This paper examines approaches to assessing the quality of fifth-generation (5G) core network software. A method is proposed that combines standard quality metrics in accordance with ISO/IEC 25010 with additional criteria relevant to 5G telecommunication systems, such as latency, scalability, energy efficiency, adaptability, and fault tolerance. A three-tier model is developed, enabling multidimensional assessment of core software considering service, functional, and infrastructure levels. Simulations were conducted to evaluate the method's effectiveness under different scenarios, including load variation, component version updates, and node failures. The proposed approach demonstrates higher precision and practical applicability compared to existing methods. Graphs, block diagrams, and experimental results are presented. Prospects for implementation and future research directions are outlined.
Software quality assessment is a critically important stage in the development lifecycle, particularly
for high-load systems such as the 5G core. One of the most widely recognized standards for
defining and evaluating quality characteristics is ISO/IEC 25010 [
Considerable attention has also been paid to the quality evaluation of cloud-based software,
which shares architectural features with the 5G core, including virtualization, microservice
architecture, and automated orchestration. In [
In the context of 5G, some studies [
Thus, an analysis of existing approaches reveals that although general software quality models
exist, they do not fully account for the specific characteristics of the 5G core. Recent research also
highlights that emerging technologies, such as software-defined radio (SDR) receivers, introduce
new cybersecurity risks in wireless environments, which should be reflected in domain-oriented
quality evaluation methods [
The objective of this paper is to develop a method for assessing the quality of 5G core software that is based on international standards but adapted to the requirements of modern telecommunication infrastructure. The paper provides an overview of existing approaches to software quality evaluation, formulates the requirements for a new assessment method, describes its structure, and presents an experimental validation of its effectiveness.
The 5G network core performs critical functions such as access control, mobility management,
traffic routing, quality of service (QoS) assurance, security, and network resource orchestration.
These functions are implemented through a set of software components deployed in a cloud
environment using Network Function Virtualization (NFV) and Software-Defined Networking
(SDN) [
Due to the architectural complexity, high workload dynamics, and strict performance guarantees, traditional software quality assessment methods are insufficient to comprehensively evaluate the quality of the 5G core. In particular, it is essential to consider parameters such as:
Data transmission latency (crucial for uRLLC-type services).
Scalability (the system's ability to adapt resources to varying loads).
Reliability (the system’s ability to function without failure under critical conditions). Energy efficiency (especially important for distributed edge infrastructures).
Fault and attack tolerance (including DDoS protection and self-healing).
Integration with cloud-native and containerized infrastructures [
Therefore, it is necessary to develop a software quality assessment method for the 5G core that: 1. Accounts for both traditional quality characteristics (based on ISO/IEC 25010) and domainspecific indicators. 2. Is adapted to the 5G Core architecture: service-oriented, containerized, and orchestrated. 3. Supports formal representation and quantitative computation of quality indicators. 4. Allows for comparison of alternative implementations of core components (e.g., AMF, SMF,
UPF, etc.).
5. Is suitable for integration into CI/CD pipelines and monitoring systems [
The proposed method for assessing the quality of 5G core software is based on adapting the ISO/IEC 25010 model to the specifics of the 5G Core architecture. The main idea is to supplement traditional quality characteristics (functionality, performance efficiency, reliability, and others) with domain-oriented metrics inherent to cloud-native and next-generation telecommunication systems.
The method comprises three hierarchical levels: 1. Basic Level (Standard): Traditional ISO/IEC 25010 characteristics—functional suitability, performance efficiency, reliability, security, maintainability, and portability. 2. Extension Level (5G-Specific): Latency, scalability, fault tolerance, NFV/SDN integration, and energy efficiency. 3. Evaluation Level: Module for collecting, weighting, and interpreting quality indicators.
Metrics are normalized to a [0;1] scale and aggregated to form a comprehensive quality score.
where Q i is the aggregated value of the i-th quality characteristic, mij is the value of the j-th metric for the i-th characteristic (Table 1), and wij is the weight of the j-th metric (determined either through expert judgment or based on Analytic Hierarchy Process (AHP) analysis).
The overall quality indicator of the 5G core software is calculated as a weighted sum of the quality characteristics:
n Q i= ∑ wij⋅ mij , j= 1
k Q total= ∑ Wi⋅ Q i , i= 1 (1) (2)
processing speed The method can be implemented as a monitoring module integrated into a CI/CD pipeline or a testing environment (e.g., OpenAirInterface, free5GC). The weighting coefficients can be determined through expert evaluation or the Analytic Hierarchy Process (AHP) method. The method can also be adapted to specific service types such as eMBB, mMTC, or uRLLC.
Stage 1: Define a comprehensive set of metrics: • • •
Functional Metrics (F): correctness, reliability, security; Non-Functional Metrics (NF): performance, scalability, energy efficiency; Contextual Metrics (C): adaptability, automation, fault tolerance.
M = { m1 , m2 , .... mn} , mi∈ [0 ; 1 ].
Stage 2: Normalize metrics.
Each metric is normalized to the interval [0;1] (if not already normalized), for example: where m^i is the normalized value, and mimin, mimax are the minimum and maximum bounds of the metric, respectively.
Stage 3: Apply weighting coefficients to metrics.
To account for the significance of different indicators, a system of weighting coefficients is applied:
Service level (Q 1): Functional level (Q 2): Infrastructure level (Q 3): m^i =
mi −mimin m max−m min , i i
n wi∈ [0 ; 1 ], ∑ wi= 1.
i= 1 k Q 1= ∑ i= 1 l Q 2= ∑ j= 1 p Q 3= ∑ z= 1 w(i1)⋅ m^i . w(j2)⋅ m^j . w(3)⋅ m^ .
z z Q total=α ⋅Q 1+ β ⋅Q 2+γ ⋅Q 3 , (3) (4) (5) (6) (7) (8) (9) where w(r ) are the weights at the corresponding level, and m^i are the metrics grouped by level. Stage 5: Calculate the integral quality index.
where α + β + γ= 1, the coefficients α, β, and γ represent the importance of each level (e.g., α = 0.4, β = 0.35, γ = 0.25).
Stage 6: Perform dynamic assessment over time.
When the version, configuration, or load changes, the metrics are tracked over timet, allowing the calculation of: and to construct graphs showing the degradation or improvement of quality over time.
All method parameters (the set of metrics, weights, levels) can be adapted to the specifics of a particular use case, such as a telecom operator’s core, a private 5G network, or a test environment.
To validate the effectiveness of the proposed method, an experimental testbed was implemented based on the open-source free5GC stack, simulating the 5G core network.
Simulation setup: 5G Core (free5GC v3.3.0): AMF, SMF, UPF, PCF, AUSF components.
User Emulator: UERANSIM v3.2.7.
Virtualization Platform: Oracle VirtualBox Version 7.1.8 r16846.
Monitoring and Metrics Collection: Prometheus + Grafana + custom scripts.
Host Environment: Ubuntu Server 22.04 LTS, Intel Xeon, 64 GB RAM.
The entire testbed was deployed inside VirtualBox virtual machines running on the host system (Figures 1–3).
Baseline load (up to 100 concurrent users).
Peak load (over 1000 eMBB sessions).
Failure condition (UPF process termination or SMF overload).
The method was compared with two other approaches:
Method A—standard quality assessment based on ISO/IEC 25010 (without 5G-specific considerations).
Method B—the model by [
The experimental results emphasize the importance of evaluating accuracy, performance, and applicability. Assessment accuracy refers to the deviation observed during repeated runs, which remained within ±3%, indicating stable metric interpretation.
Method performance is defined by the evaluation time—approximately 3 seconds per scenario with over 500 sessions—enabled by automated data collection. Applicability refers to ease of integration into CI/CD pipelines (e.g., testing new core component versions), suitability for comparing alternatives (such as UPF implementations from different vendors), and potential use as a QoS criterion in SLA agreements. The proposed method demonstrated higher flexibility and sensitivity to system parameter changes, which is critical in the dynamic environment of the 5G Core. Its main advantage lies in the ability to quantitatively account for 5G-specific factors, which are not captured by general-purpose standards. Due to its modular structure, the method can also be readily adapted to future requirements, such as integration with AI-based core optimization modules.
The histogram (Figure 4) illustrates a comparison of three 5G core software quality assessment methods across four criteria: Q total is the overall quality score; and consideration of latency, scalability, and automation—key characteristics essential for 5G systems.
The graph (Figure 5) shows the variation of the integral quality indicator Q total depending on the number of concurrent sessions (load). The proposed method demonstrates resilience to increasing load, with only a slight decrease in Q total, while Method A (ISO/IEC 25010) exhibits a significant drop in quality as the number of users increases.
The graph (Figure 6) illustrates how the Q total indicator changed during updates of 5G core components (e.g., transition from version 1.0 to 2.0). The proposed method clearly captures improvements in system-level quality following updates. Method B demonstrates stability but shows lower sensitivity to changes.
The proposed method for assessing the quality of 5G core software offers several significant advantages that distinguish it from traditional approaches, such as ISO/IEC 25010 or microserviceoriented models. Among these advantages are:
Comprehensiveness: the method accounts for both classical quality metrics (e.g., functionality, reliability, usability) and 5G-specific aspects such as latency, scalability, and energy efficiency.
Architectural adaptability: the three-level model structure enables adaptation to various 5G core implementations—ranging from monolithic to cloud-distributed systems.
Resilience under failure and load: experimental studies showed stable quality scores even under high-load conditions or failures in specific components (UPF, SMF, AMF). Automation: the method can be integrated into CI/CD pipelines for continuous quality monitoring.
A key advantage of this method compared to others is its orientation toward metrics relevant to 5G and information and communication systems (ICS) in general. This allows for the effective detection of anomalies during software operation in ICS environments and facilitates quality improvements through error elimination.
However, the method also has certain limitations, including:
Configuration complexity: full implementation requires deep integration with the operator’s telecom infrastructure, including access to telemetry data and logs.
Dependency on data completeness: the accuracy of the assessment depends on the availability and reliability of input data—particularly from KPI monitoring, tracing, and load testing results.
Need for customization: different vendor-specific 5G core implementations may require adaptations of metric weights and custom indicators.
The method can be extended to 6G systems or industrial private networks (Private 5G), which demand even higher levels of QoS/QoE. There is potential for integration with ML modules capable of automatically adjusting model weights based on changes in system load, SLA policies, or service type. Pilot deployment is recommended in a test environment, for example using platforms such as Open5GS or ONF SD-Core. Integration with monitoring systems (e.g., Prometheus, Grafana, Zabbix) should be established to collect input metrics. Gradual integration into the operator’s CI/CD pipeline is suggested for analyzing software releases within the DevOps cycle.
This paper proposes a method for assessing the quality of 5G core network software adapted to the specifics of modern telecommunication systems. The method combines ISO/IEC 25010 standards with additional metrics pertinent to 5G networks, including latency, scalability, energy efficiency, automation, and fault tolerance.
Key outcomes include:
Development of a three-level quality assessment model.
Formalization of the calculation for an integral quality index.
Experimental validation demonstrating enhanced accuracy, stability, and relevance compared to existing methods.
The method provides significant value for mobile network operators, software developers, and system integrators by enabling:
Objective quality assessment of core releases before deployment.
Identification of scalability bottlenecks.
Compliance with QoE and SLA requirements.
Integration into continuous monitoring within DevOps pipelines.
Future work will focus on expanding the model to 6G and NTN networks, integrating machine learning for dynamic metric weighting, and deploying the method in operational environments.
While preparing this work, the authors used the AI programs Grammarly Pro to correct text grammar and Strike Plagiarism to search for possible plagiarism. After using this tool, the authors reviewed and edited the content as needed and took full responsibility for the publication’s content.