5g-edge-lab 📡

GitOps-driven 5G Standalone (SA) Core & RAN on Edge Hardware.

This repository contains the Infrastructure-as-Code (IaC) to deploy a fully functional 5G network on highly constrained legacy edge hardware. Orchestrated with K3s and ArgoCD, utilizing Open5GS for the Core and UERANSIM for Radio Access Network simulation.

🎯 Project Objective (Thesis)

To demonstrate the technical viability of deploying next-generation telecom microservices (5G SA) on edge computing infrastructure with severe resource constraints (Legacy CPU without AVX support and <4GB RAM), managed via GitOps pipelines.

🏗️ Architecture & Hardware

The cluster is deployed on a Single-Node architecture optimized for "bare-metal" performance.

Edge Node Specifications

Component	Specification	Constraints / Notes
Device Type	Legacy Edge Gateway	Industrial form factor / Thin Client.
CPU	AMD G-T56N @ 1.65GHz	Bobcat Architecture (Dual Core). ⚠️ Non-AVX Instruction Set: Required a custom implementation of MongoDB 4.4 as v5.0+ is incompatible with the CPU instruction set.
RAM	4GB DDR3 (3.3GB Usable)	Aggressive memory optimization (4G/EPC components disabled).
OS	Ubuntu 20.04 LTS	Kernel 5.4+.
Orchestrator	K3s (Lightweight K8s)	CNI: Flannel.

---

🚀 Quick Start Guide

1. Prerequisites

• A running Kubernetes cluster (K3s recommended for edge).
• kubectl configured.
• [Optional] argocd CLI.

2. Clone Repository

• If the GitOps controller is not yet installed:

git clone [https://github.com/Gab02022/5g-edge-lab.git](https://github.com/Gab02022/5g-edge-lab.git)
cd 5g-edge-lab

3. Install ArgoCD

kubectl create namespace argocd
kubectl apply -n argocd -f [https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml](https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml)

4. Deploy 5G Network (The GitOps Way)

Instead of manual Helm installs, we let ArgoCD sync the desired state from this repository. Via Web UI

1. Create New App.
2. Repo URL: https://github.com/Gab02022/5g-edge-lab.git
3. Path: charts/open5gs
4. Value File: values.yaml

5. 🧪 Validation & Testing

1. Connectivity Check (Ping) Verify that the User Equipment (UE) has internet access through the GTP tunnel (Interface uesimtun0).

# Get the UE pod name
UE_POD=$(kubectl get pod -n ueransim -l app.kubernetes.io/name=ueransim-ues -o jsonpath="{.items[0].metadata.name}")

# Execute Ping from within the UE
kubectl exec -it -n open5gs $UE_POD -- ping -I uesimtun0 8.8.8.8

2. Throughput Test Speed test performed using iperf3 or direct download inside the container:

kubectl exec -it -n open5gs $UE_POD -- curl --interface uesimtun0 -o /dev/null [http://speedtest.tele2.net/100MB.zip](http://speedtest.tele2.net/100MB.zip)

📸 Evidence & Screenshots

1. ArgoCD Dashboard (Network State) Showing "Healthy" and "Synced" status for Core and RAN microservices.

2. Traffic Analysis (Wireshark) Packet capture demonstrating GTP Encapsulation. ICMP/TCP traffic is observed traveling inside the GTP-U tunnel (Port 2152).

3. Terminal (Ping Success) Confirmation of latency and PDU Session establishment.

📜 License

This project is part of academic research. Based on open-source charts by Gradiant.