WO2012167807A1

WO2012167807A1 - A method for controlling congestion within a network structure and a network structure

Info

Publication number: WO2012167807A1
Application number: PCT/EP2011/002849
Authority: WO
Inventors: Daniel CAMPS-MUR; Andreas Maeder
Original assignee: NEC Europe Ltd
Current assignee: NEC Europe Ltd
Priority date: 2011-06-10
Filing date: 2011-06-10
Publication date: 2012-12-13
Anticipated expiration: 2013-12-10

Abstract

For allowing a reduction of effects of congestion resulting in a high performance of a network structure a method for controlling congestion within a network structure is claimed, wherein the network structure is comprising a first network and a second network, wherein the first network is in communication with the second network via a gateway and a network link and wherein the first network is comprising a plurality of metering devices providing information for transmission via the gateway and the network link to the second network. The method is characterized in that the gateway is scheduling the transmission of the information from the metering devices for smoothing in time a load generated by the transmission of the information. Further, an according network structure is claimed, preferably for carrying out the above mentioned method.

Description

A METHOD FOR CONTROLLING CONGESTION WITHIN A NETWORK STRUCTURE AND A NETWORK STRUCTURE

The present invention relates to a method for controlling congestion within a network structure, wherein the network structure is comprising a first network and a second network, wherein the first network is in communication with the second network via a gateway and a network link and wherein the first network is comprising a plurality of metering devices providing information for transmission via the gateway and the network link to the second network.

Further, the present invention relates to a network structure, preferably for carrying out the above method for controlling congestion within a network structure, wherein the network structure is comprising a first network and a second network, wherein the first network is in communication with the second network via a gateway and a network link and wherein the first network is comprising a plurality of metering devices providing information for transmission via the gateway and the network link to the second network.

A method for controlling congestion within a network structure and an according network structure comprising the above mentioned features are known for example from US 7,929,430 B2. This document is showing a congestion control access gateway which interfaces between a broadband convergence network and a sensor network including sensors as metering devices. The gateway receives data from the sensor network. When a congestion situation has occurred, the congestion control access gateway detects the congestion situation based on data received in the congestion situation and broadcasts congestion notification and control messages in response to the congestion situation detection. The congestion control access gateway generates a congestion situation detection profile as a result of the congestion situation detection and performs priority queuing of the received data based on the congestion situation detection profile and service classes. The congestion control access gateway determines destination of the queued data using the congestion situation detection profile and routes the data to a corresponding service. Further, from IEEE, "IEEE 802.16p-10/0005, Machine to Machine (M2M) Communications Technical Report", Nov. 2010 is known that mobile communication systems have been identified as key candidates to implement the Advanced Metering Infrastructure (AMI) required to enable upcoming M2M applications. However, the transport of M2M traffic through cellular networks poses significant challenges that will require careful designs in order to optimize the use of the scarce radio resources.

M2M applications like smart metering transmit relatively short messages at usually long time intervals that result in a fairly low data rate per metering device. However, it is the huge amount of such metering devices - potentially in the order of several thousand within a cell - that may stress the capabilities of cellular networks. In addition, notice that in a cellular network M2M devices share resources with traditional users, e.g. data intensive smartphone applications, which may be adversely affected if the bandwidth taken by M2M communications is too high.

A solution being recently studied is that in order to avoid costly technology updates or buying new spectrum licenses, unlicensed spectrum can be used to aid M2M communications. In particular, sub-1 Ghz spectrum, like TV white spaces (TVWS) or the 900Mhz ISM (Industrial, Scientific and Medical) band, have been recently proposed as an alternative to established wireless communication networks, for reference see IEEE 802.11af, http://www.ieee802.org/11/Reports/tgaf_update.htm and IEEE 802.11 ah, http://stevencrowley.com/2010/09/30/ieee-standards-board- approves-sub-1 -ghz-802-11 -wi-fi-project/. TV white spaces are unlicensed spectrum bands in the sub-GHz band which were formerly occupied by local analogue TV operators. Due to the relatively low carrier frequencies and the resulting lower propagation losses compared to higher frequency bands, sub-1 Ghz bands enable extended coverage ranges in the order of 1.5Km² in the case of Wi- Fi, and provide a good coverage even with low powered devices and outdoor-to- indoor scenarios. One of the most promising and advanced candidates for sub- 1 Ghz band communication are the IEEE 802.11af standard, which will standardize the operation of Wi-Fi in the TVWS band, and the 802.11 ah standard, which will standardize the operation of Wi-Fi in the 900Mhz ISM band. Fig. 1 illustrates an example of the M2M architecture considered in this invention which is in agreement with architectures proposed by several standardization bodies, for reference see IEEE, "IEEE 802.16p- 10/0005, Machine to Machine (M2M) Communications Technical Report", Nov. 2010. In this case a first network is realized by an access network in the form of a Wi-Fi sub-1 Ghz network ^~1.5Km² coverage. A second network is realized by a cellular network which is in communication with M2M services via an internet connection. The first and the second network are connected via an M2M gateway. Within such a two tier M2M deployment an M2M gateway uses unlicensed spectrum (Wi-Fi TVWS) to collect data from metering devices. The M2M gateway uses a cellular network to provide access to M2M services.

As illustrated in Fig. 1 , an M2M gateway collects data from meters under its coverage using unlicensed spectrum and injects the collected data into the cellular network. However, given the scarce radio resources in the cellular network and the huge potential number of metering devices under each M2M gateway, the cellular link may suffer congestion when transporting M2M traffic.

In order to get a closer view at the previous problem we estimate a typical aggregated M2M load that each M2M gateway would inject into the cellular network in the following way:

Expected number of metering devices in a typical European city (London). Data obtained from IEEE, "IEEE 802.16p-10/0005, Machine to Machine (M2M) Communications Technical Report", Nov. 2010:

o -8,900 in a 1 Km² area,

o -35,000 in a 2Km² area.

Example of typical data rate generated by a metering device: 2,000 bytes every 10 minutes (~27bps).

o Notice that this load is only an assumption, as of today and given that potential M2M applications are not yet known, it is difficult to estimate how M2M traffic patterns will look like. Thus, the M2M gateway in Fig. 1 may have to carry an uplink aggregate traffic of:

o ~ 8,900meters x 2,000bytes/1 Omts ~ 240 Kbps

o ~ 35,000meters x 2,000bytes/1 Omts ~ 1 Mbps

Notice that cellular networks are usually dimensioned to provide higher data rates in the downlink than in the uplink direction, since traditional data traffic access patterns are asymmetric in downlink. M2M traffic though, will mainly generate upstream traffic from the meters to the infrastructure that can stress the uplink capacity of cellular systems. Notice that typical uplink peak capacities in cellular networks, which have to be shared among all devices in a cell, are 384Kbps in UMTS (Universal Mobile Telecommunications System) and a varying capacity of up to several Mbps in HSPA (High Speed Packet Access), WiMAX (Worldwide Interoperability for Microwave Access) and LTE (Long Term Evolution).

In order to address the congestion problem from 3GPP, "TR 23.888 V1.0.0, System Improvements for Machine-Type Communications", Jul. 2007 are known solutions on network level. Further, US 7,929,430 B2 is discussing a method to relieve congestion by means of priority random backoff and priority queuing within the gateway.

It is an object of the present invention to improve and further develop a method for controlling congestion within a network structure and an according network structure for allowing a reduction of effects of congestion resulting in high performance of the network structure.

In accordance with the invention the aforementioned object is accomplished by a method comprising the features of claim 1 and by a network structure comprising the features of claim 24.

According to claim 1 , the method is characterized in that the gateway is scheduling the transmission of the information from the metering devices for smoothing in time a load generated by the transmission of the information. According to claim 24 the network structure is characterized in that the gateway is comprising means for scheduling the transmission of the information from the metering devices for smoothing in time a load generated by the transmission of the information.

According to the invention it has been recognized that it is possible to allow for a reduction of effects of congestion by an appropriate activity of the gateway. Concretely and in further inventive manner the gateway is scheduling the transmission of the information from the metering devices in order to smooth in time a load which is generated by the transmission of the information. In other words, the gateway could define a time schedule for the transmission of the information so that congestion can be avoided or reduced beginning from the transmission of the information. Such a scheduling procedure could influence the provision of the information within the metering devices or the transmission start after the provision of the information within the metering devices. Anyway a reduction of congestion and thus a reduction of effects of congestion can be realized. This will result in a high performance of the according network structure.

Within a preferred embodiment the gateway could schedule the transmission by means of controlling the metering devices in such a way that the metering devices are providing the information only at predefinable time instants and/or time intervals. By such a scheduling procedure a very sophisticated transmission structure or pattern could be realized for reducing or avoiding congestion situations.

With regard to a long-term reduction of the effects of congestion the metering devices could provide the information at periodic time intervals.

With regard to a very effective and simple scheduling procedure the controlling of the metering devices by the gateway could be performed by a communication between the gateway and the metering devices at a MAC (Media Access Control) layer or layer 2. Such a communication at a MAC layer or layer 2 is providing a simple transport of information between the gateway and the metering devices without further nodes or links between the gateway and the metering devices.

For further simplifying the communication between the gateway and the metering devices the communication could be performed by means of a centralized protocol or by means of a signaling protocol or a signaling message. In other words, the transmission of the scheduling information could be involved within a signaling message for avoiding any further traffic within the first network. Thus, any additional load within the network structure can be avoided for keeping effects of congestion low.

Within a preferred embodiment the controlling of the metering devices by the gateway could comprise a transmission of a start time or a start time and a time interval value to at least one metering device for providing the information. Based on such concrete information a simple transmission pattern could be realized for reducing congestion. Having received a start time or a start time and a time interval value the metering device could transmit the requested metering information periodically after respective time intervals until the respective metering device receives different information from the gateway.

For providing a very efficient transmission pattern with regard to the reduction of congestion different metering devices could get transmitted different start times and/or time interval values by the gateway. For further refinement and efficiency the controlling of the metering devices by the gateway could comprise a definition of the start time or start time and time interval value taking into account one or several congestion-related metrics or parameters of the network link and/or of the first network and/or of an application layer and/or of the second network. Such metrics or parameters could comprise a link quality and/or data rate and/or queue size of the network link and/or of the second network. Alternatively, the metrics or parameters could comprise information about characteristics of application layer traffic conveyed to the gateway and/or information about expected congestion at the first network and/or at the second network. By taking into account such metrics or parameters an adaptation of the scheduling procedure to different demands and circumstances is possible.

Preferably, the controlling of the metering devices by the gateway could comprise a definition of the start time or start time and time interval value in such a way that a definable QoS (Quality of Service) metric of a metering device application is fulfilled. Thus, an adaptation of the scheduling procedure to application requirements is possible.

The preferred example of QoS metric could be a delay requirement or a packet drop rate.

For considering various application or user requirements at least one metering device could be prioritized within the scheduling procedure. Preferably, at least one definable metering device could be prioritized on the MAC layer of the first network according to a QoS metric. Alternatively, at least one definable metering device could be prioritized in the network link according to a QoS metric.

With regard to an efficient adaptation of the scheduling procedure to actual situations or requirements an adjustment of the scheduling pattern could be realized. Within a preferred embodiment, when congestion in the network link and/or in the second network is high, the gateway could adjust the start time or start time and time interval value of at least one metering device for increasing a spread in time of load coming from the metering devices. In a contrary situation, when congestion in the network link and/or in the second network is low, the gateway could adjust the start time or start time and time interval value of at least one metering device for decreasing a spread in time of load coming from the metering devices.

An important advantage of the method and network structure according to the invention is the proactive scheduling of the transmission of the information from the metering devices. Thus, within a preferred embodiment, the transmission of the start time or start time and time interval value could be performed proactively before congestion occurs, preferably based on a priori information such as application-level traffic patterns or metrics related to the quality of the network link and/or of the second network.

Within a further preferred concrete embodiment the gateway could transmit the start time or start time and time interval value to an application running on the respective metering device, so that the application provides the information or generates traffic only at the predefinable time instants and/or time intervals. In other words, the transmission of start time and/or time interval values could be performed to the respective metering device and/or to an application running on the respective metering device. The respective addressee could depend on the individual situation or individual user requirements.

Within a concrete and preferred embodiment the first network could be a sub- 1 Ghz network or sub-1Ghz Wi-Fi network. Further, the second network could be a cellular network and the gateway could be an M2M gateway. Thus, an M2M architecture could by realized for implementing AMI.

According to the present invention a method for scheduling the data transmissions from its associated metering devices by or within an M2M gateway is provided, in order to minimize the effects of congestion in a cellular link, for example. By means of the present invention an M2M gateway could smooth in time the load at a backhaul link generated by its associated metering devices. Within a preferred embodiment is proposed a solution on MAC level of air interface in order to optimize the buffer occupancy at the backhaul link.

Within the present invention MAC layer control mechanisms can be used as the primary method to relieve congestion. These actions can have permanent, i.e. longer than the current situation, impact on the congestion of the backhaul link. The present invention can further provide long-term time interval allocations to relieve congestion for longer periods. Contrary, known methods use one-time random backoff as a method to relieve only situational congestion. The method according to the present invention is particularly appropriate for periodic traffic patterns which are known from smart metering applications, for example.

Contrary to known methods, which are operating reactively after a congestion situation is detected, the present invention can be applied proactively to avoid congestion in beforehand, e.g. by taking cellular link and/or application level information into account.

For better understanding and summarizing important aspects of the present invention the following advantageous features are reviewed. The present invention could propose a method for an M2M gateway to proactively smooth the cumulated load of its associated devices in order to alleviate the effects of congestion in the cellular network.

The method could be performed with regard to an M2M gateway operating in a two tier M2M architecture to relieve congestion in its uplink tier 1 link, e.g. the cellular link. Details could be as follows:

The M2M gateway is used to connect a tier 1 network (access network), where the metering devices operate, with a tier 2 network (such as a cellular network).

Metering devices are controlled by the M2M gateway, in such a way that they can transmit in the access network only at certain periodic intervals, i.e. meterj may transmit in the network at times T_i = start_time_i + interval_i. Both start_time_i and intervaU are decided by the M2M gateway and signaled at MAC layer to the metering devices by means of a signaling protocol.

The M2M gateway selects start_time_i and intervaU in order to spread in time the load coming from its associated metering devices in the access network, hence achieving a reduction of the congestion in the cellular link. The M2M GW (gateway) performs the adjustment of start_time and interval values taking into account one or several congestion-related metrics or parameters of the cellular link, the access network or the application layer, including but not limited to

- link quality or data rate of the cellular link

- queue size of the cellular link

- information about the application-layer traffic characteristics conveyed to the M2M GW

- expected congestion on the access network.

The M2M GW performs said adjustments based on said parameters such that certain QoS-metrics of the M2M device applications like delay requirements or packet drop rates are fulfilled.

The metering devices could additionally be prioritized on the MAC layer of the access network according to QoS metrics. In other words, the interval values could be longer for definable metering devices as the interval values for other metering devices.

Further, the metering devices could additionally be prioritized in the cellular link according to QoS metrics, for example by priority queuing.

An important aspect of the invention is to realize that load smoothing is beneficial in a two-tier M2M scenario in order to improve the performance, in terms of reliability and delay, of upcoming M2M applications.

The present invention can be implemented for example as follows:

1. The M2M gateway can communicate the start_time and interval parameters to the application running on the metering device, so that the application only generates traffic at the designated time instants.

2. The proposed functionality can be implemented at layer 2 in order to be application agnostic. For instance the M2M gateway can use centralized protocols already available in Wi-Fi, like S-APSD and HCCA (Perez-Costa, X.; Camps-Mur, D., "IEEE 802.11 E QoS and power saving features overview and analysis of combined performance," Wireless Communications, IEEE , vol.17, no.4, pp.88-96, August 2010) or Wi-Fi Direct Notice of Absence (Wi-Fi Peer-to-Peer (P2P) Specification v1.1 , https://www.wi-fi.org/knowledge_center_overview.php?type=4 ), in order to allow transmissions in the MAC layer only at the designated time instants. Notice that all the previous protocols already allow a Wi-Fi Access Point to convey a schedule in the form of <start_time, interval> to its associated stations.

The signaling overhead incurred by the proposed mechanism is not significant because the required parameters need to be conveyed by the M2M gateway to each metering device only when this device enters the network. Since other network entry functions must be executed as well, the required information can be efficiently piggy-backed into other signaling messages. In addition, unlike the cellular network, the access network used to connect the metering devices to the M2M gateway, i.e. Wi-Fi sub-1 Ghz in Fig. 1 , should have enough available capacity to carry any excess of signaling load.

Since in order to smooth load the M2M gateway can delay the transmission of some metering devices, a more advanced implementation of the proposed method could prioritize metering devices in order to delay first metering devices carrying low priority traffic.

Finally, an even more advanced implementation could adaptively smooth the load coming from the metering devices depending on the capacity available in the cellular network. For instance when the cellular bandwidth is shared with some data intensive user like a smartphone, the M2M gateway would observe a reduction on its available cellular bandwidth and would increase the amount of load smoothing on the metering devices. On the other hand, when the available bandwidth in the cellular link increases, the M2M gateway would decrease the amount of load smoothing on the metering devices, letting them benefit from reduced latencies.

There are several ways how to design and further develop the teaching of the present invention in an advantageous way. To this end, it is to be referred to the patent claims subordinate to patent claim 1 on the one hand and to the following explanation of preferred examples of embodiments of the invention, illustrated by the drawing on the other hand. In connection with the explanation of the preferred embodiments of the invention by the aid of the drawing, generally preferred embodiments and further developments of the teaching will be explained. In the drawing

Fig. 1 is showing an example of a typical M2M architecture,

Fig. 2 is illustrating a random load pattern and a smoothed load pattern according to the invention,

Fig. 3 is illustrating delay and reliability requirements of typical M2M applications,

Fig. 4 is illustrating the result of a simulation illustrating the buffer evolution for different scenarios and

Fig. 5 is illustrating the effect of a load smoothing in time.

Fig. 1 is showing a typical example of an M2M architecture which is in agreement with architectures proposed by several standardization bodies. A first network 1 is connected to a second network 2 via a gateway 3. Concretely, the illustrated architecture is a two tier M2M deployment wherein an M2M gateway 3 uses an unlicensed spectrum (Wi-Fi TVWS) to collect data from meters 5. The M2M gateway 3 uses a cellular network 2 to provide access to M2M services. Congestion on the cellular link 4 carrying M2M traffic is usually reducing the performance of the M2M architecture.

Given that M2M applications are not expected to have very tight delay constraints, typically in the order of minutes, an M2M gateway can alleviate temporary congestion in the cellular network by smoothing in time the load coming from its associated meters. Fig. 2 is illustrating an example of such load smoothing. Obviously, one could also absorb the excess load generated by the metering devices by simply having big buffers in the M2M gateway. However, the high reliability requirements of M2M applications, typically in the order of 99.99% reliability as illustrated in Fig. 2, render this approach unpractical. The reason is that the random nature of the traffic generated by the metering devices, plus the random nature of the capacity available in the cellular network, makes it nearly impossible to dimension the buffers in the M2M gateway in order to guarantee close to zero losses as required in M2M.

Fig. 3 is illustrating delay and reliability requirements of typical M2M applications.

Achieving reliability by means of retransmissions in the higher layers, e.g. using a connection oriented protocol like TCP (Transmission Control Protocol), poses also significant challenges due to the enormous number of TCP connections, one per metering device, that should be maintained, and the low duty cycle of the metering devices, i.e. one transmission in the order of minutes or longer.

Furthermore, large buffer occupations induced by TCP would lead to a significant increase of delay for other, non-M2M devices in the system, which is generally not beneficial for the performance of most applications. In addition, TCP reacts malicious on a sharp increase in latency with spurious timeouts, which may lead to several seconds of transmission gaps. This effect is known as buffer bloating and currently draws an increasing interest from the research community, see e.g. Bufferbloat, http://gettys.wordpress.com/.

According to a preferred embodiment smoothing in 802.11 -based systems can be realized.

A possible embodiment of the present invention is the following:

1 ) A two-tier M2M system like the one illustrated in Fig. 1 is deployed.

2) A sub-1Ghz Wi-Fi technology is used as access network 1 for the metering devices 5. 3) The M2M Wi-Fi gateway 3 and the metering devices 5 implement Scheduled- APSD. This is a power saving protocol that consists in an initial frame exchange between a metering device and the M2M gateway, where the M2M gateway notifies the metering device about an initial starting time, SST_i, and an interval, T_i. Thereafter, the metering device wakes up to transmit and receive data only at times t_i (k) = SST_i + k^*T_i.

4) The M2M gateway runs a smoothing algorithm that decides which are the proper starting times for each metering device. This algorithm may take as input parameters: i) The already scheduled metering devices, ii) the QoS/energy requirements of the metering device to be scheduled.

Performance Evaluation

In order to provide an initial evaluation of the method proposed in this invention, we have simulated a simplified version of the system described in Fig. 1 , with the following characteristics from IEEE, "IEEE 802.16p-10/0005, Machine to Machine (M2M) Communications Technical Report", Nov. 2010 adapted to a sub-1 Ghz Wi- Fi network:

The traffic model assumes a packet size uniformly distributed between 3,200 and 4,080 bytes. The meters send in regular intervals of 180s, 360s or 540s. The backhaul link is assumed to be 3G UMTS with an UL (Uplink) dedicated channel of 128kbps and a buffer size of 1 MByte.

In order to illustrate the benefits of the smoothing scheme, three different scenarios are considered: Randomized: Metering devices transmit at regular intervals but the M2M gateway randomly selects start_time_i for each metering device.

Non-smoothed: Metering devices transmit at regular intervals but start all transmitting at the same time. Notice that this is a meaningful example because in many applications metering devices are expected to send traffic simultaneously when a trigger or alarm fires.

Smoothed: Metering devices transmit at regular intervals and the M2M gateway applies an algorithm that selects the start time of each device in order to spread its load.

Fig. 4 shows the buffer evolution for the different schemes under study. It can be observed that the non-smoothed scenario is a worst case where the peak buffer occupancy vastly exceeds the other cases. However, even with randomized smoothing the peak buffer occupancy is significantly higher than in the case of the smoothed approach.

This is further illustrated in Fig. 5, which shows the complementary cumulative distribution function of the queuing and transmission message delay. While the non-smoothed scenario shows that an implementation without smoothing is clearly not feasible, the comparison between the randomized and smoothed approach reveals a difference of approx. 4s for the smoothed case. Note that a packet delay higher than 5s, as observed regularly in randomized case, could already lead to spurious TCP timeouts if there is any additional traffic on the access point.

Besides the gains in delay, smoothing also leads to reduced energy consumption at the device side due to less contention and therefore less frame retransmissions. However, we have not yet evaluated this gain.

The present invention explains pro-active adjustment of starting times of M2M communication devices with regular interval traffic patterns such that inter-arrival- times are maximized. Preferably, an air interface signaling at the first tier (WiFi) can be used to optimize buffer delays at the second tier (backhaul link). It is provided an adaptive smoothing according to backhaul link buffer occupancy. The metering devices could communicate only at designated time instants specified by the M2M gateway in a way that the aggregated load is smoothed in time. The M2M gateway, which has an overall overview of the network, could explicitly signal the appropriate transmission times to each metering device.

The result is an improved reliability in M2M applications. These applications may carry highly sensitive traffic. Further, an improved energy consumption in the metering devices can be provided. Further, a reduction of congestion for non-M2M traffic sharing the same gateway can be provided.

Many modifications and other embodiments of the invention set forth herein will come to mind the one skilled in the art to which the invention pertains having the benefit of the teachings presented in the foregoing description and the associated drawings. Therefore, it is to be understood that the invention is not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of the appended claims. Although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.

Claims

C l a i m s

1. A method for controlling congestion within a network structure, wherein the network structure is comprising a first network (1 ) and a second network (2), wherein the first network (1) is in communication with the second network (2) via a gateway (3) and a network link (4) and wherein the first network (1 ) is comprising a plurality of metering devices (5) providing information for transmission via the gateway (3) and the network link (4) to the second network (2),

c h a r a c t e r i z e d in that the gateway (3) is scheduling the transmission of the information from the metering devices (5) for smoothing in time a load generated by the transmission of the information.

2. A method according to claim 1 , wherein the gateway (3) is scheduling the transmission by means of controlling the metering devices (5) in such a way that the metering devices (5) are providing the information only at predefinable time instants and/or time intervals.

3. A method according to claim 2, wherein the metering devices (5) are providing the information at periodic time intervals.

4. A method according to claim 2 or 3, wherein the controlling of the metering devices (5) by the gateway (3) will be performed by a communication between the gateway (3) and the metering devices (5) at a MAC (Media Access Control) layer or layer 2.

5. A method according to claim 4, wherein the communication will be performed by means of a centralized protocol.

6. A method according to claim 4 or 5, wherein the communication will be performed by means of a signalling protocol or a signalling message.

7. A method according to one of claims 2 to 6, wherein the controlling of the metering devices (5) by the gateway (3) is comprising a transmission of a start time or a start time and a time interval value to at least one metering device for providing the information.

8. A method according to claim 7, wherein different metering devices (5) will get transmitted different start times and/or time interval values by the gateway (3).

9. A method according to one of claims 2 to 8, wherein the controlling of the metering devices (5) by the gateway (3) is comprising a definition of the start time or start time and time interval value taking into account one or several congestion- related metrics or parameters of the network link (4) and/or of the first network (1) and/or of an application layer and/or of the second network (2).

10. A method according to claim 9, wherein the metrics or parameters are comprising a link quality and/or data rate and/or queue size of the network link (4) and/or of the second network (2).

11. A method according to claim 9 or 10, wherein the metrics or parameters are comprising information about characteristics of application layer traffic conveyed to the gateway (3).

12. A method according to one of claims 9 to 11 , wherein the metrics or parameters are comprising information about expected congestion at the first network (1) and/or at the second network (2).

13. A method according to one of claims 2 to 12, wherein the controlling of the metering devices (5) by the gateway (3) is comprising a definition of the start time or start time and time interval value in such a way that a definable QoS (Quality of Service) metric of a metering device application is fulfilled.

14. A method according to claim 13, wherein the QoS metric is a delay requirement or a packet drop rate.

15. A method according to one of claims 1 to 14, wherein at least one definable metering device will be prioritized on the MAC layer of the first network (1) according to a QoS metric.

16. A method according to one of claims 1 to 15, wherein at least one definable metering device will be prioritized in the network link (4) according to a QoS metric.

17. A method according to one of claims 7 to 16, wherein when congestion in the network link (4) and/or the second network (2) is high, the gateway (3) adjusts the start time or start time and time interval value of at least one metering device for increasing a spread in time of load coming from the metering devices (5).

18. A method according to one of claims 7 to 17, wherein when congestion in the network link (4) and/or the second network (2) is low, the gateway (3) adjusts the start time or start time and time interval value of at least one metering device for decreasing a spread in time of load coming from the metering devices (5).

19. A method according to one of claims 7 to 18, wherein the transmission of the start time or start time and time interval value will be performed proactively before congestion occurs, preferably based on a priori information such as application-level traffic patterns or metrics related to the quality of the network link (4) and/or the second network (2).

20. A method according to one of claims 7 to 19, wherein the gateway (3) transmits the start time or start time and time interval value to an application running on the respective metering device, so that the application provides the information or generates traffic only at the predefinable time instants and/or time intervals.

21. A method according to one of claims 1 to 20, wherein the first network (1) is a sub-1Ghz network or sub-1Ghz Wi-Fi network.

22. A method according to one of claims 1 to 21 , wherein the second network (2) is a cellular network.

23. A method according to one of claims 1 to 22, wherein the gateway (3) is an M2M gateway (3).

24. A network structure, preferably for carrying out the method for controlling congestion within a network structure according to any one of claims 1 to 23, wherein the network structure is comprising a first network (1 ) and a second network (2), wherein the first network (1) is in communication with the second network (2) via a gateway (3) and a network link (4) and wherein the first network

(1) is comprising a plurality of metering devices (5) providing information for transmission via the gateway (3) and the network link (4) to the second network

(2) ,

c h a r a c t e r i z e d in that the gateway (3) is comprising means for scheduling the transmission of the information from the metering devices (5) for smoothing in time a load generated by the transmission of the information.