EP3152967A1

EP3152967A1 - Method for deterministic wireless transfer of time-controlled real-time messages in a distributed real-time system

Info

Publication number: EP3152967A1
Application number: EP15737953.8A
Authority: EP
Inventors: Hermann Kopetz
Original assignee: FTS Computertechnik GmbH
Current assignee: Tttech Computertechnik AG
Priority date: 2014-06-06
Filing date: 2015-06-08
Publication date: 2017-04-12
Also published as: US10285145B2; WO2015184483A1; US20170156120A1

Abstract

The invention relates to a method for the deterministic wireless transfer of time-controlled real-time messages in a distributed real-time system comprising a plurality of node computers and one or more base stations arranged in an arena, wherein all node computers and base stations have a global time, wherein one or more real-time message sequences are periodically transferred in the arena, wherein a real-time message sequence consists of a time-controlled header message with variable length and a sequence of one or more time-controlled real-time messages with a priori known length, and wherein, at an a priori determined transmission time, the software of a T-node issues to a communication controller thereof the command for the header message to be sent, and wherein the communication controller of the T-node begins to send the header message as soon as no activity is determined in the arena during an IFS, and wherein transmission of the header message aborts at an a priori determined time-controlled abort time of the T-nodes, and wherein in a command interval before the abort time of the header message, the software of the computer node that has to send the first real-time message of the real-time message sequence issues, to the communication controller thereof, the command for the first real-time message of the real-time message sequence to be sent, and wherein, in the command interval before the first real-time message is terminated, the software of the computer node that has to send the following real-time message of the real-time message sequence issues, to the communication controller thereof, the command for the following real-time message to be sent, and wherein the procedure is repeated until all the real-time messages of a real-time message sequence are sent.

Description

METHOD FOR THE DETERMINISTIC WIRELESS TRANSMISSION OF TIME-CONTROLLED REAL-TIME MESSAGES IN A DISTRIBUTED REAL-TIME SYSTEM

The invention relates to a method for the deterministic wireless transmission of timed real-time messages in a distributed real-time system, which real-time system comprises a plurality of node computers and one or more base stations, which are arranged in an arena, wherein all node computers and base stations have a global time.

Furthermore, the invention relates to a T-node for the wireless transmission of timed messages in a distributed real-time system, which real-time system comprises a plurality of node computers and one or more base stations, which are arranged in an arena.

Finally, the invention also relates to a real-time system, in particular distributed real-time system, comprising a plurality of node computers and one or more base stations, which are arranged in an arena for carrying out such a method, wherein preferably all node computers and base stations have a global time.

In control engineering applications where a closed-loop control loop between a controller and a link is closed by means of wireless data exchange, deterministic behavior of the communications system is of particular importance.

The existing wireless data transmission protocols, such as the widespread WiFi protocol [6], use a frequency band that may be used without license by many devices simultaneously. In order to minimize the interference of the devices, according to the WIFI protocol, a device starts sending a message only when a transmission channel is not busy. This results in unpredictable delays in message transport.

It is an object of the invention to provide a solution for how real-time data can be transmitted wirelessly and deterministically in a distributed real-time system. This object is achieved with a method mentioned in the introduction in that one or more real-time message sequences are transmitted periodically in the arena, wherein a real-time message sequence consists of a time-controlled variable-length header message and a sequence of one or more timed real-time messages with a priori known length, and wherein, at an a priori scheduled send time, the software of a T node issues the command to send the header message to its communication controller, and wherein the communication controller of the T node starts sending the header message as soon as no activity is detected in the arena during an IFS, and wherein at a timed termination time determined a priori the T-node stops sending the header message, and wherein in a command interval before the termination time of the header message, the software of the computer node, the first real time In the command interval before termination of the first real-time message, in the command interval before the termination of the first real-time message, the software of the compute node which has to send the following real-time message sequence real-time message instructs its communication controller to command the real-time message sequence to send the following real-time message, and this process is repeated until all the real-time messages of a real-time message sequence have been sent.

The present invention describes a novel method for deterministic wireless transmission of timed messages between a plurality of mobile and / or stationary node computers located in a confined area, the arena.

The present new method solves this problem, in which a special node computer, a T-node (short for Time Node) that specifies the global time, periodically produces a global synchronization event in the arena that can be used by all node computers to generate the local clocks of the node computer to synchronize with the global time. Based on this global time, the hub computers can send the real-time messages according to an a priori created schedule in the form of a periodic real-time message stream. It is advantageous to externally synchronize the T-node with an external time signal, eg the GPS time. In such a T-node, a wireless transmission chipset is preferably arranged and the physical connection between the chipset and the antenna of the T-node can be interrupted by a timed switch. Preferably, a GPS signal receiver for receiving the GPS time is included in the T-node.

None of the searched patents [1-5] and the literature found anticipates the proposed method.

The present invention describes a new method for deterministic wireless transmission of timed messages between a plurality of mobile or stationary node computers located in a limited area and one or more stationary base stations. In order to ensure that the real-time data can be transmitted at a priori known times, a variable header message is sent from a special node computer (T-accounts) which has the function, in the arena, of sending messages at the a priori known time planned start of the first real-time message of a real-time message sequence are not yet completed.

Advantageous embodiments of the method or real-time system according to the invention, which can be implemented alone or in any combination, are indicated below:

> in a real-time message sequence, a real-time message appears multiple times;

> the length of the header message is longer than the longest application message that the communication controller has to transport;

> the transmission time of the complete header message is longer than the time interval between the start command for sending the header message (301) and the termination time (305) at which the sending of the header message is aborted;

> the IFS between every two messages in a real-time message sequence has the smallest value supported by the protocol; > a node computer uses the known time of arrival of the first real-time message of the real-time message sequence to synchronize its local clock;

> the first real-time message sequence real-time message is a beacon message from the network;

> the T-nodes and the base stations are synchronized externally;

> the T-nodes and the base stations are synchronized via a TTEthernet system;

> The wireless data channel is implemented according to the IEEE standard 802.11 (WiFi).

In the wired TTP protocol, it is assumed that all correct node computers adhere to the a priori created schedule. This assumption can not be realistic in an arena where messages are transmitted wirelessly.

In an arena can thus be node computers or node computers are present that are not involved in the real-time data transmission, but comply with the specification of the wireless standard, in particular the WiFi standard.

The nodes involved in the real-time data transmission are cooperative according to the given schedule, the other nodes present in the arena have no schedule and therefore can not be cooperative.

According to one aspect of the invention, a method is disclosed of how the non-cooperative but the wireless protocol, in particular the WiFi protocol, following nodes can be forced to transmit no messages, in particular no disturbing messages, during transmission of the real-time message sequence. This is preferably achieved on the one hand by the transmission of the header message and / or on the other hand by the choice of the IFS.

The following section explains new terms used in the detailed description of the invention that go beyond the standard terms in real-time processing, such as those used in textbook [7]. We call the room containing the distributed node computers, which exchange data with the base stations over a wireless connection, an arena. We refer to a message as a real-time message if it contains data whose validity is limited [7]. A real-time message is timed if the time of sending the real-time message is determined by the progression of the global time. The schedule, which specifies the times of sending the timed real-time messages, thus determines the location of the real-time messages on the time axis.

The logic of many common communication protocols is realized in highly integrated semiconductor devices, which we call chipsets. A widely used wireless data transmission protocol is the WiFi protocol standardized by the IEEE under the name IEEE 892.11. There are a variety of chipsets on the market that implement the WiFi protocol.

The WiFi protocol works on the principle of Carrier Sense Multiple Access Collision Avoidance, abbreviated CSMA / CA. In the CSMA / CA method, a node computer may not transmit a message until the transmission channel is unoccupied during a predetermined time interval. The length of this time interval, the Interframe Space (WS), is used to set the priority of the data transfer. The shorter the WS, the higher the priority. The shortest waiting time (highest priority), the minimum interface space (MWS), is determined by the characteristics of the physical transmission method, the size of the arena and the computing power of the communication controllers. The MWS is given in the protocol specification.

Under a real-time message sequence, a concatenation of timed real-time messages sent by various node computers according to a periodic schedule fixed a priori is understood. The periodic schedule indicates at which recurring times transmission of a timed real-time message begins. According to the invention, a real-time message sequence consists of a variable-length header message and a sequence of real-time time-controlled messages of known constant length, the IFS having the smallest protocol-supported value between two real-time messages. The command interval is the interval before sending a real-time message during which the node computer software must issue the command to send the following real-time message sequence real-time message to the communication controller. The CSMA / CA logic of the communication controller does not initiate transmission of a message until no activity is observed in the arena during the specified IFS after the end of the previous real-time message sequence message.

As noted in Figure 3, after the beginning of the transmission of the upstream real-time message the real-time message sequence starts and ends before the end of the transmission of the upstream real-time message sequence. The command interval for sending the first real-time message sequence real-time message begins at the time that it is ensured that the sending of the header message has safely begun.

The present invention will be explained in detail with reference to the drawings. 1 shows the construction of an arena,

Fig. 2 shows the timing of the data transmission in a control loop, and

Fig. 3 shows the detailed timing of the transmission of a real-time message sequence.

One of the many possible concrete implementations of the new method will now be described with reference to the drawings.

In Fig. 1, an arena is represented by a cuboid 100. In the arena 100 is a node computer 101, which calculates a process model in each control loop. The node computer 101 is supplied by node computers 111, 112 and 113, the sensors, via wireless connections periodically with sensor data, such as a robot. The setpoints of the node computer 100 are periodically transmitted to node computers 121 and 122, the actuators, wirelessly. Typical for this application area of process control is that the data from the sensors and the data to the actuators can be coded in a few bytes. At the bottom of Fig.l a T-node 130 is shown, which synchronizes all Node computer Arena 100 performs. It is advantageous if the T-node is synchronized with the external time standard, eg the GPS time.

Fig. 2 shows the timing of a control loop. In this picture, the cyclic time representation is used to emphasize the periodic nature of a control loop. In Fig.

2, time advances clockwise 200 and traverses the entire circle of 360 degrees during a closed loop. At time 201, the node computers 111, 112 and 113 detect the sensors, e.g. the data of a robot. From the application's point of view, it is important that all data be collected at the same time. This simultaneity of the data acquisition is realized by referring to a global time existing in the node computers 111, 112 and 113. At time 202, after the pre-processing of the sensor data, the transmission of a real-time message sequence consisting of three real-time timed messages from the node computers 111, 112 and 113 is started. At time 203, the transmission of the real-time message sequence is completed. At time 204, after the process model is computed, the transmission of another real-time message sequence consisting of two timed real-time messages to the node computers 121 and 122 is begun. At time 205, the transmission of the real-time message sequence is completed. The procedure described here is repeated periodically in each control loop. Typical for a robotic application is the duration of 1-10 msec for the passage of a control loop

From a reliability point of view, the real-time data from the sensors and to the actuators of a control loop has a different semantics than the data in the commercial data processing. In control engineering, the temporal determinism of data transmission is of great importance. The loss of a message is corrected by the next periodic message, so it makes no sense to retransmit a corrupted or lost message with the old data at a later time. However, it may be useful to have the same real-time message appear multiple times in a real-time message sequence to mask an error in a real-time message.

Fig. 3 shows the detailed timing of transmission of a real-time message sequence. In Fig.

3, the progression of time on the time axis 300 is shown. After the system has been initialized, the T-node sends the real-time message sequence schedule to all node computers in the arena 100 in a non-time-critical data message, eg message 321. In The schedule specifies at which periodic times (the global time) a node computer has to send its timed real-time message in the real-time message sequence.

The T-node 130 dictates the global time in the arena. It is advantageous if the T-node contains a receiver for the GPS signals and takes over the external time from a GPS system. The T-node 130 either takes the global time from an external time server or starts the time counting with its initialization. At the time 301 set in the schedule, the software of the T-node 130 commands the communication controller of the T-node to send the header message of the real-time message sequence.

The header message has the function of inhibiting, in the arena 100, the start of sending messages that would not have been ready at the time of the scheduled start of the first real-time message sequence real-time message. The content of the header message is meaningless. According to the invention, the length of the header message must be greater than the longest application message that the communication controller has to transport. A typical value for the longest application message is the longest Ethernet message with a length of 1538 bytes. According to the WiFi standard [6], the longest WiFi message is 2346 bytes. A header message with a length of 2000 bytes fulfills the above requirement.

In the example of Fig. 3 we assumed a communication system with a bandwidth of 54 Mbytes / sec. In this communication system, the transmission of the header message takes 296 μsec and the longest Ethernet message 228 μsec.

In the case (A) of FIG. 3, no message is transmitted in the arena 100 at the time 301. After the IFS <301,302>, the T node 130 immediately starts sending the header message 330. Since, according to the invention, the a priori determined time interval <302,305> with a length of 280 μsec must be shorter than the length of the header message 330 (transmission time of 296 becomes Transmit the header message at time 305 from the T-node 130 (at which time the sending of the header message is not yet complete). The communication controller (chipset) of the T-node 130 interrupts the antenna of the T-node 130 at time 305, since many chipsets do not support the timely termination of a message in transmission.

In case (B) of Fig. 3, the transmission of the longest Ethernet message 320 begins immediately before time 301. The transmission of this longest Ethernet message terminates at time 303, after 228 μsec. After the IFS <303,304>, in the case (B), the header message 330 can not be started until the time 304, although the software of the T node has the command to send the header message 330 at the time 301 (as in the case (A)). ) has issued to the communication controller. Also in case (B), the transmission of the header message 330 is aborted at the same time 305 as in the case (A).

According to the invention, the software of the node computer, which has to send the first real-time message 331, instructs its communication controller to send the first real-time message 331 in the command interval <304,305>. Due to the CSMA / CA logic of the communication controller, the transmission of the first real-time message 331 is started only at the time 306, after the IFS <305,306>.

During transmission of the first real-time message 331 in the interval <306,307>, which is also the command interval of the second real-time message 332, the software of the node computer that has to send the second real-time message 332 commands its communication controller to send the second real-time message 332 , Due to the CSMA / CA logic, the sending of the second real-time message 332 is started only after the IFS <307,308> at the time 308.

During transmission of the second real-time message 332 in the interval <308,309>, which is also the command interval of the third real-time message 333, the software of the node computer that has to send the third real-time message 333 instructs its communication controller to send the third real-time message 333 , Due to the CSMA / CA logic, the transmission of the third real-time message 332 is started only after the IFS <309,310> at time 310. At time 311, the transmission of the real-time message sequence ends.

In each WiFi network, a beacon message [6] is periodically sent by a base station. In the beacon message are the characteristic data of the network, such as the Network name included. In some networks, the clocks of the node computers can also be synchronized via the beacon message. According to the invention, the beacon message of a WiFi network can be scheduled as the first real-time message of a real-time message sequence in the schedule. It is advantageous if the base station transmitting the beacon message and the T-node are precisely synchronizing their clocks via a wired communication system, eg via TT Ethernet [9].

In summary, it should be noted that the time 305 at which the header message 330 is aborted by the T-node 130 represents the global synchronization event for the entire arena. Since the length and thus the transmission duration of a real-time message from this point in time 305 is known, the a priori known reception time of the first (or further) real-time message of the real-time message sequence can be used as a synchronization signal for establishing the global time in the node computers.

The present invention describes a novel method for deterministic wireless transmission of timed messages between a plurality of mobile and / or stationary node computers located in a limited area and one or more stationary base stations. In order to ensure that the real-time data can be transmitted at a priori known times, a variable header message is introduced by a special node computer which has the function, in the arena, of sending messages which at the a priori known time of the planned start of the first real-time message Real-time message sequence has not been completed to prevent.

In the wired TTP protocol, it is assumed that all correct node computers adhere to the a priori created schedule. This assumption is not realistic in an arena where messages are transmitted wirelessly.

The nodes involved in the real-time data transmission are cooperative according to the given timetable, the other nodes present in the arena do not have a timetable and therefore can not be cooperative.

Quoted literature:

[1] US Pat. 8,107,390. Budampati, et al. Apparatus and method for deterministic latency - controlled communication in process control Systems. Granted on January 31, 2012

[2] US Pat. 7,499,444. Bennett. Method for dock synchronization ofwireless 1394 bus for nodes connected via IEEE 802.11 a / b WLAN. Granted on March 3, 2009.

[3] US Pat. 8,583.777. Boyle, et al. Method and System for providing real-time end user WIFI quality data. Granted on March 27, 2012.

[4] US Patent 8,427,987. Stocks. System and method for time synchronized beacon enabled wireless personal area network communication. Granted on April 23, 2013.

[5] US Patent 8,547,958. Hundal, et al. System and Method of enhancing WIFI real-time communication. Granted October 1, 2013

[6] IEEE Standard 802.11. Wireless LANs. URL: standards.ieee.org.

[7] Kopetz, H. Real-Time Systems, Design Principles for Distributed Embedded Applications. Springer Verlag. 2011th

Hiertz, GR. The IEEE 802.11 Universe. IEEE Communication Magazine. Jan 2010. pp. 62-70. IEEE Press. 2001, M., Ed. System of Systems Engineering - Innovations for the 21st Centur., J. Wiley & Sons. In 2009. [9] SAE Standard AS6802 from TT Ethernet. URL: http://standards.sae.org/as6802

[10] RT-WiFi: Wei et al. Real-Time High-Speed Communication Protocol for Wireless Cyber-Physical Control Applications. Proc. of RTSS 2013. pp. 140-149. Vancouver. IEEE Press

Claims

A method for deterministic wireless transmission of real-time scheduled messages in a distributed real-time system, which real-time system comprises a plurality of node computers and one or more base stations arranged in an arena, all node computers and base stations having a global time, characterized in the arena, one or more real-time message sequences are periodically transmitted, wherein a real-time message sequence consists of a variable-length timed header message and a sequence of one or more timed real-time messages of a priori known length, and wherein the software of a T Node commands its communication controller to send the header message, and wherein the communication controller of the T node starts sending the header message as soon as there is no activity in the arena during an IFS is detected, and wherein at an a priori scheduled timed termination time of the T-node aborts the sending of the header message, and wherein in a command interval before the termination time of the header message, the software of the computer node that has to send the first real-time message of the real-time message sequence, the communication controller the Command to send the first real-time message string real-time message, and wherein in the command interval prior to termination of the first real-time message, the software of the computing node that is to send the following real-time message sequence real-time message commands its communication controller to send the following real-time message, and wherein This process is repeated until all real-time messages of a real-time message sequence have been sent.

2. The method according to claim 1, characterized in that in a real-time message sequence a real-time message appears multiple times.

3. The method of claim 1 or 2, characterized in that the length of the header message is longer than the longest application message that has to transport the communication controller.

4. The method according to any one of claims 1 to 3, characterized in that the transmission time of the complete header message is longer than the time interval between the start command for sending the header message (301) and the termination time (305), to which the sending of the header message is canceled ,

5. The method according to any one of claims 1 to 4, characterized in that the IFS between every two messages of a real-time message sequence has the smallest value supported by the protocol.

6. The method according to any one of claims 1 to 5, characterized in that a node computer uses the known time of arrival of the first real-time message of the real-time message sequence to synchronize its local clock.

A method according to one or more of claims 1 to 6, characterized in that the first real-time message of the real-time message sequence is a beacon message of the network.

8. The method according to any one of claims 1 to 7, characterized in that the T-nodes and the base stations are externally synchronized.

9. The method according to any one of claims 1 to 8, characterized in that the T-nodes and the base stations are synchronized via a TTEthernet system.

10. The method according to any one of claims 1 to 9, characterized in that the wireless data channel according to the IEEE standard 802.11 (WiFi) is realized.

11. T-node for wireless transmission of timed messages in a distributed real-time system, which real-time system comprises a plurality of node computers and one or more base stations, which are arranged in an arena, characterized in that in the T-node, a chipset for wireless Transmission is arranged and that the physical connection between the chipset and the antenna of the T-node can be interrupted by a timed switch.

12. T-node according to claim 11, characterized in that in the T-node, a GPS signal receiver for receiving the GPS time is included.

13. A real-time system, in particular distributed real-time system, comprising a plurality of node computers and one or more base stations, which are arranged in an arena, for carrying out a method according to one of claims 1 to 10.

14. Real-time system according to claim 13, characterized in that all node computers and base stations have a global time.

15. Real-time system according to claim 13 or 14, with at least one T-node according to claim 13 or 14.