WO2021008683A1 - Method and device for the allocation of resources in the transmission of data packets - Google Patents

Abstract

The invention relates to a method and a device which are suitable for introducing a plurality of resource models for allocation of resources and for identifying same on the basis of an ID. All resource models are based on the same generic parameters for the description of the data for said models. A maximum bandwidth and a typical latency for the class results depending on the resource model selected behind the class. All applications use the same protocol for reserving resources during the establishment of a connection. The same protocol makes interoperability in the network possible with different internal calculations and configurations of the hardware. The internal mapping onto the hardware mechanisms and the underlying resource model remain concealed from the application.

Description

description

Method and device for resource allocation for the transmission of data packets

In industrial automation, among other things, Ethernet is used as a transmission technology. The respective automation task (application) places increased demands on the maximum transmission time (latency) and the availability (downtime) of data in the network.

Depending on the respective industrial application, a special network topology is usually selected for the automation network in order to be able to technically meet the respective requirements. The network consists z. B. only from one line (devices with integrated switch) or a ring. When considering the achievable latency and quality of the connection, the previously selected restrictions were taken into account and the network was set up specifically for the implementation of the application.

Due to the widespread use of Ethernet, the network is now mostly used by several applications at the same time. A combination of several applications with different expectations of the network is, however, not possible without problems - since these have to share the resources available in the network components.

State of the art

So far, a system based on Ethernet was selected that was adapted to the respective application. This includes its own model, also called the resource model in the following, and can deliver the desired behavior within the framework of the assumptions.

In the following, we understand a resource model to be the combination of network infrastructure, transmission protocol and reservation strategies used in the network elements required for transmission. For tasks in drive technology (motion control), for example, particularly short transmission times are required. The sensors have to record measured values at a defined point in time and transmit them to the central evaluation unit (PLC). So far, this problem was addressed with the PROFINET IRT system - which is based on precise planning of all cyclical data packets in the network, the result of the planning configured in the network nodes and synchronizes the configuration during operation. The processing time in the actuators and sensors is taken into account during planning. The application and the network run in isochronous mode. For the configuration, however, detailed knowledge of all applications is required, the resulting communication relationships with the respective data volume and the current topology with the properties of all network nodes (bridges). This is not always possible. In addition, the entire system has to be recalculated and configured every time a single parameter is changed.

A combination of several independent applications in egg nem network is currently not possible - all applications must be considered in the overall ONTEXT _Z.

The PROFINET RT (RealTime) model was introduced for systems with greater flexibility that can cope with greater latency. In an environment with limiting disturbance variables (other network traffic is limited in the consideration), preferred forwarding is used based on the priority. The network itself is not taken into account when planning the application - the data is transparently forwarded by the network. Overloading the network or individual connections is not possible. During the planning of the application, the maximum bandwidth is checked in order to avoid overloading the connections to the end devices (especially the PLC, programmable logic controller, programmable logic controller of a system). There If the network is not taken into account, the system cannot give any guarantees.

If the network load is higher than that previously assumed in the planning, for example when using other applications or other than the planned network traffic, packet losses or significantly higher latencies consequently occur.

A parallel operation of several of these optimized systems, each with its own resource allocation model in a network, is not possible without interference.

The previous models for optimizing an application through a special selection of the network can no longer be used for several applications. A resource allocation model has already been developed through the further Ethernet developments within the framework of the (audio / video bridging) AVB Task Group, which makes it possible to operate multiple applications in any network.

However, the model is generic and does not take into account the restrictions in typical industrial networks, which are necessary for higher data transmission performance. As a result, the transmission performance is not sufficient for all industrial applications. The high number of nodes - network elements to be passed through, as occurs in a line topology, was not taken into account.

The uniform resource model for audio and video transmissions (AV streams) enables the calculation and verification of the resources required for secure transmission.

The configuration of the network is initiated by the end devices through MSRP. In the network nodes, the forwarding and the resources required for this are allocated based on the information from the end devices. In this way overloads in the network can be excluded and guarantees can be given for the transmission. The AV resource allocation model is optimized for the highest possible utilization of the network with the lowest possible resources required - but leads to a higher latency of individual connections.

The further development of AVB, TSN (Time-Sensitive Networking), will enable the combination of several newly introduced mechanisms in the future, but so far has not made any statements about the configuration and useful uses of the new mechanisms. The use of several different combinations in the same network has not yet been taken into account. Depending on the respective application, a (different) sensible combination of mechanisms must be selected in order to be able to technically meet the requirements.

Figure 1 shows different network configurations, for example MRP / MRPD, HSR, IRT and AVB, in which the individual network elements NE, NE1, ..., NE12, ..., NE21, ... are each shown. These can be arranged as a ring, bus or tree structure, for example, with special PLC network elements which are typical for communication in a system, as described above, and which represent a certain type of communication. Multiple transmissions can also occur, particularly in ring structures, in order to ensure the failsafe security of the data communication.

The object of the invention is to provide a possibility which, when using Ethernet-based, allows a plurality of applications to be made simultaneously with different expectations of the network using the resources in the network components that must be shared

solution

The problem described is solved by a method having the features of claim 1. Furthermore, the Problem solved by a device according to the features of the independent claim 8.

The basic idea is to introduce several resource models for the allocation of resources and to identify them using an ID. This means that the mechanisms available in the hardware can be configured differently - depending on the model used. Several classes, each with their own behavior models, can be supported in the network.

For certain applications and network topologies, an optimized analysis of the resources can be carried out. It is always a consideration to what extent one

Simplification (e.g. by area reservation for a guaranteed number of streams), or a shorter latency, or an optimization of the "Makespan" (ie the time from the beginning of the information transmission to the complete transmission of the information to the recipient) with the resulting The following losses in the possible usable resources should be preferred.

A waste of bandwidth through simplified calculation and / or the assumption of a simplified topology with simplified communication relationships is possible.

All resource models are based on the same generic parameters, already introduced with AVB, to describe their data. Depending on the selected resource model behind the class, there is a maximum bandwidth and a typical latency for this class (target latency). All applications use the same protocol to reserve resources while the connection is being established. The same protocol, with different internal calculations and hardware configurations, enables interoperability in the network.

The resource models that can be implemented in a network depend on the hardware available and, as with AVB (Audio Video Bridging), must be configured by the network management. Based on the target latency, the application can then find a class that suits its requirements choose. The internal mapping to the hardware mechanisms and the underlying resource model remain hidden from the application.

Depending on the model, the respective assumed behavior can then be checked by a configuration of the monitoring mechanisms that is adapted to it. This leads to increased robustness - since incorrect data packets can be identified and deleted before they cause further problems in the overall system. Based on the configuration of the ID of the resource model, the hardware can then be set up automatically in each network participant for the affected streams in order to monitor the behavior.

The inventive method for resource allocation for the transmission of data packets in an industrial network, which consists of individual network elements, each of the network elements having a connection to at least one further network element, with the following steps:

- Data packets are generated by at least one application (for a recipient and received from the network for this purpose)

- One of the data packets (usually the first data packet sent) contains a connection setup request which also contains information on a required transmission latency, the latency time being the maximum permissible duration of the complete transmission of the data packets (data) to the recipient.

The allocation of the resources of the network takes place in that in the individual network elements a configuration is made with regard to a storage capacity made available for the transmission in the network elements and a communication protocol used for the transmission, with

using the required latency time, a suitable resource model is selected from a set of possible resource models that meets the required latency time, and a reservation of the transmission capacities at least of the network elements required for the necessary transmission is carried out by a network manager in accordance with the selected resource model.

The inventive device for resource allocation for the transmission of data packets in an industrial network, which consists of individual network elements, each of the network element having a connection to at least one further network element, and

Data packets are generated by at least one application for a receiver and transmitted through the network, and the device determines a connection setup request for the transmission of the data packets from one of the data packets, which additionally contain information about a required latency of the transmission, the latency being the maximum permissible time for the complete transmission of the data packets to the recipient.

The device according to the invention achieves an allocation of the resources of the network in that a configuration is carried out in the individual network elements with regard to a storage capacity made available for the transmission in the network elements and a communication protocol used for the transmission, whereby beforehand based on the required latency A suitable resource model has been selected from a set of the possible resource models, which fulfills the required latency, and the transmission capacities of at least network elements required for the required transmission are reserved by the device in accordance with the selected resource model.

A model that is frequently used in industry is communication from a PLC to decentralized IO devices. The optimization consists in the utilization of the so-called wind shadow effect. In the sending direction of the PLC to all actuators, the fact that the data is sent first with a long transmission time is used. Data to closer devices are transmitted in the "slipstream" of the previously sent data and the data transmission takes place in parallel with the previously started data transmissions. In the receiving direction, this is automatically achieved with simultaneous transmission of all sensors in the network. This effect allows a higher bandwidth in the simplified flat model , however, is difficult to monitor and is only typically found in this particular application.

By adding an ID for the resource allocation model to the existing domain attribute of the existing AVB resource reservation protocol, the configuration can be carried out automatically using a protocol in the network components. The application indirectly determines the model to be used by selecting the required latency and thus ensures that the required network behavior can be checked in the network (configured by the reservation). The same reservation model is used to check in each node whether forwarding is possible. Each reservation can be checked individually according to the respective model, whereby a guarantee of transmission (low latency and no data loss) is possible.

Depending on the class and the associated assumptions about the behavior of the application, a different maximum number of streams and a different achievable latency in the network are possible. Several models can be used at the same time for different applications in a network.

The novelty lies in the introduction of different calculation and configuration variants (resource allocation model) in the same reservation protocol.

In one embodiment, the method according to the invention also has information about the maximum possible bandwidth for the transmission of data packets by means of the respective resource model. The method according to the invention advantageously enables the assigned resource model to be identified when the data packets are sent using a unique identifier. The differentiation (based on the identifier, ID) of the various models means that the various "optimizations" introduced in the industry (resource calculation model) can be applied. The appropriate model can be selected directly by the application (by selecting the stream class in The reservation based on the required latency). This enables more streams in the network, lower latency can be achieved or a completely different type of network data can be supported (e.g. flows only with bandwidth protection for alarms with maximum bandwidth) is exchanged within the network as part of the existing class description and communicated to the end devices.

Depending on the application model, the data already available through the reservation can be used in the network nodes

Streams can be monitored for compliance with the respective application model. The behavior model can be recognized by an ID of the resource model (depends on the class). It is only through monitoring that the transmission and latency can be guaranteed. These guarantees are missing in many previous systems (no monitoring possible).

The invention is further illustrated by figures, which show:

Figure 1 different network configurations according to the prior art,

Figure 2 shows a novel network configuration according to the inven tion and

Figure 3 is a flow chart according to the independent method claim.

An advantageous embodiment of the method uses a uniform reservation protocol during the reservation of the network elements required for the transmission connection establishment for all applications and resource models.

The configuration of the network elements depends on the hardware contained in the network elements.

The resource model used for the transfer is one that works according to one of the following standards:

- AVB, audio / video bridging,

- TSN, Time Sensitive Networking,

- IRT, Isochronous Real-Time,

- MRP, Media Redundancy Protocol,

- MRPD, Media Redundancy Planned Duplication,

- HSR, high-availability, seamless redundancy.

FIG. 2 shows a network structure according to the invention, where all possible network configurations already shown in FIG. 1 are found. These are structured, 1, 2, 3, 4 and suitable for the respective transmission protocols.

Figure 3 shows a flow chart of the method according to the invention. Any (any) application ANW / S generates a data stream to at least one receiver E. The connection setup request VAW is transmitted to a network manager NWM, with information about the required maximum latency, which uses the information available to provide a suitable resource model for transmitting the Selects data and sends the associated unique identification ID back to the application. The reservations AR necessary for the resource model are carried out in the network NW. For the user ANW / S it is irrelevant how and in which ways the data are transmitted; this is indicated by the dashed line around the network elements NE1, NE2, NE3. From the application's point of view, the data is transmitted transparently, ie the application has no information about the type of transport or the selected transport route. The data is then transmitted with the ID from the application and transmitted via the reserved path. The introduction of the class for a flat reservation within a configured time window (PCT / EP2017 / 066989 method for high-performance data transmission in a data network with some real-time requirements and a device for performing the method) enables z. B. a fixed statement about the number of connections that are possible in the network (e.g. backbone ring) - regardless of the location of the end devices - with a very low maximum latency - with higher resource consumption (through reservation on all links). By comparing neighboring network components (as with AVB), the limits in the network within which an application model is supported can be recognized by a protocol and taken into account when making the reservation. This means that this model, which has been optimized for latency, can be combined with other classes in the same network, e.g. with OPC UA connections outside the machine.

The support of optimized industrial models enables the use of TSN for industrial communication, as the special application requirements, such as B. a lower latency for line topologies can be achieved through special resource allocation models. Thanks to a class with area reservation (PCT / EP2017 / 056132 - method for bandwidth reservation and suitable network element), optimized mechanisms for switchover redundancy (MRP, HRP, RSTP, eRSTP, RSTP +, ...) can be used, since no new reservation is required must be carried out on the changed topology, which would increase the switchover time and prevent it from being used. This means that a class with protection for networks through switchover redundancy can be introduced - this was previously not possible.

With the relationship between a PLC with communication and the decentralized IO modules, which is typical for industrial automation, an optimization can be introduced when considering the bandwidth. By utilizing the slipstream effect through a clever transmission sequence in the PLC, more streams can be permitted in the network than with the simplified area reservation. In the simplified model with areal reservation, the latency is comparable to the latency of a network node in the general model of AVB. In this model, the optimized transmission sequence is assumed, but cannot be monitored. Without the introduction of an ID, the behavior would have to be configured separately for all bridges and there could be no standardized comparison of the model used.

A separate network for each application model that is configured accordingly and only supports one model for all connections can also meet the technical requirements - but leads to significantly higher costs and is therefore no longer accepted by the customer. A monitoring of the data connections and a comparison with the expected behavior is then only possible as a uniform system definition for all connections in the network and must be specified when the network is set up. This is probably only possible with devices from one manufacturer with a uniform configuration.

The ID of the resource model can be part of a profile (e.g. in the IEEE or IEC). This means that a uniform model can be defined for typical applications and manufacturer-independent use can be ensured.

The ID can be a manufacturer-dependent ID, such as B. an OUI in the IEEE

(http://standards.ieee.org/develop/regauth/oui/) or a label that is defined by another standardization group.

Previous invention disclosures that describe different application models:

- Area reservation to support switchover redundancy (PCT / EP2017 / 056132)

- Optimized resource allocation model for scheduled systems - Reservation takes place within a given time window with a given maximum number of hops (PCT / EP2017 / 066989) The idea of this disclosure enables the combined use of several models within the (resource) reservation protocol of a TSN network.

Claims

Claims 1. Method for resource allocation for the transmission of data packets in an industrial network (NW), the network consisting of individual network elements (NE1, NE2, NE3), each network element having a connection to at least one further network element, with - Data packets (data) generated by at least one application (Anw / S) for a receiver (E) and the data packets are received by the network (NW), and - One of the data packets (data) contains a connection setup request (VAW), which also contains information on a required latency of the transmission, the latency being the maximum permissible duration of the complete transmission of the data packets (data) to the recipient (E) , and - The allocation of the resources of the network (NW) follows that in the individual network elements (NE1, NE2, NE3) by means of a configuration with regard to a storage capacity made available for the transmission in the network elements and a communication protocol used for the transmission is set, characterized in that Based on the required latency, a suitable resource model is selected from a set of possible resource models, which fulfills the required latency, and a reservation of the transmission capacities of at least the network elements required for the necessary transmission (NE1, NE2, NE3) by a network manager (NWM) ) according to the selected resource model. 2. The method according to the features of claim 1, characterized in that the possible resource models also contain an indication of the maximum possible bandwidth for the transmission of data packets by means of the respective resource model. 3. The method according to the features of one of the claims 1 or 2, characterized in that the reservation protocol (AR) used by the network manager (NWM) for the reservation of the network elements (NE1, NE2, NE3) required for the transmission is uniform for all applications and resource models during the connection establishment. 4. The method according to the features of one of the preceding patent claims, characterized in that the assigned resource model is identified when the data packets are sent using a unique identifier (ID). 5. The method according to the features of one of the preceding Pa tent claims, characterized in that the configuration of the network elements depends on the hardware contained in the network elements (NE1, NE2, NE3). 6. The method according to the features of one of the preceding Pa tent claims, characterized in that the underlying network (NW) works according to the Ethernet standard. 7. The method according to the features of one of the preceding patent claims, characterized in that The resource model used for the transfer is one that works according to one of the following standards: - AVB, audio / video bridging, - TSN, Time Sensitive Networking, - IRT, Isochronous Real-Time, - MRP, Media Redundancy Protocol, - MRPD, Media Redundancy Planned Duplication, - HSR, high-availability, seamless redundancy. 8. Device (NWM) for resource allocation for the transmission of data packets in an industrial network (NW), whereby the network consists of individual network elements (NE1, NE2, NE3), each network element having a connection to at least one further network element, wherein - Data packets (data) are generated by at least one application (Anw / S) for a receiver (E) and are to be transmitted through the network (NW), and - The device determines a connection setup request (VAW) for the transmission of the data packets from one of the data packets (data), which additionally contains information on a required latency of the transmission, the latency time being the maximum permissible duration of the complete transmission of the data packets (data ) to the receiver (E), and - The device achieves an allocation of the resources of the network (NW) in that a configuration is carried out in the individual network elements (NE1, NE2, NE3) with regard to a storage capacity made available for the transmission in the network elements and a communication protocol used for the transmission , characterized in that the device selects a suitable resource model in advance on the basis of the required latency from a set of the possible resource models, which resource model meets the required latency, and a reservation of the transmission capacities at least of the network elements required for the required transmission (NE1, NE2, NE3) by the device (NWM) takes place in accordance with the selected resource model.