Path Difference Optimization of 5G Millimeter Wave Communication Networks in Malaysia

Abstract

The development of intelligent transport systems, mobile cellular networks, microwave links, and vehicle communications has accelerated with the use of wireless connections as a communication channel in 5G wireless technology. Weather, including rain, fog, snow, sand, and dust, impacts wireless communication channels in various ways. These effects are more pronounced at the high frequencies of millimeter-wave bands. Recently, the 5G network has made it possible to support a variety of applications with fast speeds and high-quality content. To facilitate the use of high-millimeter-wave frequencies, a recent study investigated how sand and dust affect the 5G communication channel. In this paper, we consider the impact of frequent and heavy rainfall on millimeter-wave propagation and cross-polarization of the wave at various points along the propagation path caused by rainfall in urban and highway scenarios in Malaysia. We estimate rainfall attenuation, path loss, and link margin at various millimeter-wave frequencies. From our simulation results, it is evident that rainfall attenuation, path loss, and link margin depend on the operating frequency, path difference, and rainfall rate. In this paper, we estimate and compare the optimal path difference values under urban and highway scenarios both with and without rainfall attenuation.

1. Introduction

The study and development of 5G communication networks has been inspired by the rising need for higher data rates, coverage, and capacity. Due to the scarcity of spectrum below 5 GHz, the millimeter-wave (mmWave) band above 10 GHz has been considered the primary band for 5G and beyond 5G communication networks, which are intended to support large bandwidths, low latency, and higher data rates [1]. These parameters can enhance 5G network efficiency, throughput, and spectral efficiency [2,3]. However, a serious limitation of high-frequency mmWave signals is their inability to travel over longer distances due to vulnerability to penetration losses caused by various sources of atmospheric absorption. The quality of propagating mmWave signals is affected by different propagation paths, such as line-of-sight (LOS) or non-LOS (NLOS), different regions, such as indoor, outdoor, highway, urban, semi-urban, and rural areas, and different weather conditions, such as rain, fog, snow, dust, and sand. In all these conditions, the signal energy may be absorbed, scattered, diffracted, or depolarized [4].

In wireless communication systems, LOS propagation plays a pivotal role in achieving improved data rates and quality of service. Within 5G applications such as mmWave communication, unmanned aerial vehicle (UAV) communication, and mobile satellite communication, the signal strength relies on the type of propagation medium [5]. The transmission of radio waves at millimeter wavelengths can be significantly hampered by rain-induced attenuation, which restricts path length and hinders the use of higher frequencies for LOS communication. Heavy raindrops absorb or scatter radio waves, leading to a reduction in the received signal strength. Raindrop absorption and scattering in mmWave frequency bands result in substantial transmission losses [6]. The extent of attenuation or transmission losses varies based on factors such as path length, polarization, frequency, temperature, and latitude. In Malaysia, wireless channels are notably affected by frequent and heavy rainfall [7]. Signal attenuation during propagation depends on parameters such as the rainfall rate, rain density, raindrop size, signal operating frequency, path length, and air quality [8,9]. Various physical and empirical models can be utilized to forecast rain attenuation [10]. These models take into account local environmental conditions and operating frequencies. Rain attenuation at mmWave frequencies can be estimated using models such as the ITU-R 530-16 model, Da Silva Mello model, Moupfouma’s model, Abdul Rahman model, Lin model, and differential equation model [11].

Signal attenuation increases with higher rainfall rates, signal absorption, and path differences. These factors lead to path loss and constrain the coverage area. The path loss denotes the reduction in effective transmit power of a wireless signal as it traverses the channel. It is defined as the loss of signal energy during propagation from transmitter to receiver. This path is influenced by various factors, including path differences, both small and large obstructions, climatic conditions, and antenna characteristics. The antenna’s location, tilt angle, and height determine its directivity, which is used to estimate the path loss [12]. Path loss can be estimated using LOS and NLOS prediction models. These models are further categorized into indoor and outdoor models, as well as long-distance and short-distance models. The literature extensively discusses several path loss models [13,14,15,16,17].

In [13], the influence of rainfall intensity on wave propagation at 26 GHz was measured and compared among all the states in Malaysia. Rainfall attenuation increases the overall path loss. In certain tropical regions, sand and dust particles attenuate propagating signals. The impact of dusty storms on electromagnetic waves was assessed using the Mie scattering model [18]. Sand and dust affect wireless communication channels operating at mmWave frequencies, and the resulting path loss can be evaluated using Mie scattering models [19,20]. In [21], path loss impacts in urban and highway scenarios were estimated for DSRC and mmWave frequencies. The 5G test network’s performance, including bandwidth, throughput, and latency, was compared under various weather and traffic conditions in real-life environments in [22]. The same paper discussed coverage, capacity, data rates, and throughput enhancement techniques for 5G networks [23].

Rain, fog, and precipitation affect channel characteristics and attenuate signal propagation at two mmWave frequencies: the V-band at 60 GHz, and the E-band at 73 GHz. Additionally, authors have considered the impact of temperature and humidity on wave propagation during the winter season [24]. The effect of water vapor and rainfall on E-band signals has been estimated as well [25]. The results of these studies indicate that path loss attenuation increases with higher rainfall intensity. Another study examined millimeter-wave attenuation in Nigeria during the hours with the heaviest rainfall, considering time percentages of 0.01% and 0.001% [26]. Furthermore, the impact of rain and snow on vehicle-to-vehicle communication at 60 GHz was studied in [27], where it was found that the LOS path distance was reduced to 60 m.

In vehicle-to-vehicle communication, the path distance is optimized based on the free space path loss and link margin at DSRC frequency of 5.9 GHz and 5G frequency of 28 GHz [28]. In this study, high data rates of 27 Mbps and 1 GHz were achieved at path differences of 867 m and 688 m, respectively. The GPR mixer model has been employed to measure attenuation due to various road pavement densities within the frequency range of 1.7 GHz to 2.6 GHz [29]. In [30], rain attenuation was investigated at an E-band frequency of 73.5 GHz and a K-band frequency of 21.8 GHz. The results showed enhanced microwave link back-haul performance, achieving high data rates and throughput using the E-band frequency when rainfall was less than 108 mm/h and using the K-band frequency when it was greater than 108 mm/h. Several factors, including vehicle speed, mobility, user location, and spectrum hand-off, can impact received signal strength [31,32].

In this paper, we examine the impact of heavy rainfall on the 5G mmWave wireless channel. We estimate path loss and link margin for both urban and highway scenarios. Additionally, we estimate rain attenuation and LOS and NLOS probabilities in terms of path difference and operating frequencies. We then calculate the optimal path difference values for both urban and highway scenarios both with and without attenuation due to rainfall. These optimal values can be utilized to enhance overall channel capacity, throughput, and achieve high data rates.

The remaining sections of the paper are organized as follows: Section 2 presents rain attenuation; Section 3 discusses indoor and outdoor probability models; Section 4 covers link margin estimations for urban and highway scenarios; Section 5 presents simulation results and range optimization values for both scenarios; and Section 6 concludes the paper and outlines future research prospects.

2. Related Work

The primary focus of this study revolves around rain attenuation, path loss within urban and highway scenarios, link margin, and the optimization of path differences. Within the existing literature, numerous articles have delved into topics such as path loss models, received signal strength, and the evaluation of rain attenuation across different frequencies. These references are compiled and presented in

Table 1. Summary of related works.

Author & Ref. No.	Frequency	Objectives	Key Findings
I. Shayea, et al.,  [13]	24 GHz	ITU models are used to measure the rain rate and rain attenuation in Malaysia throughout the year and estimated the error percentage.	At 0.01% of time the rain rate was 120 mm/h and the rain attenuation was 34 dB at 1.3 km are observed. The error percentage of rain rate and attenuation are 143% and 159% respectively.
S. Zang, et al., [18]	mmWave frequency band	The impact of weather like snow, fog, rain and hail on the sensors of a car such as radar, GPS, lidar and camera of an autonomous vehicles are considered.	The detection of range of mmWave radar is reduced by 45% due to heavy rainfall of 150 mm/h.
S.M. Sharif, [19]	2–100 GHz	The impact of dust particles on electromagnetic signal propagation is considered and Mie scattering approximation was used to estimate the total signal attenuation.	The values derived Mie model utilizing the Rayleigh approximation exhibit an upward trend, particularly at frequencies exceeding 40 GHz. The influence of air humidity on specific attenuation was determined to be negligible and can be safely disregarded.
A. Musa, et al., [20]	mmWave frequency band	The attenuation of electromagnetic signal due to dust storm was estimated using the Mie scattering approximation.	Attenuation was estimated mainly as a function of visibility and observed that the attenuation increases with the severity of dust storm.
A.M.M. Abuhdima, et al., [21]	mmWave frequency band	The impact of dust and sand on the 5G communication channel was considered and estimated the path loss using mie scattering models ML6363 and ML6352 in terms of visibility, particle size and frequency.	ML6363 wwas affected by dust and sand seriously when the visibility at a distance of 12 km and ML6352 was affected by dust and sand seriously when the visibility at a distance of 39 km.
E. Abuhdima, et al.,  [22]	5.9 GHz & 28–73.5 GHz	The impact of dust and sand on vehicle to vehicle communication was considered and estimated the path loss at various frequencies.	Signal attenuation increases with the operating frequency, particle size and concentration of dust and sand increases.
H.M. Hamid Dutty, et al., [25]	60 GHz and 70 GHz	The impact of weather i.e., rainy and winter seasons of Bangladesh was considered to estimate the signal attenuation.	Signal attenuation in winter season was higher than in rainy season. The dry atmosphere and cold weather increases the mmWave signal attenuation.
D. Dimce, et al., [28]	60 GHz	Impact of rain and snow on vehicle to everything was considered to estimate the signal attenuation using ITU model and cross verified the results using the NYUSIM simulator.	The weather conditions would reduce the signal propagation distance.
M. Alhilali, et al., [33]	38 GHz	The impact of rain on the signal propagation was estimated using the ITU-R model and 2D video meter.	The increase in rain rate and raindrop axial ratio increases the signal propagation loss.

.

These references reveal that rain attenuation is primarily influenced by several factors, including rainfall rate, path difference, operating frequency, and polarization. Additionally, a variety of prediction models, such as empirical, statistical, fade slope, physical, and optimization models, play a role in understanding and quantifying this phenomenon. The rainfall rate and rain attenuation directly impact the signal characteristics.

In this study, we examine the influence of rainfall and rain attenuation on signal propagation characteristics, specifically, path loss, path loss error statistics, and link margin, within urban and highway scenarios. The impact of rain attenuation in tropical regions significantly affects signal propagation. Consequently, our investigation includes an assessment of rain attenuation and the estimation and optimization of path loss and path difference, respectively, in both urban and highway scenarios.

3. Rain Attenuation

Rainfall affects wireless channels operating at 10 GHz and above, with this impact being more pronounced in tropical regions due to heavy rainfall. The size of raindrops and the rate of rainfall influence signal attenuation. In the literature, several rain attenuation prediction models exist, including empirical, statistical, fade–slope, physical, and optimization models [34,35,36,37]. Each model estimates rain attenuation based on different environmental characteristics. In this paper, we employ a statistical model to predict rain attenuation because it provides realistic values in tropical regions such as Malaysia. Malaysia’s tropical climate is characterized by uniform temperatures, frequent heavy rains, and thunderstorms. Rainfall rates primarily depend on the seasonal monsoons, which consist of the northeast monsoon from October to March and the southeast monsoon from April to September [33].

The ITU-R statistical model is utilized to estimate rain attenuation for mmWave frequencies ranging from 1 GHz to 100 GHz over path differences of up to 60 km. In this model, rain attenuation depends on the path difference, operating frequency, rain rate, polarization, and path reduction factor [38,39,40,41]. The model excludes attenuation due to snowfall, fog, and hail. Consequently, the ITU-R model proves to be the most accurate in predicting rain attenuation in Malaysia.

According to the ITU-R model, the path loss due to rainfall in the outdoor region can be defined as the product of the specific attenuation $\left({\gamma }_A\right)$ and the effective propagation path length $\left(R_{eff}\right)$ [42]. Therefore, the total attenuation $\left(\alpha \right)$ is

$$\alpha ={\gamma }_A*R_{eff}$$

(1)

The specific attenuation depends on the intensity of the rainfall, which is estimated from the rain rate $\left(A\right)$ after some percentage of time through the power law relationship, which is given as

$${\gamma }_A=K.A^{\beta }$$

(2)

where K is a constant and $\beta$ depends on the operating frequency and polarization of the signal. The K and $\beta$ values for mmWave frequencies from 1 GHz to 1000 GHZ are listed in the lookup table [41] for both vertical and horizontal polarization. The Malaysian Meteorological Department maintains a database of general weather forecast data. The rainfall rate after a particular time percentage, i.e., 0.01%, is considered in this paper when estimating the path attenuation.

The effective path length is always less than the actual path length, and is defined as the product of the distance between the transmitter and receiver $\left(R\right)$ and the path reduction factor due to rainfall $\left(p_r\right)$ after a particular time period. The path reduction factor (or path attenuation) is estimated as the difference between the received signal strength under clear skies and the same under rainy conditions. Therefore, the total signal attenuation due to rainfall is

$$\alpha =K.A^{\beta }.R.p_r,$$

(3)

while the path reduction factor of the ITU-R model can be expressed as [43]:

$$p_r=\frac{1}{0.477R^{0.633}{A}_{0.01}^{0.073\beta }{f}^{0.123}-10.579[1-exp(-0.0241R)]}$$

(4)

where R is the path difference between the transmitter and receiver in $km$ and f is the operating frequency in GHz. The path loss due to rainfall can then be estimated by substituting Equation (4) into Equation (3).

4. Indoor and Outdoor Probability Models

The performance of wireless channels is influenced by the indoor and outdoor LOS and NLOS probabilities, which are key parameters in channel estimation. The probability of an LOS path between the transmitter and receiver antennas is defined as the LOS probability [15]. When any path is obstructed by buildings, trees, or other obstacles, it is termed an NLOS path. In this paper, we consider both indoor and outdoor probability models at mmWave frequencies. In indoor environments such as shopping malls, offices, residential areas, etc., obstructions such as walls, furniture, ceilings, and partition sheets may lead to NLOS propagation, reducing the LOS probability. As the LOS probability decreases, the path loss increases, resulting in weaker signal strength. Therefore, LOS probability models play a crucial role in path loss estimation in 5G networks.

In this paper, we consider the existing ITU-R model, the WINNER-II A1 model, and our proposed model to estimate indoor probabilities in mmWave frequency bands and determine the optimal distance to maximize the probability and enhance the path loss [43]. The mathematical expressions of the three indoor probability models are provided below [43].

ITU-R Model:

$$\begin{array}{ccc}\hfill P\left(R\right)& =& 1,R\le R_a\hfill \\ \hfill & =& exp(-R(R-R_a)/a),R_a<R\le R_b\hfill \\ & =& b,R>R_b\hfill \end{array}$$

(5)

WINNER II-A1 Model:

$$\begin{array}{ccc}\hfill P\left(d\right)& =& 1,R\le R_c\hfill \end{array}$$

(6)

$$\begin{array}{ccc}& =& 1-K{\left(-{(R-Z.log10\left(R\right))}^3\right)}^{1/3},R>R_c\hfill \end{array}$$

(7)

Proposed Model:

$$\begin{array}{ccc}\hfill p\left(d\right)& =& 1,R\le R_d\hfill \end{array}$$

(8)

$$\begin{array}{ccc}& =& 1-L{\left(-{(R-Tlog10\left(R\right))}^3\right)}^{1/3},R_d<R\le R_e\hfill \end{array}$$

(9)

$$\begin{array}{ccc}& =& X.exp\left(\frac{-R(R-R_e)}{B}\right),R>R_e\hfill \end{array}$$

(10)

The outdoor probability can be estimated using 3GPP models. There are four 3GPP models in the literature [44]. The mathematical expressions of these four models are as follows.

3GPP 3D model:

$$P\left(R\right)=min\left(\frac{18}{R},1\right)\left(1-exp\left(\frac{-R}{63}\right)\right)+exp\left(\frac{-R}{63}\right)$$

(11)

3GPP R1/R2 Model:

$$P\left(R\right)=min\left(\frac{R_1}{R},1\right)\left(1-exp\left(\frac{-R}{R_2}\right)\right)+exp\left(\frac{-R}{R_2}\right)$$

(12)

Squared Model:

$$P\left(R\right)={\left(min\left(\frac{R_1}{R},1\right)\left(1-exp\left(\frac{-R}{R_2}\right)\right)+exp\left(\frac{-R}{R_2}\right)\right)}^2$$

(13)

Inverse Exponential Model:

$$P\left(R\right)=\frac{1}{1+e^{R_1(R-R_2)}}$$

(14)

In the outdoor region, the path loss probability depends on the type of geographic location, such as urban, semi-urban, or rural, along with the antenna characteristics, such as the location of the antenna, type of antenna, angle of tilt, etc. According to the location, obstructions such as ground surfaces, buildings, trees, advertising signs, bollards, etc., may change, affecting the path loss probability. In this paper, we consider urban and highway scenarios along with vehicle density to estimate the outdoor LOS and NLOS probabilities. The mathematical expressions for the outdoor LOS and NLOS probabilities for the highway and urban scenarios are shown in

Table 2. LOS and NLOS probability formulas for urban and highway scenarios under different vehicle densities.

Location	Probability	Vehicle Density	$\mathit{P}\left(\mathit{R}\right)=\mathit{min}(1,\mathit{max}(0,{\mathit{aR}}^2+\mathit{bR}+\mathit{c}$))
			a	b	c
Highway	$P_{LOS}^H\left(R\right)$	Low	$1.5*{10}^{-6}$	$-0.0015$	1
		Medium	$2.7*{10}^{-6}$	$-0.0025$	1
		High	$3.2*{10}^{-6}$	$-0.003$	1
	$P_{NLOS}^H\left(R\right)$	Low	$-2.9*{10}^{-7}$	$0.00059$	$0.0017$
		Medium	$-3.7*{10}^{-7}$	$0.00061$	$0.0150$
		High	$-4.1*{10}^{-7}$	$0.00067$	0
Urban	$P_{LOS}^U\left(R\right)$	Low	$P_l\left(R\right)=min(1,max(0,0.8548*e^{-0.0064R}))$
		Medium	$P_m\left(R\right)=min(1,max(0,0.8372*e^{-0.0114R}))$
		High	$P_h\left(R\right)=min(1,max(0,0.8962*e^{-0.017R}))$
	$P_{NLOS}^U\left(R\right)$		1-$P_{LOS}^U\left(R\right)$

[17].

In this paper, outdoor LOS and NLOS probability models are utilized to estimate path loss in urban and highway scenarios. In these scenarios, population density, vehicle traffic density, and call rates are significantly higher than in semi-urban and rural areas. Path loss and signal power are influenced by frequent and heavy rainfall. Therefore, in this paper we focus on the urban and highway scenarios in order to estimate the optimal path difference to enhance path loss.

5. Link Margin Estimation

In 5G cellular networks, accurate path loss estimation helps to determine the number of cells required to cover a given area, assess the loss due to the transmission medium, calculate the actual power required for transmission, and optimize overall system performance. Path loss is defined as the power loss or attenuation of a propagating signal in the transmission medium. In the literature, various path loss models have been introduced to estimate path loss based on specific environmental and transmission medium conditions.

In this paper, we consider the effects of rainfall in urban and highway scenarios, as well as in free-space transmission media, to estimate the path loss. Due to high population density and data requirements, we specifically examine path loss in urban $\left(P{L}^U\right)$ and highway $\left(P{L}^H\right)$ scenarios. According to the free-space path loss model, the LOS and NLOS path losses for urban and highway scenarios are estimated as follows [33].

LOS Path loss:

$$P{L}^U=38.77+16.7*log10\left(R\right)+18.2*log10\left(f\right)$$

(15)

$$P{L}^H=32.4+20*log10\left(R\right)+20*log10\left(f\right)$$

(16)

NLOS Path loss:

$$P{L}^{U/H}=36.85+30*log10\left(R\right)+18.9*log10\left(f\right)$$

(17)

The frequent heavy rainfall in Malaysia influences the path loss. The modified LOS and NLOS path loss expressions for the urban and highway scenarios are as follows:

LOS Path loss:

$$P{L}_R^U=38.77+16.7*log10\left(R\right)+18.2*log10\left(f\right)+\alpha$$

(18)

$$P{L}_R^H=32.4+20*log10\left(R\right)+20*log10\left(f\right)+\alpha$$

(19)

NLOS Path loss:

$$P{L}^{U/H}=36.85+30*log10\left(R\right)+18.9*log10\left(f\right)+\alpha$$

(20)

where $\alpha$ is the path attenuation due to rainfall, which is provided by Equation (3), R is the path difference between transmitter and receiver in m, and f is the operating frequency in GHz.

The link margin due to heavy rainfall is

$$LM=\frac{G_t{G}_r{P}_t}{K{T}_s\Gamma {\left(PL\right)}_0{\left(PL\right)}_R(E_b/N_0)},$$

(21)

where $G_t$ and $G_r$ are the respective gains of the transmitter and receiver antenna, $P_t$ is the transmitter power, K is Bolztman’s constant, ${\left(PL\right)}_0$ is the free-space path loss, $T_s$ is the noise temperature, ${\left(PL\right)}_R$ is the path loss due to rainfall, $\mathsf{\Gamma }$ is the data rate, and $E_b/N_0$ is the ratio of the energy per bit to the spectral density of the noise.

6. Simulation Results

Millimeter-wave propagation can be influenced by different propagation paths, such as LOS and NLOS, different regions, including indoor, outdoor, highway, urban, semi-urban, and rural areas, and different weather conditions, such as rain, fog, snow, dust, and sand. Several path loss models [32] are considered in our investigation of the impact of heavy and frequent rainfall on propagating mmWave signals in the Malaysia region. In this paper, we focus on LOS and NLOS propagation paths in the outdoor region with rain attenuation. Path attenuation depends on the operating frequency, path difference, and rainfall intensity. Equation (3) is used to estimate the path attenuation due to rainfall in terms of the path difference and operating frequency. Path attenuation is estimated for various mmWave frequencies (10 GHz to 100 GHz) using the ITU-R model at 0.01% of the time of the rainfall rate.

We used MATLAB R2019a software to estimate the rain attenuation, LOS and NLOS probabilities, path loss, and link margin. The estimated rain attenuation is employed to determine the path loss and link margin in the urban and highway scenarios. The path loss and link margin are compared with and without rain attenuation for both the urban and highway scenarios, and the optimal path difference in these scenarios is estimated.

Figure 1 and Figure 2 visualize the path attenuation under specific conditions: a rainfall rate of 0.01%, mmWave frequencies spanning from 10 GHz to 100 GHz, and path differences ranging from 100 m to 1 km in the former figure and 10 km to 100 km in the latter. Analyzing these two figures confirms that the path attenuation increases as the operating frequency and path difference increase.

Within the range of mmWave frequencies (10–100 GHz) and path differences between 100 m and 1000 m, it is evident from Figure 1 that the maximum path attenuation remains at only 7 dB. However, as the path difference extends from 1 km to 100 km, the path loss increases from 7 dB to 37 dB. Consequently, the path attenuation remains notably low for shorter path differences while significantly increasing for longer ones. This relationship underscores the direct correlation between path attenuation and path difference, highlighting that the impact of rainfall on signal propagation is more pronounced with greater path differences. Therefore, the intensity of rainfall has a notable influence on path attenuation, which in turn, impacts the overall performance of the communication system.

The LOS probabilities for indoor and outdoor regions based on different probability models are shown in Figure 3 and Figure 4, respectively. We estimated the indoor LOS probability for the ITU-R model, the WINNER-II A1 model, and our proposed model. From Figure 3, it can be noticed that the ITU-R models provide a better LOS probability with a path difference of (0–25 m) compared to the other two models. In addition, the LOS probability gradually decreases up to a 10 m path difference, then maintains a constant value or varies only slowly. Therefore, the optimal path difference for better performance in LOS propagation is estimated at 10 m in the indoor region. This optimal path difference reduces the path loss of an indoor communication system.

Figure 4 shows the LOS probability variations with the path difference (0–1000 m) for the four 3GPP outdoor models. From Figure 4, it can be noticed that the LOS probability gradually decreases as the path difference increases from 1 m to 1000 m. The LOS probability is not only affected by the path difference, as environmental conditions, antenna characteristics, and the geographical region all have impacts. In this paper, we estimate both the LOS and NLOS probabilities in the outdoor urban and highway scenarios.

The outdoor LOS and NLOS probabilities for the urban and highway scenarios are shown in Figure 5, Figure 6, Figure 7 and Figure 8, respectively. The LOS and NLOS probabilities in the highway and urban scenarios are affected by the path difference and vehicle density parameters. From Figure 5, it can be noticed that the LOS probability decreases gradually as the path difference (0–500 m) increases, which is because in the urban region the signal propagation path is effected by obstructions such as high-rise buildings, trees, advertising signs, etc. These obstructions cause multi-path fading, which reduces the signal’s energy. In addition, the LOS probability is influenced by the vehicle density, decreasing as the vehicle density increases. Vehicles traveling along the propagating mmWave signal path increase signal fading and absorption, thereby reducing the received signal strength. Therefore, increased path differences and vehicle density decrease the LOS probability in the outdoor urban region.

Figure 6 shows the LOS probability variation with a path difference of (0–500 m) in the highway scenario for various vehicle densities. From Figure 6, it can be noticed that the LOS probability decreases with the path difference and vehicle density. However, the LOS probability of the highway scenario is higher than that of the urban scenario, as the LOS path in the urban scenario is obstructed by a greater number of obstructions such as high-rise buildings, lamp posts, advertisement boards, etc., compared to the highway scenario. On highways, the signal is able to propagate in free space, and the optimal distance is higher in the highway scenario compared to the urban scenario.

Figure 7 and Figure 8 show the NLOS probabilities for the urban and highway scenarios, respectively. When the LOS path is obstructed by trees, buildings, or vehicles, it is termed the NLOS path. In 3GPP outdoor models [44], these distinctions are not made. For a short path difference, the path between the transmitter and receiver is assumed to be free of obstructions. As a result, in both cases the probability increases with the path difference and vehicle density. It can be observed that the NLOS probability is high in the urban scenario compared to the highway scenario, which is exactly opposite to the LOS probability seen in Figure 5 and Figure 6.

The path loss in the urban and highway scenarios for the LOS and NLOS paths with and without rainfall is estimated using Equations (15)–(20), and the results are shown in Figure 9 and Figure 10. From Figure 9 and Figure 10, it can be noticed that the path loss is influenced by the rainfall. The mmWave propagating signal loss depends on the path difference and rainfall intensity. From Figure 9, it can be observed that the path loss is higher with rainfall compared to the path loss without rainfall in both the urban and highway scenarios. Signal propagation is significantly higher in the urban scenario than the highway scenario. According to Friis’ law, propagating signals weaken over distance in a highway scenario. The observed path loss is substantially lower, pointing to a wave guide effect brought on by reflections from static obstacles in street canyons, which are more likely to occur.

Figure 10 shows the NLOS path loss of the urban and highway scenarios with and without rainfall. According to 3GPP models, there is no variation in the path loss for the urban and highway scenarios [45]. From this figure, it can be noticed that the path loss is high when there is heavy rainfall compared to a clear sky. Therefore, from all these results it is the case that the path loss depends on the path difference, rainfall intensity, and operating frequency. The path loss and path attenuation increase as the path difference, operating frequency, and rainfall intensity increase, as observed in Figure 1, Figure 2, Figure 9 and Figure 10. Along with these, the path loss performance depends on the antenna height, tilt angle, and antenna location or position [46,47].

Equation (20) is used to calculate the link margin with rainfall attenuation. The link margin is estimated at $P_t=27$ dBm, $T_s=30.62$ dB, and $K=1.38*10$–23 dB for variable antenna gains and free space path loss with mmWave frequencies ranging from 1 GHz to 100 GHz and path differences ranging from 100 m to 1 km. The link margin significantly decreases as the path difference increases, as shown in Figure 11. From this figure, it can be observed that the link margin for the urban scenario without rain fall is high compared to the remaining five scenarios. It can be further observed that the link margin is high for the LOS path compared to the NLOS path.

Figure 12 shows the variation of the link margin with the path difference for the urban and highway scenarios. From Figure 12, the optimal LOS and NLOS path difference values can be estimated to achieve link margins of 10 dB, 15 dB, and 20 dB in the urban and highway scenarios with and without rainfall, as listed in Table 2. These optimal values can be used to provide better throughput, higher data rates, and enhanced overall system performance.

In

Table 3. Optimal path difference for a given link margin.

Link Margin (dB)	Optimal Path Difference (m)
	${\mathit{R}}_{\mathit{WOR}}^{\mathit{U}}$	${\mathit{R}}_{\mathit{WR}}^{\mathit{U}}$	${\mathit{R}}_{\mathit{WOR}}^{\mathit{H}}$	${\mathit{R}}_{\mathit{WR}}^{\mathit{H}}$	$\mathit{R}{\mathbf{1}}_{\mathit{WOR}}^{\mathit{U}/\mathit{H}}$	$\mathit{R}{\mathbf{1}}_{\mathit{WR}}^{\mathit{U}/\mathit{H}}$
10	1325	870	868	615	225	180
15	1000	673	670	490	175	150
20	710	500	495	378	135	118

, $R_{WOR}^U$ and $R_{WR}^U$ are the respective optimal path differences for the urban scenario without and with rainfall attenuation, $R_{WOR}^H$ and $R_{WR}^H$ are the respective optimal path differences for the highway scenario without and with rainfall attenuation, and $R{1}_{WOR}^{U/H}$ and $R{1}_{WR}^{U/H}$ are the respective optimal path differences between the urban and highway scenarios without and with rainfall attenuation. From Table 2, it can be noticed that the optimal path difference in the urban scenario is better than that in the highway scenario. In addition, it can be seen that the optimal path difference without rainfall is greater than the optimal path difference with rainfall. These optimal path difference values improve coverage capacity, system performance, and throughput, and can boost data rates by up to 1 Gbps. The impact of rain on connected vehicle applications is reduced by these optimal distances, potentially lowering the number of crashes that occur on the roads when it rains heavily.

7. Conclusions

In this paper, we have investigated the impact of rainfall in urban and highway scenarios in 5G communication networks at mmWave frequencies. The rainfall attenuation, LOS and NLOS probability, path loss, and link margin are estimated based on the indoor and outdoor probability models. Our simulation results show that the path loss increases and the link margin decreases as the path difference and rainfall rate increase. The impact of rainfall on path loss at higher frequencies is significantly higher than at lower frequencies. The optimal path difference for urban and highway scenarios with and without rainfall is estimated and compared. The urban scenario without rainfall provides a higher optimal distance compared to the highway scenario with and without rainfall attenuation, and the urban scenario with rainfall attenuation achieves high data rates of 1 Gbps.

We considered indoor LOS probability models and estimated the path loss and link margin. The concept and estimated results are applicable to any location experiencing comparable weather conditions. In the future, indoor NLOS probability models could be used to estimate the path loss, link margin, and optimal path difference by designing a small indoor region. Notably, the impact of walls, curtains, ceilings, etc., in indoor regions remains to be addressed in future research.