Bounding the Practical Error of Path Loss Models

Table 3

Summary of results by data set.

Name	𝛼	𝜖	𝜎	𝑁	Top three performing models by SC-RMSE						Ideal RMSE
wart	1.86	9.05	13.26	15	flatedge	13.73	itu.terrain	13.89	hatao	14.03	1.96
wart/snow	1.92	9.25	13.36	15	itu.terrain	13.93	flatedge	14.16	hatao	14.19	1.87
pdx	2.25	19.53	7.8	5	allsebrook200	8.38	hatal	8.97	davidson	9.37	1.14
pdx stumble	1.79	27.08	8.96	40	allsebrook400	8.34	itur25	10.50	hatam	10.51	1.02
boulder/ptg	0.79	19.56	7.36	20	allsebrook400	7.90	ecc33m	9.38	hatam	10.47	0.94
boulder/gtp	0.27	10.88	3.67	5	allsebrook400	5.45	hatal.fc	7.15	edwards200	8.51	1.01
cost231	6.25	51.19	6.36	15	edwards200	9.23	hatam	9.99	itur25	10.55	1.23
wmp	0.62	13.74	13.92	15	flatedge	15.34	allsebrook200	16.72	egli	16.83	5.98
tfa	0.95	22.76	7.89	20	herring.atg	8.90	allsebrook200	9.03	flatedge	10.83	1.43
google	0.54	6.15	7.37	30	davidson	13.56	itu.terrain	16.12	hatal	16.83	2.93