<![CDATA[Uneven wireless signal distribution and inefficient device deployment remain common problems in multi-storey school buildings. This study proposes a traceable design selection procedure for indoor wireless networks that integrates AI Heatmap predictive simulation, a weighted decision matrix with a measurement-based scoring rubric, and segmented Quality of Experience (QoE) validation. Two scenarios were defined and simulated on the same blueprint at 2.4 GHz and 5 GHz: an efficient design with 11 access points and an ideal design with 26. Five quantitative indicators were formulated, namely coverage efficiency, marginal coverage, cost per coverage point, marginal cost per coverage point, and capacity utilisation ratio, and were converted into 1–5 scores through an explicit rubric. The efficient design achieved a coverage efficiency of 11.00–12.80 points per access point against 5.93–6.42 for the ideal design, while the marginal cost of upgrading reached IDR 1,354,736–1,560,000 per coverage point, six times the average cost, and capacity utilisation fell from 72.7% to 30.8%. The efficient design obtained a weighted score of 4.30 (Highly Feasible) against 4.05 (Feasible), and QoE testing with 128 respondents yielded 4.26 (Very Satisfied). Per-band simulation shows zero poor-category area on 2.4 GHz for both scenarios, whereas on 5 GHz 21–22% of the area falls into the Poor class, still within the coverage footprint but below the usable threshold, a shortfall absorbed by fallback to 2.4 GHz within the same service set and corroborated by client side measurements showing a mean RSSI of −61 dBm on 2.4 GHz against −74 dBm on 5 GHz. Sensitivity analysis shows that this ranking holds only while investment efficiency carries a weight above 0.182; below that threshold the ideal design prevails. The contribution therefore lies not in declaring one design universally superior, but in providing a procedure that makes design decisions traceable to measurement data and states explicitly the conditions under which each design is preferable.]]>
Copyrights © 2026
