A DISTRIBUTED NETWORK TRAFFIC DATA DECOMPOSITION METHOD

Abstract

To be able to adequately provide desired services over a 5G mobile service network, the 5G communication infrastructures requires a much-improved flexibility in resource management. Network operators are foreseen to deploy network slicing, by isolating dedicated resources and providing customised logical instances of the physical infrastructure to each service. A critical operation in performing management and orchestration of network resources is the anticipatory provisioning of isolated capacity to each network slice. Accordingly, it is necessary to obtain an estimate of service level demands. However, the estimation of such service level demands is typically obtained via deep packet inspection, which is a resource intensive and time-consuming process. Therefore, it is typically not possible to provide updated accurate estimates at a frequency suitable for use in accurate prediction of a future per-service traffic consumption, without an undesirable level of computational and time resources being required. The present invention provides a distributed network traffic data decomposition method which makes use of a neural network to provide an accurate future per-service traffic consumption prediction without deep-packet inspection or another resource intensive analysis method.

Claims

1. A distributed network traffic data decomposition method comprising the steps of: receiving input data comprising aggregate network traffic data from a plurality of distributed source locations, wherein the aggregate data includes traffic data corresponding to a plurality of services operating over the network; converting the input data into a format suitable for further analysis by re-arranging and mapping the locations of the plurality of source locations such that the source locations are arranged in a regular grid pattern and separating the aggregate network traffic data into a time-dependent sequence of snapshots; analysing the converted data with a neural network, comprising a plurality of neural layers, to extract, in a final neural layer of the plurality of neural layers, a plurality of outputs from the converted data, wherein each output corresponds to decomposed traffic volume of one service of the plurality of services operating over the network; and employing 2D convolutions to extract a plurality of outputs from the determined spatiotemporal correlations; and predicting, based on the plurality of outputs, a future per-service traffic consumption.

2. The distributed network traffic data decomposition method of claim 1, wherein the step of analysing the converted data with a neural network includes: determining a fraction of traffic that belongs to each of the plurality of services at each of the plurality of sources; employing 3D deformable convolutions to: at least partially mitigate spatial displacements introduced during the conversion of the input data; and determine at least one intermediate output from the converted data; determining spatiotemporal correlations from the determined at least one intermediate output; and employing 2D convolutions to extract a plurality of outputs from the determined spatiotemporal correlations.

3. The distributed network traffic data decomposition method of claim 1, further comprising the step of retraining the neural network by: comparing per-service traffic consumption measured at a predetermined time with a predicted per-service traffic consumption corresponding to said predetermined time; determining a prediction error by calculating a difference between the measured per-service traffic consumption and the predicted per-service traffic consumption; and amending at least one function of the neural network such that the prediction error is reduced.

4. The distributed network traffic data decomposition method of claim 3, wherein the step of amending at least one function of the neural network such that the prediction error is reduced is only undertaken if the prediction error is above a predetermined prediction error threshold value.

5. The distributed network traffic data decomposition method of claim 1, further comprising the step of retraining the neural network by: calculating a maximum likelihood estimation between a measured per-service traffic consumption for at least one of the plurality of distributed source locations and for at least one of the plurality of services operating over the network and the predicted future per-service traffic consumption; and amending at least one function of the neural network to increase the maximum likelihood estimation.

6. The distributed network traffic data decomposition method of claim 5, wherein the step of amending at least one function of the neural network to increase the maximum likelihood estimation is only undertaken if the calculated maximum likelihood estimation is below a predetermined maximum likelihood estimation threshold value.

7. The distributed network traffic data decomposition method of claim 5, wherein the step of retraining the neural network comprises calculating a maximum likelihood estimation between a measured per-service traffic consumption for each of the plurality of distributed source locations and for each of the plurality of services operating over the network and the predicted future per-service traffic consumption.

8. The distributed network traffic data decomposition method of claim 3, wherein the step of retraining the neural network is based upon a subset of input data collected at only a portion of the plurality of distributed source locations.

9. The distributed network traffic data decomposition method of claim 3, wherein the neural network is trained with a cross-entropy function.

10. The distributed network traffic data decomposition method of claim 1, wherein the measured per-service mobile traffic data is obtained from deep packet inspection.

11. The distributed network traffic data decomposition method of claim 1, wherein the step of employing 3D deformable convolutions to at least partially mitigate spatial displacements introduced during the conversion of the input data includes rearranging the converted data.

12. The distributed network traffic data decomposition method of claim 1, further comprising the step of allocating network resources based on the predicted future service-wise traffic consumption.

13. The distributed network traffic data decomposition method of claim 12, further comprising the step of reallocating network resources based on a further predicted future service-wise traffic consumption.

14. The distributed network traffic data decomposition method of claim 1, wherein the aggregate network traffic data is encrypted.

15. The distributed network traffic data decomposition method of claim 1, further comprising the step of collecting the input data.

16. The distributed network traffic data decomposition method of claim 15, wherein the step of collecting the input data comprises data collection without deep packet inspection.

17. The distributed network traffic data decomposition method of claim 1, further comprising the step of performing adaptive weighting by assigning a weight to at least one snapshot, wherein the weight applied is dependent on a time of capture of data included in said snapshot.

18. The distributed network traffic data decomposition method of claim 17, wherein a first snapshot captured at a first time is assigned a different weight when compared to a second snapshot captured at a second time, wherein the second time is more recent than the first time.

19. The distributed network traffic data decomposition method of claim 1, wherein the plurality of outputs extracted from the converted data comprises a plurality of feature maps, wherein each feature map corresponds to decomposed traffic volume of one service of the plurality of services operating over the network, and wherein the prediction of a future per-service traffic consumption is based on the plurality of feature maps.

20. The distributed network traffic data decomposition method of claim 1, wherein the network is one of a WiFi network, a mobile telecoms service network, a broadband network, an Internet of Things sensor and actuator network, a distributed electricity distribution grid, a network of electricity consumption sensors, roadways, airways, shipping lanes, a network of air quality sensors, a network of household water or gas consumption meters, or a social network.

21. The distributed network traffic data decomposition method of claim 1, wherein the step of converting the input data into a format suitable for further analysis by re-arranging and mapping the locations of the plurality of source locations such that the source locations are arranged in a regular grid pattern comprises constructing a regular grid including a number of grid points equal to the number of the plurality of distributed source locations, and performing a one-to-one source location to grid point association, such that a single grid point relates to a single source location.

22. The distributed network traffic data decomposition method of claim 21, wherein the one-to-one source location to grid point association is performed in such a manner as to minimise an average spatial displacement of a portion or all of the source locations when they are associated to a respective grid point.

23. The distributed network traffic data decomposition method of claim 21, wherein the one-to-one source location to grid point association is performed using the Hungarian Algorithm.

24. The distributed network traffic data decomposition method of claim 1, wherein the at least one intermediate output comprises at least one spatiotemporal pattern.

25. An apparatus comprising at least one processor, wherein the at least one processor is configured to be operable to execute the method comprising: receiving input data comprising aggregate network traffic data from a plurality of distributed source locations, wherein the aggregate data includes traffic data corresponding to a plurality of services operating over the network; converting the input data into a format suitable for further analysis by re-arranging and mapping the locations of the plurality of source locations such that the source locations are arranged in a regular grid pattern and separating the aggregate network traffic data into a time-dependent sequence of snapshots; analysing the converted data with a neural network, comprising a plurality of neural layers, to extract, in a final neural layer of the plurality of neural layers, a plurality of outputs from the converted data, wherein each output corresponds to decomposed traffic volume of one service of the plurality of services operating over the network; and employing 2D convolutions to extract a plurality of outputs from the determined spatiotemporal correlations; and predicting, based on the plurality of outputs, a future per-service traffic consumption.

Description

DETAILED DESCRIPTION

[0040] FIG. 1 is a schematic diagram outlining the steps of a method of decomposing data.

[0041] FIG. 1 is a schematic diagram 100 outlining the steps of a method of decomposing data. The first step 110 is to receive input data. Alternatively, the method may include a step prior to the first step 110 of collecting the input data from the distributed source locations. The input data typically comprises aggregate network traffic data collected from a plurality of distributed source locations. The aggregate data typically includes traffic data corresponding to a plurality of services operating over the network. The network may be a mobile service network, such as a 4G or 5G network. The aggregate data may therefore comprise all data moving about the network, and may include traffic data corresponding to a plurality of services operating over the mobile service network, such as mobile gaming and video streaming, among others.

[0042] The second step 120 is to convert the input data into a format suitable for further analysis. The conversion typically involves re-arranging and mapping the locations of the plurality of source locations such that the source locations are arranged in a regular grid pattern. The source locations may be the locations of a plurality of antennae which form part of the physical infrastructure of the mobile service network. The antennae may be distributed over an urban area, and may be separated by distances in the order or metres or kilometres. It is clear that the positions of the antennae will not be in a regular, grid-like, arrangement because their positioning is based on local demand and topography. Accordingly, to convert the input data into a regular, grid-like, arrangement suitable for further processing, it is clear that the positions must be rearranged. The conversion also typically involves separating the aggregate network traffic data into a time-dependent sequence of snapshots. Accordingly, the data may be given a timestamp and organised into a sequential order.

[0043] The third step 130 is to analyse the converted data with a neural network. The neural network typically includes a plurality of neural layers. The analysis is typically used to extract, in a final neural layer of the plurality of neural layers, a plurality of outputs from the converted data. Each output typically corresponds to decomposed traffic volume of one service of the plurality of services operating over the network. For example, each output may correspond to one service operating over the mobile service network. As such, a single output may be extracted for each of the services, such as mobile gaming and video streaming.

[0044] The analysis at the third step 130 typically includes several analysis steps, which may be carried out in any order.

[0045] One analysis step is to determine a fraction of traffic that belongs to each of the plurality of services at each of the plurality of sources. Accordingly, a fraction of traffic belonging to each of the plurality of services for the past collected data may be determined. For example, the fraction of the aggregate data corresponding to each of the plurality of services in a timeframe immediately preceding the present time may be determined.

[0046] Another analysis step is to employ 3D deformable convolutions to at least partially mitigate spatial displacements introduced during the conversion of the input data and determine at least one intermediate output from the converted data. The at least one intermediate output may be abstract and may not have any significance without being processed further.

[0047] Another analysis step is to determine spatiotemporal correlations from the determined at least one intermediate output. 2D convolutions may then be employed to extract a plurality of outputs from the determined spatiotemporal correlations.

[0048] The fourth step 140 is typically predicting, based on the plurality of outputs obtained via the analysis in the third step 130, a future per-service traffic consumption. The predicted future per-service traffic consumption may reflect an expected future traffic in or on the network.

[0049] Although the network described herein is a mobile service network, it is to be understood that the method described herein is equally applicable to other networks, such as a WiFi network, a mobile telecoms service network, a broadband network, an Internet of Things sensor and actuator network, a distributed electricity distribution grid, a network of electricity consumption sensors, roadways, airways, shipping lanes, a network of air quality sensors, a network of household water or gas consumption meters, or a social network. Furthermore, any or each of the other steps described herein may be incorporated into the method. In particular, the predicted future per-service traffic consumption may be used to allocate network resources. It is also to be understood that alternative analysis methods, other than a neural network, 3D deformable convolutions and 2D convolutions as described herein, may be provided. Additionally, it is to be understood that the neural network, 3D deformable convolutions and 2D convolutions as described herein may function or operate in a different manner yet still output and/or obtain the same result, which is ultimately a predicted future per-service traffic consumption.

[0050] The disclosure of the present invention may be better understood with reference to the following paragraphs from the research paper entitled “Mobile Service Traffic Decomposition for Network Slicing Using Deep Learning” which is incorporated herein in its entirety, and is not limiting on the scope of the claimed invention which is set out in the claims included herein.