Table 1.

State-of-the-art related works on path-loss models in tunnel environments.

Type	Methods	Models	Pros and Cons	Ref.
Straight tunnel	Fit measurement results using regression method	FI model	Low complexity Insufficient accuracy	[21]
	Superpose multiple modes in both near and far region	Multimode model	High accuracy Limited applicability	[22]
	Calculate Per-ray cone angle	Improved RT model	High accuracy Low computational efficiency	[23]
	Extract rectangular waveguide model using VPE	Mixed model based on waveguide and VPE	Reduced complexity Limited Validity	[24,25]
Curved tunnel	Introduce a break point distance into the CI model	Improved CI model	High accuracy Less stability	[26]
	Divide propagation region into LOS and NLOS	Two-slope model	Realistic scenario Large deviation	[27]
	Define the break point between two waveguiding effects	Improved FI model with break point	High accuracy Calculations of break point required	[28]
	Estimate the main effects of the curvature on multimode	Mixed model based on waveguide and RT	Low complexity Insufficient accuracy	[29]
	Combine RT method with neural network	Improved RT model	High applicability High complexity	[30]
Cascaded tunnel	Fit measurement results using regression method	CI model	Low complexity Insufficient accuracy	[31]
	Reconstruct a high-precision 3D model of measurement tunnel	RT model	High accuracy High-precision 3D model required	[32]
	Divide space into segments to solve stability constraint	Improved FDTD model	High accuracy high complexity	[33]