KR20030006833A - Bandwidth allocation method for real time data transmission in profibus - Google Patents

Abstract

PURPOSE: A bandwidth assigning method for transmitting data in real time in a profibus is provided to divide the bandwidth used in a field bus into a periodical interval and a non-periodical interval, transmit the data as soon as the opportunity of data transmission is provided, and perform a time synchronization between nodes to be suitable for a link time. CONSTITUTION: A sampling period of periodical real-time data is calculated using an algorithm. The first periodical real-time data sampling instant is calculated using an algorithm. A data packet length of non real-time data is calculated using an algorithm. The transmission of periodical data generated in each node is controlled using the calculated parameters.

Description

Bandwidth Allocation Method for Real-Time Data Transmission on Profibus {BANDWIDTH ALLOCATION METHOD FOR REAL TIME DATA TRANSMISSION IN PROFIBUS}

The present invention relates to a method for allocating bandwidth for real-time data transmission. In particular, the present invention relates to a requirement for a transmission delay time of real-time data that occurs periodically and sporadically in a Profibus, a type of open fieldbus used as an industrial communication network. The present invention relates to a bandwidth allocation method for real-time data transmission that satisfies and fully utilizes network bandwidth.

In the past, PROFIBUS consists of three layers: the physical layer, the data link layer, and the application layer. The dual physical layer consists of a bus topology, in which several nodes are connected to a transmission medium of one bus type. In the form of communication network, when two or more nodes transmit data at the same time, a data collision occurs in the transmission medium and the data is not transmitted successfully.

In order to prevent such data collision, the data link layer of PROFIBUS supports data transmission by token passing.

In the token-passing method, a specific frame (special data) called a token for granting media access is circulated on a transmission path in a given order on nodes connected to a bus, and a terminal that obtains the right to transmit data waiting in a transmission queue. Has After completing the transmission of the data, the nodes forward the token frame to the next node and release the token. Through this process, the nodes connected to the medium perform the data transmission cyclically. Data generated from various field devices connected to PROFIBUS are classified into three types: sporadic real-time data, periodic data and non-real-time data. Sporadic real-time data is used to transmit various events and warning signals. It is short in length and sent at the highest priority. In addition, the periodic data is mainly used to periodically sample the sensor value or to transmit the periodic control command to the actuator in the feedback control system, and the real-time data must be completed within the time limit of the sampling period. And, non-real-time data is not very limited by the transmission delay time, including a program or a data file.

In the fieldbus, the above three types of data share the same network medium. Therefore, if the traffic of the network is not properly managed, the transmission delay of sporadic real-time data or periodic data may not meet the time limit. There was a problem.

In addition, the delay time of sporadic and periodic real-time data in PROFIBUS must not exceed the maximum allowable delay time, which is defined as the queuing delay time (until the data arriving at the transmission queue reaches the end of the transmission queue). The time, which is the time it takes for all data arriving at the transmission queue to complete the transmission before the corresponding data), the waiting time for the arrival of the token (the time it takes for the data arriving at the end of the transmission queue to receive the token), It is composed of the sum of the time (the time taken by the node receiving the token to transmit the data packet) and the data packet propagation time (the time the packet propagates through the medium to the receiving end).

However, in the case of sporadic data, the number of data waiting in the transmission queue is changed randomly, resulting in irregular queuing delay time, and also because the number of data transmitted while the token is circulating all nodes once is irregular. Since the token arrival waiting time is irregular, there is a problem in that real-time data that needs to be completed within the maximum allowable delay time cannot guarantee the requirements of real-time data transmission as the data delay time changes irregularly. In order to prevent the data transmission delay time from exceeding the maximum allowable delay time, the communication network system should be designed in consideration of the worst case, but in this case, the utilization of network bandwidth is greatly reduced.

Accordingly, the present invention has been made to solve the above problems of the prior art, and the bandwidth and bandwidth available in the fieldbus to satisfy the requirements for the transmission delay time of the real-time data and to fully utilize the bandwidth of the network periodically and The object of the present invention is to divide data into aperiodic intervals so that data is transmitted as soon as an opportunity for data transmission is given, and time synchronization between nodes according to link time is provided.

1 is an exemplary view showing fieldbus bandwidth allocation in accordance with the present invention.

2 is a synchronization timing diagram of a time synchronization algorithm applied to the present invention.

Figure 3 is an exemplary view briefly showing the structure of a network system to which the present invention is applied.

4 is an exemplary view illustrating a generation pattern of periodic data according to the present invention.

5 is a table showing a performance comparison of the data delay time between the prior art and the present invention.

The bandwidth allocation method for real-time data transmission of the present invention for achieving the above object, the sampling period of the periodic real-time data To

= min [ , at i = 1 ], , , At = 2

, if , then,

Else

(here, Is the maximum allowed latency of periodic data on node i, Is the number of nodes generating periodic data, Is the number of nodes generating sporadic real-time data, Is the transmission time of periodic real-time data, Is a transmission time of sporadic real-time data, R is a token circulation time) algorithm; First periodic real-time data sampling moment at each node To

For (i = 2, l = 1: ; , )

(here, silver In section th When the slot starts, silver Including periodic data generated at the first node A second step of calculating the number of periodic data generated at the instant of " Data packet length of non real time data To

= ,

= , otherwise

If

or

Then

Overload of sporadic real-time or non-real-time data traffic ⇒ or Decrease

(here, Is the average arrival frequency of non-real-time data at node i, Is the average arrival frequency of non-real-time data packets at node i, Is the average arrival frequency of sporadic real-time data packets at node i, Is the number of nodes generating non-real-time data, Is the token rotation time, Is Obtaining a mean value of the number of periodic data generated during the third algorithm); And a fourth step of controlling transmission of periodic data generated at each node using the parameters obtained in the first, second, and third steps.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is an exemplary view showing fieldbus bandwidth allocation according to the present invention. As shown therein, periodic data is generated in a periodic section. Nodes Two windows are dynamically shared, and the bandwidth allocation technique in the periodic interval is determined by the number of periodic data generated in the periodic interval. Period of generation of periodic data at each node so as not to exceed Determine.

Vector element with maximum allowable delay time in two periodic data generation nodes A vector with elements and data generation cycles Sorts in ascending order (that is, , ).

From here, silver Minimum generation cycle among Is the maximum generation cycle. Is the maximum allowable delay time Of the minimum The creation cycle of the remaining nodes Is the maximum allowable delay time It is determined as the maximum value satisfying the following relationship within the range not exceeding, and in this case, all the production cycles have multiples of each other.

Rem ,

Therefore, the bandwidth allocation of the network Repeat with the cycle of.

The bandwidth of Profibus is Within the cycle Divided into cycles of During the periodic interval within To generate less than data The periodic data generated at the two nodes must be set to satisfy the following conditions.

here, Is The average value of the number of periodic data that is generated during the first period. Is set as follows.

: , at i = 1 , at l = 1

In the above, silver In section The start of the -th T ₁ slot, silver Including periodic data generated at the -th node The number of periodic data generated at the moment.

Periodic data Since the transmission must be completed within the interval of, the stabilization conditions of the periodic data traffic are as follows.

here, Is the maximum bandwidth that can be allocated to the sporadic real-time data transmitted during the periodic interval.

Of FIG. 1 In the bandwidth section, the remaining bandwidth except for the periodic section is allocated for the transmission of the non-real-time data. Since the non-real-time data is not greatly affected by the transmission delay time, when the periodic data is generated at any node, the non-real-time section is immediately stopped. The bandwidth of the network is switched back to periodic intervals.

During periodic intervals Of periodic data transmission queues Since only up to data is scheduled in advance, such periodic data is completed while the token is circulating through all nodes in the network. Non-real-time data, which has a relatively long data length, is divided into multiple packets at the transmitting end and transmitted, and the receiving end reassembles them.

In addition, in the case of the non-real-time data, the packet transmission time is limited as necessary in order to satisfy the transmission delay time requirements of the periodic and sporadic real-time data.

From above Is the bandwidth that can be allocated to the maximum amount of sporadic real-time data transmitted during the periodic interval.

If the maximum allowed delay time of sporadic real-time data end If smaller, the transmission time of the non-real-time data packet is limited as follows.

In order for the network system to stabilize, the traffic load of sporadic and non-real-time data must not exceed the capacity of the network, and the stabilization conditions of the real-time and non-real-time data traffic are as follows.

Sporadic real-time data In the case of several times during the time of occurrence, it rarely occurs in the actual system, but if a special case must be taken into consideration, exhaustion of transmitting all the data waiting in the transmission queue at the moment of arrival of the token instead of a single service method. Adopt service method.

However, in order to apply the bandwidth allocation technique in the PROFIBUS, two conditions, cyclic and non-real-time bandwidth separation and time synchronization between nodes must be made.

First of all, periodic and non-real-time bandwidth classifications use only bits 0 to 6 in source address (SA) and destination address (DA) of token frame, and do not use bit 7 To make a distinction. That is, when the node receiving the token needs to transmit periodic data, the 7th bit of SA and DA of the token frame is set to '1' and transmitted, and the other nodes have the 7th bit of SA and DA set to '1'. When receiving the designated token frame, it recognizes that the periodic section has started and transmits its own periodic data. Thereafter, when the node which sets bit 7 of SA and DA to '1' receives the token frame again, bit 7 of bit SA and DA is set to '0', and the aperiodic period starts again from this time.

In addition, each node has a StartPeriod state variable for recognizing that it has set a periodic section and an IsPeriod state variable for recognizing that the current section is a periodic section. That is, if the node that receives the token has periodic data and the IsPeriod variable is set to '0', the node sets the StartPeriod and IsPeriod variables to '1' and transmits only the periodic data. Each node checks whether the IsPeriod variable is '1', which indicates that the current section is a periodic section before transmitting the token frame, and if it is '1', sets the seventh bit of SA, DA to '1', and sets the IsPeriod variable to '0'. ', Set the 7th bit of SA, DA to' 0 'and transmit token frame. Then, in order to finish the periodic section and start the non-periodic section or maintain the periodic section, each node receives the token frame and checks whether the seventh bit of SA and DA is '1'. Checks whether the StartPeriod variable is' 1 'to indicate that it is a periodic interval that started, and if the StartPeriod variable is' 1', sets the IsPeriod and StartPeriod variables to '0' to indicate that it is a periodic interval and ends the periodic interval, but the StartPeriod variable is' If it is '0', the IsPeriod variable is set to '1' to maintain the periodic section.

And, in order to synchronize time between nodes, which is the second condition, the node time of each station must be synchronized first. There is a time-master node that provides a reference time in the network for time synchronization between the nodes, and this time-master node delivers time information using a predetermined frame. That is, as shown in FIG. 2, the time-master node periodically sends SD1 and SD2 frames for time synchronization successively, records the time TM1 immediately after sending the last byte of the SD1 frame, and then records them through the SD2 frame. Send to time-slave nodes. The time-slave nodes receiving the SD1 frame record the time TS1 as soon as they receive the last byte, record the time TS2 when the SD2 frame, which is the second synchronization message, is received, and then present time according to the following relationship. Determine the time.

time = TM1 + (TS2-TS1) + p-a + b

[Where p is propagation delay, a is program sequence delay and b is receive interrupt processing time for SD1 frame]

The operation of the present invention will be described in detail through the preferred embodiments as follows.

FIG. 3 is an exemplary view briefly showing the structure of a network system to which the present invention is applied, and is configured in the form of a PC interface board including a physical layer of PROFIBUS as shown in FIG. Implemented and mounted by firmware, it consists of 10 boards and a monitor board that receives all frames transmitted through the medium and delivers them to PC through RS-232 communication.

Nodes 1, 3, 5, 7, and 9 generate sporadic real-time data and periodic data, and nodes 2, 4, 6, 8, and 10 generate non-real-time data and periodic data. Assuming that it acts as a time-master node for time synchronization between nodes, the transmission speed of the Fieldbus Data Link (FDL) protocol is set to 93,750 bps, and the periodic data length is 85 bytes (header: 11). Byte, pure data: 74 bytes), and the actual data length is 11 bits per byte, so the transmission time of periodic data with 935 bit data length The processing time is set to 9.97 msec (965 bit time), which adds up to 30 bits of processing time (where processing time is the time it takes to take a message from the transmission queue, create a frame, and send it), and the length of sporadic real-time data is 14 bytes. (Header: 8 bytes, Pure Data: 6 bytes), but the actual data length is 154 bits, so the transmission time of sporadic real-time data Set the processing time to 1.96 msec (184 bit time), the length of the non-real-time data is generally variable but is set to produce 255 bytes, which is the maximum frame length on all nodes, and sporadic real-time and non- arrival frequency of the real time data is set to 0.001 msec ^{-1, -1} 0.002 msec, respectively, and token overhead is required to cycle through the 10 nodes to be 1 msec is the time it takes the sum of the token processing time and the token transmission time The token rotation time is set to 10 msec.

Maximum allowed data latency for sporadic real-time data ( ) Is 1.00 msec, which is the maximum allowable delay time for periodic data generated from 10 nodes ( ) Is set to [100, 160, 200, 240, 400, 600, 800, 1000, 1600, 2000] msec, and then applied to the present invention for such traffic conditions. ) = [100, 100, 200, 200, 400, 400, 800, 800, 1600, 1600], the first periodic real-time data sampling instant at each node ( )silver = [0, 0, 0, 0, 100, 100, 300, 300, 700, 700], data packet length of non-real-time data ( ) = 5.13 msec (41 bytes: 481 bit time with overhead).

The first shown in FIG. 4 according to the result derived above As in the data generation pattern of periodic data generated at ten nodes during a period, The number of periodic data generated within the slot No more than Is repeated in the cycle of.

In addition, if network utilization (U) is defined as the ratio of time spent in data transmission in the network per unit time, the relationship between network utilization in a fieldbus system consisting of sporadic real-time, periodic and non-real-time data traffic is as follows. It is shown as:

In this embodiment, the network utilization of sporadic real-time, periodic and non-real-time data is 1%, 39% and 31%, respectively, and the total network utilization is 71%.

This network utilization only takes into account the time it takes to transfer data, and the inevitable overhead of sending tokens ( ), So that the network load including token delivery exceeds 90% to fully utilize the resources of the network.

The performance of the data delay time by applying the respective parameters calculated above is shown in FIG. 5 by comparing the method according to the embodiment of the present invention with the existing PROFIBUS protocol as it is, but failed in FIG. 5. Frequency refers to the case where the data delay time exceeds the maximum allowable delay time.

In general, in case of periodic data, the transmission delay time may be smaller than the data generation period in order to be successfully transmitted. The transmission queue capacity of the periodic data is limited to 1, and when the transmission delay time is larger than the data generation period, the data waiting in the transmission queue is replaced by the newly generated data, causing a data transmission failure.

In FIG. 5, periodic data is set in advance so as to generate a maximum allowable delay time. In the case of periodic data, the maximum allowable transmission delay time is 167,257,478,383 msec at nodes 1,2,3,4 in the case of Profibus. Since the time exceeds 100,160,200,240 msec, the maximum allowable transmission delay time is not satisfied, and this causes about 14% of data transmission failure.

On the other hand, when applying an embodiment of the present invention as shown in FIG. The number of periodic data generated during the slot By scheduling the data generation period and the initial data generation time to be less than or equal to (s), so that the maximum transmission delay time does not exceed 95 ms in all nodes, so that no transmission failure occurs.

In the case of sporadic real-time data, it can be seen that the maximum transmission delay time does not exceed 10 msec, which is the limit of the allowable delay time, in both the above embodiment and Profibus. This is because sporadic real-time data was processed with high priority in both cases.

In the case of non-real-time data, the effect of the data delay time on the application system in the case of the above embodiment is not important, so instead of increasing the transmission delay time, sporadic real-time data and periodic data transmission requests requiring real-time data transmission are required. Satisfy the matter.

Although the invention has been particularly shown and described with reference to the preferred embodiments, those skilled in the art will appreciate that various modifications are possible without departing from the spirit and scope of the invention. Accordingly, the invention is limited only by the claims.

As described above, the bandwidth allocation method for real-time data transmission according to the present invention does not exceed the maximum allowable value of the transmission delay time of the real-time data occurring periodically and sporadically in the data link layer of PROFIBUS. At the link layer, it satisfies the transmission delay time requirement of periodic and sporadic real-time data while maximizing the bandwidth of PROFIBUS.

Claims (4)

Sampling Period of Periodic Real-Time Data To = min [ , From = 1 ], , , At = 2 , if , then, Else (here, Is the maximum allowed latency of periodic data on node i, Is the number of nodes generating periodic data, Is the number of nodes generating sporadic real-time data, Is the transmission time of periodic real-time data, Is the transmission time of sporadic real-time data, Is a token recursion time) algorithm; First periodic real-time data sampling moment at each node To For ( = 2, = 1: ; , ) (here, silver In section th When the slot starts, silver Including periodic data generated at the first node A second step of calculating the number of periodic data generated at the instant of " Data packet length of non real time data To = , = , otherwise If or Then Overload of sporadic real-time or non-real-time data traffic ⇒ or Decrease (here, Is the average arrival frequency of non-real-time data at node i, Is the average arrival frequency of non-real-time data packets at node i, Is the average arrival frequency of sporadic real-time data packets at node i, Is the number of nodes generating non-real-time data, Is the token rotation time, Is Obtaining a mean value of the number of periodic data generated during the third algorithm); And a fourth step of controlling transmission of periodic data generated at each node by using the parameters obtained in the first, second, and third steps. 2. The method of claim 1, wherein the first step is set to '1' or '0' by using bit 7 not used by the source address and destination address of the token frame. Bandwidth allocation method for real-time data transmission, characterized in that further comprising distinguishing the real-time bandwidth. 2. The method of claim 1, wherein the first step periodically sends a predetermined frame (SD1, SD2) for time synchronization periodically at the time-master, and records the time TM1 immediately after sending the last byte of the SD1 frame. Transmitting it to time-slave nodes via an SD2 frame; The time TS1 is received as soon as the last byte is received from the time-slave node receiving the SD1 frame, the time TS2 is received when the SD2 frame, which is the second synchronization message, is recorded, and then presently determined by a relational expression. Determining a time of the bandwidth allocation method for real-time data transmission. 4. The method of claim 3, wherein the current time is set to a value calculated according to the following relational expression. time = TM1 + (TS2-TS1) + p-a + b Where p is a propagation delay, a is a program sequence delay, and b is a receive interrupt processing time for SD1 frame.