as follows.

Since the weight W must reflect the delay of the changing channel, the time delayed until the node accesses the channel and the message transmission interval can be used. In the case of a node with a high frequency of channel access, synchronization occurs frequently because the transmission interval is small, so only the delay time for the medium access is considered. The nodes in the contention period go back off through random values in order to avoid collisions, which shows the degree of delay to channel access. Equation 30 below calculates an arbitrary backoff value BP. The contention node can know the channel access delay through the total BP value until the channel access success.

Here, BE (Backoff Exponent) is a variable used to determine the random delay value (Random delay value). The BP is calculated by the value of the variable aUnitBackoffPeriod (the number of symbols it takes to go through the backoff process once), which corresponds to 20 symbols, and the LR-WPAN has a communication speed of 62.5 Ksymbols / s. The fundamental frequency of the node is 32MHz, recommended for the standard, with ± 40ppm error. The number of symbols for the 1Hz error is obtained and applied to the BP to apply the 1Hz error as the weight 1.

A maximum error of 2560Hz can be seen for 32MHz, and 1 symbol corresponds to 512Hz, resulting in an error of 1 Hz per 24.414 symbols. Applying this, the weight W is obtained by Equation 31 below.

Nodes with large transmission intervals and aperiodic nodes do not have medium access for a long time, so the synchronization interval increases. Therefore, it is necessary to apply the weight W proportional to the long synchronization interval. In the above equation, apply the value of BP * SYMBOLS by changing the number of symbols in the synchronization interval.

Α obtained by the weight is applied as the maximum value of the error when it is out of the range of the frequency error. Equation 32 below is the range of α recommended in the standard.

Wired and wireless do not guarantee real-time transmission of messages between each other due to different channel access methods. CAN, which is used in industrial communication networks, satisfies the high level of real-time due to the short channel access time and the deterministic transmission. In the case of LR-WPAN, a node competes with other nodes in order to receive a message, and uses a scheme of avoiding for a certain time for channel access in the race.

In this method, when the competition fails, reception is transferred to the next superframe, and a waiting time occurs during the transmission period of the CFP and the broadcast time of the beacon. This time depends on network resources, but it takes about 7.7㎳, which is less real-time compared to CAN. In addition, due to the uncertainty of transmission, transmission failure occurs more frequently than CAN even when retransmission method is used for transmission errors. In order to improve such wireless limitations, there has been a study of transmitting a real-time message by reducing the channel access time.

The present invention ranks the nodes of each network in wired and wireless gateways for real-time message transmission between wired and wireless networks. In CAN, the channel access time is reduced by increasing the priority for transmission and reducing the latency of collision detection. In LR-WPAN, the waiting time for random time for collision avoidance is reduced regularly so that the channel access is faster than other nodes.

Message transmission and reception in the CAN network depends on the node ID and message type, and the channel access priority is different. Therefore, if a message is transmitted from the LR-WPAN to the CAN, the gateway may determine the transmission priority in the CAN at a predefined class at the node of the LR-WPAN.

In the transmission of messages from CAN to LR-WPAN, due to the nature of wireless, the gateway (coordinator) needs to inform that there is a message to send first. The node must have channel access for receiving messages, which requires competition with other nodes.

Therefore, the gateway attaches a class to the destination address of the beacon in order to send messages from nodes of a predefined class in CAN.

This applies in the standard by putting a short address and its class on a long receiving address. Nodes that receive the reception notification are limited to the maximum value of the BP differentially according to the class, thereby reducing the channel access time. If one node preempts the channel during transmission, then the node that has been restricted by the BP then increases the chance of preempting the channel.

However, due to the nature of radio, it is appropriate to have only one node in one superframe because it shares a channel and uses a scheme for avoiding it.

The following is a scenario of differential BP application.

(1) Find the minimum BP and CW values appropriate for the operation of all nodes before network initialization.

(2) When an emergency message is generated from Network B to Network A, the Gateway receives the message from Network B and assigns the highest priority message ID to Network A.

(3) When an emergency message is generated from network A to network B, the gateway notifies that there is an emergency delivery message in the extended address area of the beacon to network B with the target node ID.

(4) The target node accesses the channel by applying the predetermined minimum BP and CW for fast channel access.

(5) After the target node preempts the channel, it sends a message request to the gateway.

(6) The gateway sends an emergency message.

Although the preferred embodiment of the present invention has been described above, it is clear that the present invention may use various changes, modifications, and equivalents, and that the above embodiments may be appropriately modified to apply the same.

Accordingly, the above description does not limit the scope of the invention as defined by the limitations of the following claims.

1 is a view showing the basic principle of NTP according to an embodiment of the present invention.

2 is a diagram illustrating a simple visual information distribution method of an RBS according to an embodiment of the present invention.

3 is a diagram illustrating a network configuration serving as a basic environment according to an embodiment of the present invention.

4 is a diagram illustrating a transmission delay of a simple wired / wireless sharing network.

5 is a diagram illustrating a transmission delay between a master and a slave according to an embodiment of the present invention.

6 is a diagram illustrating a process in which a master receives a message for each node according to an embodiment of the present invention.

7 is a diagram illustrating channel efficiency reduction in the radio according to an embodiment of the present invention.

8 illustrates an RBS application scheme in 802.15.4 according to an embodiment of the present invention.

9 is a flowchart of a real time synchronization method for a hybrid network according to a preferred embodiment of the present invention.

10A illustrates bidirectional propagation delay synchronization using RBS in accordance with an embodiment of the present invention.

10B is a diagram illustrating a message transfer process from LR-WPAN to CAN according to an embodiment of the present invention.

10C is a diagram illustrating a message transfer process from CAN to LR-WPAN according to an embodiment of the present invention.

11 is a diagram illustrating a process of exchanging data and acknowledgment messages between two nodes according to an embodiment of the present invention.

12 is a flowchart of a real-time synchronization method for a hybrid network according to another preferred embodiment of the present invention.

13 is a view showing a frequency error in the LR-WPAN according to an embodiment of the present invention.

Claims (6)

(a) The coordinator broadcasts a beacon message to the coordinator's network area at the beginning of the superframe, and each node in the network area uses the beacon message to clock the corresponding node delayed by the frequency error of the vibration device. Correcting the value to a value corresponding to the clock of the coordinator; (b) a sequential node in the network area broadcasts to send a message in a time slot to a coordinator, and the contention access node in the network area obtains a channel transmission delay 1 using the message; (c) the coordinator broadcasting a beacon message in a next sequence of the superframe in which step (a) is performed, and the contention access node obtaining a channel transmission delay 2 using the broadcast beacon message; And (d) wherein the contention access node calculates an average value by adding the channel transmission delay 1 and the channel transmission delay 2 and corrects the clock of the node using the calculated average value. The method of claim 1, (e) obtaining an average channel delay between a transmitting node and a receiving node in the network by a time stamp synchronization algorithm; (f) obtaining a frequency error of each vibrating element between the transmitting node and the receiving node using the average channel delay; And (g) maintaining a mean value of the frequency errors obtained through step (f). The method of claim 1, wherein the channel transmission delay 1 and the channel transmission delay 2 are obtained by the following equation using time stamps T1 to T4. (Mathematical formula)

Here, T1 is a broadcast time of the coordinator, T2 is a broadcast reception time of a corresponding node, T3 is a broadcast time of a corresponding node, and T4 is a broadcast reception time of the coordinator. 3. The method of claim 2, wherein the frequency error [alpha] is an average channel transmission delay E [Dp], a newly generated channel transmission delay D _PN for previous delay values occurring during one superframe, and a channel delay. The real-time synchronization method for a hybrid network, characterized in that it is obtained based on a weight (W) reflecting the change of. The method of claim 2, wherein the average channel delay is obtained by the following equation. Equation

Where d is the average channel delay, ρ _S is the error due to the oscillation element at the transmitting end, ρ _R is the error due to the oscillating element at the receiving end, and C (t1) to C (t4) is the transmission node for the actual time t1 to t4. And time according to the internal clock of the receiving node 5. The method of claim 4, wherein the weight (W) is obtained by the following equation. (Mathematical formula)

Where BP is an arbitrary backoff value, N is the number of previous channel delays occurring during one superframe, and symbols is the number of symbols in the synchronization interval.

Images