Luận án Tiến sĩ Mô hình đặc tính kênh truyền cho thông tin thủy âm vùng nước nông (“Channel Modeling for Shallow Underwater Acoustic Communications”)

MINISTRY OF EDUCATION AND TRAINING HANOI UNIVERSITY OF SCIENCE AND TECHNOLOGY DO VIET HA MƠ HÌNH ĐẶC TÍNH KÊNH TRUYỀN CHO THÔNG TIN THỦY ÂM VÙNG NƯỚC NÔNG CHANNEL MODELING FOR SHALLOW UNDERWATER ACOUSTIC COMMUNICATIONS DOCTORAL THESIS OF TELECOMMUNICATIONS ENGINEERING HA NOI - 2017 MINISTRY OF EDUCATION AND TRAINING HANOI UNIVERSITY OF SCIENCE AND TECHNOLOGY DO VIET HA MƠ HÌNH ĐẶC TÍNH KÊNH TRUYỀN CHO THƠNG TIN THỦY ÂM VÙNG NƯỚC NÔNG CHANNEL MODELING FOR SHALLOW UNDERWATER ACOUSTIC COMMUNICATIONS Specialization: Telecommunications Engineering Code No: 62520208 DOCTORAL THESIS OF TELECOMMUNICATIONS ENGINEERING SUPERVISORS: Assoc.Prof Van Duc Nguyen Dr Van Tien Pham Hanoi - 2017 DECLARATION OF AUTHORSHIP I hereby declare that this dissertation titled, "Channel Modeling for Shallow Underwater Acoustic Communications”, and the work presented in it are entirely my own original work under the guidance of my supervisors I confirm that: • This work was done wholly or mainly while in candidature for a PhD research degree at Hanoi University of Science and Technology • Where any part of this dissertation has previously been submitted for a degree of any other qualification at Hanoi University of Science and Technology or any other institution, this has been clearly stated • Where I have consult the published work or others, this is always given With the exception of such quotations, this dissertation is entirely my own work • I have acknowledged all main source of help • Where the thesis is based on work done by myself jointly with others, I have made exactly what was done by others and what I have contributed myself SUPERVISORS Hanoi, August 27, 2017 PhD STUDENT Assoc.Prof Van Duc Nguyen Dr Van Tien Pham Do Viet Ha ACKNOWLEDGEMENTS First and foremost, I would like to thank my advisor Associate Prof Dr Nguyen Van Duc for for providing an excellent atmosphere for doing research, for his valuable comments, constant support and motivation His guidance helped me in all the time of research and writing of this dissertation I could not have imagined having a better advisor and mentor for my PhD I would also like to thank Dr Pham Van Tien for their advice and feedback, also for many educational and inspiring discussions My sincere gratitude goes to the members in the Wireless Communication Lab, School of Electronics and Telecommunications, Hanoi University of Science and Technology, Hanoi, Vietnam Without their support and friendship it would have been difficult to complete my PhD studies I am also thankful to Dr Nguyen Tien Hoa for his invaluable instructions in presenting my dissertation I would also like to express my deepest gratitude to my parents, my husband, my son, and my daughter They were always supporting me and encouraging me with their best wishes, they were standing by me throughout my life Hanoi, August 27, 2017 PhD STUDENT Do Viet Ha Contents TABLE OF CONTENTS ABBREVIATIONS iv LIST OF FIGURES vi LIST OF TABLES ix INTRODUCTION Chapter DESIGN OF SHALLOW UWA CHANNEL SIMULATORS 15 1.1 Introduction 16 1.2 Overview of Simulation Models for UWA Channels 1.2.1 Rayleigh and Rice channels 1.2.2 Deterministic SOS Channel Models 1.2.3 Deterministic SOC Channel Models 19 19 20 21 1.3 The Geometry-based UWA Channel Simulator 21 1.3.1 Developing the Reference Model from the Geometrical Channel Model 22 1.3.2 The Simulation Model 27 1.3.3 The Estimated Parameters of the Simulation Model 27 1.3.4 Simulation Results 28 1.4 The Measurement-based UWA Channel Simulator 1.4.1 The Reference Model from the Measurement Data 1.4.2 The Simulation Model 1.4.3 Estimated Channel Parameters of the Simulation Model 1.4.4 Comparison of the Two Channel Simulators 28 29 32 33 34 1.5 The Proposed Approach for the Static UWA Channel 1.5.1 Descriptions 1.5.2 Results and Discussions 35 36 38 i ii 1.6 The Proposed Approach for the Case of Doppler Effects 1.6.1 The Measurement Data 1.6.2 The Conventional Measurement-based Simulators 1.6.3 The Proposed Channel Simulator 39 40 41 45 1.7 Conclusions 50 Chapter MODELING OF DOPPLER POWER SPECTRUM FOR SHALLOW UWA CHANNELS 53 2.1 Introduction 53 2.2 The Proposed Doppler Spectrum Model 2.2.1 The Doppler Effects in Shallow UWA Channels 2.2.2 The Proposed Doppler Model for UWA Channels 56 56 59 2.3 The Description of Doppler Spectrum Measurements 2.3.1 Experimental Setup 2.3.2 Measurement Scenarios 2.3.3 Reference Model from the Measurement Data 63 63 64 66 2.4 Parameter Optimizations of the Proposed Model 67 2.5 Measurement and Doppler Modeling Results 2.5.1 Scenario 2.5.2 Scenario 2.5.3 Scenario 68 69 71 75 2.6 Conclusions 77 Chapter UWA-OFDM SYSTEM PERFORMANCE ANALYSIS USING THE MEASUREMENT-BASED UWA CHANNEL MODEL 78 3.1 Introduction 79 3.2 ICI Analysis of UWA-OFDM Systems 3.2.1 SIR Calculation 3.2.2 Ambient Noise Power 3.2.3 SINR Calculation 81 82 83 84 3.3 Capacity Calculation 86 iii 3.4 Numerical Results 3.4.1 The SIR 3.4.2 The SINR 3.4.3 Channel Capacity 3.4.4 Transmit Power 87 88 89 90 92 3.5 Chapter Conclusions 96 CONCLUSIONS 99 APPENDIX 103 LIST OF PUBLICATIONS 105 ABBREVIATIONS ACF Autocorrelation Function AOA Angles of Arrival AOD Angles of Departure AWGN Additive White Gaussian Noise BPSK Binary Phase Shift Keying CIR Channel Impulse Response FCF Frequency Correlation Function ICI Inter-Channel Interference INLSA Iterative Nonlinear Least Square Approximation ISI Inter-Symbol Interference LNA Low Noise Amplifier LOS Line of Sight LPNM Lp-Norm Method MESS Method of Equally Spaced Scatterers MSE Mean Square Error OFDM Orthogonal Frequency Division Multiplexing PDF Probability Density Function PDP Power Delay Profile PN Pseudo-Noise PSD Power Spectra Density Rx Receiver SINR Signal to Interference plus Noise Ratio SIR Signal-to-Interference Ratio SNR Signal to Noise Ratio SOC Sum-of-Cisoids SOS Sum-of-Sinusoids TCF Time Correlation Function iv v T-FCF Time-Frequency Correlation Function TVCIR Time Variant Channel Impulse Response TVCTF Time-Variant Channel Transfer Function Tx Transmitter UWA Underwater Acoustic WLAN Wireless Local Area Network WSSUS Wide-Sense Stationary Uncorrelated Scattering List of Figures Multipath interference in UWA communication systems 1.1 The methodology behind the geometry-based channel modelling [17, 55] 17 1.2 The methodology behind the measurement-based channel modelling [31, 56] 18 1.3 The scheme of designing the geometry-based channel simulator [17, 55] 22 1.4 The geometrical model for shallow UWA channels with randomly distributed scatterers Si,n (•) on the surface (i = 1) and the bottom (i = 2) [55] 23 1.5 The comparison between the normalized FCF of the reference model and that obtained by the geometry-based simulator 29 1.6 1.7 Illustration of the measurement setup in Halong bay 30 ˆ t)|2 for the transmission distance of 150 m 31 The measured |h(τ, 1.8 The measured and normalized PDP ρ(τ ) obtained for the transmission distance of 150 m 1.9 32 The comparison of the normalized FCF obtained by the two simulators to that of the reference model 35 1.10 The flowchart of proposed approach to design the static UWA channel simulator 36 1.11 The comparison between the normalized FCF of the reference model and that obtained by the measurement-based, the geometry-based, and the proposed simulators 38 1.12 The normalized Doppler power spectrum 41 1.13 a) The reference T-FCF derived from the measurement results b) The T-FCF of the channel simulation model designed by the conventional simulator 43 1.14 The comparison between the normalized T-FCF of the reference model and that obtained by the conventional measurement-based simulator 44 1.15 The flowchart of the proposed approach for the case of moving Rx vi 46 70 Table 2.2: The optimal and derivative parameters of the proposed model Parameters A (dB) fm (Hz) w (Hz) C fSpike (Hz) BSpike (Hz) BD (Hz) Tc (s) Vn,R (m/s) sn (m) Meanings Ratio between the Gaussian peak and the Spike peak The Doppler shift caused by the surface motion The Gaussian Doppler bandwidth Factor to determine the Spike Doppler bandwidth The Doppler shift caused by the Rx movement The Spike Doppler bandwidth The overall Doppler bandwidth The coherence time The estimated Rx speed The estimated Rx movement distance proposed model can be used to describe accurately the Doppler spectrum of the UWA channel Similar to the definition of the BSpike , the overall Doppler bandwidth BD is specified by the range of Doppler shift frequencies over which the received Doppler spectrum S (f ) is 10 dB below its maximum value Moreover, the coherence time Tc of the UWA channel can be calculated by Tc = 1/BD for each case, and given in Table 2.3 These parameters BD and Tc determine the time-varying nature of the UWA channel as well as fading effects due to the Doppler spread From the obtained results of the cases and as given in Table 2.3, we can see that the overall Doppler bandwidth BD is slightly larger than the Spike bandwidth BSpike We underline that the Spike bandwidth displays the Doppler component caused by the relative movement between the Tx and Rx, while the overall Doppler bandwidth shows the total Doppler components from both the relative movement and the surface motion The corresponding adjusting factors A obtained for cases and are −26.2 dB and −18.8 dB, respectively These results show the Spikeshape Doppler spectrum in Eq 2.9 dominates to the Gaussian-shape spectrum As depicted in Fig 2.5, the cases and correspond to the distance from B to C, when the Rx increases its speed Thus, it is logical to see that the Doppler shifts caused by the Rx movement dominate those caused by the surface motion Observing the optimal parameters obtained for the cases and as given in Table 2.3, the adjusting factors A optimized for the last two cases are significantly larger than the first ones Compared to the Spike- 71 shape contribution, the Gaussian-shape spectrum in Eq 2.9 can not be neglected As a consequence, the overall Doppler bandwidth BD is almost twice as large as the Spike bandwidth BSpike In other words, the Doppler shifts caused by the surface motion is significant in comparison with that caused by the Rx movement These results agree with the estimated results of maximum Doppler frequency and movement speed of the Rx In these two later cases, the Rx slows down its speed from C to D until it stops, as illustrated in Fig 2.5 Using the optimal results of the maximum Doppler frequency fSpike , the Rx speed Vn,R and the Rx movement distance sn over each observation interval 4Tn are calculated by using Eq 2.14 and Eq 2.15, respectively As shown in Table 2.3, the maximum Doppler shift fSpike of the case is the largest, while the Rx reaches its maximal speed of km/h After that, it decreases together with a decrease in the Rx speed for two later cases In more detail, observing the estimated Rx speeds Vn,R for the cases and as shown in Table 2.3, it can be found that the Rx increases its speed over the observation time of these cases The theoretical calculation movement distance (s1 + s2 ) = 164.6 m over this observation time meets also the real distance BC of 165 m in the fact as depicted in Fig 2.5 For the last two cases, the estimated Rx speeds Vn,R are 2.1 km/h and 0.7 km/h, respectively That means the Rx reduces its speed in this observation time, and the estimated Rx movement distance of (s3 + s4 ) = 37.5 m, which approximates well the distance CD of 38 m shown in Fig 2.5 From these results, it can be seen that the estimated Rx speeds and the movement distances are in good agreement with the measurement scenario described in Sect 2.3.2 2.5.2 Scenario In this scenario, as depicted in Fig 2.6, the Rx moves towards the fixed Tx for the observation time Tobs = 345 s At beginning, the Rx increases its speed until reaching the supposed speed of 5.5 km/h in the distance BC of 230 m Then, the Rx slows down its speed in the distance CA of 155 m until it almost stops near the Tx boat 72 Table 2.3: The optimal parameters for Doppler spectrum modeling derived from the measurement data in scenario (as plotted in Fig 2.9.) Case (n) Gaussian component fm A[dB] [Hz] Spike component w [Hz] C fSpike [Hz] BSpike [Hz] Overall Doppler BD [Hz] Tc [s] Rx movement Vn,R [km/h] 4Tn [s] sn [m] 108.9 -26.2 1.1 7.2 200.0 -8.3 3.52 3.60 0.028 3.7 105 -18.8 -1.0 20.0 200.0 -8.9 3.78 4.08 0.245 4.0 50 55.7 -4.5 -4.5 2.8 100.0 -4.7 2.82 6.08 0.164 2.1 50 29.4 -8.5 -2.2 3.5 35.7 -1.4 1.45 4.37 0.229 0.7 Total 45 8.1 250 202.2 In the similar way, by observing the time-varying measured spectrum S˜ (f, t), we classify the Doppler spectrum into six typical cases Subsequently, the optimal parameters of proposed model {A, fm , w, C, fSpike } and the derivative parameters (i.e BSpike , BD , Tc , Vn,R , and sn ) for each case are computed and listed in Table 2.4 The proposed model is plotted and compared with the reference model as shown in Fig 2.10 using these optimal parameters {A, fm , w, C, fSpike } The good curve-fitting results between the reference model and the proposed model for all cases can be observed Table 2.4: The optimal parameters for Doppler spectrum modeling derived from the measurement data in scenario (as plotted in Fig 2.10) Gaussian component Case (n) Spike component A[dB] fm [Hz] w [Hz] C -28.9 -1.0 30.0 -12.9 -1.0 30.0 -10.6 -1.1 -8.3 0.2 -8.2 -3.0 Overall Doppler Rx movement fSpike [Hz] BSpike [Hz] BD [Hz] Tc [s] Vn,R [km/h] 200.0 7.1 3.01 3.09 0.324 3.2 80 71.0 99.9 12.1 7.26 9.44 0.106 5.4 105 158.8 43.9 99.4 11.6 6.96 13.80 0.072 5.2 35 50.6 27.0 348.1 10.4 3.34 30.61 0.033 4.7 50 65.0 -0.9 8.2 187.3 7.5 3.30 15.48 0.065 3.4 25 23.5 -1.2 5.5 7.8 2.7 5.79 14.28 0.070 1.2 50 16.8 345 385.7 Total 4Tn [s] sn [m] The obtained results of first two cases given in Table 2.4 show that the optimal maximum Doppler shift fSpike increases with an increase of the Rx speed Vn,R These results correspond to the distance BC described in the measurement scenario 2, when the Rx speeds up For the latter cases, the values of optimal maximum Doppler shifts fSpike reduce together with a reduction in the Rx speed over the distance of (s3 + s4 + s5 + s6 ) = 155.9 m These estimated Rx speeds and move- Measured spectrum Reference model Proposed Model −5 −10 −15 −20 −25 −30 −35 −40 −20 −15 −10 −5 10 15 20 Doppler shift [Hz] Case Measured spectrum Reference model Proposed Model −5 −10 −15 −20 −25 −30 −35 −40 −20 −15 −10 −5 Doppler shift [Hz] 10 15 20 Normalized Doppler spectrum [dB] Case Normalized Doppler spectrum [dB] Normalized Doppler spectrum [dB] Normalized Doppler spectrum [dB] 73 Case Measured spectrum Reference model Proposed Model −5 −10 −15 −20 −25 −30 −35 −40 −20 −15 −10 −5 10 15 20 Doppler shift [Hz] Case Measured spectrum Reference model Proposed Model −5 −10 −15 −20 −25 −30 −35 −40 −20 −15 −10 −5 10 15 20 Doppler shift [Hz] Figure 2.9: The reference model S˜n (f ) compared with the proposed Doppler model S (f ) for four observed cases in scenario ment distance are well matched with the actual setup Rx movement in the measurement scenario from C to A as depicted in Fig 2.6 It is noted that the estimated Rx speed Vn,R is calculated by using Eq 2.14 with the angle-of-motion αVR of π In the first case, because the adjusting factor A is small (−28.9 dB), the Gaussian component can be neglected Therefore, the overall Doppler bandwidth BD of 3.09 Hz is almost the same as the Spike bandwidth BSpike of 3.01 Hz In other words, the time varying characteristic of the UWA channel can be mainly determined by the Spike component for this case For the other cases, the values of adjusting factor A are significant; thus, the Gaussian component should be considered As a consequence, the time varying nature of the UWA channel depends on both Spike and Gaussian components, and the results of BD are larger than the Spike bandwidths BSpike Furthermore, the values of A for the last three cases gradually increase and reach the maximal value −3.0 dB for the case These results show 74 −15 −20 −25 −30 Measured spectrum Reference model Proposed Model −35 −40 −20 −15 −10 −5 10 15 20 Doppler shift [Hz] Case −5 −10 −15 −20 −25 −30 Measured spectrum Reference model Proposed Model −35 −40 −20 −15 −10 −5 10 15 20 Doppler shift [Hz] Case −5 −10 −15 −20 −25 Measured spectrum Reference model Proposed Model −30 −35 −40 −20 −15 −10 −5 Doppler shift [Hz] 10 15 20 Normalized Doppler spectrum [dB] −5 −10 Normalized Doppler spectrum [dB] Case Normalized Doppler spectrum [dB] Normalized Doppler spectrum [dB] Normalized Doppler spectrum [dB] Normalized Doppler spectrum [dB] that the Gaussian component from the surface motion increases As depicted in Fig 2.6, these cases correspond to the distance DA, when the Rx reduces its speed until it almost stops near the Tx Therefore, the decrease in the Rx speed makes the decrease in the Spike component As a results, the Gaussian-shape spectrum in Eq 2.9 becomes significant in comparison with the Spike component Case −5 −10 −15 −20 −25 Measured spectrum Reference model Proposed Model −30 −35 −40 −20 −15 −10 −5 10 15 20 10 15 20 10 15 20 Doppler shift [Hz] Case −5 −10 −15 −20 −25 Measured spectrum Reference model Proposed Model −30 −35 −40 −20 −15 −10 −5 Doppler shift [Hz] Case −5 −10 −15 −20 −25 Measured spectrum Reference model Proposed Model −30 −35 −40 −20 −15 −10 −5 Doppler shift [Hz] Figure 2.10: The reference model S˜n (f ) compared with the proposed Doppler model S (f ) for six typical cases in scenario 75 2.5.3 Scenario In this scenario, the Rx moves around the fixed Tx for the observation time Tobs = 250 s The time-varying measured spectrum S˜ (f, t) for this scenario is observed and then classified into six typical cases Subsequently, in Table 2.5, the optimal parameters of the proposed model {A, fm , w, C, fSpike } are computed and listed The comparison between the reference model and the proposed model for each case is shown in Fig 2.11 It can be seen the proposed model is in good agreement with the reference model Similar to the previous scenarios, the Spike bandwidth BSpike , the overall Doppler bandwidth BD , and the coherence time Tc are calculated and listed in Table 2.5 As can be seen, for the cases and 6, the overall Doppler bandwidths BD are almost the same as the Spike bandwidth BSpike This is because the adjusting factors A of these two cases are relatively small, namely, −32.8 dB and −23.3 dB, respectively For the other cases, the values of adjusting factor A are significant, which shows that the Gaussian-shape Doppler component from the surface motion could not be neglected Thus, the overall Doppler bandwidths BD are greater than the Spike bandwidth BSpike due to the contribution of the Gaussian components for these cases Notice that, if the Rx had moved exactly in a circle as supposed, then the value of αVR would have been π/2 Hence, the maximum Doppler frequency shift fSpike of the Spike component calculated by Eq 2.7 will be zero In fact, the Rx did not move exactly in a circle, so the Spike Doppler shift fSpike is not equal to zero as shown in Table 2.5 Because we did not record the angle of the Rx motion αVR for this scenario, the Rx speed Vn,R and the corresponding movement distance sn could not be calculated Instead of calculating these values, we estimate the direction of Rx movement for each case using the optimal results of fSpike Namely, if the value of fSpike is positive, that means the Rx moves towards the Tx; otherwise, the Rx moves away from the Tx Based on the estimated direction movement of the Rx, we roughly predict the trajectory of the Rx movement as plotted in Fig 2.12 76 Table 2.5: The optimal parameters for Doppler spectrum modeling derived from the measurement data in scenario (as plotted in Fig 2.11) Case (n) Gaussian component Spike component Overall Doppler Rx moving direction BD [Hz] Tc [s] 4Tn [s] A[dB] fm [Hz] -32.8 -1.5 20.0 100.0 6.50 3.90 4.01 0.250 100 Towards -16.9 -1.0 7.1 0.5 -0.23 1.90 2.32 0.431 25 Far away -16.4 -2.9 5.8 60.0 -2.27 1.76 2.08 0.481 25 Far away -7.9 -3.3 5.2 34.4 -2.85 2.92 7.40 0.135 25 Far away -13.2 -3.6 5.3 3.4 -1.35 4.38 5.42 0.185 25 Far away -23.3 -1.2 6.1 11.1 1.30 2.30 2.40 0.417 50 Towards C fSpike [Hz] BSpike [Hz] w [Hz] −5 −10 −15 −20 −25 −30 Measured spectrum Reference model Proposed Model −35 −40 −20 −15 −10 −5 10 15 20 Doppler shift [Hz] Case −5 −10 −15 −20 −25 −30 Measured spectrum Reference model Proposed Model −35 −40 −20 −15 −10 −5 10 15 20 Doppler shift [Hz] Case −5 −10 −15 −20 −25 −30 Measured spectrum Reference model Proposed Model −35 −40 −20 −15 −10 −5 Doppler shift [Hz] 10 15 20 Normalized Doppler spectrum [dB] Case Normalized Doppler spectrum [dB] 250 Normalized Doppler spectrum [dB] Normalized Doppler spectrum [dB] Normalized Doppler spectrum [dB] Normalized Doppler spectrum [dB] Total Direction Case −5 −10 −15 −20 −25 −30 Measured spectrum Reference model Proposed Model −35 −40 −20 −15 −10 −5 10 15 20 10 15 20 10 15 20 Doppler shift [Hz] Case −5 −10 −15 −20 −25 −30 Measured spectrum Reference model Proposed Model −35 −40 −20 −15 −10 −5 Doppler shift [Hz] Case −5 −10 −15 −20 −25 −30 Measured spectrum Reference model Proposed Model −35 −40 −20 −15 −10 −5 Doppler shift [Hz] Figure 2.11: The reference model S˜n (f ) compared with the proposed Doppler model S (f ) for six cases in scenario 77 Figure 2.12: The estimated trajectory movement of the Rx for scenario 2.6 Conclusions In this chapter, the Doppler effect in UWA communication systems is investigated in consideration of both the Doppler components caused by the surface motion and by the Rx movement Then, the Doppler power spectrum model for the shallow UWA channel was proposed and validated using the measurement-based results for three different measurement scenarios The curve fitting results show that the proposed model matches well with the measured Doppler power spectrum of the shallow UWA channel The paper [J3] deals with the content of this chapter In the future work, the proposed Doppler model will be applied to design UWA channel simulators for the performance evaluation of UWA communication systems Furthermore, the inter-carrier interference (ICI) resulting from the Doppler effect in UWA-OFDM systems can by analyzed by using the proposed model Based on the analytical results, Doppler compensation algorithms can be proposed Chapter UWA-OFDM SYSTEM PERFORMANCE ANALYSIS USING THE MEASUREMENT-BASED UWA CHANNEL MODEL The Doppler effect is caused by the relative movement between the transmitter (Tx) and the receiver (Rx) and/or the surface motion (waves) in underwater acoustic (UWA) communication systems The inter-channel interference (ICI), as the result of the Doppler effect, degrades the performance of orthogonal frequency-division multiplexing (OFDM) systems over UWA channels [47] This chapter aims to analyze the ICI plus noise of UWA-OFDM systems over the measurement-based model for shallow UWA channels The exact calculation of the ICI power, ambient noise power, and required transmit power, as well as their effects on the performance of UWA-OFDM systems will be carried out The signalto-interference ratio (SIR) and the signal-to-interference-plus-noise ratio (SINR) performance are analyzed for different number of sub-carriers, signal bandwidths, and levels of transmit power The interference power considered in the UWA-OFDM results from the Doppler effect that destroys the orthogonality between different sub-carriers giving rise to inter-carrier interference (ICI) Namely, signals from other subcarriers that interfere with the desired sub-carrier signal The SINR of UWAOFDM in the presence of ICI and ambient noise is calculated by using the measurement-based UWA channel model The results show that the SINR not only depends on the number of sub-carriers but also on the signal bandwidth and the transmit power Furthermore, the capacity of the UWA-OFDM system is derived from the SINRs By analyzing the capacity for different numbers of sub-carriers, the numerical results show that an increase in the number of sub-carriers leads to a decrease in the SINR but an increase in OFDM system capacity However, for a 78 79 given signal bandwidth, the capacity could not be increased by simply increasing the number of sub-carriers Moreover, the SINR and also the performance of UWA-OFDM systems depend on the transmit power The connectivity may be loss if the transmit power is low, whereas a high transmit power causes unnecessary interference In this chapter, the appropriate transmit power will be determined for individual cases of setting parameters of OFDM transmission The complex relations among these parameters are thoroughly investigated Based on the simulation results, the system parameters, including the signal bandwidth, the number of sub-carrier, and the transmit power, can be optimized for the design of UWA-OFDM It should be mentioned that this chapter present the ICI plus noise analysis of UWA-OFDM systems over a measurement-based UWA channel model, which is derived from the measurement data on a specific location Therefore, the corresponding simulation results will be only valid for the system considered in the measurement location 3.1 Introduction The large delay spread caused by the low speed of sound is one of the distinctive characteristic of UWA channels To mitigate inter-symbol interference (ISI) due to the large delay spread, Orthogonal Frequency Division Multiplexing (OFDM) modulation has been widely applied to acoustic transmission [10, 22, 42, 68, 86] However, in UWA-OFDM systems, the inter-channel interference (ICI) caused by the Doppler effect becomes more serious [47] The frequency of the transmitted signal is significantly distorted by the Doppler effect and multipath propagation [21] The motion-induced distortion has far-reaching implications on the design of the synchronization unit and the channel estimation algorithm [88] Unlike other research studies, such as [47] and [22], that analyzed the ICI effect based on the assumption of the classical Jakes distribution, the uniform distribution or the two-path model, we investigated the ICI effect using the wideband shallow UWA channel simulation model that has been derived from the measurement data of the real UWA 80 channel in West lake, Hanoi, Vietnam in June 2016 for the case of moving Rx Therefore, the Doppler effect was investigated as a combination of Doppler shifts due to the relative Tx/Rx movement and the sea-surface motion Several research studies on this subject have not analyzed the noise effect in combination with ICI or they considered the noise as white noise [9, 22, 86] In this chapter, we focus on the ICI effect in combination with ambient noise on UWA-OFDM systems By using the measurementbased UWA channel model, we analyze the ICI power of UWA-OFDM systems In addition, we study the ambient noise power, the resulting signal-to-interference ratio (SIR), and the signal-to-interference-plusnoise ratio (SINR) of each sub-carrier in the UWA-OFDM system Furthermore, the system capacity is analyzed using the SINR results In other words, the ICI effect is considered for capacity calculations In UWA communication system, the relation among the capacity, the signal bandwidth and transmit power is very complex [49] Therein, the capacity is an important factor that influences design of network communication systems [80] Nevertheless, the UWA channel capacity is still an open question mainly because of lack of well-established UWA statistical channel models [11] Most current reported results analyze the capacity that not take into account the ICI effect [11, 18, 49, 57, 68, 76, 80, 87] The authors in [18, 49, 57, 87] used the time-invariant channel model to analyze the capacity of UWA channels/systems In [11, 76], the time-variant UWA channel model was used; however, the capacity is derived from the SNR, i.e., without ICI effect considerations In [68], the interference is neglected in their analysis In this chapter, we use the SINR as a figure of merit instead of SNR or SIR to determine the appreciate system parameters Using the simulation results of capacity analysis, which is calculated from the SINRs, the appropriate OFDM transmission parameters for the UWA-OFDM system such as the transmit power, the signal bandwidth, and the number of sub-carriers can be determined To obtain the desirable characteristics of the system, those parameters must be well selected Therefore, the goal of this chapter is the OFDM system performance analysis based 81 on the measurement-based UWA channel, which has been derived from the measurement data of the shallow UWA channel in Halong bay as illustrated in Chapter 1, Sect 1.6 The numerical results in this chapter are made with some of the assumptions as follows: • The Doppler effect that causes the ICI is assumed to be as a result of the surface motion and the relative Tx/Rx movement The frequency mismatch in the transmitter and the receiver oscillators is not considered here • The SINR in this chapter are derived by analyzing the effect of both ICI and ambient noise in UWA-OFDM systems before applying any Doppler compensation algorithms or noise canceling methods • The inter-symbol interference (ISI) caused by multi-path UWA fading is eliminated by using guard time Here the guard time period is chosen as more than channel delay spread The rest of this chapter is organized as follows Section 3.2 is devoted to the calculation of the ICI power, ambient noise power, SIR, and SINR of UWA-OFDM systems by using the measurement-based UWA channel model The numerical results are presented and discussed in Sect 3.4 Finally, Sect 3.5 draws some conclusions 3.2 ICI Analysis of UWA-OFDM Systems In this section, the ICI analysis of UWA-OFDM systems using the measurement-based UWA channel simulation model will be presented It is noted that the simulation model has been obtained from Sect 1.6, Chapter 1, for the consideration case of Doppler effects Moreover, we will also study the effect of the ambient noise Expressions for the ICI power, noise power, SIR, and SINR are derived in terms of the timevariant channel transfer function (TVCTF) of the measurement-based UWA channel model 82 3.2.1 SIR Calculation The lowpass equivalent baseband output of the OFDM modulator can be expressed as [35, 46] N −1 X X [fk ] ej2πnk/N , x [tn ] = √ N k=0 (3.1) where N denotes the number of sub-carriers, x [tn ] is the nth discrete time sample of the OFDM symbol in time domain, and X [fk ] is the kth data-modulated sub-carrier in the OFDM symbol In the considered UWA-OFDM system, the signal x [tn ] is transmitted over the timevariant multipath UWA channel, the received signal at the sampling time tn = nTa , where Ta denotes the sampling period, is then given by [46] N −1 X x [tn ] = √ H [fk , tn ] X [fk ] ej2πnk/N + w [tn ], N k=0 _ (3.2) where n = 0, 1, 2, , N − 1; H [fk , tn ] = H [f = k∆f, t = nTa ] is the discrete version of TVCTF of the UWA channel model as in Eq 1.31 (Chapter 1); ∆f denotes the sub-carrier bandwidth, and w [tn ] stands for the ambient noise, which heavily depends on the signal frequency At the receiver, the demodulated data at the kth sub-carrier can be written as N −1 _ X_ X [fk ] = √ x [tn ] e−j2πnk/N , (3.3) N n=0 where k = 0, 1, 2, , N − _ _ Substituting x [tn ] in Eq 3.2 to Eq 3.3, the X [fk ] in Eq 3.3 can be rewritten as N −1 X X [fk ] = N m=0 _ " N −1 X ! H [fm , tn ]ej2π(m−k)n/N # X [fm ] + W [fk ] n=0 = S [fk ] + I [fk ] + W [fk ] , (3.4) where N −1 X S [fk ] = H [fk , tn ] X [fk ], N n=0 (3.5) 83 N −1 N −1 X X H [fm , tn ]ej2π(m−k)n/N X [fm ] I [fk ] = N n=0 (3.6) m=0 m6=k In the above equations, S [fk ] is the desired signal, I [fk ] denotes the interference signal, and W [fk ] is the Fourier transform of the ambient noise w [tn ] Assuming that the transmitted symbols X [fk ] have zero mean and unit power, using Eq 3.5 and Eq 3.6, the signal-to-interference ratio (SIR) experienced by kth sub-carrier due to the ICI can be obtained as [9] SIR [fk ] =