WO2023015448A1

WO2023015448A1 - Apparatus, method, and computer program

Info

Publication number: WO2023015448A1
Application number: PCT/CN2021/111815
Authority: WO
Inventors: Tejas SUBRAMANYA; Henning Sanneck; Janne ALI-TOLPPA; Jing PING; Iris ADAM
Original assignee: Nokia Shanghai Bell Co., Ltd.; Nokia Solutions And Networks Oy; Nokia Technologies Oy
Priority date: 2021-08-10
Filing date: 2021-08-10
Publication date: 2023-02-16
Also published as: CN117813846A

Abstract

An apparatus comprising means configured to: facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.

Description

APPARATUS, METHOD, AND COMPUTER PROGRAM

Field of the disclosure

The present disclosure relates to an apparatus, a method, and a computer program for providing a framework for cross-domain trustworthy Artificial Intelligence applications suitable for but not exclusively in cognitive autonomous networks.

Background

A communication system can be seen as a facility that enables communication sessions between two or more entities such as communication devices, base stations and/or other nodes by providing carriers between the various entities involved in the communications path.

The communication system may be a wireless communication system. Examples of wireless systems comprise public land mobile networks (PLMN) operating based on radio standards such as those provided by 3GPP, satellite based communication systems and different wireless local networks, for example wireless local area networks (WLAN) . The wireless systems can typically be divided into cells and are therefore often referred to as cellular systems.

The communication system and associated devices typically operate in accordance with a given standard or specification which sets out what the various entities associated with the system are permitted to do and how that should be achieved. Communication protocols and/or parameters which shall be used for the connection are also typically defined. Examples of standard are the so-called 5G standards.

There is a need to provide control systems which enable a communications service provider (CSP) to control and optimise a complex network of communications system elements.

One of current approaches being employed is closed-loop automation and machine learning which can be built into self-organizing networks (SON) enabling an operator to automatically optimize every cell in the radio access network.

Summary

According to a first aspect there is provided an apparatus comprising means configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.

The means configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may be configured to facilitate at least one of: defining cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness; configuring cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness; measuring cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness; and reporting cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness.

The cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function may comprise at least one of: fairness; explainability; and robustness.

The means configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may be configured to: obtain a cross domain machine learning or artificial intelligence quality of trustworthiness, the cross-domain machine learning or artificial intelligence quality of trustworthiness configured to cover domain-specific machine learning or artificial intelligence quality of trustworthiness requirements and constraints of cross-domain network service-related machine learning or artificial intelligence pipelines; translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains; and provide the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains.

The means configured to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may be configured to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness based on a risk-level of the cross-domain network service.

The means configured to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may be configured to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness based on at least one service level agreement requirement for the cross-domain network, wherein the at least one service level agreement comprises at least one of: service type; service priority; and at least one key performance indicator metric.

The means configured to provide the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may be configured to: generate and pass to at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request configured to control the at least one domain specific machine learning or artificial intelligence trustworthiness function to configure a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and obtain from the at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation response based on an implementation of the configuration of the machine learning or artificial intelligence pipeline for the at least one of the at least two domains.

The means configured to provide the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may be configured to: generate and pass to at least one domain specific policy manager a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request configured to control at least one domain specific machine learning or artificial intelligence trustworthiness function to configure a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and obtain from the at least one domain policy manager a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation response based on an implementation of the configuration of the machine learning or artificial intelligence pipeline for the at least one of the at least two domains.

The cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request may comprise: a domain scope parameter configured to identify a domain the request is addressing; a pipeline identification parameter configured to identify a domain-specific machine learning or artificial intelligence pipeline the request is addressing; a class identification parameter configured to identify the at least one domain specific machine learning or artificial intelligence quality of trustworthiness.

The cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request may further comprise at least one of: a desired fairness parameter configured to indicate a relative fairness level for the domain-specific machine learning or artificial intelligence pipeline; a desired explainability parameter configured to indicate a desired explainability level for the domain-specific machine learning or artificial intelligence pipeline; a desired technical robustness parameter configured to indicate a desired technical robustness level for the domain-specific machine learning or artificial intelligence pipeline; and desired adversarial robustness parameter configured to indicate a desired adversarial robustness level for the domain-specific machine learning or artificial intelligence pipeline.

The means configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may be configured to:generate and pass to at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence capability information request configured to control the at least one domain specific machine learning or artificial intelligence trustworthiness function to implement a capability discovery for a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and obtain from the at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence capability information response reporting the capability discovery for a machine learning or artificial intelligence pipeline for the at least one of the at least two domains.

The cross-domain trustworthiness machine learning or artificial intelligence capability information request may comprise: a domain scope parameter configured to identify a domain the request is addressing; and a scope parameter configured to identify a domain-specific machine learning or artificial intelligence pipeline the request is addressing.

The cross-domain trustworthiness machine learning or artificial intelligence capability information request may further comprise a pipeline phase parameter configured to identify a phase of the domain-specific machine learning or artificial intelligence pipeline the request is addressing.

The means configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may be further configured to: obtain a cross-domain trustworthiness artificial intelligence report request or subscription from a network operator; generate and pass to at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function a domain specified trustworthiness artificial intelligence report request or subscription based on the cross-domain trustworthiness artificial intelligence report request or subscription from a network operator, wherein the domain specified trustworthiness artificial intelligence report request or subscription is configured to control the at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function to provide at least one domain specific machine learning or artificial intelligence pipeline report response or notification; receive from the at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function machine learning or artificial intelligence capability information and/or the at least one domain specific machine learning or artificial intelligence pipeline report response or notification; store machine learning or artificial intelligence capability information and/or the at least one domain specific machine learning or artificial intelligence pipeline report received from the the at least one domain specific machine learning or artificial intelligence pipeline in a cross-domain trust knowledge database; generate and pass a cross-domain trustworthiness artificial intelligence report based on the at least one domain specific machine learning or artificial intelligence pipeline report response or notification.

The cross-domain trustworthiness artificial intelligence report request may comprises: a domain scope parameter configured to identify a domain the request is addressing; a scope parameter configured to identify a domain-specific machine learning or artificial intelligence pipeline the request is addressing; and a pipeline phase parameter configured to identify a phase of the domain-specific machine learning or artificial intelligence pipeline the request is addressing. The cross-domain trustworthiness artificial intelligence report request further may comprise at least one of: a list of fairness metrics parameter configured to identify fairness metrics to be reported; a list of fairness metric explanations configured to identify fairness metric explanations to be reported; a list of explainability metrics configured to identify explainability metrics to be reported; a list of explanations configured to identify explanations to be reported; a list of technical robustness metrics configured to identify technical robustness metrics to be reported; a list of technical robustness metric explanations configured to identify technical robustness metric explanations to be reported; a list of adversarial robustness metrics configured to identify adversarial robustness metrics to be reported; a list of adversarial robustness metric explanations configured to identify adversarial robustness metric explanations to be reported; a start time parameter configured to identify a start time for reporting; an end time parameter configured to identify an end time for reporting; and a report interval parameter configured to identify a periodicity interval for reporting.

The cross-domain trustworthiness artificial intelligence report subscription may comprise: a domain scope parameter configured to identify a domain the subscription is addressing; a scope parameter configured to identify a domain-specific machine learning or artificial intelligence pipeline the subscription is addressing; and a pipeline phase parameter configured to identify a phase of the domain-specific machine learning or artificial intelligence pipeline the subscription is addressing.

The cross-domain trustworthiness artificial intelligence report subscription may further comprise at least one of: a list of fairness metrics parameter configured to identify fairness metrics to be reported; a list of explainability metrics configured to identify explainability metrics to be reported; a list of technical robustness metrics configured to identify technical robustness metrics to be reported; a list of adversarial robustness metrics configured to identify adversarial robustness metrics to be reported; and a crossed reporting threshold parameter configured to identify a threshold for which a metric or metric explanation is reported.

According to a second aspect there is provided a method comprising: facilitating a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.

Facilitating a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may comprise facilitating at least one of: defining cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness; configuring cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness; measuring cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness; and reporting cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness.

Facilitating a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may comprise: obtaining a cross domain machine learning or artificial intelligence quality of trustworthiness, the cross-domain machine learning or artificial intelligence quality of trustworthiness configured to cover domain-specific machine learning or artificial intelligence quality of trustworthiness requirements and constraints of cross-domain network service-related machine learning or artificial intelligence pipelines; translating the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains; and providing the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains.

Translating the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may comprise translating the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness based on a risk-level of the cross-domain network service.

Translating the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may comprise translating the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness based on at least one service level agreement requirement for the cross-domain network, wherein the at least one service level agreement comprises at least one of: service type; service priority; and at least one key performance indicator metric.

Providing the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may comprise: generating and pass to at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request configured to control the at least one domain specific machine learning or artificial intelligence trustworthiness function to configure a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and obtaining from the at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation response based on an implementation of the configuration of the machine learning or artificial intelligence pipeline for the at least one of the at least two domains.

Providing the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may comprise: generating and passing to at least one domain specific policy manager a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request configured to control at least one domain specific machine learning or artificial intelligence trustworthiness function to configure a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and obtaining from the at least one domain policy manager a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation response based on an implementation of the configuration of the machine learning or artificial intelligence pipeline for the at least one of the at least two domains.

Facilitating a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may comprise: generating and passing to at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence capability information request configured to control the at least one domain specific machine learning or artificial intelligence trustworthiness function to implement a capability discovery for a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and obtaining from the at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence capability information response reporting the capability discovery for a machine learning or artificial intelligence pipeline for the at least one of the at least two domains.

Facilitating a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may further comprise: obtaining a cross-domain trustworthiness artificial intelligence report request or subscription from a network operator; generating and passing to at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function a domain specified trustworthiness artificial intelligence report request or subscription based on the cross-domain trustworthiness artificial intelligence report request or subscription from a network operator, wherein the domain specified trustworthiness artificial intelligence report request or subscription may control the at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function to provide at least one domain specific machine learning or artificial intelligence pipeline report response or notification; receiving from the at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function machine learning or artificial intelligence capability information and/or the at least one domain specific machine learning or artificial intelligence pipeline report response or notification; storing machine learning or artificial intelligence capability information and/or the at least one domain specific machine learning or artificial intelligence pipeline report received from the the at least one domain specific machine learning or artificial intelligence pipeline in a cross-domain trust knowledge database; generating and passing a cross-domain trustworthiness artificial intelligence report based on the at least one domain specific machine learning or artificial intelligence pipeline report response or notification.

The cross-domain trustworthiness artificial intelligence report request may comprise: a domain scope parameter configured to identify a domain the request is addressing; a scope parameter configured to identify a domain-specific machine learning or artificial intelligence pipeline the request is addressing; and a pipeline phase parameter configured to identify a phase of the domain-specific machine learning or artificial intelligence pipeline the request is addressing. The cross-domain trustworthiness artificial intelligence report request further may comprise at least one of: a list of fairness metrics parameter configured to identify fairness metrics to be reported; a list of fairness metric explanations configured to identify fairness metric explanations to be reported; a list of explainability metrics configured to identify explainability metrics to be reported; a list of explanations configured to identify explanations to be reported; a list of technical robustness metrics configured to identify technical robustness metrics to be reported; a list of technical robustness metric explanations configured to identify technical robustness metric explanations to be reported; a list of adversarial robustness metrics configured to identify adversarial robustness metrics to be reported; a list of adversarial robustness metric explanations configured to identify adversarial robustness metric explanations to be reported; a start time parameter configured to identify a start time for reporting; an end time parameter configured to identify an end time for reporting; and a report interval parameter configured to identify a periodicity interval for reporting.

According to a third aspect there is provided an apparatus comprising at least one processor and at least one memory including computer code for one or more programs, the at least one memory and the computer code configured, with the at least one processor, to cause the apparatus at least to: facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.

The apparatus caused to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may be caused to facilitate at least one of: defining cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness; configuring cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness; measuring cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness; and reporting cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness.

The apparatus caused to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may be caused to: obtain a cross domain machine learning or artificial intelligence quality of trustworthiness, the cross-domain machine learning or artificial intelligence quality of trustworthiness configured to cover domain-specific machine learning or artificial intelligence quality of trustworthiness requirements and constraints of cross-domain network service-related machine learning or artificial intelligence pipelines domain pipelines; translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains; and provide the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains.

The apparatus caused to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may be caused to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness based on a risk-level of the cross-domain network service.

The apparatus caused to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may be caused to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness based on at least one service level agreement requirement for the cross-domain network, wherein the at least one service level agreement comprises at least one of: service type; service priority; and at least one key performance indicator metric.

The apparatus caused to provide the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may be caused to: generate and pass to at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request configured to control the at least one domain specific machine learning or artificial intelligence trustworthiness function to configure a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and obtain from the at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation response based on an implementation of the configuration of the machine learning or artificial intelligence pipeline for the at least one of the at least two domains.

The apparatus caused to provide the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains may be caused to: generate and pass to at least one domain specific policy manager a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request configured to control at least one domain specific machine learning or artificial intelligence trustworthiness function to configure a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and obtain from the at least one domain policy manager a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation response based on an implementation of the configuration of the machine learning or artificial intelligence pipeline for the at least one of the at least two domains.

The apparatus caused to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may be caused to: generate and pass to at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence capability information request configured to control the at least one domain specific machine learning or artificial intelligence trustworthiness function to implement a capability discovery for a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and obtain from the at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence capability information response reporting the capability discovery for a machine learning or artificial intelligence pipeline for the at least one of the at least two domains.

The apparatus caused to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains may be caused to: obtain a cross-domain trustworthiness artificial intelligence report request or subscription from a network operator; generate and pass to at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function a domain specified trustworthiness artificial intelligence report request or subscription based on the cross-domain trustworthiness artificial intelligence report request or subscription from a network operator, wherein the domain specified trustworthiness artificial intelligence report request or subscription is configured to control the at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function to provide at least one domain specific machine learning or artificial intelligence pipeline report response or notification; receive from the at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function machine learning or artificial intelligence capability information and/or the at least one domain specific machine learning or artificial intelligence pipeline report response or notification; store machine learning or artificial intelligence capability information and/or the at least one domain specific machine learning or artificial intelligence pipeline report received from the the at least one domain specific machine learning or artificial intelligence pipeline in a cross-domain trust knowledge database; generate and pass a cross-domain trustworthiness artificial intelligence report based on the at least one domain specific machine learning or artificial intelligence pipeline report response or notification.

The cross-domain trustworthiness artificial intelligence report request may comprises: a domain scope parameter configured to identify a domain the request is addressing; a scope parameter configured to identify a domain-specific machine learning or artificial intelligence pipeline the request is addressing; and a pipeline phase parameter configured to identify a phase of the domain-specific machine learning or artificial intelligence pipeline the request is addressing.

The cross-domain trustworthiness artificial intelligence report request further may comprise at least one of: a list of fairness metrics parameter configured to identify fairness metrics to be reported; a list of fairness metric explanations configured to identify fairness metric explanations to be reported; a list of explainability metrics configured to identify explainability metrics to be reported; a list of explanations configured to identify explanations to be reported; a list of technical robustness metrics configured to identify technical robustness metrics to be reported; a list of technical robustness metric explanations configured to identify technical robustness metric explanations to be reported; a list of adversarial robustness metrics configured to identify adversarial robustness metrics to be reported; a list of adversarial robustness metric explanations configured to identify adversarial robustness metric explanations to be reported; a start time parameter configured to identify a start time for reporting; an end time parameter configured to identify an end time for reporting; and a report interval parameter configured to identify a periodicity interval for reporting.

According to a fourth aspect there is provided an apparatus comprising: circuitry configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.

According to a fifth aspect there is provided a computer program comprising computer executable code which when run on at least one processor is configured to: facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.

According to a sixth aspect there is provided a computer program comprising instructions [or a computer readable medium comprising program instructions] for causing an apparatus to perform at least the following: facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.

According to a seventh aspect there is provided a non-transitory computer readable medium comprising program instructions for causing an apparatus to perform at least the following: facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.

According to an eighth aspect there is provided a non-transitory computer readable medium comprising program instructions for causing an apparatus to perform at least the following: facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.

According to a ninth aspect there is provided a computer readable medium comprising program instructions for causing an apparatus to perform at least the following: facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.

An apparatus comprising means for performing the actions of the method as described above.

An apparatus configured to perform the actions of the method as described above.

A computer program comprising program instructions for causing a computer to perform the method as described above.

A computer program product stored on a medium may cause an apparatus to perform the method as described herein.

An electronic device may comprise apparatus as described herein.

A chipset may comprise apparatus as described herein.

Embodiments of the present application aim to address problems associated with the state of the art.

In the above, many different aspects have been described. It should be appreciated that further aspects may be provided by the combination of any two or more of the aspects described above.

Various other aspects are also described in the following detailed description and in the attached claims.

List of abbreviations

AI: Artificial Intelligence

CAN: Cognitive Autonomous Network

CD: Cross-Domain

CN: Core Network

CNF: Cognitive Network Function

CRUD: Create, Read, Update, Delete

CDSMD: Cross-Domain Service Management Domain

CU: Centralized Unit

DU: Distributed Unit

E2E: End-to-End

E2ESMD: End-to-End Service Management Domain

HLEG: High Level Expert Group

MANO: Management and Orchestration

MD: Management Domain

ML: Machine Learning

QCI: QoS Class Identifier

QoE: Quality of Experience

QoS: Quality of Service

QoT: Quality of Trustworthiness

RAN: Radio Access Network

RRU: Remote Radio Unit

SLA: Service Level Agreement

TAI: Trustworthy Artificial Intelligence

TAIF: TAI Framework

TN: Transport Network

UMTS: Universal Mobile Telecommunication System

URLLC: Ultra-Reliable Low Latency Communication

VNF: Virtual Network Function

V2X: Vehicle to Everything

WI: Work Item

3GPP: 3 ^rd Generation Partnership Project

5G: 5 ^th Generation

5GC: 5G Core network

5GS: 5G System

Brief Description of the Figures

Embodiments will now be described, by way of example only, with reference to the accompanying Figures in which:

Figure 1 shows a schematic representation of a 5G communications system;

Figure 2 shows a schematic representation of a control apparatus;

Figure 3 shows a schematic representation of a terminal;

Figure 4 shows a schematic representation of an artificial intelligence/machine learning pipeline;

Figure 5 shows a schematic representation of an example Trustworthy Artificial Intelligence Framework for Cognitive Autonomous Networks;

Figure 6 shows a workflow diagram with respect to the example Trustworthy Artificial Intelligence Framework for Cognitive Autonomous Networks as shown in Figure 5;

Figure 7 shows a schematic representation of an example Cross-Domain Management and Orchestration Architecture leveraging the domain-specific Trustworthy Artificial Intelligence Framework;

Figure 8 shows a schematic representation of an example Cross-Domain Management and Orchestration Architecture leveraging the domain-specific Trustworthy Artificial Intelligence Framework according to some embodiments;

Figure 9 shows a workflow diagram of an example Cross-Domain Trustworthy Artificial Intelligence application programming interface offered by the domain-specific Artificial Intelligence Trust Engine to the Cross-Domain Artificial Intelligence Trust Engine; and

Figure 10 shows a schematic representation of a non-volatile memory medium storing instructions which when executed by a processor allow a processor to perform one or more of the steps of the methods described herein.

Detailed Description of the Figures

In the following certain embodiments are explained with reference to mobile communication devices capable of communication via a wireless cellular system and mobile communication systems serving such mobile communication devices. Before explaining in detail the exemplifying embodiments, certain general principles of a wireless communication system, access systems thereof, and mobile communication devices are briefly explained with reference to Figures 1, 2 and 3 to assist in understanding the technology underlying the described examples.

Figure 1 shows a schematic representation of a 5G system (5GS) . The 5GS may comprises a terminal, a (radio) access network ( (R) AN) , a 5G core network (5GC) , one or more application functions (AF) and one or more data networks (DN) .

The 5G (R) AN may comprise one or more gNodeB (gNB) distributed unit (DU) functions connected to one or more gNodeB (gNB) centralized unit (CU) functions.

The 5GC may comprise an access and mobility management function (AMF) , a session management function (SMF) , an authentication server function (AUSF) , a user data management (UDM) , a user plane function (UPF) , a network exposure function (NEF) and/or other network functions (NFs) not represented such as an operation administration and maintenance (OAM) NF.

Figure 2 illustrates an example of a control apparatus 200 for controlling a function of the (R) AN or the 5GC as illustrated on Figure 1. The control apparatus may comprise at least one random access memory (RAM) 211a, at least on read only memory (ROM) 211 b, at least one

processor

212, 213 and an input/output interface 214. The at least one

processor

212, 213 may be coupled to the RAM 211a and the ROM 211 b. The at least one

processor

212, 213 may be configured to execute an appropriate software code 215. The software code 215 may for example allow to perform one or more steps to perform one or more of the present aspects. The software code 215 may be stored in the ROM 211 b. The control apparatus 200 may be interconnected with another control apparatus 200 controlling another function of the 5G (R) AN or the 5GC. In some embodiments, each function of the (R)AN or the 5GC comprises a control apparatus 200. In alternative embodiments, two or more functions of the (R) AN or the 5GC may share a control apparatus.

Figure 3 illustrates an example of a terminal 300, such as the terminal illustrated on Figure 1. The terminal 300 may be provided by any device capable of sending and receiving radio signals. Non-limiting examples comprise a user equipment, a mobile station (MS) or mobile device such as a mobile phone or what is known as a ’smart phone’ , a computer provided with a wireless interface card or other wireless interface facility (e.g., USB dongle) , a personal data assistant (PDA) or a tablet provided with wireless communication capabilities, a machine-type communications (MTC) device, a Cellular Internet of things (CIoT) device or any combinations of these or the like. The terminal 300 may provide, for example, communication of data for carrying communications. The communications may be one or more of voice, electronic mail (email) , text message, multimedia, data, machine data and so on.

The terminal 300 may receive signals over an air or radio interface 307 via appropriate apparatus for receiving and may transmit signals via appropriate apparatus for transmitting radio signals. In Figure 3, transceiver apparatus is designated schematically by block 306. The transceiver apparatus 306 may be provided for example by means of a radio part and associated antenna arrangement. The antenna arrangement may be arranged internally or externally to the mobile device.

The terminal 300 may be provided with at least one processor 301, at least one memory ROM 302a, at least one RAM 302b and other possible components 303 for use in software and hardware aided execution of tasks it is designed to perform, including control of access to and communications with access systems and other communication devices. The at least one processor 301 is coupled to the RAM 302b and the ROM 302a. The at least one processor 301 may be configured to execute an appropriate software code 308. The software code 308 may for example allow to perform one or more of the present aspects. The software code 308 may be stored in the ROM 302a.

The processor, storage and other relevant control apparatus can be provided on an appropriate circuit board and/or in chipsets. This feature is denoted by reference 304. The device may optionally have a user interface such as keypad 305, touch sensitive screen or pad, combinations thereof or the like. Optionally one or more of a display, a speaker and a microphone may be provided depending on the type of the device.

As described previously closed-loop automation and machine learning (ML) (also known as Artificial Intelligence or AI) can be built into self-organizing networks (SON) or cognitive autonomous networks (CAN) enabling an operator to automatically optimize every cell in the radio access network.

An Artificial Intelligence (AI) or Machine Learning (ML) pipeline helps to automate AI/ML workflows by splitting them into independent, reusable and modular components that can then be pipelined together to create a model. The AI/ML Pipeline is iterative where each step is repeated to continuously improve the accuracy of the model.

As shown in Figure 4 an example AI/ML workflow comprises the following three components:

Data Source Manager 403. The data source manager 403 is configured to implement such functions as data collection and data preparation.

Model Training Manager 405. The model training manager 405 is configured to implement functions such as hyperparameter tuning) .

Model Inference Manager 407. The model inference manager 407 is configured to implements functions such as model evaluation.

With AI/ML pipelining and the recent push for microservices architecture (e.g., containers) , each AI/ML workflow component is abstracted into an independent service that relevant stakeholders (for example data engineers, data scientists) can independently work on. Furthermore the AI/ML Pipeline Orchestrator 401 (an example of which is provided by Kubeflow) can manage the AI/ML Pipelines' lifecycle. For example managing the stages in the lifecycle of commissioning, scaling, decommissioning.

For AI/ML systems to be widely accepted, they should be trustworthy in addition to their performance (e.g., accuracy) . Legal bodies are proposing frameworks on AI/ML applications, for example the European Commission has proposed the first-ever legal framework on AI. This legal framework presents new rules for AI to be Trustworthy and which mission-critical AI-based systems must adhere to in the near future. The High-level Expert Group (HLEG) group on AI has developed the European Commission's Trustworthy AI (TAI) strategy. In the deliverable 'Ethics Guidelines for Trustworthy AI' released in April 2019, the group has listed seven critical requirements that the AI systems should meet to be considered trustworthy. Below are the requirements:

Transparency. The transparency requirement includes traceability, explainability and communication.

Diversity, non-discrimination and fairness. This requirement includes the avoidance of unfair bias, accessibility and universal design, and stakeholder participation.

Technical robustness and safety. This requirement includes resilience to attack and security, fall back plan and general safety, accuracy, reliability and reproducibility.

Privacy and data governance. This requirement includes respect for privacy, quality and integrity of data, and access to data.

Accountability. The accountability requirement includes auditability, minimization and reporting of negative impact, trade-offs and redress.

Human agency and oversight. The human agency and oversight requirement includes fundamental rights, human agency and human oversight.

Societal and environmental wellbeing. This requirement includes sustainability and environmental friendliness, social impact, society and democracy.

Additionally, the International Organization for Standardization and the International Electrotechnical Commission (ISO/IEC) has also published a technical report on ‘Overview of trustworthiness in artificial intelligence’ . Early efforts in the open-source community are also visible towards developing TAI frameworks/tools/libraries such as IBM AI360, Google Explainable AI and TensorFlow Responsible AI.

In the following, we introduce some key TAI definitions/algorithms/metrics described in the AI/ML research community.

Fairness: Fairness is the process of understanding bias introduced in the data, and ensuring the model provides equitable predictions across all demographic groups. It is important to apply fairness analysis throughout the entire AI/ML Pipeline, making sure to continuously re-evaluate the models from the perspective of fairness and inclusion. This is especially important when AI/ML is deployed in critical business processes that affect a wide range of end users. There are three broad approaches to detect bias in the AI/ML model:

1. Pre-processing fairness -To detect bias in the AI/ML training data using algorithms such as Reweighing and Disparate impact remover.

2. In-processing fairness -To detect bias in the AI/ML model generation using algorithms such as Prejudice Remover and Adversarial debiasing.

3. Post-processing fairness -To detect bias in the AI/ML model decisions using algorithms such as Odds-equalizing and Reject option classification.

Quantification of Fairness –There are several metrics that measure individual and group fairness. For example, Statistical Parity Difference, Average Odds Difference, Disparate Impact and Theil Index.

Explainability: Explainability of an AI/ML model refers to unveiling of the black box model which just makes the prediction or gives the recommendation to the White box which actually gives the details of the underlying mechanism and pattern identified by the model for a particular dataset. There are multiple reasons why it is necessary to understand the underlying mechanism of an AI/ML model such as human readability, justifiability, interpretability and bias mitigation. There are three broad approaches to design an ML model to be explainable:

1. Pre-modelling explainability –To understand or describe data used to develop AI/ML models. For example, using algorithms such as ProtoDash and Disentangled Inferred Prior Variational Autoencoder Explainer.

2. Explainable modelling/Interpretable modelling –To develop more explainable AI/ML models, e.g., ML models with joint prediction and explanation or surrogate explainable models. For example, using algorithms such as Generalized Linear Rule Models and Teaching Explainable Decisions (TED) .

3. Post-modelling explainability -To extract explanations from pre-developed AI/ML models. For example, using algorithms such as ProtoDash, Contrastive Explanations Method, Profweight, LIME and SHAP.

Furthermore, explanations can be local (i.e., explaining a single instance/prediction or global (i.e., explaining the global AI/ML model structure/predictions, e.g., based on combining many local explanations of each prediction) ) .

Quantification of Explainability -Although it is ultimately the consumer who determines the quality of an explanation, the research community has proposed quantitative metrics as proxies for explainability. There are several metrics that measure explainability such as Faithfulness and Monotonicity.

Robustness (adversarial) : There are four adversarial threats that any AI/ML model developers/scientists need to consider defending and evaluating their AI/ML models and applications.

1. Evasion: Evasion attacks involves carefully perturbing the input samples at test time to have them misclassified. For example, using techniques such as Shadow attack and Threshold attack.

2. Poisoning: Poisoning is adversarial contamination of training data. Machine learning systems can be re-trained using data collected during operations. An attacker may poison this data by injecting malicious samples during operation that subsequently disrupt retraining. For example, using techniques such as Backdoor attack and Adversarial backdoor embedding.

3. Extraction: Extraction attacks aim to duplicate a machine learning model through query access to a target model. For example, using techniques such as KnockoffNets and Functionally equivalent extraction.

4. Inference: Inference attacks determine if a sample of data was used in the training dataset of an AI/ML model. For example, using techniques such as Membership inference black-box and attribute inference black-box.

In addition to adversarial robustness, there are other aspects of AI/ML technical robustness such as dealing with missing data, erroneous data, confidence levels of data, fallback plans etc., which needs to be addressed.

There are a number of approaches to defend AI/ML models against such adversarial attacks at each stage of the AI/ML design:

Preprocessor –For example, using techniques such as InverseGAN and DefenseGAN.

Postprocessor –For example, using techniques such as Reverse sigmoid and Rounding.

Trainer –For example, using techniques such as General adversarial training and Madry’s protocol.

Transformer –For example, using techniques such as Defensive distillation and Neural cleanse.

Detector –For example, using techniques such as Detection based on activations analysis and Detection based on spectral signatures.

Quantification of Robustness: There are several metrics that measure robustness of ML models such as Empirical Robustness and Loss Sensitivity.

An example Trustworthy Artificial Intelligence Framework (TAIF) for Cognitive Autonomous Networks (CAN) to facilitate the definition, configuration, monitoring and measuring of AI/ML model trustworthiness (e.g., fairness, explainability, robustness) for interoperable and multi-vendor environments, is shown in Figure 5 and further detail can be found from PCT/EP2021/062396.

In this example there is shown a network operator 501 which is configured to pass information to a policy manager 533 and AI trust engine 503.

The policy manager 533 is configured to receive information from the network operator 501. Furthermore the policy manager 533 is configured to receive or otherwise obtain a service definition or a business/customer intent. The service definition or the business/customer intent may include AI/ML trustworthiness requirements in addition to the Network/AI Quality of Service (QoS) requirements, and the TAIF is used to configure the requested AI/ML trustworthiness and to monitor and assure its fulfilment.

Thus for example the system can comprise a service management and orchestration 527 function configured to receive the service quality of service (QoS) from the policy manager 533. The service management and orchestration 527 is configured to control an element manager or virtual network function (VNF) manager or resource manager 531 based on the output of the service management and orchestration function 527.

Additionally the system can comprise a AI pipeline orchestrator 525. The AI pipeline orchestrator 525 is configured to obtain or receive an AI QoS rom the policy manager 533 and based on this, and in a manner similar as shown with respect to Figure 4, be configured to control the operations of the

Data Source Manager

509, 519,

Model Training Manager

511, 521 and

Model Inference Manager

513, 523 for the AI pipeline 1 505 and AI pipeline 2 515.

The TAIF introduced two further management functions, the AI Trust Engine (trustworthiness function) 503 (one per management domain) and the AI Trust Manager 507, 617 (one per AI/ML Pipeline 505, 515) and six new interfaces (T1-T6) that are configured to support the interactions in the TAIF. The AI Trust Engine 503 is configured to function as a centre for managing all AI trustworthiness related components in the network, whereas the

AI Trust Managers

507, 517 are use case and often vendor specific, with knowledge of the AI use case and how it is implemented.

Furthermore the example TAIF also employs the concept of AI Quality of Trustworthiness (AI QoT) to define AI/ML model trustworthiness in a unified way covering three factors, i.e., fairness, explainability and robustness. The AI QoT for example is passed from the policy manager 533 to the AI trust engine function 503 and is similar to how QoS is used for network performance.

An example QoT can be shown by the following table

A high-level generic workflow within the example TAIF is shown in Figure 6.

In this example workflow is shown the customer intent being provided to the policy manager function 533 as shown in Figure 6 by step 601.

Additionally is shown the network operator (via the policy manager function 533) specifying, for example over the T1 interface, the required AI QoT (use case-specific based on risk levels) to the AI Trust Engine 503 as shown in Figure 6 by step 603.

The AI Trust Engine 503 translates the AI QoT into specific AI trustworthy (i.e., fairness, explainability and robustness) requirements and identifies the affected use-case-specific AI Trust Manager (s) . Using the T2 interface, The AI Trust Engine 503 configures the AI Trust Managers 507 as shown in Figure 6 by step 605.

The use case specific and implementation-aware AI Trust Manager 507 is configured to configure, monitor, and measure AI trustworthy requirements for AI Data Source Manager 509 over the T3 interface as shown in Figure 6 by step 607.

Further the use case specific and implementation-aware AI Trust Manager 507 is configured to configure, monitor, and measure AI trustworthy requirements for AI Training Manager 511 over the T4 interface as shown in Figure 6 by step 609.

Additionally the use case specific and implementation-aware AI Trust Manager 507 is configured to configure, monitor, and measure AI trustworthy requirements for AI Inference Manager 513 over the T5 interface as shown in Figure 6 by step 611.

The measured or collected TAI metrics and/or TAI explanations from the AI Data Source Manager 509, AI Training Manager 511 and AI Inference Manager 513 regarding the AI Pipeline are pushed to the AI Trust Manager 507 over T3, T4 and T5 interfaces, respectively as shown in Figure 6 by steps 613, 615, and 617 respectively.

The AI Trust Manager 507 pushes the TAI metrics and/or TAI explanations to the AI Trust Engine 503, over the T2 interface, based on the reporting mechanisms configured by the AI Trust Engine (trustworthiness function) as shown in Figure 6 by step 619.

Finally, the network operator 501 can request (as shown in Figure 6 by step 621) and receive (as shown in Figure 6 by step 623) the TAI metrics/explanations of an AI Pipeline from the AI Trust Engine over the T6 interface.

Based on the information retrieved, the Network Operator may decide to update the policy via Policy/Intent Manager as shown in Figure 6 by step 625.

The example TAI Framework thus enables various telco stakeholders (e.g., Cognitive Network Function vendors, network operators, regulators, end users) to trust the decisions/predictions made by AI/ML models in the network.

An example Cross-domain management and orchestration architecture is furthermore shown in Figure 7. This example shows Cross-domain End-to-End (E2E) network service scenario, but other cross-domain non-E2E scenarios (i.e., within each domain) are possible. For example a core domain can recursively embed 3GPP defined Network Function (NF) domain and virtualization domain, a RAN domain can include Centralized Unit (CU) , Distributed Unit (DU) , Remote Radio Unit (RRU) , midhaul and fronthaul domains provided by different vendors.

In the example Cross-domain E2E network service scenario shown in Figure 7, the Cross-Domain Service Management Domain (CDSMD) (e.g., E2E Service Management Domain) 705 is located between a cross-domain policy/intent manager 703 and network operator 701 which is configured to control the cross-domain service MD 705 and the domain management domains below. For example as shown in Figure 7 there are shown a (first) domain 1 MD (e.g. RAN MD) 712, a (second) domain 2 MD (e.g. Transport MD) 722 and a (third) domain 3 MD (e.g. Core network MD) 732. The Cross-Domain Service Management Domain (CDSMD) 705 is configured to decompose the Cross-Domain E2E network service request (as per the service level agreement (SLA) ) , received from the network operator 701 or the customer (for example via the Cross-Domain Policy/Intent Manager 703) , into domain-specific (e.g., RAN, transport, core) network resource/service requirements and communicate them to the corresponding individual Management Domains (MDs) 712, 722, 732.

The

individual MDs

712, 722, 732 are furthermore configured to be responsible for ensuring that the domain-specific resource/service requirements are fulfilled, within their corresponding domains, by continuously monitoring the resource/service related Key Performance Indicators (KPIs) and reporting them to the CDSMD 705.

In the example shown in Figure 7 there is shown the (first) domain 1 MD (e.g. RAN MD) 712. The (first) domain 1 MD (e.g. RAN MD) 712 can furthermore be controlled/monitored by a domain specific policy/intent manager 711 (which is similar to the policy/intent manager shown in Figures 5 and 6) . The (first) domain 1 MD (e.g. RAN MD) 712 can comprise a domain 1 AI Trust engine 713 (or trustworthiness function) and domain 1 AI pipeline orchestrator 714 both of which manage and control the domain 1 AI pipeline 1 715. The domain 1 AI pipeline 1 715 can then control or configure the domain specific aspects, the domain 1 resources (RAN resources 716) of the cross-domain network service 717.

Furthermore the example shown in Figure 7 shows the (second) domain 2 MD (e.g. Transport MD) 722. The (second) domain 2 MD 722 can furthermore be controlled/monitored by a domain specific policy/intent manager 721 (which is similar to the policy/intent manager shown in Figures 5 and 6) . The (second) domain 2 MD 722 can comprise a domain 2 AI Trust engine 723 (or trustworthiness function) and domain 2 AI pipeline orchestrator 724 both of which manage and control the domain 2 AI pipeline 2 725. The domain 2 AI pipeline 2 725 can then control or configure the domain specific aspects, the domain 2 resources (Transport resources 726) of the cross-domain network service 717.

Additionally as shown in the example of Figure 7 there can be employed a (third) domain 3 MD (e.g. Core network MD) 732. The (third) domain 3 MD 732 can furthermore be controlled/monitored by a domain specific policy/intent manager 731 (which is similar to the policy/intent manager shown in Figures 5 and 6) . The (third) domain 3 MD 732 can comprise a domain 3 AI Trust engine 733 (or trustworthiness function) and domain 3 AI pipeline orchestrator 734 both of which manage and control the domain 3 AI pipeline 3 735. The domain 3 AI pipeline 3 735 can then control or configure the domain specific aspects, the domain 3 resources (Core network resources 736) of the cross-domain network service 717.

Thus the requested/instantiated Cross-Domain E2E network service e.g., covering RAN, transport and core domains (represented by the box 717) , are managed by their corresponding AI pipelines (or CNFs) in the respective MDs (represented in

boxes

715, 725, 735) . It is to be noted that depending on the use case the AI pipeline may be instantiated either in the domain-specific MDs (e.g., for proactive resource autoscaling) or within the domain itself (e.g., for proactive mobility handover in RAN domain) . Then, leveraging the domain-specific AI Trust Engine (trustworthiness function) and the AI pipeline-specific AI Trust Manager, as described and referenced above, the AI pipeline trustworthiness (i.e., AI QoT comprising of fairness, explainability, robustness) for the domain-specific AI pipelines may be defined, configured, measured and reported within the corresponding MD.

However the network as shown in Figure 7 can be problematic as there is no management function in the CDSMD 703 configured to receive the desired Cross-Domain AI QoT (i.e., defined by the Cross-Domain Policy/Intent Manager based on the risk level of the Cross-Domain E2E network service) for the Cross-Domain E2E network service. Consequently, there is no way for the CDSMD to:

Translate the Cross-Domain AI QoT into domain-specific AI QoT;

Discover the TAI capability information from the domain-specific AI Trust Engine (s) ;

Communicate the translated domain-specific AI QoT to the domain-specific AI Trust Engine (s) ; and

Collect/request the Cross-Domain TAI metrics/explanations from the domain-specific AI Trust Engine (s) .

Furthermore even resolving this the CDSMD has no way to address (e.g., performing root-cause analysis) the TAI-related escalations (i.e., in terms of TAI metrics/explanations) , belonging to a Cross-Domain E2E network service, received from the domain-specific AI Trust Engine (s) and there is no way to delegate the relevant TAI escalation-related information received from the domain-specific AI Trust Engine of one MD to another MD so that the other MD may take preventive measures to avoid Cross-Domain E2E network service SLA violations (in our case Cross-Domain AI QoT) .

Additionally such systems fail to provide the CDSMD with a way to aggregate the TAI-related escalation metrics/explanations received from the AI Trust Engine (s) of individual MDs to provide a global view of the problem to the network operator or the customer. In this example, the TAI-related escalation metrics/explanations can comprise Cross-Domain AI QoT violations.

The concept as further discussed in the following embodiments is the introduction of a cross-domain TAI Framework (which extends the domain-specific TAI Framework discussed above and introduced in PCT/EP2021/062396 for Cognitive Autonomous Networks to facilitate the definition, configuration, measurement and reporting of Cross-Domain network service-related AI pipelines trustworthiness (i.e., fairness, explainability, robustness) for interoperable and multi-vendor environments.

In these embodiments a customer intent corresponding to a network service may include Cross-Domain AI trustworthy requirements in addition to the Cross-Domain QoS requirements, and the cross-domain TAI Framework is used to ensure the fulfilment of desired Cross-Domain AI trustworthy requirements.

As shown in Figure 8, the cross-domain TAI Framework comprises a novel management function, the Cross-Domain AI Trust Engine (trustworthiness function) 801 (or trustworthiness function) . The Cross-Domain AI Trust Engine in some embodiments is employed within the Cross-Domain Service Management Domain (CDSMD) . Additionally in some embodiments there is implemented a new interface (which in the example shown in Figure 8 is designated TCD-1) 805 that is configured to support the interactions between the Cross-Domain AI Trust Engine 801 and the domain-specific AI Trust Engine (s) 813, 823, 833 (or trustworthiness functions) . Furthermore in some embodiments there is implemented another new interface (which in the example shown in Figure 8 is designated TCD-2) 803 between the Cross-Domain AI Trust Engine 801 and the domain-specific Policy/Intent Manager (s) 811, 821, 831.

Furthermore in the embodiments described in detail herein a concept of Cross-Domain AI QoT is introduced to define the Cross-Domain AI trustworthiness in a unified way covering the domain-specific AI QoT requirements and constraints of Cross-Domain network service-related AI pipelines.

In some embodiments the Cross-Domain AI Trust Engine 801 is configured to support the following operations:

Translation of the Cross-Domain AI QoT requirements to domain-specific AI QoT requirements (e.g., RAN domain AI QoT, transport domain AI QoT and core domain AI QoT) depending on the risk-level of the Cross-Domain network service;

Discovery/determination of the TAI methods and/or TAI metrics and/or TAI explanations that the domain-specific AI Trust Engine is capable to configure in the domain-specific AI pipeline belonging to the Cross-Domain network service;

Verification of whether the Cross-Domain AI QoT and/or the domain-specific AI QoT requirements for the Cross-Domain network service are satisfied;

Configuration/delegation of the desired/updated AI QoT (derived from the Cross-Domain AI QoT) that the domain-specific AI Trust Engine is required to meet in the domain-specific AI pipeline belonging to the Cross-Domain network service;

Requesting/subscription for TAI metrics and/or TAI explanations that the domain-specific

AI Trust Engine

813, 823, 833 is capable of measuring/reporting/escalating in the domain-specific AI pipeline belonging to the Cross-Domain network service;

Storing all the TAI capability information and/or TAI reports received from the domain-specific AI Trust Engine (s) 813, 823, 833 about the domain-

specific AI pipelines

715, 725, 735 belonging to the Cross-Domain network service in a Cross-Domain Trust Knowledge Database;

Performing root-cause analysis of the TAI reports received from the domain-specific AI Trust Engine (s) 813, 823, 833. Also, if needed, updating the domain-specific AI QoT requirements based on the TAI reports;

Providing a global view of the problem/escalation with respect to the Cross-Domain network service (e.g., aggregated Cross-Domain network service-related TAI report) –which may be in this example Cross-Domain AI QoT violations -to the network operator.

In some embodiments there can be implemented Cross-Domain TAI APIs (which can be produced by domain-specific AI Trust Engine (s) and consumed by Cross-Domain AI Trust Engine) . These can for example be the following:

1. Cross-Domain TAI Capability Discovery API (Req/Resp) –The Cross-Domain TAI Capability Discovery API is configured to allow the Cross-Domain AI Trust Engine, via the TCD-1 interface, to discover TAI methods and/or TAI metrics and/or TAI explanations that the domain-specific AI Trust Engine is capable of configuring in the domain-specific AI pipeline belonging to the Cross-Domain network service.

2. Cross-Domain TAI Configuration API or Cross-Domain TAI Delegation API (Req/Resp) . The Cross-Domain TAI Configuration API or Cross-Domain TAI Delegation API is configured to allow the Cross-Domain AI Trust Engine, via the TCD-1 interface, to configure/delegate the desired/updated AI QoT (derived from the Cross-Domain AI QoT) that the domain-specific AI Trust Engine is required to meet in the domain-specific AI pipeline belonging to the Cross-Domain network service. Alternatively in some embodiments the Cross-Domain TAI Configuration API or Cross-Domain TAI Delegation API is configured to allow the Cross-Domain AI Trust Engine, via the TCD-2 interface, to notify the desired/updated AI QoT (derived from the Cross-Domain AI QoT) that the domain-specific Policy/Intent Manager (via domain-specific AI Trust Engine) is required to configure in the domain-specific AI pipeline belonging to the Cross-Domain network service.

3. Cross-Domain TAI Reporting API or Cross-Domain TAI Escalation API (Req/Resp &Subscribe/Notify) . The Cross-Domain TAI Reporting API or Cross-Domain TAI Escalation API is configured to allow the Cross-Domain AI Trust Engine, via the TCD-1 interface, to request/subscribe for TAI metrics and/or TAI explanations that the domain-specific AI Trust Engine is capable of measuring/reporting/escalating in the domain-specific AI pipeline belonging to the Cross-Domain network service.

Figure 9 shows an example workflow diagram showing the implementation of the Cross-Domain AI Trust Engine 801 according to some embodiments. Furthermore the Figure also shows the application of the APIs described above and offered by the domain-specific AI Trust Engine (s) over the TCD-1 interface to discover, configure, measure and query/collect TAI methods and/or TAI metrics and/or TAI explanations from the domain-specific AI pipeline belonging to the Cross-Domain network service.

Furthermore is also shown a further example alternative implementation showing the interaction between the Cross-Domain AI Trust Engine and the domain-specific Policy/Intent Manager, over the TCD-2 interface.

As shown in Figure 9 by step 901, the network operator 701 informs the Cross-Domain Policy/Intent Manager 703 (within the cross-domain service management domain 705) , for example over the T1 interface, the Intent for the Cross-Domain network service.

Then the Cross-Domain Policy/Intent Manager 703 is configured to translate the Intent for the Cross-Domain network service into a Cross-Domain AI QoT and informs the Cross-Domain AI Trust Engine 801 as shown in Figure 9 by step 903.

The Cross-Domain AI Trust Engine 801 can then be configured to translate the obtained Cross-Domain AI QoT requirements into domain-specific AI QoT requirements as shown in Figure 9 by step 905. In other words the Cross-Domain AI Trust Engine 801 is configured to generate the parameters RAN domain AI QoT, transport domain AI QoT and core domain AI QoT depending on the risk-level of the Cross-Domain network service. The generation of the parameters RAN domain AI QoT, transport domain AI QoT and core domain AI QoT depending on the risk-level of the Cross-Domain network service can be shown for example by the following table:

In some embodiments the translation/mapping logic may take into account the SLA requirements (e.g., service type, service priority, KPI metrics) for the Cross-Domain network service and, optionally, also the domain-specific TAI capability information. In some embodiments the translation may be performed after step 911.

The operations shown in Figure 9 as

steps

907, 909 and 911 can be implemented as part of a Cross-Domain TAI Capability Discovery API as shown in Figure 9 as reference 906.

For example the cross-domain AI Trust engine 801 can be configured to generate a Cross-Domain TAI Capability Information Request (CDTAICIReq) . The request can be sent as shown in Figure 9 by step 907 from the Cross-Domain AI Trust Engine 801 to the domain-specific AI Trust Engine (s) . In this example the domain specific AI Trust Engine is a RAN management domain 712 AI Trust Engine 813. The request is for requesting information concerning TAI methods and/or TAI metrics and/or TAI explanations that the AI Trust Engine (s) 813 is capable of configuring in the domain-specific AI pipeline belonging to the Cross-Domain network service.

In some embodiments an example request (CDTAICIReq) comprises the following parameters:

In some embodiments the domain-specific AI Trust Engine (s) , which in this example is the RAN management domain 712 AI Trust Engine 813, is configured to determine all the information requested in the received request (CDTAICIReq) by interacting with the domain-specific AI pipeline as shown in Figure 9 by step 907. The interaction in this example is with the RAN management domain 712 AI pipeline 715 (for example an AI Trust Manager 902, AI Data Source Manager 903, AI Training Manager 904, AI Inference Manager 905) belonging to the Cross-Domain network service. The details of this operation can be implemented according to any suitable manner, for example as described in PCT/EP2021/071044.

In some embodiments the domain-specific AI Trust Engine, which in this example is the RAN management domain 712 AI Trust Engine 813, sends a Cross-Domain TAI Capability Information Response (CDTAICIResp) back to the requesting cross domain AI Trust engine 801 as shown in Figure 9 by step 911. The Cross-Domain TAI Capability Information Response (CDTAICIResp) in some embodiments comprises the information about the domain-specific AI pipeline (belonging to the Cross-Domain network service) on the supported TAI methods and/or TAI metrics and/or TAI explanations.

The CDTAICIResp in some embodiments comprises the following parameters:

In some embodiments the Cross-Domain AI Trust Engine is configured to store the Cross-Domain TAI Capability Information within the Cross-Domain Cross-Domain Trust Knowledge Database. The database may be updated if there is a capability update in one of the domains.

In some embodiments the Cross-Domain AI Trust Engine 801 is configured to determine whether the Cross-Domain AI QoT and/or the domain-specific AI QoT requirements can be satisfied based on the Cross-Domain TAI Capability Information Response or the Cross-Domain TAI Capability Information stored in the Cross-Domain Trust Knowledge Database as shown in Figure 9 by step 913. Additionally in some embodiments the Cross-Domain AI Trust Engine 801 is configured to translate the domain-specific AI QoT to fairness, explainability and robustness requirements, as shown and described previously, for determining the Cross-Domain AI QoT and/or the domain-specific AI QoT satisfaction outcome.

The Cross-Domain AI Trust Engine 801 can furthermore be configured, if the requirement can not be satisfied, to generate and send, as shown in Figure 9 by step 915, a negative acknowledgement (Cross-Domain AI QoT NACK) to the Cross-Domain Policy/Intent Manager 703. This is sent in response to the received cross-domain AI QoT from step 903.

The operations shown in Figure 9 as

steps

917, 919 and 921 show a first alternative as part of a Cross-Domain TAI Configuration API or Cross-Domain TAI Delegation API as shown in Figure 9 as reference 916.

For example a Cross-Domain TAI Configuration/Delegation (CRUD) Request (CDTAIConReq) is sent from the Cross-Domain AI Trust Engine 801 to the domain-specific AI Trust Engine (s) , which in this example is the RAN management domain 712 AI Trust Engine 813, to notify the translated domain-specific AI QoT required to be met in the domain-specific AI pipeline belonging to the Cross-Domain network service. This is shown in Figure 9 as step 917.

In some embodiments the Cross-Domain AI Trust Engine 801 is configured to also translate the domain-specific AI QoT into fairness, explainability and robustness requirements, as shown and described above, and include this information in the Cross-Domain TAI Configuration/Delegation (CRUD) Request. The CDTAIConReq in some embodiments comprises the following parameters:

In some embodiments the domain-specific AI Trust Engine is configured to configure the corresponding (i.e., based on the desired domain-specific AI QoT) fairness and/or explainability and/or robustness methods/metrics/explanations in the domain-specific AI pipeline belonging to the Cross-Domain network service based on the Cross-Domain TAI Configuration/Delegation CRUD Request. The configuration is shown in Figure 9 by step 919. The implementation can be any suitable implementation, for example, such as described in PCT/EP2021/071044.

Depending on whether the configuration process in the previous step was successful or not, the domain-specific AI Trust Engine is configured to respond to the Cross-Domain AI Trust Engine with a Cross-Domain TAI Fairness Configuration/Delegation CRUD Response (CDTAIConResp) . The response in some embodiments comprises an ACK/NACK for satisfying the domain-specific AI QoT in the domain-specific AI pipeline belonging to the Cross-Domain network service. The generation and sending the response is shown in Figure 9 by step 921.

The operations shown in Figure 9 as

steps

923, 925, 927, 929, and 931 show a second alternative as part of a Cross-Domain TAI Configuration API or Cross-Domain TAI Delegation API as shown in Figure 9 as reference 922.

The Cross-Domain AI Trust Engine is configured to generate and pass a Cross-Domain TAI Configuration/Delegation CRUD Request (CDTAIConReq) to the domain-specific Policy/Intent Manager 811 to notify the domain-specific Policy/Intent Manager 811 about the translated domain-specific AI QoT required to be met in the domain-specific AI pipeline belonging to the Cross-Domain network service. The operation of generating and passing the CDTAIConReq is shown in Figure 9 by step 923. In some embodiments the CDTAIConReq may comprise the following parameters:

The domain-specific Policy/Intent Manager 811 is configured to send the desired AI QoT information to the domain-specific AI Trust Engine 813 as shown in Figure 9 by step 925.

The domain-specific AI Trust Engine 813 can then, based on the Cross-Domain TAI Configuration/Delegation CRUD Request, translate the AI QoT into fairness, explainability and robustness requirements and configure the corresponding fairness and/or explainability and/or robustness methods/metrics/explanations in the domain-specific AI pipeline belonging to the Cross-Domain network service. The configuration is shown in Figure 9 by step 927. The implementation can be any suitable implementation, for example, such as described in PCT/EP2021/071044] .

The domain-specific AI Trust Engine 813 can then be configured to send the ACK/NACK for satisfying the desired AI QoT in the domain-specific AI pipeline belonging to the Cross-Domain network service to the domain-specific Policy/Intent Manager 811. The operation of sending the ACK/NACK is shown in Figure 9 by step 929.

Then based on whether the domain-specific AI QoT was satisfied or not, the domain-specific Policy/Intent Manager 811 is configured to respond to the Cross-Domain AI Trust Engine with the Cross-Domain TAI Fairness Configuration/Delegation CRUD Response (CDTAIConResp) containing an ACK/NACK for satisfying the domain-specific AI QoT in the domain-specific AI pipeline belonging to the Cross-Domain network service. The generation and sending the Cross-Domain TAI Fairness Configuration/Delegation CRUD Response (CDTAIConResp) is shown in Figure 9 by step 931.

The operations shown in Figure 9 as

steps

933, 935, 937, 939, and 941 show an example Cross-Domain TAI Reporting API or Cross-Domain TAI Escalation API implementation as shown in Figure 9 as reference 932.

In some embodiments the domain-specific AI Trust Engine 933 is configured to collect the TAI reports (metrics and/or explanations) , configured in step 919 or step 927, from the domain-specific AI pipeline belonging to the Cross-Domain network service. The collection of the reports is shown in Figure 9 by step 933. The implementation can be any suitable implementation, for example, such as described in PCT/EP2021/071044.

The network operator 701 in some embodiments generates and passes a request or subscribes for an Cross-Domain TAI Report as shown in Figure 9 by step 935.

A Cross-Domain TAI Report Request/Subscribe (CDTAIRReq/CDTAIRSub) is generated and sent, as shown in Figure 9 by step 937, from the Cross-Domain AI Trust Engine 801 to the domain-specific AI Trust Engine 813 containing the reporting/subscription configuration for domain-specific AI pipeline belonging to the Cross-Domain network service. In some embodiments the CDTAIRReq in some embodiments comprises the following parameters:

In some embodiments the CDTAIRSub comprises the following parameters:

In some embodiments the Cross-Domain AI Trust Engine may store the Cross-Domain TAI Reports within the Cross-Domain Trust Knowledge Database.

In some embodiments, when one or more reporting characteristics (i.e., periodic or on-demand) is met, then the domain-specific AI Trust Engine 813 is configured to send a Cross-Domain TAI Report Response (TAIRResp) to the Cross-Domain AI Trust Engine 801 as per the reporting configuration specified in the CDTAIRReq. In some embodiments when one or more reporting thresholds are met for the applicable TAI metrics then the domain-specific AI Trust Engine 813 is configured to generate and send the Cross-Domain TAI Report Notify (TAIRNot) message to the Cross-Domain AI Trust Engine 801 comprising the actual TAI reports. The generation of the Cross-Domain TAI Report Response (TAIRResp) or the he Cross-Domain TAI Report Notify (TAIRNot) is shown in Figure 9 by step 939.

The Cross-Domain AI Trust Engine 801 then performs the root-cause analysis of the TAI reports coming from domain-specific AI Trust Engine (s) and provides a global view of the problem with respect to the Cross-Domain network service (e.g., aggregated Cross-Domain network service-related TAI report) . For example a Cross-Domain AI QoT violations which can be sent to the network operator 701 as shown in Figure 9 by step 941. The aggregation logic may consider combining/aggregating data/training/inference related local/global explanations received from individual domain-specific AI Trust Engine (s) .

Figure 10 shows a schematic representation of non-volatile memory media 1000a (e.g. computer disc (CD) or digital versatile disc (DVD) ) and 1000b (e.g. universal serial bus (USB) memory stick) storing instructions and/or parameters 1002 which when executed by a processor allow the processor to perform one or more of the steps of the methods as described above.

It is noted that while the above describes example embodiments, there are several variations and modifications which may be made to the disclosed solution without departing from the scope of the present invention.

It will be understood that although the above concepts have been discussed in the context of a 5GS, one or more of these concepts may be applied to other cellular systems.

The embodiments may thus vary within the scope of the attached claims. In general, some embodiments may be implemented in hardware or special purpose circuits, software, logic or any combination thereof. For example, some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device, although embodiments are not limited thereto. While various embodiments may be illustrated and described as block diagrams, flow charts, or using some other pictorial representation, it is well understood that these blocks, apparatus, systems, techniques or methods described herein may be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.

The embodiments may be implemented by computer software stored in a memory and executable by at least one data processor of the involved entities or by hardware, or by a combination of software and hardware. Further in this regard it should be noted that any procedures, e.g., as in Figures 7 and 8, may represent program steps, or interconnected logic circuits, blocks and functions, or a combination of program steps and logic circuits, blocks and functions. The software may be stored on such physical media as memory chips, or memory blocks implemented within the processor, magnetic media such as hard disk or floppy disks, and optical media such as for example DVD and the data variants thereof, CD.

The memory may be of any type suitable to the local technical environment and may be implemented using any suitable data storage technology, such as semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory. The data processors may be of any type suitable to the local technical environment, and may include one or more of general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs) , application specific integrated circuits (ASIC) , gate level circuits and processors based on multi-core processor architecture, as non-limiting examples.

Alternatively or additionally, some embodiments may be implemented using circuitry. The circuitry may be configured to perform one or more of the functions and/or method steps previously described. That circuitry may be provided in the base station and/or in the communications device.

As used in this application, the term “circuitry” may refer to one or more or all of the following:

(a) hardware-only circuit implementations (such as implementations in only analogue and/or digital circuitry) ;

(b) combinations of hardware circuits and software, such as:

(i) a combination of analogue and/or digital hardware circuit (s) with software/firmware and

(ii) any portions of hardware processor (s) with software (including digital signal processor (s) ) , software, and memory (ies) that work together to cause an apparatus, such as the communications device or base station to perform the various functions previously described; and

(c) hardware circuit (s) and or processor (s) , such as a microprocessor (s) or a portion of a microprocessor (s) , that requires software (e.g., firmware) for operation, but the software may not be present when it is not needed for operation.

This definition of circuitry applies to all uses of this term in this application, including in any claims. As a further example, as used in this application, the term circuitry also covers an implementation of merely a hardware circuit or processor (or multiple processors) or portion of a hardware circuit or processor and its (or their) accompanying software and/or firmware. The term circuitry also covers, for example integrated device.

The foregoing description has provided by way of exemplary and non-limiting examples a full and informative description of some embodiments. However, various modifications and adaptations may become apparent to those skilled in the relevant arts in view of the foregoing description, when read in conjunction with the accompanying drawings and the appended claims. However, all such and similar modifications of the teachings will still fall within the scope as defined in the appended claims.

Claims

An apparatus comprising means configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.
The apparatus as claimed in claim 1, wherein the means configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains is configured to facilitate at least one of:

defining cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness;

configuring cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness;

measuring cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness; and

reporting cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness.
The apparatus as claimed in any of claims 1 or 2 wherein the cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function comprises at least one of:

fairness;

explainability; and

robustness.
The apparatus as claimed in any of claims 1 to 3, wherein the means configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains is configured to:

obtain a cross domain machine learning or artificial intelligence quality of trustworthiness, the cross-domain machine learning or artificial intelligence quality of trustworthiness configured to cover domain-specific machine learning or artificial intelligence quality of trustworthiness requirements and constraints of cross-domain network service-related machine learning or artificial intelligence pipelines;

translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains; and

provide the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains.
The apparatus as claimed in claim 4, wherein the means configured to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains is configured to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness based on a risk-level of the cross-domain network service.
The apparatus as claimed in any of claims 4 or 5, wherein the means configured to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains is configured to translate the cross-domain machine learning or artificial intelligence quality of trustworthiness into at least one domain specific machine learning or artificial intelligence quality of trustworthiness based on at least one service level agreement requirement for the cross-domain network, wherein the at least one service level agreement comprises at least one of: service type; service priority; and at least one key performance indicator metric.
The apparatus as claimed in any of claims 4 to 6, wherein the means configured to provide the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains is configured to:

generate and pass to at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request configured to control the at least one domain specific machine learning or artificial intelligence trustworthiness function to configure a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and

obtain from the at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation response based on an implementation of the configuration of the machine learning or artificial intelligence pipeline for the at least one of the at least two domains.
The apparatus as claimed in any of claims 4 to 6, wherein the means configured to provide the at least one domain specific machine learning or artificial intelligence quality of trustworthiness for at least one of the at least two domains is configured to:

generate and pass to at least one domain specific policy manager a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request configured to control at least one domain specific machine learning or artificial intelligence trustworthiness function to configure a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and

obtain from the at least one domain policy manager a cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation response based on an implementation of the configuration of the machine learning or artificial intelligence pipeline for the at least one of the at least two domains.
The apparatus as claimed in any of claims 7 or 8, wherein the cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request comprises:

a domain scope parameter configured to identify a domain the request is addressing;

a pipeline identification parameter configured to identify a domain-specific machine learning or artificial intelligence pipeline the request is addressing;

a class identification parameter configured to identify the at least one domain specific machine learning or artificial intelligence quality of trustworthiness.
The apparatus as claimed in claim 9, when dependent on claim 7, wherein the cross-domain trustworthiness machine learning or artificial intelligence configuration or delegation request further comprises at least one of:

a desired fairness parameter configured to indicate a relative fairness level for the domain-specific machine learning or artificial intelligence pipeline;

a desired explainability parameter configured to indicate a desired explainability level for the domain-specific machine learning or artificial intelligence pipeline;

a desired technical robustness parameter configured to indicate a desired technical robustness level for the domain-specific machine learning or artificial intelligence pipeline; and

desired adversarial robustness parameter configured to indicate a desired adversarial robustness level for the domain-specific machine learning or artificial intelligence pipeline.
The apparatus as claimed in any of claims 4 to 10, the means configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains is configured to:

generate and pass to at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence capability information request configured to control the at least one domain specific machine learning or artificial intelligence trustworthiness function to implement a capability discovery for a machine learning or artificial intelligence pipeline for the at least one of the at least two domains; and

obtain from the at least one domain specific machine learning or artificial intelligence trustworthiness function a cross-domain trustworthiness machine learning or artificial intelligence capability information response reporting the capability discovery for a machine learning or artificial intelligence pipeline for the at least one of the at least two domains.
The apparatus as claimed in claim 11, wherein the cross-domain trustworthiness machine learning or artificial intelligence capability information request comprises:

a domain scope parameter configured to identify a domain the request is addressing; and

a scope parameter configured to identify a domain-specific machine learning or artificial intelligence pipeline the request is addressing.
The apparatus as claimed in claim 12, wherein the cross-domain trustworthiness machine learning or artificial intelligence capability information request further comprises a pipeline phase parameter configured to identify a phase of the domain-specific machine learning or artificial intelligence pipeline the request is addressing.
The apparatus as claimed in any of claims 4 to 13, wherein the means configured to facilitate a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains is further configured to:

obtain a cross-domain trustworthiness artificial intelligence report request or subscription from a network operator;

generate and pass to at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function a domain specified trustworthiness artificial intelligence report request or subscription based on the cross-domain trustworthiness artificial intelligence report request or subscription from a network operator, wherein the domain specified trustworthiness artificial intelligence report request or subscription is configured to control the at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function to provide at least one domain specific machine learning or artificial intelligence pipeline report response or notification;

receive from the at least one domain specific machine learning or artificial intelligence pipeline trustworthiness function machine learning or artificial intelligence capability information and/or the at least one domain specific machine learning or artificial intelligence pipeline report response or notification;

store machine learning or artificial intelligence capability information and/or the at least one domain specific machine learning or artificial intelligence pipeline report received from the at least one domain specific machine learning or artificial intelligence pipeline in a cross-domain trust knowledge database;

generate and pass a cross-domain trustworthiness artificial intelligence report based on the at least one domain specific machine learning or artificial intelligence pipeline report response or notification.
The apparatus as claimed in claim 14, wherein the cross-domain trustworthiness artificial intelligence report request comprises:

a domain scope parameter configured to identify a domain the request is addressing;

a scope parameter configured to identify a domain-specific machine learning or artificial intelligence pipeline the request is addressing; and

a pipeline phase parameter configured to identify a phase of the domain-specific machine learning or artificial intelligence pipeline the request is addressing.
The apparatus as claimed in claim 15, wherein the cross-domain trustworthiness artificial intelligence report request further comprises at least one of:

a list of fairness metrics parameter configured to identify fairness metrics to be reported;

a list of fairness metric explanations configured to identify fairness metric explanations to be reported;

a list of explainability metrics configured to identify explainability metrics to be reported;

a list of explanations configured to identify explanations to be reported;

a list of technical robustness metrics configured to identify technical robustness metrics to be reported;

a list of technical robustness metric explanations configured to identify technical robustness metric explanations to be reported;

a list of adversarial robustness metrics configured to identify adversarial robustness metrics to be reported;

a list of adversarial robustness metric explanations configured to identify adversarial robustness metric explanations to be reported;

a start time parameter configured to identify a start time for reporting;

an end time parameter configured to identify an end time for reporting; and

a report interval parameter configured to identify a periodicity interval for reporting.
The apparatus as claimed in claim 14, wherein the cross-domain trustworthiness artificial intelligence report subscription comprises:

a domain scope parameter configured to identify a domain the subscription is addressing;

a scope parameter configured to identify a domain-specific machine learning or artificial intelligence pipeline the subscription is addressing; and

a pipeline phase parameter configured to identify a phase of the domain-specific machine learning or artificial intelligence pipeline the subscription is addressing.
The apparatus as claimed in claim 17, wherein the cross-domain trustworthiness artificial intelligence report subscription further comprises at least one of:

a list of fairness metrics parameter configured to identify fairness metrics to be reported;

a list of explainability metrics configured to identify explainability metrics to be reported;

a list of technical robustness metrics configured to identify technical robustness metrics to be reported;

a list of adversarial robustness metrics configured to identify adversarial robustness metrics to be reported; and

a crossed reporting threshold parameter configured to identify a threshold for which a metric or metric explanation is reported.
A method comprising:

facilitating a cross-domain network service-related machine learning or artificial intelligence pipeline trustworthiness function, wherein the cross-domain machine learning or artificial intelligence pipeline is for control of a cognitive autonomous network comprising at least two domains.
A computer program comprising computer executable instructions which when run on one or more processors perform the steps of the method of claim 19.