Received 26 November 2024, accepted 1 February 2025, date of publication 10 February 2025, date of current version 20 February 2025.

Digital Object Identifier 10.1109/ACCESS.2025.3540713

Performance Evaluation of Cross M-QAM

Modulation Over Standardized RF2 5G
Frequency Bands Using the
η-µ Fading Model
WILIAN EURÍPEDES VIEIRA 1 , PEDRO LUIZ LIMA BERTARINI 2 ,
KARINE BARBOSA CARBONARO2 , GILBERTO ARANTES CARRIJO 3,

EDSON AGUSTINI 4 , AND ANDRÉ ANTÔNIO DOS ANJOS 2

1 IME-Institute of Mathematics and Statistics, UFU-Federal University of Uberlandia, Campus Patos de Minas, Patos de Minas, Minas Gerais 38701-002, Brazil
2 FEELT-Faculty of Electrical Engineering, UFU-Federal University of Uberlandia, Campus Patos de Minas, Patos de Minas, Minas Gerais 38701-002, Brazil
3 FEELT-Faculty of Electrical Engineering, UFU-Federal University of Uberlandia, Campus Uberlandia, Uberlandia, Minas Gerais 38408-100, Brazil
4 IME-Institute of Mathematics and Statistics, UFU-Federal University of Uberlandia, Campus Uberlandia, Uberlandia, Minas Gerais 38408-100, Brazil

Corresponding author: André Antônio dos Anjos ([Link]@[Link])

This work was supported in part by Minas Gerais Research Foundation (FAPEMIG) through Minas Gerais Research Network in 5G and
6G Project under Grant RED-00194-23.

ABSTRACT This paper delves into the study of the generalized short-term fading model η-µ. It introduces,
for the first time in the literature, closed-form expressions for the Symbol Error Probability (SEP) of cross
M -QAM modulation operating under the η-µ fading channel. Furthermore, the SEP of the aforementioned
modulation is evaluated across selected frequency bands within the standardized Range Frequency 2
(RF2) millimeter-wave (mm-wave) for 5G systems, specifically at 26, 28, and 39 GHz. The assessment
is based on channel modeling results available in the literature, derived from an extensive measurement
campaign conducted under both line-of-sight (LoS) and non-line-of-sight (nLoS) conditions within an indoor
environment, and considering different Tx-Rx distances. The findings provide valuable information on the
performance of cross M -QAM modulation across different modulation orders, frequencies, line-of-sight
conditions, and Tx-Rx distances in realistic mm-wave 5G scenarios, enhancing our understanding of its
suitability in various real deployment conditions.

INDEX TERMS Wireless communication, 5G mobile communication, channel estimation, cross M -QAM,
millimeter wave, fading channels.

I. INTRODUCTION are already investigating the next wireless systems that will
The constant evolution of telecommunications systems is replace them (6G and Wi-Fi 7).
driven by advancements in hardware, software, and the Quadrature Amplitude Modulation (QAM) is a key
widespread adoption of high-speed broadband networks. technique in the context of modern wireless communication,
The convergence of artificial intelligence, edge computing, offering efficient data transmission. In square QAM, symbols
and the Internet of Things (IoT) further accelerates this on a grid represent unique amplitude and phase combinations,
evolution [1]. It is for these reasons that systems such as 5G or enabling high data rates in applications like broadband and
Wi-Fi 6 are still in the implementation phase, and researchers Wi-Fi [2]. Non-square QAM variants, like cross QAM,
adapt symbol spacing for increased efficiency. This is
especially valuable in challenging wireless environments
The associate editor coordinating the review of this manuscript and with interference and fading, optimizing spectral efficiency
approving it for publication was Ahmed Almradi . and data throughput in diverse communication scenarios.
2025 The Authors. This work is licensed under a Creative Commons Attribution 4.0 License.
31024 For more information, see [Link] VOLUME 13, 2025
W. E. Vieira et al.: Performance Evaluation of Cross M-QAM Modulation

In wireless environments, obstacles can block the direct the performance of cross M -QAM modulation in realistic
radio path, causing shadowing or long-term fading. On the mm-wave 5G scenarios, enhancing our understanding of its
other hand, multipath effects are caused by reflection and suitability in diverse real-world deployment conditions and
scattering of the transmitted signal, leading to short-term revealing the operational challenges of future telecommuni-
fading. These phenomena can result in severe signal variation, cation systems in the mm-wave range.
making communication impractical in extreme cases [3]. The remainder of this article is organized as follows.
Therefore, the development of reliable systems necessitates Section II presents fundamental concepts of cross M -QAM
accurate channel models. modulation, the influence of fading on the SEP, and an
The diverse scenarios anticipated for advanced wireless overview of the η-µ fading model. In Section III, a novel
network applications, particularly in the mm-wave band closed-form expression for the SEP considering the η-µ
for 5G or even 6G systems, will undoubtedly result in model is derived. Section IV revisits the field measurements
varied propagation conditions. Traditional fading models and channel modeling used in this work and conducted for
such as Rayleigh [4], Hoyt [5], Rice [6], Nakagami-m the standardized RF2 millimeter-wave bands for 5G systems.
[7], and Weibull [8] may still prove useful but exhibit Section V describes computational simulations for validating
limited flexibility in addressing more complex situations. the proposed expression and its results, analyzing the SEP
Alternatively, more adaptable and generalized models like performance of cross M -QAM modulation under various
α-µ [9], κ-µ, and η-µ [10] can be useful. propagation conditions, modulation orders, and frequencies.
Generalized models have been widely applied in Finally, Section VI provides concluding remarks and insights
the literature for short-term fading modeling. They are drawn from this study.
particularly useful in practical scenarios such as vehicular-
to-vehicular networks (V2V) [11], device-to-device com- II. BASIC REVIEW
munications (D2D) [12], body area networks (BAN) [13], A. CROSS M-QAM
ultrasonic underwater channels [14], 5G RF2 networks [15], Quadrature Amplitude Modulation is a widely used modula-
narrow-band communication [16], and mm-wave propaga- tion scheme in digital communication where both amplitude
tion [17], [18], [19], [20]. A recent study [21] evaluated the and phase are utilized to carry information. The constellation
α-µ model using practical measurements, demonstrating its of QAM has M = 2b symbols, where b is the number
superior capability in modeling composite fading compared of bits of each constellation symbol. In the cross M -QAM
to traditional models, thus highlighting its versatility. The constellation, b ⩾ 5 is an odd integer. Its symbol alphabet is
body of research on these generalized fading models is defined as follows [22]:
extensive, with numerous studies illustrating their practical
usability. Many of these studies conclude that the modeling s = ± (2u − 1) ± (2v − 1) j, (1)
capabilities of generalized models exceed those of simpler where
models [15], [17], [18], providing a more accurate repre-
sentation of various wireless communication environments. u = 1, . . . , 3 · 2(b−5)/2 ,
The traditional models may not suffice in some specific (
1, . . . , 3 · 2(b−5)/2 ; if u ⩽ 2(b−3)/2
scenarios, such as some encountered in mm-waves [17], v= ,
[18], [19], [20]. In these situations, the received signal’s 1, . . . , 2(b−3)/2 ; if u > 2(b−3)/2
envelope probability density function (PDF) may exhibit and j is the imaginary unit of complex numbers.
bimodality effects, especially in challenging severe fading In addition, there is also the so-called scale factor, defined
conditions, necessitating more generalized models to capture by:
all peculiarities.
This work explores the study of the generalized fading k= √1 , (2)
EM
model η-µ [10], known for its mathematical simplicity and
its ability to model nonlinearities and correlations imposed where EM is the average symbol energy of the M -QAM
by the propagation medium. We derived, for the first time constellation, which is used to re-scale the constellation in
in the literature, a closed-form SEP expression for cross terms of energy. Specifically, this average symbol energy of
M -QAM modulation operating in the η-µ fading channel. cross M -QAM constellations is defined as [22]:
The proposed expression was validated through performance  q 
simulations in MATLAB. The digital communication sim- 1
Es = 128 82M + M 128 M − 128 . (3)
ulations were conducted across selected frequency bands
within the standardized RF2 mm-wave range for 5G systems, The points of the M -QAM constellation are classified
specifically at 26, 28, and 39 GHz. This assessment is according to the number of neighboring symbols horizontally
based on channel modeling results derived from an extensive and vertically. The quantities of each type of point√are: (a)
measurement campaign conducted under both LoS and nLoS Eight points with two neighboring symbols; (b) 3 √2M −
conditions within an indoor environment, as documented in 16 points with three neighboring symbols; (c) M − 3 2M +
the literature [15]. The findings provide valuable insights into 8 points with four neighboring symbols (interior points).

VOLUME 13, 2025 31025

 W. E. Vieira et al.: Performance Evaluation of Cross M-QAM Modulation

of γ , considering η-µ fading, is given by [24]:

√
2 π µµ+ 2 hµ γ µ− 2
1 1

p(γ ) =
0(µ)H µ− 2 γ µ+ 2
1 1

γ γ
   
· exp −2µh I 1 2µH , (6)
γ µ− 2 γ
where h and H are constants, 0 is the Gamma function, given
by:
Z ∞
0(z) = t z−1 exp(−t)dt; Re(z) > 0, (7)
0
and I 1 is the modified Bessel function of first kind and
µ− 2
order µ − 12 .
According to [25] (p. 375, Eq. 9.6.7) we have:
FIGURE 1. Cross 32-QAM constellation.
z µ−

1
2
Iµ− 1 (z) ≈ 2  (8)
2
0 µ + 12
As an example, in Fig. 1 we have the representation of the
cross 32-QAM modulation scheme with its types of points. where µ − 21 is not a negative integer.
Therefore, considering only additive white Gaussian noise By substituting (8) into (6), considering z = 2µH γγ ,
(AWGN) the cross QAM SEP is given by [22]: we obtain:
√
2 π µ2µ hµ γ 2µ−1
exp −2µh γγ .
 
p (γ ) ≈
 q 
√ (9)
Pe (γ ) = 2 − 32 M2 erfc(k γ )
 
0 (µ) 0 µ + 12 γ 2µ
 q 
√ The parameter µ is related to the number of multipath
− 1 + M2 − 23 M2 erfc2 (k γ ), (4)
clusters, while γ denotes the average normalized signal-to-
noise ratio.
log (M )E
where γ = NE0s = 2N0 b is the instantaneous normalized In Format 1,
signal-to-noise ratio at the receiver, Eb is the average bit 2+η−1 +η
h= 4 , (10)
energy in joules, and N0 is the noise power spectral density at
the receiver, given in [watts/Hz]. where 0 < η < ∞ is the scattered wave power ratio between
the in-phase and quadrature components of each cluster.
B. DIGITAL MODULATION IN FADING CHANNELS In Format 2,
In a propagation scenario with fading, the instantaneous value h= 1
, (11)
of γ will vary randomly and its PDF will depend on the 1−η2
considered type of fading. Thus, the average SEP can be where −1 < η < 1 is the correlation coefficient between the
calculated as [23]: scattered wave in-phase and quadrature components of each
cluster of multipath.
Z ∞
Pe = Pe (γ ) p (γ ) dγ (5) III. CLOSED-FORM SEP EXPRESSION FOR CROSS M-QAM
0
UNDER η-µ FADING
where, Pe (γ ) is the SEP conditioned on a fixed γ and p (γ ) Applying the Pe (γ ) for the cross M -QAM constellation given
is the PDF of γ . in (4), and p (γ ) for the η-µ fading model presented in (9),
to Pe given in (5), we obtain
√
C. THE η-µ FADING MODEL 2 π hµ µ2µ
Pe = (AX −BY ) , (12)
The η-µ fading model, proposed in [10], is grounded in
 
0 (µ) 0 µ + 12 γ 2µ
physical propagation phenomena and is developed in two
distinct formats. Format 1 considers a signal composed where:
of clusters of multipath waves propagating in a non-  q 
homogeneous environment, while Format 2 also considers A = 2 − 23 M2 , (13)
a signal composed of clusters of multipath waves, but  q 
propagating in a homogeneous environment, where the in- B = 1 + M − 2 M2 ,
2 3
(14)
phase and quadrature components are correlated. The PDF

31026 VOLUME 13, 2025

W. E. Vieira et al.: Performance Evaluation of Cross M-QAM Modulation

and X and Y are respectively defined as: and we apply the approximation to the equation (23) to obtain
Z ∞ Z ∞
√

erfc k γ γ 2µ−1 exp −2µh γγ dγ ,
2γ
   
X= (15) Y = 18k14µ x 4µ−1 exp − 2µh+2k
k γ
2 x 2
dx
Z0 ∞ 0
Z ∞
√

erfc2 k γ γ 2µ−1 exp −2µh γγ dγ ,
2γ
   
Y = (16) + 3k14µ x 4µ−1 exp − 6µh+7k
3k 2 γ
x 2
dx
0
Z0 ∞
2γ
 
in which k is a scaling factor defined in (2). + 2k14µ x 4µ−1 exp − 6µh+8k x 2
dx. (25)
Initially, the term X is determined by applying the 3k γ
2
0
substitution from equation (17), which involves a change of By using the definite integral in equation (21) again,
variables: equation (23) results in:
x2
γ = 2x
(17)
  2µ 
k2
⇒ dγ = k2
dx. 0(2µ)γ 2µ 1 1
Y = 2 18 2µh+2 k 2 γ
This substitution changes the integration limits from γ = 0   2µ 
(corresponding to x = 0) to γ = ∞ (corresponding to x = + 0(2µ)γ 2µ 1 3
2 3 6µh+7 k 2 γ
∞). Thus, we can then rewrite the term X as:
Z ∞   2µ 
+ 0(2µ)γ
2µ
1 3
.
 
X = k 24µ erfc (x) x 4µ−1 exp − 2µh
k2γ
x 2
dx. (18) 2 2 6µh+8 k 2 γ (26)
0
By applying equations (26) and (22) to (12), we obtain
As stated by [26], the complementary error function is
the closed-form average SEP expression for a cross M -QAM
approximated to:
  constellation in the η-µ fading channel as presented as in (27),
exp(−x 2 ) exp − 43 x 2 shown at the bottom of the page.
erfc(x) ≈ 6 + 2 . (19)
Hence, IV. FIELD MEASUREMENTS AND CHANNEL MODELING
Z ∞ In this work, we utilize the measurement results obtained
γ 2 2
 
X= 1
3k 4µ
x 4µ−1 exp − 2µh+k
k2γ
x dx from [15] to validate the performance of cross QAM schemes
0 Z ∞ in mm-wave bands. The channel measurements in [15]
2γ
 
+ k 14µ x 4µ−1 exp − 6µh+4k
3k 2 γ
x 2 dx. (20) were carried out in a laboratory of the iTEAM Research
0 Institute at the University Polytechnic of Valencia, Spain.
We use the following definite integral from the integral table The dimensions of the laboratory are 13.5-m-long, 7-m-width
in [27]: and 2.6-m-high. The environment is characterized by rich-
Z ∞ multipath, with scattered and reflected contributions, due
0(v)
x 2m exp(−βx 2n )dx = 2nβ v, (21) to radio-frequency equipment, metallic cupboards, multiple
0
work tables, and benches equipped with electronic devices
where v = 2m+1
2n . and chairs, among others. The channel’s frequency response
By comparing equations (20) and (21), we obtain: n = was quantified in the frequency domain using a Keysight
2γ
1, v = (4µ−1)+1
2 = 2µ and β = 2µh+k k2γ
. Therefore, PNA N5227 VNA-based channel sounder, employing Q-PAR
equation (18) results in: ultra-wideband antennas with vertical polarization and an
  2µ  2µ  omnidirectional pattern. The receiver antenna, positioned to
0(2µ)γ 2µ 1
X= 2
1
3 2µh+k 2 γ
3
+ 6µh+4k 2 γ . (22) simulate a uniform rectangular array, was part of a system
for detailed channel behavior assessment. The measurements
The term Y can be found by considering the same variable enable the examination and comparison of channel behavior
substitution presented in (17) and can be rewritten as: across the 26, 28, and 39 GHz bands under identical
Z ∞   conditions.
Y = k 24µ erfc2 (x) x 4µ−1 exp − 2µh
k2γ
x 2
dx. (23) In [15], measurements were conducted at various distinct
0
positions across six different scenarios, and short-term fading
We expand the approximation of [26] given in (19), modeling was performed for several distributions using the
nonlinear least square method, including the η-µ distribution,
   
exp(−2x 2 ) exp − 73 x 2 exp − 83 x 2
2
erfc (x) ≈ 36 + 6 + 4 , (24) which is of particular interest for this work. In this study,

√  q   2µ 2µ 
πhµ µ2µ 0(2µ)

3 2 1 1 3
Pe =  2− 2 M 3 2µh+k 2 γ
+ 6µh+4k 2 γ
0(µ)0 µ+ 12
√  q  2µ 2µ 2µ 
πhµ µ2µ 0(2µ)
  
2 3 2 1 1 1 3 1 3
−  1+ M − 2 M 18 2µh+2k 2 γ
+ 3 6µh+7k 2 γ
+ 2 6µh+8k 2 γ
(27)
0(µ)0 µ+ 12

VOLUME 13, 2025 31027

 W. E. Vieira et al.: Performance Evaluation of Cross M-QAM Modulation

we specifically focus on the results obtained for three of V. PERFORMANCE EVALUATION

these six scenarios (2, 3, and 5). Scenarios 2 and 3 involve a To evaluate the performance of cross M -QAM modulation
spacing of 7.78 meters between the transmitting and receiving operating in an indoor mm-wave environment, a simulated
antennas, with measurements carried out under LoS and nLoS cross M -QAM communication system was developed in
conditions, respectively. Scenario 5 involves LoS conditions MATLAB. The system includes a transmitter that generates
with a spacing of 4.91 meters between the antennas. These N random symbols from the M -QAM constellation. The
three scenarios were selected due to their coverage of channel between Tx and Rx is modeled as a slow η-µ
both LoS and nLoS conditions, as well as representing the fading channel, assumed not to vary significantly during the
maximum and minimum Tx-Rx distances in the measurement interval of each symbol, thus introducing a multiplicative
campaign. Table 1, partially extracted from [15], presents the gain to the envelope of the received symbols.1 A vector
fitting results for scenarios 2, 3, and 5 using the η-µ fading of circularly symmetric complex Gaussian noise with zero
model. mean and variance N20 , dependent on the configured signal-
to-noise ratio, is added to the received symbol vector, already
TABLE 1. Fitting results for Scenarios 2, 3, and 5 considering η-µ weighted by the channel gain. The receiver is implemented
distribution extracted from [15]. with a zero-forcing equalizer at the input, followed by a
maximum likelihood detection system. The SEP estimation
is performed by dividing the number of incorrectly detected
symbols by the total number of transmitted symbols. All
performance curves presented in this section were generated
with NEb0 varying from 0 dB to 50 dB and for each point
considering the transmission of N = 108 symbols.
The developed system was initially simulated for M =
{32, 128, 512}, considering an η-µ channel with the η and µ
values presented in Table 1 for Scenarios 2 (LoS) and 3
(nLoS). The estimated Pe (marks) at the frequencies of
The results in Table 1 demonstrate that the η-µ distribution 26, 28, and 39 GHz are shown in Figs. 2, 3 and 4, along
provides good adherence to practical data, with the normal- with the theoretical curves for Scenario 2 (solid lines) and
ized mean square error (NMSE) consistently below −20 dB Scenario 3 (dashed lines), obtained using equation (27) for
and, in some cases, even lower than −36 dB. By drawing each evaluated case.
new conclusions from the results presented in Table 1 (not Analyzing the results presented in Figs. 2, 3 and 4,
included in [15]), it is possible to verify that in almost all it can be observed that the theoretical and simulated curves
cases, η ≈ 1, indicating a balance of power between the exhibit strong agreement in all simulated cases, validating the
in-phase and quadrature components of the received signal. proposed equation (27) in this work. As expected, for both
The µ value, which is related to the number of multipath scenarios and all frequencies, increasing the modulation order
clusters, tends to be higher for Scenario 2 (LoS) compared to increases Pe . It is also noticeable that performance becomes
Scenario 3 (nLoS), despite both having a Tx-Rx separation of quite critical across all evaluated cases, since achieving Pe <
7.78 m. Additionally, in Scenario 5, which also features LoS 10−6 is not feasible without an NEb0 > 33 dB for M = 32 and
conditions but with a smaller Tx-Rx separation (4.91 m), the
N0 > 43 dB for M = 512. Such high Eb /N0 values are
Eb
average µ values further increase compared to the previous rarely attainable in real-world situations, highlighting the
two scenarios. This suggests that while the η-µ distribution importance of employing robust channel coding to mitigate
does not explicitly model the LoS path, it effectively captures this issue and guarantee quality of service for the final user.
the LoS effect by increasing the µ factor. Comparing the results obtained for LoS and nLoS condi-
Interestingly, in Scenario 2 at 28 GHz, despite being a tions, it is evident that, for all values of M , the LoS scenarios
LoS condition, the µ value of 1.08 was lower than for consistently provide significantly better performance than the
the other two frequencies evaluated, and the η value of nLoS scenarios. This is expected, as under LoS conditions,
1.39 indicates a power imbalance between the in-phase and signal variations are less pronounced, maintaining a more
quadrature components in the environment. This suggests stable average received signal level. An exception was
a more challenging propagation condition compared to the observed at 28 GHz, where the performance in Scenario 2
26 GHz and 39 GHz frequencies in the same scenario. (LoS) was similar to that in Scenario 3 (nLoS). This similarity
It is conjectured that in this specific case (Scenario 2, is attributed to the channel modeling in Scenario 2 at 28 GHz,
28 GHz), despite the presence of a direct line-of-sight path, which captures the effect of a power imbalance between the
multipath components caused by reflections, diffractions, in-phase and quadrature components of the signal (η = 1.39)
and scattering from objects in the measurement environment
likely created secondary paths with intensities comparable to 1 For the sake of simplicity, the random samples corresponding to the
the main path but with different phases, leading to this power channel gain are generated in MATLAB using the Inverse Transform
imbalance. sampling method.

31028 VOLUME 13, 2025

W. E. Vieira et al.: Performance Evaluation of Cross M-QAM Modulation

FIGURE 2. Scenarios 2 (LoS) and 3 (nLoS) both with Tx-Rx separation of FIGURE 4. Scenarios 2 (LoS) and 3 (nLoS) both with Tx-Rx separation of
7.78m and 26 GHz. 7.78m and 39 GHz.

the same modulation scheme in the same scenario across the

three plots (26, 28, and 39GHz). For example, considering
the curves for 128-QAM modulation in Scenario 2 (with
the theoretical curve represented by the solid black line),
it can be seen that to achieve a Pe < 10−6 , an NEb0 greater
than ≈ 39 dB is required at 26 GHz, while at 28 GHz, NEb0
must be greater than ≈ 41 dB, and at 39 GHz, NEb0 must be
greater than ≈ 38 dB, indicating that, in this specific case, the
best performance is obtained at the highest frequency. Thus,
although in practice the increase in frequency leads to higher
attenuation, it does not have a direct impact on the severity of
fading.
To compare the Pe results between two LoS scenarios
with different Tx-Rx separation distances, simulations were
conducted for Scenarios 2 (Tx-Rx separation of 7.78 m) and 5
(Tx-Rx separation of 4.91 m) at a frequency of 26 GHz.
The estimated Pe for M = {32, 128, 512}, along with
the theoretical expressions for Scenario 2 (solid lines) and
Scenario 5 (dashed lines), are presented in Fig. 5.
FIGURE 3. Scenarios 2 (LoS) and 3 (nLoS) both with Tx-Rx separation of
Analyzing Fig. 5, it can be concluded that, for any value
7.78m and 28 GHz. of M , operation in Scenario 5, with a shorter LoS distance,
results in a lower Pe compared to Scenario 2. This is because
the closer the transmitter is to the receiver without obstacles
and a low µ value of 1.08, rendering this scenario as critical between them, the stronger the main path’s power is relative
as the nLoS condition. This was previously discussed in to other secondary paths in the environment. Consequently,
the analysis of Table 1. Another important and non-obvious this reduces signal variation at the receiver and improves
conclusion that can be drawn from a careful analysis of the performance. Although not shown here, the authors also
results in Figs. 2, 3 and 4 is that increasing the frequency compared Scenarios 2 and 5 at 28 GHz and 39 GHz through
does not necessarily lead to an increase in the probability simulations, and the overall conclusions aligned with those
of error. This can be observed by evaluating the curves for obtained for the presented case of 26 GHz.
VOLUME 13, 2025 31029
 W. E. Vieira et al.: Performance Evaluation of Cross M-QAM Modulation

[5] R. S. Hoyt, ‘‘Probability functions for the modulus and angle of the normal
complex variate,’’ Bell Syst. Tech. J., vol. 26, no. 2, pp. 318–359, Apr. 1947.
[6] S. O. Rice, ‘‘Statistical properties of random noise,’’ Bell Syst. Tech. J.,
vol. 24, no. 1, pp. 146–156, 1945.
[7] M. Nakagami, ‘‘The m-distribution-A general formula of intensity
distribution of rapid fading,’’ Stat. Methods Radio Wave Propag., vol. 26,
no. 2, pp. 3–36, Apr. 1960.
[8] W. Weibull, ‘‘A statistical distribution function of wide applicability,’’
J. Appl. Mech., vol. 18, no. 3, pp. 293–297, Sep. 1951.
[9] M. D. Yacoub, ‘‘The α-µ distribution: A physical fading model for the
Stacy distribution,’’ IEEE Trans. Veh. Technol., vol. 56, no. 1, pp. 27–34,
Jan. 2007.
[10] M. D. Yacoub, ‘‘The κ-µ distribution and the η-µ distribution,’’ IEEE
Antennas Propag. Mag., vol. 49, no. 1, pp. 68–81, Mar. 2007.
[11] Q. Wu, D. W. Matolak, and I. Sen, ‘‘5-GHz-band vehicle-to-vehicle
channels: Models for multiple values of channel bandwidth,’’ IEEE Trans.
Veh. Technol., vol. 59, no. 5, pp. 2620–2625, Jun. 2010.
[12] S. L. Cotton, ‘‘Second-order statistics of κ-µ shadowed fading channels,’’
IEEE Trans. Veh. Technol., vol. 65, no. 10, pp. 8715–8720, Oct. 2016.
[13] N. Bhargav, S. L. Cotton, and D. B. Smith, ‘‘An experimental-based
analysis of inter-BAN co-channel interference using the κ–µ fading
model,’’ IEEE Trans. Antennas Propag., vol. 65, no. 2, pp. 983–988,
Feb. 2017.
[14] F. Cañete, J. López-Fernández, C. García-Corrales, A. Sánchez, E. Robles,
F. Rodrigo, and J. Paris, ‘‘Measurement and modeling of narrowband
channels for ultrasonic underwater communications,’’ Sensors, vol. 16,
no. 2, p. 256, Feb. 2016. [Online]. Available: [Link]
8220/16/2/256
[15] T. R. R. Marins, A. A. D. Anjos, C. R. N. D. Silva, V. M. R. Peñarrocha,
FIGURE 5. Scenarios 2 (LoS with Tx-Rx separation 7.78m) and Scenario 5
L. Rubio, J. Reig, R. A. A. De Souza, and M. D. Yacoub, ‘‘Fading
(LoS with Tx-Rx separation 4.91m), at 26 GHz.
evaluation in standardized 5G millimeter-wave band,’’ IEEE Access, vol. 9,
pp. 67268–67280, 2021.
[16] P. K. Chong, S.-E. Yoo, S. H. Kim, and D. Kim, ‘‘Wind-blown foliage
VI. CONCLUSION and human-induced fading in ground-surface narrowband communications
This paper presented, for the first time in the literature, at 400 MHz,’’ IEEE Trans. Veh. Technol., vol. 60, no. 4, pp. 1326–1336,
May 2011.
closed-form expressions for the SEP of cross M -QAM [17] T. R. R. Marins, A. A. dos Anjos, V. M. R. Peñarrocha, L. Rubio, J. Reig,
modulation operating in η-µ fading channels. Through a R. A. A. de Souza, and M. D. Yacoub, ‘‘Fading evaluation in the mm-wave
MATLAB simulation, the system performance was evaluated band,’’ IEEE Trans. Commun., vol. 67, no. 12, pp. 8725–8738, Dec. 2019.
[18] A. A. Dos Anjos, T. R. R. Marins, C. R. N. Da Silva, V. M. Rodrigo
in standardized mm-wave RF2 bands for 5G systems, Peñarrocha, L. Rubio, J. Reig, R. A. A. De Souza, and M. D. Yacoub,
specifically at frequencies of 26, 28, and 39 GHz, under ‘‘Higher order statistics in a mmWave propagation environment,’’ IEEE
LoS and nLoS conditions, as well as different Tx-Rx Access, vol. 7, pp. 103876–103892, 2019.
[19] J. Reig, M.-T. Martínez-Inglés, L. Rubio, V.-M. Rodrigo-Peñarrocha, and
distances and modulation order, where it was possible to J.-M. Molina-García-Pardo, ‘‘Fading evaluation in the 60 GHz band in line-
validate the proposed theoretical expression. Additionally, of-sight conditions,’’ Int. J. Antennas Propag., vol. 2014, no. 1, pp. 1–12,
a series of conclusions were drawn regarding the system’s 2014.
[20] J. Blumenstein, A. Prokes, A. Chandra, T. Mikulasek, R. Marsalek,
performance, where the difficulty of achieving low error rates T. Zemen, and C. Mecklenbrauker, ‘‘In-vehicle channel measurement,
without high signal-to-noise ratio values was demonstrated. characterization, and spatial consistency comparison of 30-11 GHz and
An important and non-obvious conclusion was that increasing 55–65 GHz frequency bands,’’ IEEE Trans. Veh. Technol., vol. 66, no. 5,
pp. 3526–3537, May 2017.
the frequency does not necessarily lead to performance [21] A. A. dos Anjos, G. K. Tenkorang, F. C. Martins, E. J. Leonardo, and
degradation. The results presented contribute to a better M. D. Yacoub, ‘‘Suitability of α-µ for composite fading modeling,’’ IEEE
understanding of the challenges future systems will face Trans. Antennas Propag., vol. 72, pp. 3670–3679, 2024.
[22] K. B. Carbonaro and G. Arantes Carrijo, ‘‘Performance analysis of M-
in ensuring reliable communication within these mm-wave QAM/OFDM systems for PLC under Gaussiano and impulsive noise,’’
bands. IEEE Latin Amer. Trans., vol. 14, no. 1, pp. 109–114, Jan. 2016.
[23] F. Xiong, Digital Modulation Techniques, 2nd ed., Boston, MA, USA:
Artech House, 2006.
REFERENCES [24] D. Da Costa and M. Yacoub, ‘‘Moment generating functions of generalized
[1] P. Newman, ‘‘How Internet of Things technology growth is fading distributions and applications,’’ IEEE Commun. Lett., vol. 12, no. 2,
reaching mainstream companies and consumers,’’ Bus. Insider, pp. 112–114, Feb. 2008.
New York, NY, USA, Tech. Rep., 2020. [Online]. Available: [25] M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions
[Link] With Formulas, Graphs and Mathematical Tables. New York, NY, USA:
[2] A. Ghosh, A. Maeder, M. Baker, and D. Chandramouli, ‘‘5G evolution: A Dover, 1964.
view on 5G cellular technology beyond 3GPP release 15,’’ IEEE Access, [26] M. Chiani, D. Dardari, and M. K. Simon, ‘‘New exponential
vol. 7, pp. 127639–127651, 2019. bounds and approximations for the computation of error
[3] S. Haykin, Communication Systems, 4th ed., New York, NY, USA: Wiley, probability in fading channels,’’ IEEE Trans. Wireless Commun.,
2000. vol. 24, no. 5, pp. 840–845, May 2003. [Online]. Available:
[4] L. Rayleigh, ‘‘On the resultant of a large number of vibrations of the same [Link]
pitch and of arbitrary phase,’’ Philos. Mag., vol. 10, no. 60, pp. 73–78, [27] I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and Products,
1880. 8th ed., Amsterdam, The Netherlands: Elsevier, 2014.

31030 VOLUME 13, 2025

W. E. Vieira et al.: Performance Evaluation of Cross M-QAM Modulation

WILIAN EURÍPEDES VIEIRA received the GILBERTO ARANTES CARRIJO received the
degree (Full Degree) and master’s degrees in degree in electrical engineering from the Uni-
mathematics from the Federal University of Uber- versity of Brasília, in 1972, the master’s degree
landia, in 1996 and 2009, respectively, where he in electronic engineering and telecommunication
is currently pursuing the Ph.D. degree in electrical and the Ph.D. degree in electronic and computer
engineering. He is a Professor with the Federal engineering from the Technological Institute of
University of Uberlandia. He has experience in the Aeronautics, in 1976 and 1983, respectively, and
area of mathematics, working mainly in the area of the Ph.D. degree from The Western Australia
differential geometry. University, in 1990. He is currently a Full Pro-
fessor with the Federal University of Uberlandia.
He has experience in electrical engineering, with an emphasis on the
digital processing of voice and image signals, working mainly on the
following topics: UHF propagation, image compression, spectral spread,
facial recognition, and codes.

PEDRO LUIZ LIMA BERTARINI received the

degree and the Ph.D. degree in electrical engi-
neering from the University of São Paulo (USP),
EDSON AGUSTINI received the degree in math-
in 2006 and 2012, respectively. He is currently
ematics from São Paulo State University-Unesp,
a Professor and a Researcher with the Faculty
Campus São José do Rio Preto, São Paulo, in 1995,
of Electrical Engineering, Federal University of
and the master’s and Ph.D. degrees in mathe-
Uberlandia (UFU), Campus Patos de Minas.
matics from the State University of Campinas-
He coordinates the UFU Laboratory of Urban and
Unicamp, in 1998 and 2002, respectively. He is
Rural Technologies (LATUR), Campus Patos de
currently a Professor with the Federal University
Minas, where he leads multidisciplinary projects
of Uberlandia-UFU, Minas Gerais. His research
that involve: 1) developing devices based on the Internet of Things (IoT)
interests include geometry/topology and informa-
technology; 2) image analysis through computer vision techniques; and
tion and coding theory.
3) data analysis using artificial intelligence (AI), machine learning, and
multivariate statistics methods.

ANDRÉ ANTÔNIO DOS ANJOS received the

Ph.D. degree from the State University of Camp-
inas, Brazil, in 2021. He was an Engineer with the
KARINE BARBOSA CARBONARO received the Inatel Competence Center, Brazil, for more than
degree in electrical engineering from the National 11 years, developing projects in telecommunica-
Institute of Telecommunications-INATEL, Brazil, tions and digital signal processing. He is currently
in 2001, and the master’s and Ph.D. degrees in a Professor with the Electronic and Telecommuni-
electrical engineering from the Federal University cations Engineering Program, Federal University
of Uberlandia, in 2004 and 2015, respectively. She of Uberlandia, Campus Patos de Minas, Brazil.
is currently a Professor with the Federal University His research interests include wireless communi-
of Uberlandia. She has experience in the area of cations, fading channel modeling and simulation, nonlinear systems, and
telecommunications. spectrum sensing.

VOLUME 13, 2025 31031