US-12621270-B2 - Masking of privacy related information for network services

US12621270B2US 12621270 B2US12621270 B2US 12621270B2US-12621270-B2

Abstract

A method for operating a service consumer which is requesting to utilize a network service provided by a service provider in a cellular network. The method includes, at the service consumer, transmitting a service request to the service provider, the service request including a privacy indication indicating that a privacy related information necessary as input for the network service is requested to be privacy protected when used outside the service consumer, receiving a service response from the service provider, the service response comprising a privacy model and an indication how to use the privacy model, and processing the privacy model at the service consumer based on the indication.

Inventors

Abdulrahman Alabbasi
Massimo CONDOLUCI

Assignees

TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)

Dates

Publication Date: 20260505
Application Date: 20210818

Claims (16)

1 . A method for operating a service consumer which is requesting to utilize a network service provided by a service provider in a cellular network, the method comprising, at the service consumer: transmitting a service request to the service provider, the service request comprising a privacy indication indicating that a privacy related information necessary as input for the network service is requested to be privacy protected when used outside the service consumer; receiving a service response from the service provider, the service response comprising a privacy model and an indication how to use the privacy model, the privacy model comprising a protection model configured to receive as input the privacy related information without privacy protection and configured to generate as output the privacy related information with privacy protection; processing the privacy model at the service consumer based on the indication, processing the privacy model comprising inputting the privacy related information without privacy protection to the protection model and determining as output of the protection model the privacy related information with privacy protection; transmitting a further service request to the service provider including the privacy related information with privacy protection, the further service request comprising a latent variable generated by an encoder-decoder neural network system provided at the service consumer, the latent variable comprising the privacy related information with privacy protection, the privacy model received with the service response used at the service consumer as an encoder part of the encoder-decoder neural network system; and receiving a further service response from the service provider including the requested network service as generated based on the privacy protected privacy information.
2 . The method of claim 1 , wherein the privacy model comprises a service model configured to provide the requested network service to the service consumer when carried out at the service consumer, the service model being configured to receive as an input the privacy related information without privacy protection and configured to provide as output the requested network service, wherein processing the privacy model comprises inputting the privacy related information without privacy protection to the service model and determining the output of the service model as requested network service.
3 . The method of claim 1 , wherein processing the privacy model comprises inputting the privacy related information without privacy protection to the encoder part and determining the privacy related information with privacy protection as an encoded latent variable output by the encoder part.
4 . The method of claim 1 , wherein the received encoder part is an aggregated encoder part that has been locally trained at a plurality of different service consumers and was aggregated at the service provider.
5 . The method of claim 1 , wherein the received privacy model is used for distributed learning, further comprising: training the received privacy model based on training data provided locally at the service consumer in order to generate an updated privacy model; and transmitting the updated privacy model to the service provider.
6 . The method of claim 5 , wherein the indication received in the service response indicates whether the privacy model is to be updated or not before use at the service consumer, wherein the privacy model is updated or not based on the received indication.
7 . The method of claim 5 , wherein the updated privacy model is generated based on at least a part of the privacy related information.
8 . The method of claim 1 , wherein the privacy related information comprises a mobility information indicating a degree of mobility of the service consumer.
9 . The method of claim 1 , wherein the privacy related information comprises at least one of a time dependent location of the service consumer, and a time dependent moving trajectory of the service consumer.
10 . The method of claim 1 , wherein the indication how to use the privacy model indicates whether the service consumer is to use the received privacy model in order to directly generate the requested network service at the service consumer or whether the privacy model is to be used to generate as output the privacy related information with privacy protection.
11 . The method of claim 1 , wherein the network service comprises at least one of the following: Quality of Service (QOS) Sustainability Analytics; Observed Service Experience Analytics; Network function load analytics; consumer mobility analytics; and a data network performance analytics.
12 . A method for operating a service provider configured to provide a network service in a cellular network, the method comprising, at the service provider: receiving a service request from a service consumer configured to utilize the network service, the service request comprising a privacy indication indicating that a privacy related information necessary as input for the network service is requested to be privacy protected when used outside the service consumer; selecting a privacy model for the service consumer taking into account the privacy indication, the privacy model comprising a protection model configured to receive as input the privacy related information without privacy protection and configured to generate as output the privacy related information with privacy protection; determining how to use the selected privacy model at the service consumer, determining how to use the selected privacy model comprising inputting the privacy related information without privacy protection to the protection model and determining as output of the protection model the privacy related information with privacy protection; and transmitting a service response to the service consumer, the service response comprising the selected privacy model and an indication how to use the selected privacy model at the service consumer; receiving a further service request from the service consumer including the privacy related information with privacy protection, the further service request comprising a latent variable generated by an encoder-decoder neural network system provided at the service consumer, the latent variable comprising the privacy related information with privacy protection, the privacy model received with the service response used at the service consumer as an encoder part of the encoder-decoder neural network system; and transmitting a further service response to the service consumer including the requested network service as generated based on the privacy protected privacy information.
13 . The method of claim 12 , wherein determining how to use the privacy model comprises determining whether the service consumer is to use the received privacy model in order to directly generate the requested network service at the service consumer or whether the privacy model is to be used at the service consumer to generate as output the privacy related information with privacy protection which is to be transmitted to the service provider.
14 . The method of claim 12 , wherein the privacy model comprises a service model configured to provide the requested network service to the service consumer when carried out at the service consumer, the service model being configured to receive as an input the privacy related information without privacy protection and configured to provide as output the requested network service.
15 . A service consumer configured to request utilization of a network service provided by a service provider in a cellular network, the service consumer comprising a memory and at least one processing unit, the memory containing instructions executable by the at least one processing unit to cause the service consumer to: transmit a service request to the service provider, the service request comprising a privacy indication indicating that a privacy related information necessary as input for the network service is requested to be privacy protected when used outside the service consumer; receive a service response from the service provider, the service response comprising a privacy model and an indication how to use the privacy model, the privacy model comprising a protection model configured to receive as input the privacy related information without privacy protection and configured to generate as output the privacy related information with privacy protection; process the privacy model at the service consumer based on the indication, processing the privacy model comprising inputting the privacy related information without privacy protection to the protection model and determining as output of the protection model the privacy related information with privacy protection; transmit a further service request to the service provider including the privacy related information with privacy protection, the further service request comprising a latent variable generated by an encoder-decoder neural network system provided at the service consumer, the latent variable comprising the privacy related information with privacy protection, the privacy model received with the service response used at the service consumer as an encoder part of the encoder-decoder neural network system; and receive a further service response from the service provider including the requested network service as generated based on the privacy protected privacy information.
16 . A service provider configured to provide a network service in a cellular network, the service provider comprising a memory and at least one processing unit, the memory containing instructions executable by the at least one processing unit to cause the service consumer to: receive a service request from a service consumer configured to utilize the network service, the service request comprising a privacy indication indicating that a privacy related information necessary as input for the network service is requested to be privacy protected when used outside the service consumer; select a privacy model for the service consumer taking into account the privacy indication, the privacy model comprising a protection model configured to receive as input the privacy related information without privacy protection and configured to generate as output the privacy related information with privacy protection; determine how to use the selected privacy model at the service consumer, determining how to use the selected privacy model comprising inputting the privacy related information without privacy protection to the protection model and determining as output of the protection model the privacy related information with privacy protection; transmit a service response to the service consumer, the service response comprising the selected privacy model and an indication how to use the selected privacy model at the service consumer; receive a further service request from the service consumer including the privacy related information with privacy protection, the further service request comprising a latent variable generated by an encoder-decoder neural network system provided at the service consumer, the latent variable comprising the privacy related information with privacy protection, the privacy model received with the service response used at the service consumer as an encoder part of the encoder-decoder neural network system; and transmit a further service response to the service consumer including the requested network service as generated based on the privacy protected privacy information.

Description

CROSS-REFERENCE TO RELATED APPLICATION This application is a Submission Under 35 U.S.C. § 371 for U.S. National Stage Patent Application of International Application Number: PCT/EP2021/072936, filed Aug. 18, 2021 entitled “MASKING OF PRIVACY RELATED INFORMATION FOR NETWORK SERVICES,” the entirety of which is incorporated herein by reference. TECHNICAL FIELD The present application relates to a method for operating a service consumer and for operating a service provider. The invention furthermore relates to the corresponding service provider and service consumer. Further, a computer program and a carrier comprising the computer program is provided and a system comprising the service provider and service consumer. BACKGROUND 5G Standard introduced the possibility to request several services offered based on the availability of UE-related information, such as location. As example, the Network Data Analytics Function (NWDAF, defined in 3GPP TS 23.288) is a network function introduced in 5G core network to provide analytics (including AI-based ones) to several consumers, these being other network functions of the core network (e.g., AMF (Access and Mobility Management Function), SMF (Session Management Function) or service-related functions such as Application Function (AF) and consequently application layer. Considering 3GPP TS 23.288, the following analytics deal with location information: QoS Sustainability Analytics, which generates analytics based on the (mandatory) location information including in the consumer request, where the location information can be an area or a path of interest also reflecting a list of waypoints. The output contains a List of QoS (Quality of Service) sustainability Analytics (RAN (Radio Access Network) UE Throughput and/or QoS Flow Retainability, Applicable Area, Applicable Time Period, Crossed Reporting Threshold(s) (in case of prediction).Observed Service Experience Analytics, which used the UE location as input provided by the AF when generating the analytics. Observed Service Experience refers to average of observed Service MoS (Mean Opinion Socre) and/or variance of observed Service MoS indicating service MOS distribution for services such as audio-visual streaming as well as services that are not audio-visual streaming such as V2X and Web Browsing services analytics.NF (Network Function) load Analytics, which uses as inputs UE information such as destination (expected final location of UE), Route (planned path of UE movement), time of arrival at destination. The output contains information such as NF type, NF instance ID, NF status (availability status of the NF on the Analytics target period, expressed as a percentage of time per status value such as registered, suspended, undiscoverable), NF resource usage (average usage of assigned resources such as CPU, memory, disk), NF load (average load of the NF instance over the Analytics target period), NF peak load (maximum load of the NF instance over the Analytics target period), etc.UE mobility Analytics, which uses as inputs UE trajectory (timestamped UE geographical positions). The output contains information such as UE location statistics for a certain UE group ID or UE ID, where statistics are in the form of list of UE location (tracking areas or cells which the UE stays) and associated ratio (percentage of UEs in the group in case of an UE group)DN Performance Analytics, which used the UE location as input provided by the AF when generating the analytics. DN Performance Analytics provide analytics for user plane performance for certain S-NSSAI, DNN, UPF, DNAI, where analytics include average traffic rate, maximum traffic rate, average packet delay, maximum packet delay, average packet loss rate. It should be understood that other network services, in addition to those provided by NWDAF, could also rely on information such as geographical location of UEs and similar types of information. On top of NWDAF related standardization, 3GPP is also focused on RAN-centric data collection for mobility optimization use-cases as described in 37.816 section 5.3. 3GPP TR 22.874 is investigating aspects related to traffic characteristics and performance requirements for AI/ML (Artificial Intelligence/Machine Learning) model transfer in a 5GS (5G System). Section 7 focuses on distributed/federated learning over 5G system. The current state-of-the-art considers federated learning as a novel machine learning tool that competes with regular ML methods that train on large aggregations of data collected over multiple data sources. FIG. 1 shows an architectural overview of a system with federated learning. Different clients/UEs 20 to 24 have corresponding Machine learning modules 30 to 34 in order to train locally trained models 40 to 44. The models 40 to 44 are uploaded to a central server 50 which is configured to generate an aggregated model 70. As described in FIG. 1, Federated Learning (FL) comprises: Clients 20-24 (e.g. UEs) that