Dynamic network slice selection and switching in 5G Android devices using URSP
Loading...
Date
2026-07-08
Authors
Έλλις, Κρίστοφερ
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
5G networks are designed to support heterogeneous services with different performance requirements, such as enhanced mobile broadband, low-latency applications and massive machine-type communications. Network slicing is one of the key mechanisms that enables this flexibility, allowing different logical network slices to coexist over the same physical infrastructure. In this Diploma Thesis, a dynamic network slice selection mechanism for Android devices is designed and evaluated using UE Route Selection Policy (URSP). The proposed system focuses on Android Enterprise slicing and investigates how application traffic can be mapped to specific enterprise slices through the Android Device Policy Manager and URSP rules. A custom Android DPC application was developed to enable preferential network service, map selected applications to enterprise slice identifiers and trigger user driven slice requests. On the network side, an Amarisoft Callbox was used as a private 5G Standalone test environment, providing both gNB and 5G Core functionality. URSP policies were sent to the UE through the Amarisoft Remote API, while UE registration state, PDU sessions, DNNs, S-NSSAI values and core metrics were monitored through automated backend services. A cloud-based backend architecture was implemented using a REST API, Docker containers, Prometheus and Grafana. The backend communicates with the Amarisoft Callbox through SSH and WebSocket-based remote API commands, collects telemetry from the UE and network, and supports both manual and congestion driven slice switching. The monitoring mechanism evaluates selected metrics, such as downlink traffic counters and slice related session information, in order to trigger URSP policy updates when predefined conditions are met. The experimental results demonstrate that Android application traffic can be associated with enterprise URSP categories and routed through different PDU sessions based on URSP policies. The work also identifies practical implementation limitations in Android’s telephony stack, including the restricted set of supported OS/App IDs and the reuse of similar data profiles when APN configurations overlap. Overall, the thesis provides a practical implementation and analysis of URSP based dynamic slice selection on real Android devices in a controlled 5G SA testbed.
Description
Keywords
5G networks, Network slicing, UE route selection policy (URSP), Android enterprise, Custom TestDPC, Rest API, Amarisoft callbox, PDU session, S-NSSAI, Prometheus, Grafana, Dynamic slice selection