Method and network device for determining causes of network slice performance degradation in an open radio access network, O-RAN

1 . A computer-implemented method for determining causes of network slice performance degradation in an open radio access network (O-RAN), the O-RAN being under Central Unit-Distributed Unit (CU-DU) functional split, the method comprising:

in a setup phase:

obtaining telemetries that contain multivariate observations of DUs,

aggregating, per CU per network slice, telemetry data obtained from involved DUs,

extracting, for a specific time slot, high order features from correlations between the multivariate observations across the involved DUs, and

based on the telemetries and the extracted high order features, learning a k-time slots ahead performance prediction model; and

in a detection phase:

using, upon receiving new telemetries, the performance prediction model to forecast the network slice performance,

based on a performance degradation being forecasted, sampling a set of performance prediction models and computing impact scores based on gradients of all telemetries and the extracted high order features using the sampled performance prediction models, and

selecting telemetries and high order features with impact scores exceeding a configurable threshold and identifying network slice entities related to the selected telemetries as root causes of the forecasted performance degradation.

2 . The method according to claim 1 , wherein the multivariate observations of DUs in the obtained telemetries of DUs include observations such as traffic per transmission/reception point (TRP), channel quality indicator, Multiple Input Multiple Output (MIMO) usage rate, and/or other telemetries related to SLA of interest.

3 . The method according to claim 1 , wherein the high order features extraction is performed by means of a CNN based component with multiple kernels with nonlinear activation functions, wherein each kernel defines a functional relationship between the multivariate observations.

4 . The method according to claim 1 , wherein sampling the set of performance prediction models is performed by applying a dropout based sampling method.

5 . The method according to claim 1 , further comprising:

computing, for each sampled performance prediction model of the set of performance prediction models, a gradient as a partial derivative of a prediction function of the respective performance prediction model with respect to telemetries.

6 . The method according to claim 1 , further comprising:

computing, for each single telemetry of a DU at a particular time, the impact score of the telemetry as a mean of the product of sampled gradients and the telemetry.

7 . The method according to claim 1 , further comprising,

upon identifying a network slice entity as the cause of the performance degradation, proactively changing the settings of the respective network slice entity to prevent SLA violations.

8 . A network device for determining causes of network slice performance degradation in an open radio access network (O-RAN), the O-RAN being under Central Unit-Distributed Unit (CU-DU) functional split, the network device comprising a functional unit having one or more computational processors with access to memory, which, alone or in combination, are configured to provide for execution of the following steps:

in a setup phase:

obtaining telemetries that contain multivariate observations of DUs,

aggregating, per CU per network slice, telemetry data obtained from involved DUs,

extracting, for a specific time slot, high order features from correlations between the multivariate observations across the involved DUs, and

based on the telemetries and the extracted high order features, learning a k-time slots ahead performance prediction model; and

in a detection phase:

using, upon receiving new telemetries, the performance prediction model to forecast the network slice performance,

based on a performance degradation being forecasted, sampling a set of performance prediction models and computing impact scores based on gradients of all telemetries and the extracted high order features using the sampled performance prediction models, and

selecting telemetries and high order features with impact scores exceeding a configurable threshold and identifying network slice entities related to the selected telemetries as root causes of the forecasted performance degradation.

9 . The network device according to claim 8 , wherein the functional unit includes a feature extractor component that is configured to perform the high order features extraction by means of a CNN based component with multiple kernels with nonlinear activation functions, wherein each kernel defines a functional relationship between the multivariate observations.

10 . The network device according to claim 8 , wherein the functional unit includes a root cause detection component that is configured to sample the set of performance prediction models by means of a dropout based sampling method.

11 . The network device according to claim 10 , wherein the root cause detection component is further configured to compute, for each single telemetry of a DU at a particular time, the impact score of the telemetry as a mean of a product of sampled gradients and the telemetry.

12 . The network device according to claim 10 , wherein the root cause detection component is further configured to compute, for each sampled performance prediction model, the gradient as a partial derivative of the prediction function of the respective model with respect to telemetries.

13 . The network device according to claim 8 , wherein the functional unit is locally deployed per CU per slice.

14 . The network device according to claim 8 , wherein the functional unit is deployed as an extra functionality for a Near Real-Time RAN Intelligent Controller (Near-RT RIC) of the open radio access network (O-RAN).

15 . A non-transitory, computer-readable storage medium having instructions thereon which, upon execution on one or more processors, provide for execution of the following steps:

in a setup phase:

obtaining telemetries that contain multivariate observations of DUs of an open radio access network (O-RAN), the O-RAN being under Central Unit-Distributed Unit (CU-DU) functional split,

aggregating, per CU per network slice, telemetry data obtained from involved DUS,

extracting, for a specific time slot, high order features from correlations between the multivariate observations across the involved DUs, and

based on the telemetries and the extracted high order features, learning a k-time slots ahead performance prediction model; and

in a detection phase:

using, upon receiving new telemetries, the performance prediction model to forecast the network slice performance,

based on a performance degradation being forecasted, sampling a set of performance prediction models and computing impact scores based on gradients of all telemetries and the extracted high order features using the sampled performance prediction models, and

selecting telemetries and high order features with impact scores exceeding a configurable threshold and identifying network slice entities related to the selected telemetries as root causes of the forecasted performance degradation.

16 . The method according to claim 1 , wherein the high order features represent relations between one or more telemetries of the telemetry data and one or more DUs of the DUs.