5G Zero Trust A Zerotrust Architecture For Telecom
Figure 1 presents the logical components of a policy framework. The essential logical entities are the policy decision point (PDP) and policy enforcement point (PEP). To access the specific resource, a subject requests permission from the PDP and provides the information needed to perform authentication and authorization.
Policies are created to reflect an organization’s processes and acceptable level of risk as well as the sensitivity of the targeted asset. A policy specifies the required level of protection for an object, privileges of a subject and environmental conditions that can change the allowed behavior of the subject toward the object.
The policy engine is part of the PDP. It runs a trust evaluation algorithm to calculate a subject’s trust score, which is used to determine whether the subject is allowed to access a resource. The trust algorithm may only use information provided by the subject or it may utilize additional metadata (geographic location of the subject, historical resource usage and behavior).
The PEP is a component that is responsible for setting up a micro-perimeter to protect a resource. Where possible, the PEP is integrated into the resource or placed as close as possible to it, and it forms a logical demarcation point between security zones. The PEP provides access control of connections between the subject and resource based on access control decisions from the PDP.
Policy frameworks are employed in 3GPP-based systems to manage access to resources in different security domains. For example, to gain access to the 5G network services (T1), the user equipment (UE) contacts an Access and Mobility Management Function (AMF) that takes a PEP role. A PDP role can be represented by multiple NFs where Unified Data Management (UDM) and the Policy Control Function (PCF) may be highlighted, among others.
The AMF transmits the UE’s access request to the UDM to validate the UE’s identity and trigger authentication and authorization procedures to establish a secure channel (T2, T6). The PCF feeds the AMF with access and mobility policies that may affect UE authorization to access 5G network resources due to, for example, mobility restrictions (T4) [10, 11, 12, 13].
Another example describes how ZT principles apply in 5G SBA. In reference to T1, the SBA identifies NF service consumers and NF service producers. Communication security between core NFs has improved significantly in comparison with previous generations of mobile networks. SBA security specification requires the performance of Transport Layer Security (TLS) based mutual authentication and OAuth 2.0 token-based authorization for any NF that wants to communicate with another NF (T2, T6). The network repository function (NRF) takes the role of authorization server, which makes the NRF act as the PDP.
The introduction of the service communication proxy (SCP) allows indirect communication between NFs. The SCP can take the role of the PEP and provide access control functionality by requesting authorization decisions from the NRF. This makes it possible to implement the zero trust model in the 5G Core, where an NF service consumer (subject) requests access to an NF service producer (resource) through the SCP (PEP), and the NRF (PDP) grants or denies access [10, 13, 14]. With regard to T4, to support decision-making about requested access to resources, the NRF can store additional information, defining the actions allowed for an NF service consumer to specific NF producers .
Security monitoring supports the detection of threats and measuring the security posture of network assets and compliance with security policies. Monitoring and evaluation of subjects, resources compliance, trustworthiness and state are important when deciding whether to permit access to resources.
The European Telecommunications Standards Institute (ETSI) defines security and trust guidance for NFs . With guidelines emphasizing that compliance and state measurements must be continually monitored to effectively evaluate the level of trust of an NF, ETSI’s guidance adheres with the principles of zero trust design.
In line with T3 and T4, the security posture of the requesting entity must be evaluated by dynamic access control policies before access is granted to the requested resource. Additionally, to satisfy T5, all owned assets in a telecom network should be monitored and their security posture should be evaluated continuously. These assets include, but are not limited to, devices accessing the network, RAN NFs, core NFs and management functions.
There are different ways to implement a trust evaluation algorithm. Identifying which trust algorithm implementation to adopt depends on two characteristics:
1. How different parameters are evaluated (as binary decisions or as weighted parts of a whole score or confidence level)
2. How requests are evaluated in relation to other historical requests by the same subject.
Parameters can be evaluated either based on criteria or score . Score-based evaluation computes a confidence level based on values from every data source, recognizing that there may be various levels of trust between different subjects. Criteria-based evaluation relies on a set of statically configured attributes that must be met before access is granted to a resource or an action is allowed. Moreover, requests can be evaluated either singularly or contextually. Singular evaluation treats each request individually, which risks that an attack can go undetected. Unlike singular evaluations, contextual evaluation takes the subject’s history into consideration when evaluating access requests.
The implementation of a trust evaluation algorithm that combines contextual, score-based characteristics would make it possible to offer dynamic and granular access control, since the score provides a confidence level for the requesting account and adapts to changing factors more quickly than static policies.
With respect to T7, there are multiple parameters  that can be taken into consideration for evaluating trust that are relevant for telecom networks. Examples include geographical location, NF location, software capabilities (such as patch level, software versions), execution history of an instance, configuration compliance and the appropriate use of encryption techniques.
Future telecom networks should not only consider how to handle trust in subjects, but also trust in resources – particularly in multi-vendor deployments or in cloud where services are provided by a third party.