Implementasi dan Evaluasi Network Experience Black-Box Micro-UXI Berbasis Arduino UNO Q Menggunakan Pendekatan Event-Driven Monitoring

Rizqy, Muhammad Faishal (2026) Implementasi dan Evaluasi Network Experience Black-Box Micro-UXI Berbasis Arduino UNO Q Menggunakan Pendekatan Event-Driven Monitoring. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027221026-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Pemantauan infrastruktur jaringan modern, perlu memastikan ketersediaan layanan dan menangkap kualitas pengalaman pengguna secara proaktif. Solusi Digital Experience Monitoring komersial seperti Aruba UXI mampu melakukan pemantauan tersebut, tetapi umumnya memiliki biaya implementasi yang tinggi dan ditujukan untuk skala enterprise. Penelitian ini mengimplementasikan dan mengevaluasi Micro-UXI, yaitu sistem network experience black-box event recorder berbasis Arduino UNO Q dengan pendekatan event-driven monitoring. Sistem melakukan pengukuran aktif terhadap beberapa metrik pengalaman jaringan, seperti DNS, ping, RTT, packet loss, HTTP total, dan HTTP TTFB. Pengujian dilakukan terhadap enam skenario gangguan, yaitu DNS Degraded, DNS Timeout Burst, Loss Burst, High RTT, HTTP Slow, dan Connectivity Flap, menggunakan fault injection terkontrol dengan ground truth. Metode event-driven monitoring dibandingkan dengan baseline monitoring berbasis threshold statis menggunakan metrik precision, recall, F1-score, False Alarm Rate (FAR), dan Mean Time To Detect (MTTD). Hasil pengujian menunjukkan bahwa baseline monitoring memperoleh rata-rata precision 100%, recall 95%, F1-score 0,973, FAR 0%, dan MTTD 21,83 detik. Sementara itu, event-driven monitoring memperoleh rata-rata precision 91,1%, recall 88,3%, F1-score 0,891, dan MTTD 19,24 detik. Hasil tersebut menunjukkan bahwa baseline monitoring lebih stabil dari sisi akurasi, sedangkan event-driven monitoring cenderung lebih cepat dalam waktu deteksi. Pengukuran overhead menunjukkan penggunaan sumber daya yang rendah, sehingga sistem layak dijalankan pada perangkat dengan sumber daya terbatas. Evidence bundle berhasil terbentuk pada seluruh skenario, meskipun komponen pre-event window masih bersifat parsial karena mengacu pada waktu sebelum deteksi, bukan sepenuhnya sebelum fault terjadi.
=====================================================================================================================================
Modern network infrastructure monitoring needs to ensure service availability and proactively capture the quality of user experience. Commercial Digital Experience Monitoring solutions such as Aruba UXI are capable of performing such monitoring, but they generally have high implementation costs and are intended for enterprise-scale environments. This study implements and evaluates Micro-UXI, a network experience black-box event recorder system based on Arduino UNO Q using an event-driven monitoring approach. The system performs active measurements on several network experience metrics, such as DNS, ping, RTT, packet loss, HTTP total time, and HTTP TTFB. The evaluation is conducted on six fault scenarios, namely DNS Degraded, DNS Timeout Burst, Loss Burst, High RTT, HTTP Slow, and Connectivity Flap, using controlled fault injection with ground truth. The event-driven monitoring method is compared with static-threshold-based baseline monitoring using precision, recall, F1-score, False Alarm Rate (FAR), and Mean Time To Detect (MTTD). The results show that baseline monitoring achieves an average precision of 100%, recall of 95%, F1-score of 0.973, FAR of 0%, and MTTD of 21.83 seconds. Meanwhile, event-driven monitoring achieves an average precision of 91.1%, recall of 88.3%, F1-score of 0.891, and MTTD of 19.24 seconds. These results indicate that baseline monitoring is more stable in terms of accuracy, while event-driven monitoring tends to provide faster detection time. The overhead measurement shows low resource usage, indicating that the system is feasible to run on resource-constrained devices. The evidence bundle is successfully generated in all scenarios, although the pre-event window component remains partial because it refers to the period before detection, rather than entirely before the fault occurs.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Quality of Experience, event-driven monitoring, fault injection, Arduino UNO Q, evidence bundle Quality of Experience, event-driven monitoring, fault injection, Arduino UNO Q, evidence bundle
Subjects:	T Technology > T Technology (General) > T58.5 Information technology. IT--Auditing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Faishal Rizqy
Date Deposited:	01 Aug 2026 01:48
Last Modified:	01 Aug 2026 01:48
URI:	http://repository.its.ac.id/id/eprint/140580

Actions (login required)

View Item