Kisong Lee, Hyun-Ho Choi, Woongsup Lee, Victor C. M. Leung
©SHUTTERSTOCK.COM/DOUCEFLEUR
Interference is usually regarded as a detrimental factor that must be avoided or suppressed to achieve higher performance in traditional wireless communications. Wireless energy harvesting (EH) technologies have been found to be capable of converting such harmful interference into a feasible energy source for low-powered Internet of Things (IoT) devices that otherwise have limited lifetimes. In this context, we introduce a wireless-powered interference network (WPIN) in which interference is proactively controlled, considering the two opposing concepts of signal jammers and energy sources to improve the bidirectional transmission rate of IoT devices. First, an overview of WPIN applications is provided in various wireless topologies with complex cochannel interference. Then, a wireless interference harvesting protocol is presented to manage this cochannel interference for bidirectional communications in WPINs. We investigate coordinated resource management and beamforming schemes based on this interference harvesting protocol and demonstrate how these schemes improve the performance of WPINs. Simulation results show that the proper utilization of interference according to the channel structure decreases interference’s negative effects on information decoding and increases the amount of harvested energy, thereby simultaneously improving the downlink and uplink capacities. Finally, imminent research challenges and directions with regard to making WPINs more practical and useful are outlined.
Recent advances in information and communication technology have led to an increase in the number of wireless devices connected to the Internet, heralding the emergence of the IoT era. To provide seamless IoT services, it is necessary to replace batteries when they are depleted. Although many IoT devices consume only small amounts of energy, battery replacement is inconvenient, costly, and may even cause environmental problems [1]. With the demand for green communications for 6G, radio-frequency (RF) EH communication systems, in which wireless devices harvest energy from the received RF signals, have been the subject of considerable attention as a more convenient and sustainable approach. In particular, wireless power transfer (WPT) enables energy to be transmitted over a wireless medium to target IoT devices, making it a plausible solution for supplying energy to such devices [2].
Because RF signals can be used to transfer both information and energy, the integration of wireless communication and WPT has become an emerging research field that has attracted great interest. This field has two main research directions. The first is simultaneous wireless information and power transfer (SWIPT), in which the same RF signal is utilized to transmit both the energy that is harvested by the receiver and information that is decoded by the receiver. However, it is not currently possible for practical receiver circuits to perform both functions simultaneously. Therefore, receiver architectures for SWIPT, such as time switching (TS) and power splitting (PS), have been proposed [3]. These architectures divide the received signal into two circuits, one for EH and the other for information decoding. Recently, a hybrid technique that combines TS and PS has also been suggested [4]. SWIPT has since been investigated extensively, and several studies have examined different aspects of SWIPT, such as circuit designs, protocols, resource allocation, and network architectures [5].
The second research direction focuses on the design of a new type of wireless network called a wireless-powered communication network (WPCN). In a WPCN, the hybrid access point (HAP) broadcasts an RF signal for WPT during the downlink phase. User devices then harvest energy from this RF signal, which is utilized for data transmission in the subsequent uplink phase. Similar to SWIPT, WPCNs have been extensively studied under various network architectures, such as cellular, random-access, and multihop networks [6].
Recent studies have begun to consider SWIPT and WPCNs jointly for bidirectional links. The main targets of these investigations are the throughput maximization of each link and minimization of the outage probability in SWIPT–WPCN hybrid networks. Therein, the SWIPT functionality (e.g., the transmit power, length of the WPT period, and beamforming vector) is considered for the downlink, and the WPCN functionality (e.g., the EH ratio and time allocation for each user) is considered for the uplink resources [7], [8], [9].
Despite extensive studies, the realization of such RF-based EH technology in practical communication systems remains hindered by the significant attenuation of the signal strength over physical distances. Specifically, the amount of energy that can be harvested by users located far from the HAP may be insufficient for subsequent transmissions. Therefore, in relation to user fairness, severe performance disparities may arise for users at different locations. Moreover, with regard to the entire network, both the service coverage and the network lifetime can be limited.
To overcome these problems, the notion of interference harvesting has recently been introduced to allow energy-deprived users, e.g., cell edge users, to scavenge energy consistently from the signals sent by nearby transmitters, e.g., neighboring base stations (BSs), in addition to the signals sent by their associated BS [10]. To illustrate the benefits of interference harvesting compared to signal-only harvesting, the amounts of harvested energy in both cases are compared in a simulation, as shown in Figure 1. For this simulation, 25 microcell BSs with a transmit power of 30 dBm are randomly deployed within the coverage of a macrocell BS with a transmit power of 43 dBm [10]. The received power is increased by 10 dBm on average, and the corresponding variance is significantly reduced from 62 to 26 dBm when interference harvesting is employed. Such substantial performance gains are mainly attributable to the harvesting of energy from interference due to neighboring BSs by cell edge users.
Figure 1 A comparison of the received power in heterogeneous cellular networks: (a) signal harvesting only (μ = −39.08, v2 = 62.94) and (b) signal-plus-interference harvesting (μ = −30.55, v2 = 26.48).
It should be emphasized that the key difference between the interference harvesting system and conventional SWIPT and WPCN systems is that the former can actively control the interference while the latter cannot [11]. Interference is usually regarded as an intrinsically destructive phenomenon that must be avoided in the design of conventional wireless systems [12]. However, the interference in the WPINs considered here can be a valuable resource for supplying wireless energy to IoT devices that improves the overall network performance as long as it can be managed efficiently. Therefore, the receivers in WPINs harvest interference in addition to the desired signal, while the harvesting target of both SWIPT and WPCNs is mainly the desired signal of the receivers. To control the amount of interference provided to the receivers in WPINs, the transmitters perform transmit power control and/or beamforming, and the receivers adjust the EH ratio to balance the transmission rate and the amount of harvested energy. Thus, the amount of interference can be deliberately increased in WPINs to improve the EH capability of IoT devices, which seems absurd in conventional wireless networks.
The aims of this article are to reexamine the notion of interference in wireless networks and to introduce a new perspective in which interference is considered a valuable energy source for enhancing the performance of WPINs. To this end, first, we provide an overview of the topology and applications of WPINs, after which we present a protocol for bidirectional communication in WPINs. Thereafter, we present two coordination schemes to handle interference in WPINs and demonstrate the degree of performance improvement that can be achieved. Finally, related research challenges and approaches are presented for those attempting to introduce the advanced capabilities of WPINs and increase their practicality.
The use cases of WPINs and the issues they face in diverse network topologies are explained in this section, followed by a description of a novel two-way communication protocol for WPINs.
Figure 2 illustrates the use cases of WPINs in various types of network topologies. Future network topologies (e.g., beyond 5G or 6G) are expected to be more diverse and denser than current topologies, due to the continuous increase in wireless data transmission levels and demand for various wireless services. Thus, as shown in Figure 2, the introduction of macrocell, picocell, femtocell, and integrated access backhaul networks will lead to more hierarchical cell levels in WPINs. Moreover, multihop ad hoc networks, proximate device-to-device (D2D) communications, and wireless sensor networks (WSNs) for the IoT will be overlaid on top of existing baseline networks. Such a complex network topology results in more overlapping RF signals, which can be either desired data signals or interference.
Figure 2 Use cases of WPINs in various network topologies. IAB: integrated access backhaul; WIT: wireless information transfer; WSN: wireless sensor network; D2D: device to device.
In traditional wireless networks, cochannel interference due to simultaneous transmissions by multiple devices has long been regarded as the dominant factor that limits the performance of wireless systems because it adversely affects information signal decoding [12]. Moreover, such cochannel interference increases significantly as networks become more complex and diverse. However, this cochannel interference can be a useful energy source in WPINs if the end devices (e.g., low-power IoT devices) harvest the interference instead of decoding the signals. For example, the low-power sensors made by Texas Instruments (e.g., CC 430 and RF 430) consume tens of $\mu{\text{W}}$ when in sleep mode and hundreds of microwatts when used in active mode. This is a level that can be operated through wireless EH in practice. This allows the devices to harvest more RF energy, at the expense of the data rate. The harvested energy can then be used for the next transmission to increase the total data rate of the end device.
In the WPIN considered here, any RF signal can be turned into an energy source that can be used to improve different aspects of the system performance, such as the energy efficiency, network lifetime, and even transmission rate. In particular, the WPIN allows for bidirectional communication in end devices that are not plugged into power sources through the use of SWIPT in the downlink and wireless information transfer (WIT) in the uplink. In other words, the transmitters control their transmit power to send signals, while the receivers adjust the EH ratio to receive both information and energy simultaneously in the downlink. Subsequently, the receivers perform WIT in the uplink using the harvested energy in response to the downlink message.
This controllable SWIPT and WIT operation can provide various use cases in practical WPIN environments. With regard to downlink-oriented services, users receive a long data message and transmit a short feedback message (e.g., channel information or acknowledgment) on the uplink using the harvested energy. This case enables users to receive information from most downlink signals by adjusting the EH ratio and to transmit short response information on the uplink with the energy harvested from the remaining downlink signal. For uplink-oriented services, users harvest most of the energy while receiving the minimum information from the downlink signal and then transmit as much information as possible on the uplink using the harvested energy. This case applies when the sinker node transmits a short control message on the downlink, and sensor nodes transmit their maximum amounts of sensing data on the uplink in WSNs. For two-way data transmission services, such as voice and video calls, users properly adjust the EH ratio to balance the downlink and uplink transmission rates according to their requirements. On the other hand, when users do not need to send and receive data, they can collect the full energy on the downlink without consuming energy during the uplink transmission. This allows the devices to charge their batteries wirelessly and to prolong their usage times conveniently for all WPIN use cases. To support these use cases in WPIN, a SWIPT-then-WIT protocol was devised [9]. It is described in detail in the following section.
Figure 3 exhibits the data transmission and EH by N transmitter–receiver pairs that use the SWIPT-then-WIT protocol in a WPIN. Figure 4 presents the frame structure of the SWIPT-then-WIT protocol for the representative SWIPT policies at the receiving nodes, i.e., TS and PS, in which synchronized time-division duplexing with the same bandwidth and time resources is considered.
Figure 3 The data transmission and EH in the SWIPT-then-WIT protocol. (a) Phase 1: SWIPT on the downlink. (b) Phase 2: WIT on the uplink. EIT: energy and information transmitter; EIR: energy and information receiver.
Figure 4 The frame structure of the SWIPT-then-WIT protocol. The (a) TS and (b) PS.
As shown in Figure 3, the system operation consists of two phases in the SWIPT-then-WIT protocol. In the first downlink phase, each transmitting node sends a signal using the transmit power ${p}_{i}$ as an energy and information transmitter (EIT), while each receiving node harvests energy and receives information simultaneously from the received signal as an energy and information receiver (EIR). The desired signals in each pair can be used adaptively for both information decoding and EH; in contrast, the interfering signals interfere with information decoding in other pairs but can still be used for EH. This ambivalence in interference is considered in the EIRs, as they control the ratio of information decoding to EH. Specifically, under the TS policy, each EIR i divides the time for receiving signals from the EITs into two portions such that the portion ${\alpha}_{i}\,{\in}\,{[}{0},{1}{]}$ is used for EH, and the remaining portion ${(}{1}{-}{\alpha}_{i}{)}$ is used for information decoding. In contrast, under the PS policy, each EIR i splits the received signal power by adjusting the PS ratio ${\alpha}_{i}$ in the SWIPT functionality.
In the second uplink phase, the EITs and EIRs switch roles. The EIRs now act as ITs and transmit information to the EITs using the energy ${e}_{i}$ collected in the first phase, while the EITs now act as IRs that receive this information. It should be noted that the EH circuit has a relatively low power sensitivity, from −10 to approximately −30 dBm, in the corresponding energy harvesters compared to that of the communication circuit, which ranges from −60 to −80 dBm in IRs [13]. Therefore, the received signal strength at the uplink, which depends on the harvested energy in the ITs, is likely to be significantly weaker than that at the downlink. Consequently, it is reasonable to consider only WIT without EH in the uplink because the received signal strength at the uplink is insufficient to activate the EH circuit for the IRs.
Two coordination strategies related to the SWIPT-then-WIT protocol in WPINs, coordinated resource management and coordinated beamforming, are discussed in this section. Through simulations, we show that the two coordination schemes can properly manage interference according to the network environment and that they have the potential to extend the downlink and uplink rate region.
The SWIPT operation in the first phase of the SWIPT-then-WIT protocol directly affects the WIT performance in the second phase. Thus, the proper selection of the transmit power of the EITs ${(}{p}_{i}{)}$ and the EH ratio of the EIRs $({\alpha}_{i})$ is critical for effectively utilizing cochannel interference to achieve the objectives of the WPIN application [8], [9]. To investigate the effect of resource management on the bidirectional transmission performance, we consider the application of a WPIN to a D2D network in which all the nodes are distributed randomly, resulting in frequent asymmetric channel conditions. We describe the downlink and uplink rate region (${R}^{\text{DL}}{-}{R}^{\text{UL}}$ region) for this scenario in Figure 5 and indicate the optimal point for maximizing the sum rate of the downlinks and uplinks for each method. For comparison, two baseline schemes are considered: a scheme with the optimal ${p}^{\ast}$ and fixed ${\alpha} = {0}{.}{5}$ and a scheme with the optimal ${\alpha}^{\ast}$ and fixed ${p} = {1}$ for both the TS and PS policies, which demonstrate the impact of optimizing each control parameter on the performance. In the scheme with the optimal ${p}^{\ast}$ and fixed ${\alpha} = {0}{.}{5}$, the EITs determine the optimal transmit power by considering the effect of interference, while the EIRs use a fixed EH ratio of 0.5. In the scheme with the optimal ${\alpha}^{\ast}$ and fixed ${p} = {1}$, the EITs use the maximum transmit power, whereas the EIRs use the optimal EH ratio determined by considering the balance between EH and information decoding.
Figure 5 The downlink and uplink rate region of the SWIPT-then-WIT protocol based on coordinated resource management.
The channel conditions experienced by each user pair in the considered D2D scenario can be completely different from those experienced by other pairs. Thus, we consider an asymmetric two-user pair scenario, which is a major obstacle that limits D2D performance. In our simulation, the downlink signal-to-interference-plus-noise ratios (SINRs) of the two user pairs are set to ${\gamma}_{1}^{\text{DL}} = {-}{10}\,{\text{dB}}$ and ${\gamma}_{2}^{\text{DL}} = {10}\,{\text{dB}}$, while the uplink SINRs of the two user pairs are set to ${\gamma}_{1}^{\text{UL}} = {\gamma}_{2}^{\text{UL}} = {0}\,{\text{dB}}$, considering a linear EH model with ${\eta} = {0}{.}{5}$ [8], [9]. In the linear EH model, the amount of harvested energy is affected by the energy conversion efficiency ${(}{\eta}{)}$ and the received RF power. However, this depends on the various parameters of EH circuits as well as the received RF power in the nonlinear EH model [14]. In our simulation, we adopt the linear EH model, assuming that the received RF power is sufficiently low for the EH circuits to exhibit linear behavior in a practical wireless environment. In this case, although user pair 1 experiences difficulty in decoding information, owing to the strong interfering signal, the strength of the interfering signal is advantageous for EH. In comparison, the high downlink SINR of user pair 2 is advantageous for information interpretation, but the relatively low interference limits the amount of energy that can be harvested.
The simulation result clearly shows that the rate region can be extended through coordinated resource management compared to the conventional methods in which either ${\alpha}$ or p is fixed. Specifically, the scheme with the optimal ${p}^{\ast}$ and fixed ${\alpha} = {0}{.}{5}$ cannot provide a high ${R}^{\text{UL}}$ because the harvested energy, which is used for transmission in the uplink, is not optimally determined. Moreover, the scheme with the optimal ${\alpha}^{\ast}$ and fixed ${p} = {1}$ cannot ensure a high ${R}^{\text{DL}}$ because the transmit power, which directly affects ${R}^{\text{DL}}$, is not optimally determined. In contrast, the use of the optimal ${p}^{\ast}$ and ${\alpha}^{\ast}$ values in the proposed schemes improves both ${R}^{\text{DL}}$ and ${R}^{\text{UL}}$ by optimizing the transmit power and EH ratio jointly and thereby extending the rate regions of both PS and TS considerably. It is also observed that the TS performance is lower than the PS performance, which is consistent with the observations in [8]. The reason for this can be inferred from [8, eqs. (1) and (23)]. Specifically, the TS policy temporally divides the received signal to accommodate EH, thereby resulting in the direct scaling down of the downlink rate by the EH ratio. In contrast, the PS policy partitions the incoming signal power for EH, which not only reduces the received signal power but also decreases the interference power, resulting in a smaller decrease in the downlink rate.
Moreover, the simulation result in Figure 5 shows that the optimal parameters for maximizing the sum rate are ${[}{p}_{1},{\alpha}_{1}{;}\,\,{p}_{2},{\alpha}_{2}{]} = {[}{0},{0}{;}\,\,{1},{0}{.}{38}{]}$ for PS and ${[}{p}_{1},{\alpha}_{1}{;}\,\,{p}_{2},{\alpha}_{2}{]} = {[}{0}{,}\,\,{1}{;}\,\,{1}{,}\,\,{0}{]}$ for TS. These parameters indicate that for PS, user pair 1, which has a low SINR, should be silent on both links, while user pair 2, which has a high SINR, should use both the downlink and uplink exclusively. For TS, user pair 2 should use the full transmit power and not harvest energy at the receiver (i.e., it should perform only downlink transmission), while user pair 1 should not transmit power and should harvest energy from the downlink transmission of user pair 2 (i.e., it should perform only uplink transmission). This observation indicates that concurrent transmission by two user pairs does not guarantee high downlink and uplink rates for both user pairs and, instead, leads only to interference with the transmission of the other user pair. Therefore, it is conjectured that each user pair should allocate resources such that the downlink or uplink interference is thoroughly suppressed in the asymmetric channel condition.
It was observed in the previous section that in D2D networks under asymmetric channel conditions (i.e., one of two user pairs experiences a severely poor SINR), the user pair with a poor SINR should silence itself to avoid interference with the other pair and harvest the received interference fully to collect energy for the subsequent transmission. It should be noted that the communication environment in cellular networks differs significantly from that in D2D networks because in the former, the interference is rarely stronger than the desired signal when the cell partitions are well designed. Moreover, it is more affordable to equip BSs with multiple antennas than to do so for D2D devices. Coordinated beamforming can, then, be used in this context. In coordinated beamforming, each BS generates a directional beam in which the power is concentrated on the corresponding user, as demonstrated in Figure 6. This allows each user to receive the desired signal without severe interference or to harvest energy from the aggregated signal and interference [7] because the amount of harvested energy and the ability to receive information are simultaneously increased by the beamforming gain of the antennas. In other words, coordinated beamforming can significantly improve the SINR of data transmissions even under poor channel conditions, such as those faced by cell edge users. The use of coordinated beamforming can thus increase the data rate while maintaining the amount of energy harvested. It is also important to note that increasing the SWIPT capability in the downlink also enhances the WIT performance in the uplink and therefore improves the performance of the WPIN.
Figure 6 The SWIPT-then-WIT protocol based on coordinated beamforming.
Figure 7 shows the downlink and uplink rate region (${R}^{\text{DL}}{-}{R}^{\text{UL}}$ region) for different numbers of cooperating BSs to illustrate the performance improvement that can be achieved through coordinated beamforming. This rate region was obtained by considering a hexagonal cell layout, with a cell radius of 50 m, and an EH-enabled receiver positioned 35 m away from its serving BS [11]. In addition, a typical distance-based path loss model, with a path loss exponent of 2.3, was employed, and the beamforming gain was assumed to be 6.7 dB, considering the average beamwidth of 40° [7]. We fixed the number of cooperative BSs transmitting the same signal to the EH-enabled receiver to two or three and varied the number of interfering BSs from four to 30 to investigate the changes in the downlink and uplink transmission rates.
Figure 7 The downlink and uplink rate region of the SWIPT-then-WIT protocol based on coordinated beamforming.
As the number of cooperating BSs increases, the ${R}^{\text{DL}}{-}{R}^{\text{UL}}$ region is extended, as expected. Moreover, as the number of noncooperating BSs (i.e., interferers) increases, the receiving nodes are interfered with to decode information but harvest more energy from the interfering signals in the downlink. This eventually decreases ${R}^{\text{DL}}$ but increases ${R}^{\text{UL}}$, reflecting the duality of interference in WPINs. This result demonstrates the viability of using interference to improve the bidirectional transmission performance through the use of coordinated beamforming, which increases the uplink transmission capability by increasing the strength of the desired signal such that it overwhelms the interference. In other words, cochannel interference can be proactively utilized as a valuable energy resource in cellular networks for effective WPINs by adopting a coordinated beamforming strategy. Our results show that there are significant differences among interference management strategies for D2D and cellular networks; therefore, interference should be appropriately managed according to the network environment, that is, the interfering channel. It should also be noted that a practical beamforming design that can operate under the constraints imposed by limited data exchanges and the characteristics of WPINs is necessary to fully realize the performance gains of coordinated beamforming. Moreover, reconfigurable intelligent surfaces that artificially reconstruct the propagation environment of electromagnetic waves can be adopted to enhance the efficiency of WPINs.
This section discusses several research challenges that can arise when applying WPINs to future wireless communications systems.
Figure 8 illustrates an example of cross-link interference in WPINs. When the downlink and uplink transmissions are not synchronized in the time domain, i.e., when one transmitter–receiver pair performs downlink SWIPT while the other transmitter–receiver pair performs uplink WIT within the same time period, cross-link interference can occur between different transmitter–receiver pairs. Cross-link interference commonly occurs during the dynamic deployment of heterogeneous networks for WPINs and can seriously compromise the reliability of the entire network. However, it also significantly increases the amount of energy that can be harvested from interfering devices. Therefore, this ambivalent effect of cross-link interference on the uplink and downlink operations should be considered when designing the framework, protocol, scheduling, and resource management strategies for WPINs.
Figure 8 The cross-link interference in WPINs.
Compared to conventional wireless-powered communication systems with fixed positions, unmanned aerial vehicles (UAVs) can adjust their locations over time to establish short-distance line-of-sight links with ground nodes, effectively owing to their high mobility. This allows UAVs to transmit RF energy efficiently and control information to ground nodes and to collect data from these nodes, which harvest energy from the signals transmitted by the UAVs. The suitability of wireless-powered two-way communication for UAV-assisted WPINs mandates that strategies for UAVs regarding the characteristics of UAV-assisted two-way WPINs should be investigated.
The topology of WPINs can change dynamically because they consist of diverse and dense networks. This may result in the inability of some users to receive sufficient energy to operate. Therefore, flexible designs and flexible operation are required to cope with frequent topology changes. Moreover, heavyweight centralized approaches may not work well in WPINs with dozens or more devices [8]. Therefore, it is necessary to develop lightweight distributed approaches with low computational overhead to enable WPINs to adapt flexibly to topology changes in the surrounding environment.
Because a large number of communication and interference links must be considered simultaneously in WPINs, the resulting signaling and computational overhead can be overwhelming, which limits the applicability of WPINs in practice. Therefore, a distributed approach should be considered to reduce the signaling overhead and computational complexity for WPINs. For example, a recent deep learning-based technique, such as federated learning [15], which trains algorithms on multiple decentralized edge devices with local data samples, can be a promising candidate to solve this problem.
In WPINs, the energy receivers must be able to receive higher signal power levels than the IRs, due to the low power sensitivity of the EH circuit. Hence, the energy receivers are more likely to be located near the transmitters than the IRs. The energy receivers therefore have the potential to become powerful eavesdroppers if they attempt to decode information in the signals sent by the transmitters instead of harvesting energy, such that they are perfectly capable of eavesdropping. It is therefore necessary to devise a security strategy to protect information from such potential eavesdroppers for secure and robust communications in WPINs while ensuring that a sufficient amount of power is transmitted to the energy receivers.
Fairness among wireless devices is a crucial performance metric in WPINs to gauge the quality of service (QoS) of users. Since WPINs are entirely distinct from conventional EH-enabled networks, new utility functions must be developed properly to improve both the efficiency and fairness with respect to the characteristics of WPINs. Therefore, it is necessary to propose a fairness-aware resource management approach that considers user scheduling, power/bandwidth allocation, and cooperation methods. This approach should optimize these utility functions and ensure the QoS requirements for each WPIN application.
To truly realize the SWIPT-then-WIT protocol in practice, performance verification through prototyping is required. To this end, it is necessary fully to consider the physical characteristics of the wireless channels and carefully select the proper attributes of the operating environments, such as the frequency and transmit power. Moreover, novel antenna designs with a metasurface and efficient power amplifiers should be considered important factors for improving EH. Furthermore, the nonlinear characteristics of EH circuits must be taken into account in designing prototypes to examine the real amount of energy harvested in dense WPINs. Additionally, it is crucial to examine the impact of electromagnetic interference emissions, which may pose a risk to human health.
In this article, WPINs for the efficient use of complex cochannel interference under various topologies in future wireless networks have been introduced. To utilize this interference as a feasible energy source, we introduced the SWIPT-then-WIT protocol and presented coordinated resource management and beamforming schemes to enhance the bidirectional transmission rates in different interference environments. Performance evaluation results showed that the performance of WPINs can be significantly improved if the interference is adaptively controlled to increase the energy harvested while reducing the adverse effects on information decoding. Furthermore, we discussed the technical challenges and research directions for the further development of WPINs. We expect that this research will contribute to the realization of practical EH-enabled wireless networks in the future.
This work was supported by the National Research Foundation of Korea, funded by the Korea government (Grants 2021R1A2C4002024 and 2022R1A2C1011901). The corresponding authors are Hyun-Ho Choi and Woongsup Lee.
Kisong Lee (kisonglee@dongguk.edu) is an associate professor with the Department of Information and Communication Engineering, Dongguk University, Seoul 04620, South Korea. He received the Ph.D. degree in electrical engineering from the Korea Advanced Institute of Science and Technology, Daejeon, South Korea, in 2013. His research interests include network optimization, wireless power transfer, energy harvesting networks, information security, deep learning, and mobility optimization. He is a Member of IEEE.
Hyun-Ho Choi (hhchoi@hknu.ac.kr) is a professor with the school of Information and Communications Technology, Robotics, and Mechanical Engineering and the Research Center for Hyper-Connected Convergence Technology, Hankyong National University, Anseong 17579, South Korea. His research interests include wireless communication, wireless energy harvesting, bioinspired algorithms, distributed optimization, and machine learning. He is a Senior Member of IEEE.
Woongsup Lee (woongsup.lee@yonsei.ac.kr) is an associate professor with the Graduate School of Information, Yonsei University, Seoul 03722, South Korea. His research interests include deep learning, cognitive radio, future wireless communication systems, and smart grid systems. He is a Member of IEEE.
Victor C. M. Leung (vleung@ieee.org) is a distinguished professor with the College of Computer Science and Software Engineering, Shenzhen University, Shenzhen 518060, China. He is also an emeritus professor with the Department of Electrical and Computer Engineering, University of British Columbia, Vancouver, BC, Canada. His research interests include wireless networks and mobile systems. He is a Life Fellow of IEEE.
[1] X. Liu and N. Ansari, “Toward green IoT: Energy solutions and key challenges,” IEEE Commun. Mag., vol. 57, no. 3, pp. 104–110, Mar. 2019, doi: 10.1109/MCOM.2019.1800175.
[2] Z. Zhang, H. Pang, A. Georgiadis, and C. Cecati, “Wireless power transfer — An overview,” IEEE Trans. Ind. Electron., vol. 66, no. 2, pp. 1044–1058, Feb. 2019, doi: 10.1109/TIE.2018.2835378.
[3] J. Huang, J. Cui, C.-C. Xing, and H. Gharavi, “Energy-efficient SWIPT-empowered D2D mode selection,” IEEE Trans. Veh. Technol., vol. 69, no. 4, pp. 3903–3915, Apr. 2020, doi: 10.1109/TVT.2020.2970235.
[4] G. Zhang et al., “Hybrid time-switching and power-splitting EH relaying for RIS-NOMA downlink,” IEEE Trans. Cogn. Commun. Netw., vol. 9, no. 1, pp. 146–158, Feb. 2023, doi: 10.1109/TCCN.2022.3216406.
[5] T. D. P. Perera, D. N. K. Jayakody, I. Pitas, and S. Garg, “Age of information in SWIPT-enabled wireless communication system for 5GB,” IEEE Wireless Commun., vol. 27, no. 5, pp. 162–167, Oct. 2020, doi: 10.1109/MWC.001.2000001.
[6] C. Jeong and H. Son, “Cooperative transmission of energy-constrained IoT devices in wireless-powered communication networks,” IEEE Internet Things J., vol. 8, no. 5, pp. 3972–3982, Mar. 2021, doi: 10.1109/JIOT.2020.3027101.
[7] S. Gong, S. Ma, C. Xing, and G. Yang, “Optimal beamforming and time allocation for partially wireless powered sensor networks with downlink SWIPT,” IEEE Trans. Signal Process., vol. 67, no. 12, pp. 3197–3212, Jun. 2019, doi: 10.1109/TSP.2019.2912876.
[8] W. Lee, K. Lee, H. H. Choi, and V. C. Leung, “Deep learning for SWIPT: Optimization of transmit-harvest-respond in wireless-powered interference channel,” IEEE Trans. Wireless Commun., vol. 20, no. 8, pp. 5018–5033, Aug. 2021, doi: 10.1109/TWC.2021.3065029.
[9] K. Lee and H. H. Choi, “Weighted sum rate maximization for two-way wireless powered interference channels,” IEEE Trans. Veh. Technol., vol. 71, no. 5, pp. 5627–5632, May 2022, doi: 10.1109/TVT.2022.3154020.
[10] W. Shin, M. Vaezi, J. Lee, and H. V. Poor, “Cooperative wireless powered communication networks with interference harvesting,” IEEE Trans. Veh. Technol., vol. 67, no. 4, pp. 3701–3705, Apr. 2018, doi: 10.1109/TVT.2017.2785318.
[11] J. Liang, D. Feng, C. He, G. Qian, C. Guo, and N. Zhang, “Joint time and power allocation in multi-cell wireless powered communication networks,” IEEE Access, vol. 7, pp. 43,555–43,563, Mar. 2019, doi: 10.1109/ACCESS.2019.2908240.
[12] W. Liu, “Harmonious wireless networks: Perspective of interference management,” Intell. Converged Netw., vol. 2, no. 3, pp. 198–204, Sep. 2021, doi: 10.23919/ICN.2021.0013.
[13] P. Xu, D. Flandre, and D. Bol, “Analysis, modeling, and design of a 2.45-GHz RF energy harvester for SWIPT IoT smart sensors,” IEEE J. Solid-State Circuits, vol. 54, no. 10, pp. 2717–2729, Oct. 2019, doi: 10.1109/JSSC.2019.2914581.
[14] E. Boshkovska, D. W. K. Ng, N. Zlatanov, and R. Schober, “Practical non-linear energy harvesting model and resource allocation for SWIPT systems,” IEEE Commun. Lett., vol. 19, no. 12, pp. 2082–2085, Dec. 2015, doi: 10.1109/LCOMM.2015.2478460.
[15] A. M. Elbir and S. Coleri, “Federated learning for channel estimation in conventional and RIS-assisted massive MIMO,” IEEE Trans. Wireless Commun., vol. 21, no. 6, pp. 4255–4268, Jun. 2022, doi: 10.1109/TWC.2021.3128392.
Digital Object Identifier 10.1109/MVT.2023.3306552