Resource allocation for wireless powered iot network doctor of philosophy major electronics engineering

Dissertation Resource Allocation for Wireless Powered IoT Network 무선전력 IoT 네트워크를 위한 자원할당 최적화 기법 Graduate School, Myongji University Department of Electronics Engineering Nguyen Tien Tung Dissertation Advisor Yong-Hwa Kim February, 2021 Resource Allocation for Wireless Powered IoT Network Submitted in partial fulfillment of the requirements for the Ph.D degree in Electronics Engineering February, 2021 Graduate School, Myongji University Department of Electronics Engineering Nguyen Tien Tung Resource Allocation for Wireless Powered IoT Network Graduate School, Myongji University Department of Electronics Engineering Nguyen Tien Tung We hereby recommend that the dissertation by the above candidate for the Ph.D degree in Electronics Engineering be accepted Chair, Evaluation Committee Prof JungKuk Kim Name Signature Member, Evaluation Committee Prof JaeMin Kim Name Signature Member, Evaluation Committee Prof Yong-Hwa Kim Name Signature Member, Evaluation Committee Prof Jong-Ho Lee Name Signature Member, Evaluation Committee Prof SeongWook Lee Name Signature February, 2021 (별표 2-4) 인준서 Dissertation Title(22∼26p) Acknowledgement First and foremost, I would like to express my sincere gratitude to my advisor, Professor Yong-Hwa Kim, for guiding, supporting and valuable advice during my study I was very fortunate and happy to have such a great advisor and mentor for my Ph.D study I have obtained a great deal of experiences from my advisor I would also like to extend my appreciation to the rest of my dissertation committee members, including Prof JungKuk Kim, Prof JaeMin Kim, Prof Jong-Ho Lee, Prof SeongWook Lee for their encouragement and valuable comments The valuable comments and feedbacks helped me to improve my dissertation I would like to thank all Professors at the Department of Electronics Engineering, Myongji University who taught and helped me to complete this dissertation I would also like to thank Dr Nguyen Van Dinh, Dr Pham Quoc Viet for their very kind support Special thanks to all my colleagues at the ICT Information Technology Convergence Technology for supporting and sharing everything Last but not the least, I would like to dedicate this dissertation and show my deepest gratitude and appreciation to my family, my beloved wife, Nguyen Thi Nhu Quynh and my two lovely daughters, Nguyen Ngoc Dan Khue, Nguyen Ngoc Hoa Nhien Table of Contents List of Figures iv List of Tables vi Abstract vii Chapter Introduction 1.1 Background 1.2 Contribution 1.3 Dissertation Outline Chapter Sum Rate Maximization for Multi-user Wireless Powered IoT Network with Non-linear Energy Harvester: Time and Power Allocation 2.1 Introduction 2.2 Description of the System Model and Energy Harvesting Models 2.2.1 System Model 2.2.2 Energy Harvesting Model 10 2.2.3 TDMA-enabled WPCN 11 2.2.4 OFDMA-enable WPCN 12 2.3 Joint Energy Harvesting Time and Power Allocation for TDMA-enable WPCN 15 i 2.4 Joint Energy Harvesting Time, Subcarrier Allocation and Power Allocation for OFDMA-enabled WPCN 20 2.5 Simulation Results 25 2.6 Conclusion 38 Chapter Resource Allocation for Energy Efficiency in OFDMA-Enabled WPCN 39 3.1 Introduction 39 3.2 System Model and Power Consumption Model 42 3.2.1 System Model 42 3.2.2 Power Consumption Model 44 3.2.3 Problem Formulation 45 3.3 Solution to Energy Efficiency of OFDMA-enabled WPCN 46 3.4 Numerical Results 52 3.5 Conclusion 56 Chapter Resource Allocation for AF Relaying Wireless-Powered Networks with Nonlinear Energy Harvester 58 4.1 Introduction 58 4.2 System Model and Problem Description 60 4.2.1 System Model 60 4.2.2 WET Phase 61 4.2.3 WIT Phase 62 4.3 Problem Formulation 63 ii 4.4 Proposed Solution 64 4.5 Numerical Results 71 4.6 Conclusion 76 Chapter Summary and Conclusions 77 References 79 Abstract in Korean 86 iii List of Figures Fig 2.1 System model Fig 2.2 (a) Frame structure of TDMA-enabled WPCN (b) Frame structure of OFDMAenabled WPCN 11 Fig 2.3 Convergence behavior of the proposed algorithms 27 Fig 2.4 Average sum-rate versus the transmit power of the PB for different numbers of antennas at the AP 28 Fig 2.5 Average sum-rate versus the distance between the PB and AP 29 Fig 2.6 EH time versus the distance between the AP and PB 30 Fig Performance comparison between the proposed algorithms and the equal time allocation (ETA) algorithm 31 Fig 2.8 Performance comparison between the proposed algorithms and the fixed EH timebased algorithm 32 Fig 2.9 Performance comparison between the proposed algorithms and the fixed EH timebased algorithm 33 Fig 2.10 EH time versus the transmit power of the PB 34 Fig 2.11 Average sum-rate versus the number of users 36 Fig 2.12 Average sum-rate versus the energy conversion efficiency 37 Fig 2.13 Average sum-rate versus the EH time 38 Fig 3.1 System model 43 Fig 3.2 Convergence of the proposed algorithm with different number of antennas at the AP 54 iv Fig 3.3 Energy efficiency for all schemes and EH durations of the proposed scheme versus PS's transmit power 55 Fig 3.4 Energy efficiency versus number of users 56 Fig Illustration of an RWPCN 61 Fig 4.2 Convergence of the proposed algorithm with different numbers of antennas at the BS and different power levels at the PS 72 Fig 4.3 The e2e sum throughput of the system for different schemes versus the transmit power of the BS 73 Fig 4 The e2e sum throughput versus the number of users 74 Fig End-to-end sum throughput versus the number of antennas at the BS 75 v List of Tables Table 2.1 Simulation parameters 26 Table 3.1 Simulation parameters 52 Table 3.2 Complexity analysis for different schemes 53 Table 4.1 Complexity analyses for different schemes 72 Table 4.2 Evaluation of user fairness issue 76 vi Initialize:  l = , u =  , m = ;  is a small value for the stopping criterion Repeat:  (m) = (l + u ) / 2; Obtaining optimal pk*,n based on (2.29); Obtaining optimal x* based on (2.33); Let x(m) = x* and p(m) = p* ; If F (p(m), x(m), (m) )  0, u =  (m); Else  l =  ( m); 10 End 11 m = m + 1; 12 Until: u − l   ; 13 Output: 14 Optimal power allocation: p* = p* (m) ; 15 Optimal subcarrier allocation: x* = x( m) ; 2.5 Simulation Results In the simulations, all the channels are assumed to be experiencing a Rayleigh fading, −3 − and channel gains are modeled as 10  d [36], where d represents the link distance and  is the exponential distribution Rayleigh with unit mean,  = 2.2 [36] is the pathloss exponent The positions of the PB and the AP are {0m,0m} and {0m,10m} -25- respectively The users are randomly distributed inside an area with −50m,50 m , −50m,50 m Several other important simulation parameters are given in Table 2.1 Table 2.1 Simulation parameters Parameter Value System bandwidth, B 180 kHz [37] Number of subcarriers 12 [37] Noise special density,  -117 dBm/Hz Energy conversion efficiency,  0.8 M 24 mW [33] a 0.014 [33] b 150 [33] In Figs 3-13, we refer our proposed solutions as ``TDMA-EH'', ``TDMA-nonEH'', ``OFDMA-EH'', and ``OFDMA-nonEH'' for the TDMA-enabled schemes and OFDMAenabled schemes using a linear EH model and a non-linear EH model, respectively -26- (a) PB = 30 dBm (b) PB = 40 dBm Fig 2.3 Convergence behavior of the proposed algorithms As shown in Fig 2.3 the convergence rate of Algorithm and Algorithm is plotted for all schemes, NT = , K = , and d x = 10 m We observed that the proposed algorithms for the TDMA-enabled schemes converge to the maximum SR faster than that for the OFDMA-enabled schemes Clearly, the SR of the TDMA-enabled WPCNs nearly converges after three iterations, while that of the OFDMA-enabled WPCNs converges after ten iterations This difference is due to the uncertain values of binary variables on the SC of the OFDMA-enabled WPCNs -27- Fig 2.4 Average sum-rate versus the transmit power of the PB for different numbers of antennas at the AP Fig 2.4 shows the change in the SR in accordance with the transmit power of the PB for various schemes with different number of antennas at the AP with K = and d x = 10 m In this scenario, the distance between the PB and the AP is set as d x = 10m and the AP serves K = IoT users Fig 2.4 shows that when the number of antennas at the AP and the transmit power of the PB increase, the SR of all schemes increases This is because increase in its number of antennas, the AP harvests more energy, resulting in an improvement in the transmission information quality In addition, all schemes that adopted TDMA technique are more effective than those that adopted the OFDMA technique and the performance gap between the TDMA-enabled schemes and OFDMA-enabled schemes increases as the number of antennas at the AP increases The performance of the schemes deploying the non-linear EH model is worst than that of the schemes deploying the linear EH model owing to the limitations of EH circuits in the non-linear EH model -28- Fig 2.5 Average sum-rate versus the distance between the PB and AP Fig 2.5 illustrates the SR as a function of the distance between the PB and the AP when transmit power of the PB are set PB =30 dBm and PB = 40 dBm with NT = and K = It is observed that the SR decreases significantly as this distance increases This means that the greater is the distance between the AP and the PB, the longer is the EH duration of the AP as shown in Fig 2.6 Correspondingly, there is less time for information transmission from the AP to multiple users, thus the SR reduces In addition, as the distance increases, the path loss becomes the dominant factor leading to a lower amount of energy received at the AP The advantage of the schemes using the linear EH model is lost as the AP is placed farther away from the PB In addition, the duration of EH of the TDMA-enabled schemes is always greater than that of the OFDMA-enabled schemes as shown in Fig 2.6 -29- Fig 2.6 EH time versus the distance between the AP and PB In Fig 2.7, we compare the performance of the proposed algorithms with that of the ETA algorithm studied in [22], [23] under the different transmit powers of the PB with NT = and K = For the ETA algorithm, from K  k =0 k = , we have  k = , k  K , while the K +1 power allocation is similar to that of the TPA algorithm As can be seen that the performance of the proposed algorithms outperforms that of the ETA algorithm for all schemes This higher performance is because some users in the ETA algorithm may be allocated insufficient time to communicate with the AP, leading to a strictly suboptimal solution This phenomenon further confirms the effectiveness of jointly optimizing  k and pk in the considered system -30- (a) Linear EH models (b) Non-linear EH models Fig Performance comparison between the proposed algorithms and the equal time allocation (ETA) algorithm Fig 2.8 and 2.9 compare the performance of the proposed optimal algorithms with the fixed EH time (FEHT) algorithm with different values of  [21], [25], NT = , K = and d x = 10 m For the FEHT algorithm, we only fix the EH duration  but use the same method for TPA at the AP, similar to the proposed optimal algorithms As expected, from -31- Fig 2.8, the proposed optimal algorithms outperform the FEHT for the TDMA-enabled schemes because of the use of optimal jointing time and TPA at the AP These results verify the effectiveness of our proposed optimal solutions (a) TDMA-EH-enabled schemes (b) TDMA-nonEH-enabled schemes Fig 2.8 Performance comparison between the proposed algorithms and the fixed EH time-based algorithm -32- However, from Fig 2.9, the performance of the FEHT-based OFDMA-enabled schemes with  = 0.5 is close to that of the proposed OFDMA-enabled schemes, using both the linear EH model and non-linear EH model (a) OFDMA-EH-enabled schemes (b) OFDMA-EH-enabled schemes Fig 2.9 Performance comparison between the proposed algorithms and the fixed EH time-based algorithm -33- From Fig 2.10, we can see that  of the proposed OFDMA-enabled schemes is approximately 0.5( s) (with NT = , K = and d x = 10m ); hence, the proposed OFDMA-enabled schemes and FEHT-based OFDMA-enabled schemes exhibit similar performances for  = 0.5( s ) Fig 2.10 EH time versus the transmit power of the PB The affect of the number of users on the SR of all schemes for PB =30 dBm and PB = 40 dBm with d x = 10m at the PB and two cases of the number of antennas at the AP, i.e., NT = , NT = is illustrated in Fig 2.11, respectively We can observe that the all schemes tend to reach their maximum SRs when the number of users is more than Specially, the OFDMA-enabled schemes outperform the TDMA-enabled ones when the number of users increases from to 10 for NT = However, the advantage of OFDMAenabled schemes compared with that of TDMA-enabled schemes is not significant for NT = The same phenomenon is observed as in the Fig for NT = and PB = 40 dBm In addition, the TDMA-enabled schemes can harvest more energy due to taking advantage of having more antennas, i.e., NT = , and therefore, their SRs reach saturate values faster -34- than that of the OFDMA-enabled schemes Moreover, the gap between TDMA-enabled schemes and OFDMA-enabled schemes becomes smaller as the number of users increases for PB = 40 dBm In the case of the small number of users, the performance of the TDMAenabled schemes is only better than that of the OFDMA-enabled schemes when the number of antennas at the AP as well as the transmit power at the PB are high enough In other words, the OFDMA-enabled schemes can be a suitable option when the AP is equipped with a small number of antennas and the transmit power of the PB is also small -35- (a) PB = 30 dBm (b) PB = 30 dBm Fig 2.11 Average sum-rate versus the number of users In Fig 2.12, we evaluate the performance of the schemes that adopt the linear EH model and non-linear EH model as energy conversion efficiency is varied, where NT = , K = , and d x = 10 m Clearly, the SR of the linear EH-based schemes increases when the energy conversion efficiency varies from 0.1 to 0.9, while that of the non-linear EH-based -36- schemes is constant In addition, the Fig shows that the performance of the linear EHbased schemes can be equivalent to that of the non-linear EH-based schemes as energy conversion efficiency of the EH circuit is between 0.3 and 0.4 Fig 2.12 Average sum-rate versus the energy conversion efficiency Fig 2.13 depicts the SR as a function of EH duration  for the considered resource allocation schemes with NT = , K = , PB = 40 dBm and d x = 10 m From the Fig , one can observe that the SR function is a concave function with respect to  This further confirms that we can use 1-D line search, e.g., golden search, to obtain the optimal  0* for all the TDMA-enabled schemes -37- Fig 2.13 Average sum-rate versus the EH time 2.6 Conclusion We have studied the problem of maximizing the EE of the OFDMA-enabled WPCN The MINLP problem is efficiently solved by the iterative based Dinkelbach method The novelty of the proposed algorithm is attributed to the fact that in each iteration closed-form solutions are found The desirability of jointly optimizing EH time, SC assignment, and power allocation is justified by simulation results -38- Chapter Resource Allocation for Energy Efficiency in OFDMA-Enabled WPCN 3.1 Introduction Communication plays an important role in disaster situations where communication infrastructure and power disruptions result in low reliability and reduced availability of the network [38] In many cases, communication networks in disaster areas are unavailable because of power failure Equipment such as base stations and access points are not functional due to this loss of power Further, Internet of Things (IoT) devices are typically with low-power and low-cost requirements with a finite-capacity battery, and the replacement is not always available Therefore, a vital task during disasters is the restoration of power supplies To overcome this issue, energy harvesting (EH) through renewable energy sources (e.g., wind, solar, geothermal, and hydropower) is advocated as a promising solution to prolong the life-time of IoT devices However, in disaster situations, such energy sources may not be a viable approach due to destruction Therefore, wireless power transmission using radio frequency (RF) to charge for energy-constrained devices is a promising solution for low-power wireless networks [6], [28], [30], [38], [39] In [30], a hybrid power transfer architecture was proposed where airships, helicopters, and balloons are -39-