Design of an Intelligent Trust Management Architecture for 5G Service

Deployment

Samra Bouakkaz

1,2 a

, Luis Su

arez

1 b

, Nora Cuppens

1 c

and Fr

eric Cuppens

2 d

Computer Engineering and Software Engineering, Polytechnique Montreal, Canada

Ericsson, Saint-Laurent, Canada

{samra.bouakkaz, luis.suarez}@ericsson.com, {nora.cuppens, frederic.cuppens}@polymtl.ca

Keywords:

5G Security, Trust Management Architecture, Trustworthiness Assessment, Trust Model.

Abstract:

The security of 5G networks relies on trust, but managing this is challenging due to their dynamic nature

and lack of a uniﬁed trust framework. Current research focuses on trust evaluation mechanisms, neglecting a

comprehensive architecture. Hyperscale Cloud Providers (HCPs) are crucial in securing 5G network deploy-

ment, especially with virtualization. To enhance cloud service adoption, HCPs must demonstrate trust and

security while addressing end-user and industry concerns. This paper’s primary contribution is the design of a

comprehensive intelligent trust management architecture rather than focusing solely on speciﬁc implementa-

tions for particular 5G use cases. It serves as a blueprint for integrating and managing various trust evaluation

methods within a single framework, making it ﬂexible and adaptable for successful 5G service deployment.

We instantiate our architecture within 5G networks to tailor it to speciﬁc methods and techniques best suited

for different scenarios. Furthermore, we animate its dynamic adaptation to various 5G use cases, showcasing

real-time changes in trust levels and strategies to ensure secure and reliable service delivery.

1 INTRODUCTION

The 5G network is a technology platform that en-

ables new services and business models across var-

ious sectors. Open-source innovation, utilizing vir-

tualization technologies, automation, and elasticity,

has driven the expansion of cloud infrastructure ser-

vices. Initially focusing on the enterprise market,

cloud providers like Amazon Web Services, Mi-

crosoft Azure, and Google Cloud Platform offer

economies of scale and ﬂexibility in resource alloca-

tion. Communication Service Providers (CSPs) adopt

a hybrid and multi-cloud approach to lower owner-

ship costs and accelerate their market entry (Alonso

et al., 2022). As cloud providers evolve into Hyper-

scale Cloud Providers (HCPs), they deliver faster ser-

vices, greater capacity, and massive scalability, which

allows network operators to optimize their service

delivery and business models (Zhang et al., 2019).

Telecommunication partnerships must build trust to

enforce security requirements, protect information

sharing, and comply with policies, while HCPs must

https://orcid.org/0000-0001-5852-4956

https://orcid.org/0000-0002-7831-9252

https://orcid.org/0000-0001-8792-0413

https://orcid.org/0000-0003-1124-2200

address end-user and industry concerns in cloud ser-

vice adoption. Trust management in the evolving net-

work landscape faces challenges such as a lack of a

uniﬁed architecture, inconsistencies in trust models,

and difﬁculties accommodating emerging technolo-

gies like AI-driven automation and edge computing.

Scalability issues in large-scale 5G environments and

a lack of real-time solutions to security threats are

also signiﬁcant issues. However, trust is conﬁdence

in others’ abilities, which helps facilitate smooth col-

laboration within a network. Trustworthiness refers

to an entity’s demonstrated qualiﬁcations, capabili-

ties, and reliability (Scarfone and Hoffman, 2007).

To address these, we propose a new architecture for

real-time monitoring, dynamic adjustments, and scal-

ability in the 5G ecosystem, ensuring the trustworthi-

ness of the 5G ecosystem. To our knowledge, this pa-

per is the ﬁrst to design a comprehensive Intelligent

Trust Management Architecture for 5G services, us-

ing HCPs as the primary shared infrastructure. The

5G-TMA modules addresses four key contributions,

which are addressed by speciﬁc modules as follows:

1. Propose a trustworthiness assessment model for

5G stakeholders by collecting real-time data to

evaluate trust in their relationships.

2. Propose a trust assessment framework that gener-

ates a dynamic trust model fostering adaptable re-

310

Bouakkaz, S., Suárez, L., Cuppens, N. and Cuppens, F.

Design of an Intelligent Trust Management Architecture for 5G Service Deployment.

DOI: 10.5220/0013252200003899

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 11th International Conference on Information Systems Security and Privacy (ICISSP 2025) - Volume 1, pages 310-317

ISBN: 978-989-758-735-1; ISSN: 2184-4356

lationships and permitting only trustworthy stake-

holders to maintain high trust.

3. Propose Trust Lifecycle Management (TLM)

among trustworthy entities by establishing, as-

sessing, and revoking trust relationships, ensuring

trust stays dynamic and responsive.

4. Finally, it enhances decision-making, entity be-

havior understanding, system reliability by en-

abling proactive responses to trust issues.

This paper is organized as follows: Section 2

offers an overview of related research. Section 3

presents a use-case scenario. In Section 4, we explain

the proposed 5G-TMA; in this section, 5, we discover

the integration of 5G-TMA within the 5G architec-

ture. Section 6 provides an example of how this 5G-

TMA can be instantiated, followed by a demonstra-

tion of its animation in Section 7. We discuss future

directions in Section 8, and Section 9 presents our ﬁ-

nal thoughts.

2 RELATED WORK

Network operators can collaborate with cloud infras-

tructure and specialized service providers to enhance

service delivery and make network capabilities avail-

able to third-party providers (Maman et al., 2021).

This collaboration will signiﬁcantly impact business

models and the dynamics of the mobile market, neces-

sitating the establishment of new trust relationships in

security design. Trust, deﬁned as the belief that one

entity excels in a speciﬁc task or action, is essential

in communication and networking to mitigate risks

(Svare et al., 2020). Zhang et al. (Zhang et al., 2019)

highlighted the importance of trust management in

5G systems, emphasizing collaboration between net-

work operators and vertical service providers to cre-

ate and manage trust. This collaboration is essen-

tial for enabling secure and efﬁcient identity man-

agement. The new trust model requires additional

security measures, including authentication, account-

ability, and non-repudiation. Valero et al. (Valero

et al., 2023) developed a reputation-based trust frame-

work using PeerTrust that predicts trust scores based

on stakeholder behavior patterns. These scores are

adjusted according to breach predictions, detections,

and violations. It emphasizes trust in deployment ser-

vices, highlighting its importance in 5G adoption, re-

liability, transparency, and stakeholder accountability,

as it is a strategic goal for the ecosystem’s success.

Wary et al. (Wary et al., 2019) point out the com-

plexity of managing trust in 5G, as it involves mon-

itoring both the network’s and the service’s reliabil-

Figure 1: Use case: instantiating an rAPP within an SMO.

ity. The competition among providers offering vari-

ous guarantees and trust levels makes implementing

a dynamic and intelligent system necessary. Collab-

oration between CSPs and HCPs is crucial for de-

ploying cloud-native 5G Network Functions (NFs),

focusing on trust to enhance stakeholder cooperation

and data exchange. Current trust management ap-

proaches are inconsistency-prone and lack scalability

for multi-tenant and multi-domain scenarios. A more

sophisticated trust management architecture and scal-

able mechanisms are needed to address these issues to

adapt to evolving conditions and emerging threats.

3 USE CASE: rApp IN SMO

In this motivating scenario (refer to Figure 1), a Mo-

bile Network Operator (MNO), which is a telecom-

munications service provider, utilizes a Service Man-

agement and Orchestration (SMO) system to manage

its Radio Access Network (RAN). The SMO hosts

rApps, software applications designed to operate on

the Non-Real Time RAN Intelligent Controller (Non-

RT RIC). These applications aim to automate various

RAN management and optimization tasks, function-

ing with control loops that take effect over a time scale

of one second or longer. The MNO plans to deploy a

rApp developed by a software provider through its in-

stantiation over the HCP. The SMO components are

instantiated in the HCP, allowing the rApp to access

the Network Functions (NFs) within the RAN. In this

context, several interesting questions arise:

• How to assess trustworthiness evidence from

HCPs?

• How does the MNO choose the appropriate HCP

based on high conﬁdence?

• How can a relationship of trust can be built con-

tinuously between trustworthy parties?

Design of an Intelligent Trust Management Architecture for 5G Service Deployment

311

4 THE PROPOSED 5G-TMA

We design a generic trust management architecture

for a 5G network (5G-TMA) to support various use

cases requiring different levels of trust, security, and

reliability, consisting of four main modules (M1, M2,

M3, and M4), each making speciﬁc contributions, as

illustrated in Figure 2.

Figure 2: The Proposed 5G-TMA.

1. The Trustworthiness Assessment Module. The

5G ecosystem requires meticulous collection and

analysis of trust evidence to promote global technol-

ogy adoption and mitigate security and privacy risks.

We begin with the ﬁrst sub-contribution, which is

proposing the trustworthiness model of 5G stakehold-

ers. The model uses a trust-based fuzzy logic system

to gather, assess, and organize evidence on trust met-

rics, security, and data sources through various inter-

faces to create trust proﬁles with all scores related to

trust for each stakeholder. The trustworthiness assess-

ment module consists of three interfaces: Trust Data

Submission Interface, Data Source Listing Interface,

and Interface Security Mechanisms. The Trust Data

Submission Interface allows nodes to share informa-

tion about their interactions with network providers or

services, enabling third-party recommendations and

indirect trust assessments. The Data Source Listing

Interface presents available data sources related to

trust, helping stakeholders determine how to utilize

the trust framework effectively. The Interface Secu-

rity Mechanisms ensure secure interactions between

network functions using secure API access tokens and

endpoint protection. However, the model’s effective-

ness in identifying trust strategies and relationships

among HCPs for 5G service deployment is limited,

necessitating an additional assessment for improve-

ment. The ﬁrst module collects trust information from

each 5G stakeholder, compiles it into a trust proﬁle,

and sends it to the second module for further analysis.

2. The Trust Assessment Module. We propose a

trust assessment framework that integrates game the-

ory (GT) predictive capabilities with reinforcement

learning (RL) adaptability. The framework evalu-

ates the trust relationships among 5G stakeholders

and creates a trust model to determine their eligibil-

ity for partnerships in deploying 5G services. This

model adapts to the evolving landscape of 5G tech-

nology, allowing stakeholders to adjust trust strategies

in real-time for successful deployment and operation.

It supports a dynamic trust decision-making process,

fostering strong stakeholder relationships. It includes

several key interfaces. The UseCase interface for 5G-

TMA demonstrates how different layers and compo-

nents interact with 5G-TMA for trust validation, scor-

ing, and policy enforcement. The Cross-layer Trust

interface also allows communication between the ser-

vice layer and 5G-TMA, facilitating dynamic policy

enforcement and adjustments to trust scores.

3. The Trust Lifecycle Management Module.

Based on the generated trust model, stakeholders en-

tering into partnerships are considered trustee part-

ners. It is essential to assess how trust information

is exchanged among these partners, which occurs in

several phases known as trust lifecycle management

(TLM). This process involves managing the trust rela-

tionships among stakeholders participating in the trust

model, including the phases of trust establishment,

evaluation, maintenance, and revocation, as detailed

in Section 6.3.

4. The Trust Decision-Making Module.In trust re-

lationships, reward and punishment mechanisms are

used to monitor stakeholders’ behaviors in real-time.

Trust scores are recalculated after new events, such as

security threats, policy changes, service-level agree-

ment (SLA) violations, service execution failures, and

time decay. Negative events decrease previous trust

scores, helping to identify more reliable stakehold-

ers, while positive events contribute to increased trust

scores. Implementing effective internal policies can

help maintain trust in established relationships.

5 INTEGRATION THE 5G-TMA

IN 5G ARCHITECTURE

The Business Support System (BSS) is a crucial com-

ponent of a telecommunications network owned by

the Network Operator. It manages customer busi-

ness operations. Additionally, the BSS oversees Ra-

dio Access Network (RAN) applications (referred to

as rApps) to enhance operational capabilities and im-

prove network efﬁciency. The Control Plane Network

Function (NF) is responsible for managing network

ICISSP 2025 - 11th International Conference on Information Systems Security and Privacy

312

Figure 3: Integration of 5G-TMA in 5G Architecture.

operations and data ﬂow, while the Access and Mo-

bility Management Function (AMF) focuses on user

mobility and session management. The BSS initiates

workﬂows to deploy necessary network functions,

with the Management and Orchestration (MANO)

system handling the orchestration of this deployment

process. The BSS establishes a well-coordinated

workﬂow to meet speciﬁc service requirements, with

the MANO system coordinating these requests for

optimal performance. The Network Operator over-

sees the BSS, Operations Support System (OSS),

and MANO, implementing a strategic plan to deploy

rApps within the SMO and NF in the Control Plane

(CP). The 5G-TMA is designed to collect and process

information from the Operations, Administration, and

Management (OAM) entities distributed across vari-

ous HCPs. This data collection takes place when the

5G-TMA modules (M1, M2, M3, and M4 as shown

in Figure 2) perform their speciﬁc functions within

the proposed 5G-TMA architecture. M2 is particu-

larly important in this process, as it determines which

HCP(s) should be selected based on the current re-

quirements and conditions. Once M2 has made the

selection, it communicates with the OSS to instruct

it to deploy and instantiate the relevant applications,

known as rApps, along with the AMF. This deploy-

ment occurs within the selected HCPs, ensuring that

the necessary resources and functionalities are avail-

able to support optimal network performance and ser-

vice delivery.

6 5G-TMA INSTANTIATION

This example demonstrates how to instantiate our ar-

chitecture in 5G networks, adapting it to the appropri-

ate methodologies and techniques for each use case’s

unique requirements and challenges.

6.1 Instantiate Trustworthiness

Assessment Module

Trustworthiness is often evaluated using qualitative

and ambiguous metrics like reliability and reputation,

which are challenging to measure with traditional

logic systems. Conventional probabilistic methods,

such as Bayesian inference (Wong, 2019) or Markov

models, struggle to accurately quantify trust relation-

ships (Nefti et al., 2005). The fuzzy logic system

effectively addresses uncertainty and imprecise in-

formation, which are the challenges of trustworthi-

ness. It allows for evaluating trust metrics that are not

strictly binary (such as low, medium, and high), of-

fering a more nuanced and continuous output. Inputs

like reputation, reliability, historical behavior, and se-

curity metrics are assessed using fuzzy rules to gen-

erate a trust score. These rules mimic human reason-

ing by skillfully handling vague and incomplete data

(Ampririt et al., 2021).

Figure 4: Module I: Trustworthiness Assessment based on

Fuzzy Logic.

The fuzzy system as illustrated in Figure 4 con-

sists of four main components: System Inputs, Fuzzi-

ﬁcation, Fuzzy Inference Engine, and Defuzziﬁca-

tion. System Inputs are variables analyzed to make

decisions in uncertain scenarios concerning trust met-

rics. Fuzziﬁcation determines the degree of mem-

bership for each input within various trust sets. The

Fuzzy Inference Engine processes fuzzy inputs based

on deﬁned trust rules, producing trust membership de-

grees. Defuzziﬁcation converts these outputs into a

deterministic format called a trust proﬁle. Fuzzy trust

sets represent varying levels of trust among 5G stake-

holders, with their corresponding membership func-

tions. A trust proﬁle structure is used to evaluate a

stakeholder’s trustworthiness, including stakeholder

identiﬁcation, overall trust score, trust metrics, con-

textual details, recommendations, and a timestamp.

The context highlights strengths and areas for im-

provement, while the recommendations offer strategic

enhanced trustworthiness and mitigate risks.

6.2 Instantiate Trust Assessment

Trust in network systems often relies on rational ex-

pectations and cooperation from trusted entities; in-

complete information can complicate the conﬁrma-

tion process. We propose a framework that combines

GT and RL to address these trust issues. GT analyzes

Design of an Intelligent Trust Management Architecture for 5G Service Deployment

313

strategic decision-making interactions that can lead to

conﬂicting outcomes, while RL helps agents under-

stand the consequences of their actions and enhances

their decision-making abilities. Trust-based RL opti-

mizes interactions by adjusting trust levels based on

the actions and decisions of others. Deep Reinforce-

ment Learning (DRL) (Xiong et al., 2019) and GT

together can tackle complex problems in dynamic en-

vironments characterized by high stakeholder mobil-

ity. We propose a dynamic trust model speciﬁcally de-

signed for 5G stakeholders. This model is tailored to

various use cases and effectively addresses the com-

plexities of trust relationships in advanced 5G net-

works. It allows stakeholders to adjust their strate-

gies in real-time, fostering trust and cooperation that

are essential for the successful deployment and op-

eration of 5G networks. As illustrated in Figure 5,

the trust assessment module evaluates the trust lev-

els among various 5G stakeholders and their relation-

ships within a partnership. Its primary goal is to iden-

tify effective strategies for successfully deploying 5G

services. This module creates a dynamic trust model

by employing predictive GT and adaptive RL. It uti-

lizes trust proﬁles, network metrics, and global data

to identify stakeholders involved in a trust-based part-

nership. Furthermore, it assesses their trust relation-

ships to develop a comprehensive trust model for col-

laboration.

Figure 5: Module II: Trust Assessment Based on GT & RL.

Step 1- Setup of Bayesian Game. Bayesian game

theory predicts trust strategies among 5G stakeholders

aligned with partnership objectives, considering their

private information and beliefs to optimize utility and

foster trusting relationships.

• The 5G ecosystem comprises stakeholders de-

noted as P = {p

, p

, . . . , p

}, including network

operators and HCPs, each characterized by a type

, encompassing their capabilities, strategic pref-

erences, and trust proﬁle. Each stakeholder P

has a trust strategy set, S

, based on quantitative

measures for decision-making, with the distribu-

tion function F

(θ

) representing initial assump-

tions about other stakeholders’ private informa-

tion in space S .

• Each stakeholder has a payoff function, repre-

sented by u

: S

× S

× ... × S

→ R, which maps

a combination of trust strategies to a real num-

ber representing the payoff to the stakeholder, de-

noted as player p

. This function considers the

individual stakeholder and others’ strategies, con-

sidering trust-related aspects.

Step 2: Setup the DRL. The State Space S repre-

sents all possible trust states observed by RL agents

within 5G stakeholders, including public perceptions

of different trust types and conditions. Each state

s ∈ S represents a vector of trust-related variables

s = {v

, v

, ..., v

}, and the Reward Function R(s, a)

offers immediate rewards for actions a and states s,

combining economic beneﬁts and trust-related out-

comes.

Step 4: DRL Algorithm. DRL is a machine learn-

ing technique that enables agents to make optimal de-

cisions without prior knowledge of the system or en-

vironment. It aims to maximize long-term rewards

by learning from past experiences. A trust model

is developed by integrating RL to optimize dynamic

trust strategies, reacting to current situations, and an-

ticipating future states based on stakeholder trust be-

haviours. GT provides structured insights into effec-

tive trust strategies, enhancing the learning process

within the RL process.

6.3 Instantiate TLM

The Trust Life Cycle Management (TLM) in Figure 6

facilitates collaboration among partners by establish-

ing trust agreements for data exchange and interaction

among 5G stakeholders. The MNO identiﬁes poten-

tial HCPs like AWS, Google Cloud, and Microsoft

Azure, establishes trust requirements based on regula-

tory standards, business needs, and security policies,

issues digital certiﬁcates, manages credentials, imple-

ments access control policies, and conducts continu-

ous checks to ensure only authorized HCPs can access

the network and sensitive data.

Maintaining trust requires continuously monitor-

ing and analyzing behavior, collecting feedback, and

enforcing strict access controls. We assess risks to

identify potential threats and vulnerabilities associ-

ated with each HCP, using quantitative trust metrics to

evaluate their trustworthiness. Additionally, we con-

duct compliance checks to ensure that HCPs adhere

to deﬁned trust policies and regulatory standards. We

also respect ZTA to verify every access request. Trust

evaluation involves updating policies based on trust

assessments, adjusting reputation scores, and consid-

ering thresholds. If an HCP is found to be non-

compliant or poses a security risk, their credentials

and access rights will immediately be revoked. Inci-

ICISSP 2025 - 11th International Conference on Information Systems Security and Privacy

314

Figure 6: Trust Life Cycle Management Phases.

dent response procedures will be activated to address

any breaches of trust and mitigate potential impacts

on the network and services. Revoking trust means

terminating the services provided by HCPs that are

no longer trusted or needed and ensuring that all data

is handled securely. Speciﬁc procedures are in place

to ensure that HCPs leave the system without com-

promising network security or data integrity.

7 5G-TMA ANIMATION

The 5G-TMA is animated to adapt to various scenar-

ios and demonstrate its real-world applicability. It

offers context-speciﬁc trust establishment, monitor-

ing, evaluation, and enforcement methodologies in re-

sponse to 5G stakeholder changes, speciﬁcally sup-

porting trust relationships between CSPs, MNOs, and

HCPs.

1. Animate Trustworthiness Assessment Process.

Data collection in O-RAN and 5G architectures in-

volves multiple stakeholders, each responsible for

speciﬁc data types. Efﬁcient data collection and

analysis are crucial for optimizing network perfor-

mance, ensuring security, and delivering high-quality

services. In O-RAN architecture, the Radio Intel-

ligent Controller (RIC), including both Near-Real-

Time (Near-RT) and Non-Real-Time (Non-RT) com-

ponents, along with interfaces such as E2 and O1,

play vital roles in data collection (Bonati et al., 2022).

In contrast, traditional 5G architectures rely on Op-

erations, Administration, and Maintenance (OAM)

systems, Network Management Systems (NMS), and

various network interfaces (F1, N2, N3, S1) for data

collection and management (Lei et al., 2021; 3GPP,

2020; Sharma et al., 2021). These components and

interfaces are designed to gather data efﬁciently, sup-

porting the optimization of network performance, the

maintenance of quality of service, and the facilitation

of advanced network management capabilities. Trust-

based fuzzy logic measures trustworthiness among

partners by adjusting trust attributes and normaliza-

tion processes based on speciﬁc requirements. This

results in a trust proﬁle for each partner, generated

using trust contribution rules from trust metrics, as

shown in Figure 7. The fuzzy set F

= {L

, M

, H

}

categorizes trust metric scores into low L

, medium

, and high H

, with membership functions 1, 2,

and 3 representing the minimum l, medium m, and

maximum h trust metric scores for each category.

(x) =







1 , (x <= l)

h−x

h−l

, (l < x < h)

0 , (x >= l)

(1)

(x) =







0 , (x <= l)or(x >= h)

x−l

m−l

, (l < x < m)

h−x

h−m

, (m < x < h))

(2)

(x) =







0 , (x <= m)

x−m

h−m

, (m < x < h)

1 , (m < x <= h))

(3)

To effectively deploy 5G services in HCPs, it is es-

sential to consider several trust attributes: security,

privacy, reliability, performance, compliance, trans-

parency, and interoperability. Security guarantees the

conﬁdentiality of data, while privacy safeguards user

information. Reliability ensures uninterrupted ser-

vice. Transparency involves clear data handling poli-

cies, and interoperability allows for seamless integra-

tion with existing systems and vendor equipment. The

rules determine how trust metrics such as reputation

scores R

, trust levels L

, behavioral trust B

, reliabil-

ity scores E

, security posture P

, and trustworthiness

index I

contribute to trust attributes by considering

expert knowledge, domain-speciﬁc requirements, and

trust metric characteristics, for instance, If R

is high

AND L

is high, Then Transparency is high. Attribute

values are calculated based on their corresponding

trust metrics, which are aggregated, normalized, or

weighted to create a trust attribute vector for each

stakeholder. This vector compiles with other informa-

tion into a structure called a trust proﬁle, as shown in

Figure 7. Additionally, We proposed Table 1 to illus-

trate the contribution of trust metrics in trust attributes

for deﬁning rules related to them.

2. Animate Trust Assessment Process. The trust

model, generated by GT and RL, considers stake-

holder OAM roles as DRL agents, enabling MNOs

to select HCPs for service deployment based on trust

metrics like reliability and data security.

Design of an Intelligent Trust Management Architecture for 5G Service Deployment

315

Table 1: Trust Metrics involved in the Trust Attribute.

Trust Metrics

Trust Attributes R

Security

Privacy

Performance

Transparency

Interoperability

Figure 7: Generate a Trust Proﬁle using Fuzzy Logic.

Step 1: Initial Setup. The process involves deﬁning

trust requirements, establishing an initial trust level,

and forming initial beliefs based on historical perfor-

mance or previous interactions. The initial beliefs of

the MNO regarding the trustworthiness of each HCP

range from 0 to 1, as follows:

• HCP1: Security: 0.7, Performance: 0.6, Interop-

erability: 0.8

• HCP2: Security: 0.5, Performance: 0.5, Interop-

erability: 0.7

• HCP3: Security: 0.9, Performance: 0.85, Interop-

erability: 0.9

Step 2: GT Model. MNOs and HCPs use GT to en-

hance their decision-making processes, with an em-

phasis on security, performance, and interoperability.

MNOs focus on prioritizing robust security protocols,

ensuring low latency, maintaining high reliability, and

achieving seamless integration with existing systems.

On the other hand, HCPs aim to improve interoper-

ability, realize cost savings, and allocate resources

effectively to boost security and performance, while

also ensuring strong interoperability.

Step 3: Deﬁne Strategies. The MNO and HCP are

strategizing to maximize their utility functions based

on trust levels. The MNO chooses HCPs with high

Table 2: Generate trust model based ﬁnal trust levels.

Cloud Provider Security Performance Interoperability Selected

HCP1 0.72 0.65 0.85 YES

HCP2 0.52 0.55 0.73 NO

HCP3 0.93 0.88 0.92 YES

trust ratings. HCP1 invests in advanced interoper-

ability standards, which moderately improve secu-

rity. In contrast, HCP2 focuses on maintaining base-

line performance and cost-effectiveness without sig-

niﬁcant investments. Meanwhile, HCP3 implements

advanced security solutions.

Step 4: Run RL. In Trust Relationships, RL involves

three components: state space, action space, and a

reward function. The initial state considers security,

performance, and interoperability levels. MNOs se-

lect HCPs based on trust levels, adjust investments,

and reward high-score HCPs.

• HCP1: Security increases to 0.72, Performance

rises to 0.65, Interoperability improves to 0.85.

• HCP2: It remains stable, with Security at 0.52,

Performance at 0.55, and Interoperability at 0.73.

• HCP3: Security increases to 0.93, Performance

rises to 0.88, Interoperability improves to 0.92.

The MNO’s rewards are determined by the trust lev-

els assigned to each HCPs. HCP1 receives a mod-

erate reward for improving interoperability, HCP2 is

given a lower reward due to minimal enhancements,

and HCP3 is awarded the highest reward for making

substantial investments across all areas.

Step 5: Final Decision. Table 2 shows that the model

integrates GT and RL to dynamically adjust trust lev-

els. This capability allows MNOs to make informed

decisions, ensuring reliable service deployment for

HCP1 and HCP3.

8 FUTURE DIRECTIONS

The novelty of this work is unlike existing trust evalu-

ation mechanisms that operate in silos; it is the ﬁrst to

construct a comprehensive 5G trust management ar-

chitecture. Our 5G-TMA integrates trust across the

RAN, Core Network, and Orchestration/Management

layers, ensuring end-to-end trust. The initiative aims

to foster trust among various 5G stakeholders, includ-

ing CSPs, service providers like HCPs, mobile net-

work operators, and end-users. It is designed for com-

patibility with heterogeneous hardware, diverse ven-

dors, and global standards (e.g., 3GPP, O-RAN). It

also incorporates dynamic and context-aware trust as-

sessment using AI/ML models that adapt to evolving

network behaviors, user activities, and threat land-

scapes.

ICISSP 2025 - 11th International Conference on Information Systems Security and Privacy

316

Trustworthiness is crucial for 5G and future 6G

telecommunication services, encompassing factors

like security, reliability, and safety. However, eval-

uating trust in a 5G environment is complex, inﬂu-

enced by factors like device reliability, communi-

cation behavior, and past interactions (Yang et al.,

2021). Future research should explore new parame-

ters and methods to manage trust among parties, en-

suring high-trust components in telecommunication

services. Development of comprehensive trust eval-

uation metrics and the use of machine learning and

artiﬁcial intelligence are expected to improve adap-

tive and accurate trust evidence collection. Interoper-

ability between 3GPP standards and O-RAN architec-

tures can be challenging, especially when incorporat-

ing Zero Trust principles. Future research should fo-

cus on establishing standardized protocols and inter-

faces for trust management, including uniformly ap-

plying trust metrics and security policies across net-

work components (Sun et al., 2022). Our plan aims

to enhance the trust life cycle management in 5G

and 6G networks by integrating detailed trust metrics,

AI, blockchain technologies, and privacy-preserving

techniques. This integration will improve trust assess-

ments, interoperability, security, and compliance with

international standards. Furthermore, we plan to es-

tablish a test bed to evaluate the framework’s secu-

rity, trust management, and network slice establish-

ment performance.

9 CONCLUSION

This paper presents an intelligent trust management

architecture for stakeholders in 5G services, utilizing

HCPs as the primary shared infrastructure. It is de-

signed to handle trust constraints, quantify 5G stake-

holder trustworthiness, and address trust management

issues in new 5G technologies and business environ-

ments. It aims to create trustworthy networks with

ﬂexibility, reliability, and scalability. We then present

a scenario demonstrating our architecture’s support

for trust relationships between CSPs and HCPs in 5G

service deployment. The architecture is extensible,

accommodating multiple 5G stakeholders and provid-

ing a generic approach to assessing trustworthiness.

REFERENCES

3GPP (2020). 5g system overview. Accessed: 2024-07-02.

Alonso, A., Persson, H. S., and Kassaei, H. (2022). 5g

architecture for hybrid and multi-cloud environments.

Ericsson Technology Review, 2022(3):2–12.

Ampririt, P., Ohara, S., Qafzezi, E., Ikeda, M., Matsuo,

K., and Barolli, L. (2021). An integrated fuzzy-based

admission control system (ifacs) for 5g wireless net-

works: Its implementation and performance evalua-

tion. Internet of Things, 13:100351.

Bonati, L., Polese, M., D’Oro, S., Basagni, S., and Melodia,

T. (2022). Openran gym: An open toolbox for data

collection and experimentation with ai in o-ran. In

2022 IEEE Wireless Communications and Networking

Conference (WCNC), pages 518–523. IEEE.

Lei, W., Soong, A. C., Jianghua, L., Yong, W., Classon, B.,

Xiao, W., Mazzarese, D., Yang, Z., Saboorian, T., Lei,

W., et al. (2021). 5g system architecture. 5G System

Design: An End to End Perspective, pages 297–339.

Maman, M., Calvanese-Strinati, E., Dinh, L. N., Haustein,

T., Keusgen, W., Wittig, S., Schmieder, M., Bar-

barossa, S., Merluzzi, M., Costanzo, F., et al. (2021).

Beyond private 5g networks: applications, architec-

tures, operator models and technological enablers.

EURASIP Journal on Wireless Communications and

Networking, 2021:1–46.

Nefti, S., Meziane, F., and Kasiran, K. (2005). A fuzzy

trust model for e-commerce. In Seventh IEEE In-

ternational Conference on E-Commerce Technology

(CEC’05), pages 401–404. IEEE.

Scarfone, K. and Hoffman, P. (2007). Guide to secure web

services. Technical report, National Institute of Stan-

dards and Technology (NIST).

Sharma, P., Jain, S., Gupta, S., and Chamola, V. (2021).

Role of machine learning and deep learning in secur-

ing 5g-driven industrial iot applications. Ad Hoc Net-

works, 123:102685.

Sun, S., Repeta, M., Healy, M., Nandall, V., Fung, E., and

Thomas, C. (2022). Towards 5g zero trusted air inter-

face architecture. arXiv preprint arXiv:2211.03776.

Svare, H., Gausdal, A. H., and M

ollering, G. (2020). The

function of ability, benevolence, and integrity-based

trust in innovation networks. Industry and Innovation,

27(6):585–604.

Valero, J. M. J., S

anchez, P. M. S., P

erez, M. G., Celdr

an,

A. H., and P

erez, G. M. (2023). Trust-as-a-service:

A reputation-enabled trust framework for 5g network

resource provisioning. Computer Communications,

211:229–238.

Wary, J.-P., Gaber, C., Lacoste, M., Vilchez, J. S., Chopin,

M., de Oca, E. M., UMU, N. P. P., Alemany, P., Vi-

lalta, R., Mu

noz, R., et al. (2019). Intelligent security

and pervasive trust for 5g and beyond.

Wong, S. (2019). The ﬁfth generation (5g) trust model. In

2019 IEEE Wireless Communications and Networking

Conference (WCNC), pages 1–5. IEEE.

Xiong, Z., Zhang, Y., Niyato, D., Deng, R., Wang, P., and

Wang, L.-C. (2019). Deep reinforcement learning for

mobile 5g and beyond: Fundamentals, applications,

and challenges. IEEE Vehicular Technology Maga-

zine, 14(2):44–52.

Yang, L., Yu, K., Yang, S. X., Chakraborty, C., Lu, Y., and

Guo, T. (2021). An intelligent trust cloud management

method for secure clustering in 5g enabled iomt. IEEE

Transactions on Industrial Informatics, 18(12):8864–

8875.

Zhang, S., Wang, Y., and Zhou, W. (2019). Towards se-

cure 5g networks: A survey. Computer Networks,

162:106871.

Design of an Intelligent Trust Management Architecture for 5G Service Deployment

317