Impact of inter-channel interference on shallow underwater acoustic ofdm systems

h should be considered in UWA-
 Figure 2 shows the SINR results of the UWA-OFDM OFDM system design is the number of subcarriers . For
system with the number of subcarriers = 1024. It is a given bandwidth, the subcarrier spacing is narrower with
noted that the SINR is calculated for each subcarrier and a larger . Consequently, the ICI effect on the OFDM
the results in Fig. 2 is the average of all subcarriers. We system is more severe, and then the SINR decreases. As
see the strong Doppler effect on the SINR when the signal the results show in Fig. 3, the SINR is lower in the case
bandwidth is small. With a given number of subcarriers, of larger number of subcarriers . Using these results,
the smaller signal bandwidth is, the subcarrier spacing the appropriate values of bandwidth and the number of
Δ = / is narrower that causes more serious ICI effect sub-carrier can be determined to achieve a required
and decreasing the SINR. On the contrary, when increasing SINR of the UWA-OFDM system. For even a very small
the signal bandwidth, the larger carrier spacing mitigates value of Doppler shift = 1 Hz, the maximum number
the ICI effect. If bandwidth is large enough, the ICI of subcarriers of = 1024 and the minimum bandwidth
effect can be neglected and the SINR results approach to of 10 kHz should be selected to avoid the ICI effect. For
the SNR=20 dB. = 2048, the minimum bandwidth is required to be
 greater than 20 kHz. However, on one hand, a small number
 From the SINR results in Fig. 2, it is observed that with
 of subcarriers results in mitigating the Doppler effect; on
a bandwidth greater than 10 kHz, the ICI effect for the
 the other hand, it makes the efficiency of the spectrum
Doppler frequencies of = 1 Hz and 2 Hz is negligible.
 and the system capacity decrease. The next section will
For the case of = 4 Hz, a bandwidth greater than 20 kHz
 evaluate the capacity to determine the appropriate number
is required to significantly reduce the ICI effect. Therefore,
 of subcarriers for the UWA-OFDM system.
depending on the hardware capabilities of the system, the
impact of ICI on system performance can be significantly
reduced if a wide bandwidth is chosen.
 46
 Vol. 2020, No. 01, September
 200 6.5
 N =512
 c
 6 N =1024
 150 c
 N =2048
 c
 5.5
 100
 5
 Capacity (kbps) Capacity
 Capacity (bit/s/Hz) Capacity
 N =512
 50 c 4.5
 N =1024
 c
 N =2048
 c
 4
 0 0 200 400 600 800 1000
 0 5 10 15 20 25 30
 T (ms)
 Bandwidth (kHz) S
Figure 4. System capacity (Kbps) versus bandwidth for different Figure 5. Spectral efficiency / (b/s/Hz) versus symbol length 푆 for
numbers of subcarriers ( = 1Hz). different numbers of subcarriers.
3. Capacity Results 6.4
 f =1 Hz
 6.2 d
 f = 2 Hz
 The system capacity of the UWA-OFDM system versus d
 6 f = 4 Hz
the signal bandwidth for different numbers of subcarriers d
 5.8
(with = 1Hz) is shown in Fig. 4. For the bandwidth range
of < 10 kHz, the system capacity is almost the same value 5.6
for different numbers of subcarriers of 512, 1024, and 2048. 5.4
In this case, the lowest number of subcarriers = 512 5.2
 Capacity (b/s/Hz) Capacity
should be chosen to reduce complexity of receiver. The
 5
reason can be explained by using Eq. (28). For the narrow
 4.8
bandwidth (i.e. less than 10 kHz for the considered case),
 4.6
the increase in the number of sub-carriers leads to the 100 200 300 400 500 600 700 800 900 1000
decrease in the SINR as shown in Fig. 3. However, that T (ms)
 S
makes the OFDM symbol 푆 = / larger. As a result,
the bandwidth efficiency 훽 = 푆/( 푆 + ), in which Figure 6. Spectral efficiency / (b/s/Hz) versus symbol length 푆 for
 different Doppler frequencies.
 = 휏 is fixed, will be increased. Therefore, if 
increases, the bandwidth efficiency will increases, while the
SINR decreases. From this argument along with Eq. (28),
 different numbers of subcarriers for the case of = 1 Hz.
we also observed that the capacity remains unchanged for 
 For the small symbol duration range of < 204.8ms, a
the larger numbers of subcarriers as shown in Fig. 4 with 푆
 larger results in a higher spectral efficiency / . This
the bandwidth range < 10 kHz. In other words, with the 푆
 is because a larger makes a higher bandwidth efficiency
limited bandwidth of UWA channels, it may be impossible 푆
 coefficient 훽. Consequently, the spectral efficiency / 
to increase the capacity by simply increasing the number
 will be increased as shown in Eq. (29). On the contrary, an
of sub-carriers.
 increase in 푆 will get a decrease in the spectral efficiency
 In a similar way, for the larger bandwidth of range from
 / for the larger symbol range 푆 > 204.8 ms. The reason
10 kHz to 15 kHz, one should choose = 1024 to ensure
 is that the larger 푆 (i.e. the smaller subcarrier spacing Δ )
that the system capacity is not significantly reduced, but to makes the ICI effect more serious on the UWA channel.
limit the ICI effect. For a bandwidth of more than 20 kHz,
 Hence, 푆 > 204.8 ms also results in the decrease of
 = 2048 is suitable. both SINR and the spectral efficiency / as shown in
 Besides the system capacity (Kbps), the spectrum Eq. (29). As observed in Fig. 5, the spectral efficiency all
efficiency / (b/s/Hz) is also an important factor in the achieves the maximal value / max = 6.265 (b/s/Hz) at
OFDM system design due to the limited bandwidth of 푆 = 204.8 ms. Using this result, we can determine the
UWA channels. Figure 5 shows the results of the spectrum optimal bandwidth for each different number of subcarriers
efficiency / (b/s/Hz) versus the symbol duration 푆 for .
 47
Research and Development on Information and Communication Technology
 TABLE I for OFDM-Based Cognitive Radio Systems,” in Pro-
 THE UWA-OFDM SYSTEM PARAMETERS FOR DIFFERENT DOPPLER ceedings of the 7th International Conference on Ubiq-
 FREQUENCIES.
 uitous Information Management and Communication,
 ( / ) (kHz)
 max (ms) ICUIMC ’13, (New York, NY, USA), Association for
 (b/s/Hz) 푆 N=512 N= 1024 N =2048
 1 Hz 6.265 204.8 2.5 5.0 10.0 Computing Machinery, 2013.
 2 Hz 6.059 128.0 4.1 8.2 16.4 [4] T. Ebihara and K. Mizutani, “Experimental Study of
 4 Hz 5.757 78.8 6.5 13.0 26.0 Doppler Effect for Underwater Acoustic Communi-
 cation Using Orthogonal Signal Division Multiplex-
 ing,” Japanese Journal of Applied Physics, vol. 51,
 Moreover, depending on the Doppler shift , an appro-
 p. 07GG04, jul 2012.
priate should be chosen to achieve the maximum / .
 푆 [5] J. Tao, “DFT-Precoded MIMO OFDM Underwater
Figure 6 depicts the / results versus for different
 푆 Acoustic Communications,” IEEE Journal of Oceanic
Doppler shifts for the case of = 1024. When the
 Engineering, vol. 43, no. 3, pp. 805–819, 2018.
Doppler frequency increases, the optimal value of to
 푆 [6] L. Zhang, T. Kang, H. C. Song, W. S. Hodgkiss, and
achieve maximum / is decreased. The optimal and
 푆 X. Xu, “MIMO-OFDM Acoustic Communication in
 values for different Doppler frequencies are shown in
 Shallow Water,” in 2013 OCEANS - San Diego, pp. 1–
Table I. Based on these results, one can determined the
 4, 2013.
parameters , , , for UWA-OFDM systems to not only
 푆 [7] H. Tran Minh, S. Rie, S. Taisuki, and T. Wada, “A
achieve maximum spectral efficiency and channel capacity
 Transceiver Architecture for Ultrasonic OFDM with
but also limit the ICI effect and minimize complexity at the
 Adaptive Doppler Compensation,” in OCEANS 2015
receiver under different transmission conditions.
 - MTS/IEEE Washington, pp. 1–6, 2015.
 [8] S. Yoshizawa, H. Tanimoto, and T. Saito, “Experimen-
V. CONCLUSIONS tal results of OFDM rake reception for shallow wa-
 ter acoustic communication,” in 2016 Techno-Ocean
 This paper has investigated the capacity of the UWA-
 (Techno-Ocean), pp. 185–188, 2016.
OFDM system under the impact of ICI effect. Using the
 [9] H. Do Viet, T. Chien, and V. Nguyen, “Propos-
time variant geometry-based channel model for the shallow
 als of Multipath Time-Variant Channel and Additive
water environment, the SINR of each sub-carrier has been
 Coloured Noise Modelling for Underwater Acous-
derived. The channel capacity is computed from the SINR,
 tic OFDM-Based Systems,” International Journal of
which takes the Doppler effect into account. The numerical
 Wireless and Mobile Computing, vol. 11, pp. 329–338,
results of SINR, channel capacity, and spectral efficiency
 Jan. 2016.
give us guidance in UWA-OFDM system design, specifi-
 [10] G. Qiao, Z. Babar, L. Ma, L. Wan, X. Qing, X. Li, and
cally in choosing the transmission parameters including the
 M. Bilal, “Shallow water acoustic channel modeling
number of subcarriers, the symbol length, and the signal
 and mimo-ofdm simulations,” in 2018 15th Interna-
bandwidth for the different Doppler shifts.
 tional Bhurban Conference on Applied Sciences and
 Technology (IBCAST), pp. 709–715, 2018.
ACKNOWLEDGEMENT [11] D. Nguyen, H. Nguyen, and H. Ho, “Methods to Esti-
 This study was funded by the Vietnam National Foun- mate the Channel Delay Profile and Doppler Spectrum
dation for Science and Technology Development (NAFOS- of Shallow Underwater Acoustic Channels,” Archives
TED) under the project number 102.04-2018.12. of Acoustics, vol. 44, pp. 375–383, Jan. 2019.
 [12] X. Wang, J. Wang, L. He, and J. Song, “Doubly Se-
 lective Underwater Acoustic Channel Estimation with
REFERENCES Basis Expansion Model,” in 2017 IEEE International
 [1] G. Marani, S. K. Choi, and J. Yuh, “Underwater Conference on Communications (ICC), pp. 1–6, 2017.
 Autonomous Manipulation for Intervention Missions [13] M. Naderi, D. V. Ha, V. D. Nguyen, and M. Patzold,
 AUVs,” Ocean Engineering, vol. 36, no. 1, pp. 15–23, “Modelling The Doppler Power Spectrum of Non-
 2009. Stationary Underwater Acoustic Channels Based on
 [2] D. W. Kim, “Tracking of REMUS Autonomous Un- Doppler Measurements,” in OCEANS 2017 - Ab-
 derwater Vehicles with Actuator Saturations,” Auto- erdeen, pp. 1–6, 2017.
 matica, vol. 58, no. C, pp. 15–21, 2015. [14] H. Nouri, M. Uysal, E. Panayirci, and H. Senol, “In-
 [3] N. T. Hoa, N. T. Hieu, N. Van Duc, G. Gelle, and formation Theoretical Performance Limits of Single-
 H. Choo, “Second Order Suboptimal Power Allocation carrier Underwater Acoustic Systems,” IET Commu-
 48
 Vol. 2020, No. 01, September
 nications, vol. 8, no. 15, pp. 2599–2610, 2014. Do Viet Ha received her B.S, M.Sc., and
[15] P. Bouvet and A. Loussert, “Capacity Analysis of Ph.D. degrees in Electronics and Telecom-
 Underwater Acoustic MIMO Communications,” in munications Engineering from Hanoi Uni-
 OCEANS’10 IEEE SYDNEY, pp. 1–8, 2010. versity of Science and Technology (HUST),
[16] D. E. Lucani, M. Stojanovic, and M. Medard, “On Hanoi, Vietnam, in 2001, 2007, and 2017,
 respectively. She is currently working with
 the Relationship between Transmission Power and
 the Department of Electronics Engineering,
 Capacity of an Underwater Acoustic Communication University of Transport and Communica-
 Channel,” pp. 1–6, 2008. tions, Hanoi, Viet Nam, as a lecturer. Her main areas of re-
[17] Y. M. Aval, S. K. Wilson, and M. Stojanovic, “On the search interest are mobile channel modeling, especially underwa-
 Achievable Rate of A Class Of Acoustic Channels and ter acoustic channels, underwater acoustic OFDM systems, and
 Practical Power Allocation Strategies for OFDM Sys- mobile-to-mobile communications
 tems,” IEEE Journal of Oceanic Engineering, vol. 40,
 no. 4, pp. 785–795, 2015.
[18] A. Radosevic, J. G. Proakis, and M. Stojanovic, “Sta- Nguyen Tien Hoa graduated with a Dipl.-
 tistical Characterization and Capacity of Shallow Wa- Ing. in Electronics and Communication En-
 ter Acoustic Channels,” in OCEANS 2009-EUROPE, gineering from Hanover University. He has
 worked in the R&D department of image
 pp. 1–8, 2009.
 processing and in the development of SDR-
[19] Ye Li and L. J. Cimini, “Bounds on The Inter- based drivers in Bosch, Germany. He de-
 channel Interference Of OFDM in Time-Varying Im- voted three years ot experiment with MI-
 pairments,” IEEE Transactions on Communications, MOon’s R&D team to develop embedded
 vol. 49, no. 3, pp. 401–404, 2001. signal processing and radio modules for LTE-A/4G. He worked
[20] A. S. I. R. Capoglu, Y. Li, “Effect of Doppler spread as a senior expert at Viettel IC Design Center (VIC) and VinS-
 in OFDM-based UWB systems,” IEEE Transactions mart for development of advanced solutions for aggregating and
 on Wireless Communications, vol. 4, no. 5, pp. 2559– splitting/steering traffic at the PDCP layer to provide robust and
 2567, 2005. QoS/QoE guaranteeing integration between heterogeneous link
[21] P. J. Bouvet and Y. Auffret, “On the Achievable types, as well as Hybrid Beamforming for Millimetre-Wave 5G
 systems. Currently, he is a lecturer at the School of Electronics
 Rate of Multiple-Input–Multiple-Output Underwater
 and Telecommunications, Hanoi University of Science and Tech-
 Acoustic Communications,” IEEE Journal of Oceanic nology. His research interests are resource allocation in B5G, and
 Engineering, vol. 45, no. 3, pp. 1126–1137, 2020. vehicular communication systems.
[22] D. V. Ha, V. D. Nguyen, and Q. K. Nguyen, “Mod-
 eling of Doppler power spectrum for underwater
 acoustic channels,” Journal of Communications and Nguyen Van Duc received his B.S and
 Networks, vol. 19, no. 3, pp. 270–281, 2017. M.Sc degrees in Electronics and Telecom-
[23] A. G. Zajic, “Statistical Modeling of MIMO Mobile- munications engineering from Hanoi Uni-
 to-Mobile Underwater Channels,” IEEE Transactions versity of Science and Technology (HUST),
 on Vehicular Technology, vol. 60, no. 4, pp. 1337– Hanoi, Vietnam, in 1995 and 1997, re-
 1351, 2011. spectively. In 2003, he received Ph.D. in
 Electronics and Telecommunications Engi-
[24] Y. Widiarti, Suwadi, Wirawan, and T. Suryani, “A
 neering from Leibniz University Hannover,
 Geometry-Based Underwater Acoustic Channel Model Germany. Since 2006, he has been with the Department of
 for Time Reversal Acoustic Communication,” in 2018 Communications Engineering, HUST, where he is currently an
 International Seminar on Intelligent Technology and Associate Professor. His current research interests include under-
 Its Applications (ISITIA), pp. 345–350, 2018. water communications, intelligent transport systems, and cognitive
[25] A. Withamana and H. Kondo, “Wideband Time- radio networks.
 Varying Underwater Acoustic Channel Emulator,” in
 2017 IEEE Underwater Technology (UT), pp. 1–5,
 2017.
[26] M. Naderi, M. Patzold,¨ and A. G. Zajic, “A Geometry-
 Based Channel Model for Shallow Underwater Acous-
 tic Channels Under Rough Surface And Bottom
 Scattering Conditions,” in 2014 IEEE Fifth Interna-
 tional Conference on Communications and Electronics
 (ICCE), pp. 112–117, 2014.
 49