Scheduling network usage under the threat of unexpected bandwidth shortages

Abstract

Embodiments of the present disclosure relate to orchestrating network usage under the threat of unexpected bandwidth shortage. A method (100) for orchestrating usage of at least one communication network (1) for transmitting a plurality of data streams (31-36) transmitted by a plurality of applications (21-23) comprises the steps of attributing (110) importance levels (31a-36a) to individual data streams (31-36) and / or to groups of data streams (31-36); determining (120) an ordered list (4) of data streams (31-36) to be scaled back or stopped in case of a bandwidth shortage in the communication network (1) based at least partly on the importance levels (31a-36a) of the data streams (31-36) and network capacity requirements (31b-36b); and providing (140) the ordered list (4) to one network management entity (1a).

Description

TECHNICAL FIELD

[0001] The present invention relates to the field of orchestrating network communication in industrial facilities. BACKGROUND

[0002] Operating complex industrial facilities and / or executing industrial processes on such facilities depends on a multitude of data streams that need to be communicated on the facility. For example, a distributed control system, DCS, needs to issue control actions to low-level controllers and actuators in the facility and needs to read out process variables as feedback for control loops.

[0003] As these data streams can be critical for the safe and reliable operation of the facility, wired networks have been preferred over wireless networks in the past due to their higher reliability. However, it is also desirable to exploit the increasing potential of wireless networks for critical communication of industrial facilities.

[0004] WO 2018 / 095507 A1 discloses a method for scheduling communication of field devices in a wireless network of an industrial process system. SUMMARY

[0005] It is an object of the present invention to improve the reliability of communication in an industrial environment where a more diverse spectrum of applications needs to transmit critical data streams over wireless networks.

[0006] This object is achieved by the method according to the independent claims. Further advantageous embodiments are detailed in the dependent claims. DETAILED DESCRIPTION

[0007] The present invention provides a method for orchestrating the use of at least one communication network for transmitting a plurality of data streams transmitted by a plurality of applications. The network can be a wired network or a wireless network. The object of the method is to orchestrate the use of the network in case of unexpected shortage of bandwidth. For several reasons, wireless networks tend to be more prone to such unexpected shortages. The most important reason lies in the nature of the radio communication medium as opposed to cables or wires. Typically, the capacity of a specific channel between communication partners depends on the physical environment, including distance, obstacles, reflections, and other factors down to air humidity. For some technologies, the radio frequency spectrum is also a medium that the operator of an industrial facility has to share with other entities. In particular, the unlicensed spectrum such as WiFi tends to be crowded, especially in urban areas. Furthermore, 5 GHz WiFi has to stop transmitting and change frequency as soon as an attempt of a primary user, e.g. a radar, to use the channel is detected. Network slicing in 5G networks logically separates the public 5G network into multiple logical networks for exclusive use by their owners, but the physical infrastructure still has to be shared with other users. Even private 5G networks cannot be completely exempted from unexpected bandwidth shortages, as the spectrum can be impaired by malfunctioning equipment of third parties or even by intentional radio jamming (interference).

[0008] During the course of the method, importance levels are attributed to individual data streams and / or groups of data streams. Based at least partly on these importance levels and network capacity requirements of the data streams, an ordered list of data streams to be scaled back or to be stopped in case of a bandwidth shortage in the communication network is determined. Here, the network capacity requirements can be formulated in terms of any suitable quantity, such as one or more of: bandwidth, latency, reliability, and redundancy level. In particular, the ordered list can specify specific actions to be taken to scale back and / or stop the data streams, such as "scale down resolution of video data stream" or "convert video data stream to black and white". The ordered list is provided to a network management entity. The management entity is configured to:

[0009] • monitor the available bandwidth in the communication network;

[0010] • compare the available bandwidth to the combined bandwidth requirement of all data streams currently being transmitted; and

[0011] • in response to determining that the available bandwidth is less than or is about to become less than the combined bandwidth requirement, scale back or stop the data streams in the order given by the list in order to bring the combined bandwidth requirement back to or below the available bandwidth.

[0012] Previously, data streams could carry individual quality of service, QoS, information about priority, latency, or bandwidth requirements. When dealing with multiple data streams with different QoS information, one data stream could be prioritized over another data stream based on the QoS information. But the QoS information could not accommodate any additional knowledge about how important each data stream is for a specific application domain. For example, all video data streams or all control data streams carry the same QoS information, as they are technically identical and only differ in their content. But different content can have different importance for the operation of the entire industrial facility. For example, a video data stream that monitors the actual execution of a process on the industrial facility is much more important than a video data stream that monitors the parking lot of the facility. The new, more granular assignment of importance levels allows to globally distinguish individual data streams with the same priority from each other.

[0013] This information can then be utilized to produce an explicit degrading response plan in the form of an ordered list. For each level of network capacity, the network manager knows how to degrade which data stream and which application's integrity level. Likewise, on the facility operator side, it is known in advance in which order which degradations are expected to be made in case of a bandwidth shortage.

[0014] In the toy example of a professional drone flying in a facility, the link between the remote control and the drone can be used to transmit a video stream captured by the drone in the downlink in multiple directions to the remote control and to transmit a control data stream in the uplink to the drone. Of course, in order to maintain control of the drone, it is important to maintain the control data stream. However, if the drone is flying out of the pilot's direct visual range, it is equally important to have at least a video data stream captured in the current flight direction in front of the drone. Without that video data stream, the pilot is flying blind and the risk of a collision increases. Therefore, it is advantageous to maintain at least that video data stream and the control traffic, and to sacrifice video feeds from directions in which the drone is not currently flying.

[0015] In a particularly advantageous embodiment, the ordered list contains at least a first entry and a later second entry affecting the same data stream, wherein the first entry indicates that the data stream is to be scaled down and the second entry indicates that the data stream is to be scaled down further or stopped. In this way, the degradation of the service can become more granular and graceful. For example, the ordered list can provide that, before the video data stream is finally completely disabled, first the resolution of the video is scaled down, then it is degraded from color to black and white, and then the frame rate is reduced.

[0016] In another particularly advantageous embodiment, the network management entity is further configured to, in response to determining that the available bandwidth is higher than the combined bandwidth requirement, scale down or stop at least one data stream: unscale down or restart at least one scaled down or stopped data stream. That is, once the bandwidth shortage is remedied, the overall service is gradually returned. Most unexpected bandwidth shortages are only temporary, i.e. they disappear as quickly as they came. If a selection is to be made among the multiple scaled down or stopped data streams, the selection can be made according to any suitable policy. For example, the last scaled down or stopped service can be the first candidate for unscale down or restart. This would restore more important data streams earlier. Another possible policy is to select the first scaled down or stopped service as the first candidate for unscale down or restart. This would restore the data stream that has been affected the longest time. This can avoid, for example, transmission timeouts in applications.

[0017] To facilitate this, in another advantageous embodiment, a second ordered list of data streams is determined based at least in part on the importance level of the data streams and the network capacity requirements. This second ordered list indicates the order in which the data streams are to be unscaled down or restarted. In this way, if the services are returned in a different order than they were scaled down and / or stopped, the operation of the facility and / or the execution of the processes can benefit more.

[0018] Any suitable "ordering indicator" can be used to compile an ordered list of resource (de-)allocation. For example, a general strategy can provide that applications are first scaled down in priority order before stopping the application in priority order. For example, video surveillance can have a lower priority than process condition monitoring, and resolution is first reduced. But if possible, condition monitoring can also be scaled down first before stopping the video completely. Such a general strategy can be augmented and / or replaced by a specific strategy that can put different key performance indicators, KPIs, in a specific order, which can optionally be combined with application or signal priority. In this way, a specific allocation order of a set of connections, e.g. individual connections, partial applications, one or more applications, or any other partitioning.

[0019] Scaling down and / or stopping of one or more data streams can advantageously be combined with de-scaling and / or restarting of other data streams. For example, if some data streams related to condition monitoring have been stopped first, and then a larger bandwidth block is freed by scaling down process critical video from UHD to FHD, it can be larger than the current bandwidth deficit after this block is freed. That is, some bandwidth can actually be left over after this block is freed. In this case, the data streams related to condition monitoring that were stopped previously can be restarted in order to take advantage of the excess bandwidth that now becomes available by freeing up the large bandwidth block.

[0020] Scaling down or stopping of at least one data stream can be performed in any suitable manner. For example, resources previously allocated for the data stream can be de-allocated at the network side. At least one network infrastructure device can be instructed to throttle bandwidth usage of the data stream, or to block forwarding of the data stream. Furthermore, at least one network infrastructure device can be instructed to compress and / or transcode the data stream. In particular, this can free up part of the bandwidth occupied by the video data stream.

[0021] Alternatively or in combination therewith, a source of the data stream can be instructed to reduce bandwidth usage of the data stream, or to stop the data stream. The source is closer to the actual application, and can therefore have more information about which part of the information in the data stream is more dispensable than other parts of the information. For example, in some types of video, color information can be more important, while in other types of video, pixel resolution or frame rate can be more important.

[0022] In another particularly advantageous embodiment, attributing importance to the data stream can comprise classifying at least one data stream as

[0023] • a non-negotiable data stream that is not scaled down or stopped;

[0024] • a scalable data stream that can be scaled down to a given extent; or

[0025] • best-effort data flows that can be stopped.

[0026] These are the main categories of data flows that are relevant during the execution of an industrial process on an industrial facility.

[0027] Thus, in another particularly advantageous embodiment, the application that transmits the data flows participates in the operation of the industrial facility, and / or in the execution of at least one production process on the industrial facility, and the importance level is attributed to the data flows and / or groups of data flows at least partly based on these data flow importance for the operation of the facility and / or for the execution of the process. In this way, in case of bandwidth shortage, knowledge about the process domain can be transferred into the plan for graceful degradation. Essentially, such domain knowledge decides what is considered "graceful". Examples of data flows related to a production process in an industrial facility include I / O signals such as measurement values, status information, condition data, actuation data or other traffic exchanged within a distributed control system (DCS). But relevant data flows can also for example include video data flows, whether they are automatically evaluated by machines (for the purpose of the DCS) or not.

[0028] In one particularly advantageous embodiment, at least one data flow comprises a critical video data flow that is required for the continuation of the operation of the facility and / or the execution of the process, at least in degraded form. Most of the time, multiple video data flows are transmitted in an industrial facility, but only few of them are important. It is then advantageous to not treat all video data flows equally (in terms of QoS framework), but to concentrate the bandwidth on the video data flows that are most important for the execution of the process.

[0029] In particular, the critical video data flow can be classified as a non-negotiable data flow or a scalable data flow. This ensures that the critical video data flow is always transmitted at least in degraded form. In contrast, if the critical video data flow is classified as a best-effort data flow, it can be stopped completely. This would contradict the requirement of always transmitting at least in degraded form.

[0030] In one example, the critical video data flow is a live video stream that monitors the execution of the process or is analyzed to assess the quality or quality problems of the manufactured product. For example, in a waste incineration facility, a video data flow can monitor the combustion. In a chemical production process, a video data flow can monitor a reaction vessel for signs that the process is about to get out of control. Such signs can for example include a change of color or the formation of bubbles or foam in the reaction vessel content.

[0031] In principle, control services belong to the most important services for executing a process in an industrial facility. Such important control services comprise the measurement values of process quantities used as feedback in control loops. However, the urgency of updating these process quantities differs between the process quantities. In another particularly advantageous embodiment, at least one data stream with measurement values of a first process quantity is classified as non-negotiable data stream. At the same time, at least one data stream with measurement values of a second process quantity is classified as scalable data stream. During process execution, the first process quantity changes on a faster time scale than the second process quantity. By possibly introducing a delay into the updating of the measurement values of the second process quantity, some error can be caused. However, if the second process quantity changes slowly, the error is small.

[0032] As an illustrative example, consider an industrial process that normally uses the following data streams:

[0033] • a process critical video data stream;

[0034] • a first video data stream that is less important for the operation of the process than the process critical video data stream;

[0035] • a second video data stream that is also less important for the operation of the process than the process critical video data stream;

[0036] • an asset management service;

[0037] • a network control data stream; and

[0038] • a process control data stream.

[0039] One possible strategy to compile an ordered list of actions to be taken to reduce and / or stop data streams is "scale before fail". That is, first exhaust all possibilities to scale down the bandwidth requirements of data streams that do not lead to a complete failure of these data streams before the first data stream is completely stopped. For example,

[0040] • first, scale down the first video data stream (or the second video data stream) stepwise from UHD resolution to FHD resolution and then to SD resolution;

[0041] • next, scale down the second video data stream (respectively the first video data stream) stepwise in the same way;

[0042] • next, reduce the bandwidth requirements of the process critical video data stream, the resource management service, the network control service and the process control service in that order; and

[0043] • Finally, the first (or second) video data stream, the second (respectively first) video data stream, the process critical video data stream, the resource management traffic and the network control traffic are stopped in this order.

[0044] Then, the process control traffic can remain as the only service which is never completely stopped.

[0045] The order mentioned in this example does not mean that the process will run unhindered or continue completely until the point where only the process control traffic remains available in a simplified form. Rather, there can be one or more points where a further reduction or even a stop of a particular data stream can result in a quality of execution of the process being impaired or can even cause the process to stop. That is, "graceful degradation" can not be available for all data streams. For example,

[0046] • If the quality of the process critical video is degraded or stopped, the product quality cannot be monitored sufficiently and will be affected;

[0047] • If the network control traffic is degraded, this will result in a degradation of the network services on which the process relies (e.g. the TSN guarantees can no longer be met); and

[0048] • If the process control traffic is degraded, only the basic control of the process devices is still possible, but it is no longer possible to run the process in an optimal manner.

[0049] However, the order in which the faults occur is still different. For example, to the extent that the process devices can still be controlled, they can be maneuvered into a safe state to protect them from damage.

[0050] A modification of the "fault before scale" strategy is also possible. For example, one of the first video data stream and the second video data stream can be stopped before the other is scaled down to SD resolution.

[0051] In a particularly advantageous embodiment, the first process quantity is pressure and the second process quantity is temperature. Changes in temperature are usually damped by some thermal mass of the device or some thermal mass of the contents of such a device, e.g. process educts or products. However, there is nothing like a "pressure mass" which can slow down a change in pressure which has a certain inertia. Therefore, the pressure in a vessel can change much faster than the temperature. In addition, there are specific physical processes which result in a very rapid increase in pressure, e.g. the boiling of a substance into a vapor which increases the volume of the same amount of substance by a factor of 1000 or more.

[0052] In another advantageous embodiment, an ordered list of data flows to be scaled down or stopped is determined in a consistent manner among a plurality of entities and the result of this negotiation is provided to a network management entity. In this way, the best solution on how to cope with a bandwidth shortage can be derived distributively, in a way similar to the auto-configuration of a wireless mesh network based on a negotiation among the mesh nodes.

[0053] The network management entity can be any entity that is authorized to allocate network resources to data flows. For example, in a 5G network, the 5G network exposure interface can serve as the entity to which the ordered list is provided. In another example, the network management entity can be a controller of a software defined network, SDN. In a SDN, the architecture of the network can be changed in order to prioritize data flows between certain endpoints in the network.

[0054] The method can be implemented in whole or in part by a computer. Therefore, the present invention also relates to one or more computer programs having machine-readable instructions which, when executed on one or more computers and / or computing instances, cause the one or more computers to perform the above-mentioned method. In this context, a virtualization platform, a hardware controller, a network infrastructure device such as a switch, a bridge, a router or a wireless access point, and a terminal device in the network such as a sensor, an actuator or other industrial field device are also considered to be a computer which is able to execute machine-readable instructions.

[0055] Therefore, the present invention also relates to a non-transitory storage medium and / or a download product having one or more computer programs. A download product is a product which can be sold in an online store for immediate fulfillment by download. The present invention also provides one or more computers and / or computing instances having one or more computer programs and / or having one or more non-transitory machine-readable storage media and / or a download product. BRIEF DESCRIPTION OF DRAWINGS

[0056] In the following, the present invention is explained using the drawings, which are not intended to reduce the scope of the invention. The drawings show:

[0057] Figure 1 : Exemplary embodiment of a method 100 for orchestrating the use of a network 1 by data flows 31-36 from applications 21-23;

[0058] Figure 2 : Exemplary use case of the method 100 in an industrial facility 6.

[0059] Figure 1 is a schematic flow chart of an embodiment of the method 100 for orchestrating the use of a network 1 by data flows 31-36. The data flows 31-36 are transmitted by different applications 21-23.

[0060] In step 110, an importance level 31a-36a is assigned to each individual data stream 31-36. As discussed before, this can be more granular than the previous QoS information. In particular, the importance level 31a-36a can be motivated by the specific application 21-23, so they can even differ greatly between data streams 31-36, while having the same or similar QoS information due to their similar technical structure, such as a video stream or control traffic.

[0061] In step 120, an ordered list 4 of data streams 31-36 to be scaled back or stopped in case of a bandwidth shortage in the communication network 1 is determined based at least partly on the importance level 31a-36a and the bandwidth requirement 31b-36b of the data streams 31-36.

[0062] In step 130, a second ordered list 5 of data streams 31-36 is established based at least partly on the importance level 31a-36a and the network capacity requirement 31b-36b of the data streams 31-36 as a corresponding part of the ordered list 4, the second ordered list 5 indicating an order in which the data streams 31-36 are to be unscaled or restarted.

[0063] In step 140, the ordered lists 4 and 5 are provided to a network management entity 1a of the communication network 1. The network management entity 1a is configured to:

[0064] • monitor an available bandwidth in the communication network 1;

[0065] • compare the available bandwidth to a combined bandwidth requirement of all data streams 31-36 currently transmitted;

[0066] • in response to determining that the available bandwidth is less than or will become less than the combined bandwidth requirement, scale back or stop data streams 31-36 in order to bring the combined bandwidth requirement back to or below the available bandwidth; and

[0067] • in response to determining that the available bandwidth is higher than the combined bandwidth requirement, scale back or stop the at least one data stream 31-36: unscale or restart at least one scaled back or stopped data stream 31-36.

[0068] According to block 111, any data stream 31-36 can be classified as

[0069] • a non-negotiable data stream 31-36 that is not to be scaled back or stopped;

[0070] • a scalable data stream 31-36 that can be scaled back to a given extent; or

[0071] • best-effort data streams 31-36 that can be stopped.

[0072] In particular, the applications 21-23 that send the data streams 31-36 can be involved in the operation of the industrial facility 6 and / or in the execution of at least one production process on the industrial facility 6. According to block 112, the importance levels 31-36 can then be attributed to the data streams 31-36 and / or to groups of data streams 31-36 based at least partly on the importance of the data streams 31-36 for the operation of the facility 6 and / or for the execution of the process.

[0073] According to block 113, at least one data stream 31-36 can comprise measurements of a process quantity. According to block 121, the reduction of this data stream can then comprise extending the interval at which new measurements of the process quantity are sent.

[0074] In particular, according to block 113a, at least one data stream 31-36 with measurements of a first process quantity can be classified as a non-negotiable data stream, while according to block 113b, at least one data stream 31-36 with measurements of a second process quantity is classified as a scalable data stream. The difference between the data streams is that the first process quantity (e.g. pressure) varies on a faster time scale than the second process quantity (e.g. temperature).

[0075] According to block 122, the determination 120 of the ordered list 4 of data streams 31-36 to be reduced or stopped can be negotiated 122 between multiple entities in a consistent manner. According to block 141, the result of this negotiation can then be provided to the network management entity la.

[0076] Figure 2 An exemplary use case of the method 100 in an industrial facility 6 is shown. On the side of the industrial facility 6, the facility management entity 6a collects from the facility 6 knowledge about which data streams 31-36 are of their respective importance 31a-36a. That is, the importance 31a-36a stems from knowledge about the facility domain. On the side of the network 1, the network management entity la collects from the network 1 knowledge about which data streams 31-36 are of their respective importance 31b-36b. That is, the importance 31b-36b stems from knowledge about the network domain. Figure 2 In the shown example, the applications themselves inform the facility management entity about the respective network capacity requirements 31b-36b of the data streams 31-36 they need to send over the network 1.

[0077] An ordered list 4 of data streams 31-36 to be scaled down or stopped in case of bandwidth shortage in the communication network 1 is determined based on network capacity requirements 31b-36b in combination with importance 31a-36a of the data streams 31-36. The list 4 can have entries of the following type: "scale this video data stream from UHD to FHD to save 6 Mbit / s", and "stop this video stream altogether to save 15 Mbit / s". It is communicated to the network management entity 1a. The network management entity 1a can then interact with the applications 21-23, and / or with any suitable network devices (not shown in Fig. 1) in the network 1 to perform scaling down and / or stopping of data streams when needed. Figure 2

[0078] Reference symbol list

[0079] 1 Communication network

[0080] 1a Network management entity of 1 in the communication network

[0081] 21-23 Application

[0082] 31-36 Data stream transmitted by the application 21-23

[0083] 31a-36a Importance of the data stream 31-36

[0084] 31b-36b Network capacity (e.g. bandwidth) requirement of the data stream 31-36

[0085] 4 Ordered list of data streams 31-36 to be scaled down or stopped

[0086] 5 Ordered list of data streams 31-36 to be unscaled or restarted

[0087] 6 Industrial facility

[0088] 6a Facility management entity

[0089] 100 Method for orchestrating usage of a communication network 1

[0090] 110 Attributing importance 31a-36a to data streams 31-36

[0091] 111 Classifying data streams 31-36 into specific categories

[0092] 112 Determining importance 31a-36a based on facility 6 knowledge

[0093] 113 Selecting data streams 31-36 with process quantity measurement values

[0094] 113a Classifying first process quantity data streams 31-36 as "non-negotiable"​

[0095] 113b classifying the second process quantity data stream 31-36 as "scalable"

[0096] 120 determining the ordered list 4

[0097] 121 extended transmission interval of the measurement value

[0098] 130 determining the ordered list 5

[0099] 140 providing the ordered lists 4, 5 to the management entity 1a

[0100] 141 providing the negotiation result to the management entity 1a

Claims

1. A method (100) for orchestrating usage of at least one communication network (1) for transmitting a plurality of data streams (31-36) transmitted by a plurality of applications (21-23), comprising the steps of: ascribing (110) importance levels (31a-36a) to individual data streams (31-36) and / or to groups of data streams (31-36); determining (120) an ordered list (4) of data streams (31-36) to be scaled back or to be stopped in case of a bandwidth shortage in the communication network (1) based at least partly on the importance levels (31a-36a) of the data streams (31-36) and network capacity requirements (31b-36b); and providing (140) the ordered list (4) to a network management entity (1a) configured to: monitor available bandwidth in the communication network (1); compare the available bandwidth to a combined bandwidth requirement of all data streams (31-36) currently being transmitted; and in response to determining that the available bandwidth is less than or is about to become less than the combined bandwidth requirement, scale back or stop data streams (31-36) in the order given by the list (4) in order to bring the combined bandwidth requirement back to or below the available bandwidth, wherein the applications (21-23) transmitting the data streams (31-36) participate in operation of an industrial facility (6) and / or in execution of at least one production process on the industrial facility (6), and the importance levels (31-36) are ascribed (112) to data streams (31-36) and / or to groups of data streams (31-36) based at least partly on importance of the data streams (31-36) for the operation of the facility (6) and / or for the execution of the process, and wherein: at least one data stream (31-36) comprises a critical video data stream which is required for continuing the operation of the facility (6) and / or the execution of the process at least in a degraded form, and which is a live video stream monitoring the execution of the process or being analyzed to assess quality or quality problems of manufactured products; and / or at least one data stream (31-36) comprises (113) measurements of a process quantity, and scaling back of the data stream comprises (121) extending intervals at which new measurements of the process quantity are transmitted; and / or at least one data stream (31-36) having measurements of a first process quantity is classified (113a) as a non-negotiable data stream (31) which is not to be scaled back or stopped, at least one data stream (31-36) having measurements of a second process quantity is classified (113b) as a scalable data stream (31-36) which can be scaled back to a given extent, and the first process quantity changes on a faster time scale than the second process quantity during execution of the process.

2. The method (100) of claim 1, wherein the first process quantity is pressure and the second process quantity is temperature.

3. The method (100) of any one of claims 1 or 2, wherein the ordered list (4) contains at least a first entry and a later second entry affecting the same data flow (31-36), wherein the first entry indicates that the data flow (31-36) is to be scaled down and the second entry indicates that the data flow (31-36) is to be further scaled down or to be stopped.

4. The method (100) of any one of claims 1 to 2, wherein the network management entity (la) is further configured to, in response to determining that the available bandwidth is higher than the combined bandwidth requirement and at least one data flow (31-36) is scaled down or stopped: unscale or restart at least one scaled down or stopped data flow (31-36), unscale or restart at least one scaled down or stopped data flow (31-36) does not result in exceeding the available bandwidth.

5. The method (100) of claim 4, further comprising: determine (130) a second ordered list (5) of data flows (31-36) based at least in part on the importance levels (31a-36a) and network capacity requirements (31b-36b) of the data flows (31-36), the second ordered list (5) indicating an order in which data flows (31-36) are to be un-scaled or to be restarted.

6. The method (100) of any one of claims 1 to 2, wherein the scaling down or stopping of at least one data flow (31-36) comprises one or more of: de-allocating resources intended to be used by the data flow (31-36); indicating a source of the data flow (31-36) to scale down bandwidth usage of the data flow (31-36) or to stop the data flow (31-36); indicating at least one network infrastructure device to throttle bandwidth usage of the data flow (31-36) or to block forwarding of the data flow (31-36); and indicating at least one network infrastructure device to compress and / or transcode the data flow (31-36).

7. The method (100) of any one of claims 1 to 2, wherein the determination (120) of the ordered list (4) of data flows (31-36) to be scaled down or to be stopped is negotiated (122) between multiple entities in a consistent manner and the result of the negotiation is provided (141) to the network management entity (la).

8. The method (100) of any one of claims 1 to 2, wherein the network management entity (la) is a software defined network, SDN, controller or a 5G network exposure interface.

9. A computer program product comprising machine readable instructions which, when executed on one or more computers, cause the one or more computers to perform the method (100) of any one of claims 1 to 8.

10. A non-transitory storage medium having machine-readable instructions contained in the computer program product of claim 9, which when executed perform the method of any one of claims 1 to 8.

11. One or more computers having the computer program product of claim 9 and / or having the non-transitory storage medium of claim 10, which when executing the computer program perform the method of any one of claims 1 to 8.

Images