<![CDATA[Quality of Service (QoS) management using Simple Queue on MikroTik routers often faces allocation inefficiencies due to manual, static, and reactive configurations. This study proposes a closed-loop autonomous automation architecture based on Intent-Based Networking (IBN) by integrating the Qwen2.5-Coder-7B Large Language Model (LLM) and the n8n orchestration engine. Procedurally, the system workflow encompasses a pre-execution monitoring phase via SSH speedtest, text denoising using structured prompt engineering, and automated configuration injection through the MikroTik RouterOS v7 REST API. Experimental results across 10 test cases demonstrated that the model's cognitive component achieved a 100% Exact Match Rate in translating unstructured natural language commands into deterministic JSON objects. The state-aware logic within n8n effectively acted as a network safety valve by autonomously capping the child bandwidth allocation to a maximum threshold of 70% of the actual ISP capacity to prevent bufferbloat. Furthermore, the system successfully isolated command executions using HTTP PATCH to minimize the control-plane overhead on the resource-constrained MikroTik hAP Lite router. The measured end-to-end operational latency ranged from 27.66 seconds to 43.98 seconds, which was predominantly driven by the active probing sampling time on the physical ISP link. This integration successfully delivered an adaptive, precise, and secure autonomous network quality management system for SOHO environments while eliminating human configuration errors.]]>
Copyrights © 2026
