Implementasi Platform Open-Source Untuk Monitoring User Experience Insight (UXI) Jaringan Nirkabel Menggunakan Perangkat Raspberry Pi 4

Baswara, Jonathan (2026) Implementasi Platform Open-Source Untuk Monitoring User Experience Insight (UXI) Jaringan Nirkabel Menggunakan Perangkat Raspberry Pi 4. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027221062-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Kebutuhan pemantauan user experience insight (UXI) pada jaringan nirkabel semakin meningkat seiring ketergantungan institusi terhadap layanan digital berbasis internet. Namun, sensor UXI komersial seperti HPE Aruba UXI relatif mahal sehingga tidak selalu terjangkau bagi organisasi dengan anggaran terbatas. Penelitian ini mengembangkan prototipe sensor UXI berbasis open-source menggunakan Raspberry Pi 4 sebagai alternatif berbiaya rendah dan membandingkan performanya terhadap Aruba UXI pada kondisi jaringan yang sama. Prototipe dibangun dengan arsitektur pemantauan berbasis exporter, Prometheus, dan Grafana untuk menghasilkan metrik pengujian sintetik yang merepresentasikan pengalaman pengguna, meliputi throughput (Fast.com), round-trip time (RTT) ICMP ping, durasi respons HTTP/HTTPS, serta kualitas sinyal Wi-Fi (RSSI dan tx rate). Pengujian dilakukan pada jaringanWi-Fi myITS-WiFi band 5 GHz di ruang divisi IKTI pada 22 Januari 2026, dengan penempatan Raspberry Pi dan Aruba UXI berdekatan agar menerima kondisi radio dan beban jaringan yang serupa. Timestamp Aruba UXI (UTC) dinormalisasi ke WIB (UTC+7), lalu kedua dataset difilter pada jendela uji 08:30–14:00 WIB. Karena interval pengukuran kedua perangkat berbeda, perbandingan dilakukan dengan nearest timestamp pairing pada kategori tes yang sebanding. Evaluasi dilaporkan menggunakan statistik P50 (median) dan P95, serta metrik MAE, bias (UXI-Lite − Aruba), korelasi Spearman (ρ), dan analisis Bland - Altman untuk menilai agreement pengukuran. Secara kuantitatif, metrik ping latency menunjukkan kesesuaian paling kuat dengan MAE 4,96 ms, bias +1,11 ms, dan Spearman ρ = 0,798. Metrik HTTP GET menunjukkan selisih yang lebih besar dengan MAE 0,300 s, bias +0,248 s, dan Spearman ρ = 0,475, sementara throughput berbeda lebih jauh (MAE 21,09 Mbps, bias -11,44 Mbps, Spearman ρ = 0,152). Untuk metrik link, RSSI memiliki MAE 2,935 dBm, bias +2,632 dBm, dan Spearman ρ = 0,124, sehingga lebih tepat digunakan sebagai indikator relatif (trend) daripada nilai absolut. Temuan ini menunjukkan UXI-Lite layak sebagai alternatif low-cost untuk pemantauan dasar dan indikasi degradasi, terutama pada metrik konektivitas (ping), namun belum ideal untuk penggantian penuh pada metrik aplikasi/throughput tanpa kalibrasi metode uji dan stabilisasi operasional.
====================================================================================================================================
User experience insight (UXI) monitoring over wireless networks is increasingly important as institutions rely on internet-based services. However, commercial UXI sensors such as HPE Aruba UXI can be costly, motivating the development of a low-cost alternative. This study develops UXI-Lite, an open-source UXI sensor based on a Raspberry Pi 4 Model B, and evaluates its measurement parity against Aruba UXI (UX-G6E) on the same Wi-Fi network. UXI-Lite performs client-side synthetic measurements covering connectivity and service access, including ICMP ping latency, HTTP/HTTPS GET elapsed time, throughput testing, and Wi-Fi link telemetry (e.g., RSSI/channel). Experiments were conducted on myITS-WiFi (5 GHz) on 22 January 2026 from 08:30–14:00 WIB (UTC+7), with both sensors placed side-by-side. Aruba UXI timestamps recorded in UTC were converted to WIB, and both datasets were filtered to the same test window. Because the two sensors use different sampling patterns, comparisons were performed using nearest-timestamp pairing per test category and target/resolver/channel. Performance was summarized using P50 (median) and P95, and measurement parity was evaluated using MAE, bias (UXI-Lite − Aruba), Spearman correlation (ρ) for trend agreement, and Bland - Altman analysis for measurement agreement. Quantitatively, ping latency showed the strongest agreement (MAE 4.96 ms, bias +1.11 ms, Spearman ρ = 0.798). HTTP GET exhibited larger discrepancies (MAE 0.300 s, bias +0.248 s, ρ = 0.475), while throughput differed substantially (MAE 21.09 Mbps, bias -11.44 Mbps, ρ = 0.152). For Wi-Fi telemetry, RSSI showed small absolute error but weak trend agreement (MAE 2.935 dBm, bias +2.632 dBm, ρ = 0.124), suggesting it is more reliable for relative indication than absolute calibration. Overall, UXI-Lite is feasible as a low-cost UXI monitoring alternative for detecting connectivity degradation, with limitations for full replacement on application/throughput metrics without further calibration.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	UXI-Lite; Raspberry Pi 4; Aruba UXI; synthetic monitoring; Bland - Altman.
Subjects:	T Technology > T Technology (General) > T58.5 Information technology. IT--Auditing T Technology > T Technology (General) > T58.62 Decision support systems T Technology > T Technology (General) > T58.8 Productivity. Efficiency Z Bibliography. Library Science. Information Resources > ZA Information resources > ZA4450 Databases
Divisions:	Faculty of Information and Communication Technology > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Jonathan Adithya Baswara
Date Deposited:	03 Feb 2026 07:38
Last Modified:	03 Feb 2026 07:38
URI:	http://repository.its.ac.id/id/eprint/132018

Actions (login required)

View Item