Wall parameters estimation based onsupport vector regression for through wall radar sensing

EURASIP Journal on Advances in Signal Processing

Table 1 RMSEs of the estimates

RMSE	Train^b	Test^b	20a^c	20b^c	20c	20d	10a	10b	10c	10d
[A] ε _r ^a	0.03	0.05	0.02	0.02	0.02	0.02	0.02	0.02	0.02	0.02
[A] d ^a	0.27	0.09	1.25	0.58	0.16	0.09	0.88	0.18	0.09	0.09
[B] ε _r ^a	0.08	0.20	1.45	0.87	1.01	0.56	1.09	0.70	0.80	0.46
[B] d	0.35	1.40	7.03	2.29	1.85	1.13	6.38	1.82	1.39	0.88
			5a	5b	5c	5d	2a	2b	2c	2d
			0.02	0.02	0.02	0.02	0.02	0.02	0.02	0.02
			0.33	0.18	0.09	0.09	0	0.16	0.09	0.09
			0.71	0.55	0.69	0.43	0.30	0.41	0.57	0.41
			5.09	1.43	1.16	0.80	3.41	1.08	0.96	0.77

^a[A] represents the new proposed approach. [B] represents the regular SVR method. Permittivity and thickness are denoted as ε _r and d, respectively
^b“Train” refers to the training data. “Test” refers to the test data (with no target)
^c“20a” represents the test data in which the number “20” indicates that the radius of the target is 20 cm and the letter “a” refers to the distance between the target and the wall. Here, “a” =1 cm “b” =1×d _j. “c” =2×d _j. “d” =3×d _j