Future Communications and Energy Management in the Internet of Vehicles (IoV) PDF

FUTURE COMMUNICATION TRENDS TOWARDS INTERNET OF THINGS SERVICES AND APPLICATIONS Future Communications and Energy Management in the Internet of Vehicles: Toward Intelligent Energy-Harvesting Xiaojie Wang, Zhaolong Ning, Xipin...

FUTURE COMMUNICATION TRENDS TOWARDS INTERNET OF THINGS SERVICES AND APPLICATIONS Future Communications and Energy Management in the Internet of Vehicles: Toward Intelligent Energy-Harvesting Xiaojie Wang, Zhaolong Ning, Xiping Hu, Lei Wang, Lei Guo, Bin Hu, and Xinyu Wu Abstract hours, respectively. IoV supports several types of communication patterns, such as vehicle-to-ve- As an emerging communication platform in hicle (V2V), vehicle-to-infrastructure (V2I), vehi- the Internet of Things, IoV is promising to pave cle-to-sensor (V2S), and vehicle-to-pedestrian the way for the establishment of smart cities and (V2P). Currently, automobile exhaust emission is provide support for various kinds of applications a major factor affecting human environments and and services. Energy management in IoV has been air conditions. As early as 1970, a Los Angeles attracting an upsurge of interest in both academia photochemical smog episode occurred, caused and industry. Currently, green IoV mainly focus- by the exhaust emission of more than 2.5 billion es on two aspects: energy management of bat- vehicles. After that, many countries have issued tery-enabled RSUs and EVs. However, these two and implemented relevant laws as well as regu- issues are always resolved separately while ignor- lations to strengthen the management of auto- ing their interactions. This standalone design may mobile exhaust emission. For example, Europe cause energy underutilization, a mismatch between has enforced the regulation of Euro I-VI, with the traffic demands and energy supplies, as well as purpose of limiting emissions of NOx, HC, and high deployment and sustainable costs for RSUs. CO. For CO 2 emissions, a target of 130 g/km Therefore, the integration of energy management was realized in 2015 by the European Commis- between battery-enabled RSUs and EVs calls for sion, and 95 g/km will be reached in 2021. comprehensive investigation. This article first pro- Other countries, including China, Japan, and the vides an overview of several promising research United States, have also made similar policies in fields for energy management in green IoV sys- their automotive markets. Therefore, an upsurge tems. Given the significance of efficient commu- of interest has arisen in the establishment of green nications and energy management, we construct IoV systems all over the world, aiming to relieve an intelligent energy-harvesting framework based the environmental pollution by taking advantage on V2I communications in green IoV communica- of electric vehicles (EVs) and renewable energy tion systems. Specifically, we develop a three-stage sources. Stackelberg game to maximize the utilities of both Since IoV systems can provide drivers and RSUs and EVs in V2I communications. After that, passengers with various kinds of vehicular appli- a real-world trajectory-based performance evalua- cations, such as location-aware road services, tion is provided to demonstrate the effectiveness of autonomous driving, and in-car entertainment, our scheme. Finally, we identify and discuss some wireless traffic demands from vehicles to the Inter- research challenges and open issues for energy net are increasing rapidly. A number of battery-en- management in green IoV systems. abled roadside units (RSUs) have been deployed along main roads to boost network capacity. For Introduction example, some RSUs can cache contents in their With the increasing number of intelligent vehi- local buffers and deliver the required contents to cles and the integration of sensors, the Internet passing vehicles without fetching through back- of Vehicles (IoV) has become a promising com- hauls. Due to the disconnection between smart munication platform of the Internet of Things, and grid and RSUs in some rural areas, energy har- is acknowledged as the fundamental technology vesting policies need to be designed for RSUs to to construct intelligent transportation systems for serve the huge service requests from vehicles. A smart cities. Modern transportation systems are promising solution is to enable wind or solar-pow- promising to bring convenience to citizens’ daily ered RSUs in an energy-constrained vehicular life through services and applications of enhanced environment. The Department of Transportation mobility and improved safety on roads. Accord- of the United States has predicted that solar-pow- ing to the results published by Juniper Research ered RSUs will dominate 40 percent of all rural UK 2018, smart cities can help citizens save freeway RSUs by 2050. 125 hours per year on average, among which Some studies have investigated downlink traffic mobility and public safety can save 60 and 35 scheduling and service request routing strategies Xiaojie Wang, Zhaolong Ning, Xiping Hu, and Bin Hu are with Lanzhou University; Xiaojie Wang, Zhaolong Ning, and Lei Wang are with Dalian University of Technology; Zhaolong Ning and Lei Guo are with Chongqing University of Posts and Telecommunications; Digital Object Identifier: Xiping Hu and Xinyu Wu are with Chinese Academy of Science. 10.1109/MWC.001.1900009 IEEE Wireless Communications December 2019 1536-1284/19/$25.00 © 2019 IEEE 87 NING_LAYOUT.indd 87 12/19/19 3:09 PM  EVs will play a to reduce energy consumption of RSUs. Howev- sumption optimization by scheduling the require- dominant role in er, these schemes merely consider the situation in ments among different RSUs is necessary when a the next-generation which RSUs are a unique kind of energy consump- set of RSUs coexist. Another promising solution vehicular system for tion components in the whole network, regardless is to make RSUs turn on and off in a periodic of other energy consumption terminals, such as EVs manner so that the overall energy consumption smart and green city and sensors. The highly dynamic traffic demands can be reduced. However, a minimum number construction. Charging and intermittent renewable energy supplies may of active RSUs should be set to maintain the net- management for EVs is cause energy underutilization, a mismatch between work operation and connectivity. important to achieve traffic demands and energy supplies, as well as high efficient energy man- deployment and sustainable costs for RSUs. Energy Harvesting for RSUs Therefore, the combined energy management pol- Since smart grid cannot be connected with RSUs agement. Its challenges icy for various energy-consumption components in in some rural areas, taking advantage of renew- mainly contain two IoV systems calls for deep investigation. able energy sources is an attractive approach to aspects: selecting a This article mainly focuses on two energy support the operation of RSUs. Specifically, wind charging station to consumption components in IoV systems: bat- and solar energy can be commonly used for elec- design a reasonable tery-enabled RSUs and EVs. To the best of our tronic power generation. Solar-powered batteries knowledge, this is the first work to provide a com- can be embedded in RSUs to convert solar ener- charging plan, and con- prehensive energy management strategy by joint- gy to electricity. A rechargeable large battery structing an efficient ly integrating these two network components. is necessary since solar panels cannot directly pro- communication frame- Specifically, an energy-harvesting framework is vide energy for RSUs. In the discharge circle, a work between EVs and constructed to support green IoV systems. We downlink scheduling scheme is carried out to con- the power grid. first provide an overview of several promising trol the energy consumption for communications research aspects for energy management in IoV between RSUs and vehicles. Wind-powered RSUs systems, and classify them into four categories. are utilized in , and the minimum size of stor- Following that, we establish an energy-harvesting age battery is analyzed based on a power con- framework based on V2I communications in IoV sumption model. When the energy is insufficient, systems, by which RSUs can make intelligent deci- some RSUs will go into a sleep cycle. In addition, sions on energy harvesting to satisfy the required it is cost-effective to leverage radio frequency services. In order to satisfy the utilities of both EVs (RF) energy transfer technology to transfer energy and RSUs, we further design an energy-harvesting from passing vehicles to RSUs. strategy between EVs and RSUs based on a three- stage Stackelberg game to balance the benefits EV Charging Based on Vehicle-Grid Technology among participants. A real-world trajectory-based EVs will play a dominant role in the next-gener- case study is performed to evaluate the effective- ation vehicular system for smart and green city ness of our framework. Finally, we discuss some construction. Charging management for EVs is open issues and directions for energy manage- important to achieve efficient energy manage- ment in green IoV systems to provide a guideline ment. Its challenges mainly contain two aspects: for future studies. selecting a charging station to design a reason- able charging plan, and constructing an efficient Overview of Energy Management in communication framework between EVs and the power grid. A communication framework based Green IoV Systems on a publish/subscribe mode is designed in. Energy management in IoV systems contains var- Considering traffic conditions and drivers’ pref- ious technologies, for example, power plants, erences, charging stations periodically broadcast energy storage systems, efficient communication their conditions (e.g., the number of charging EVs protocols, and flexible power management. In and residual electric power) to RSUs. EVs can the following, we mainly discuss four promising obtain the published information when traveling research aspects for energy management in green into the communication range of an RSU and IoV systems. choose a suitable charging station. An optimiza- tion problem is formulated to configure electric Energy-Saving in RSU Scheduling power capacities for both EVs and the smart grid RSUs are fundamental infrastructures in vehicu- in , where EVs can not only purchase energy lar networks because they can provide network from the power grid, but also sell energy in turn. access for vehicles in both highway and urban A cooperative optimization scheme based on environments. Currently, battery-enabled RSUs improved particle swarm optimization is proposed are deployed along roads in many countries. to balance the benefits of EVs and the power grid, With the limited number of RSUs and contin- considering different prices of electric power in uous traffic requirements of vehicles, how to different space and time. make RSUs conserve their battery power until the next charging cycle to serve vehicular com- Wireless Power Transfer for EVs munication requirements deserves investigation. Traditional power transfer stations are based on For battery-enabled RSUs, downlink traffic sched- wired charging technologies, making drivers stop uling is promising to reduce energy consumption their EVs and spend several hours to charge.. The energy consumption can be significantly Many drivers cannot tolerate the long charging lowered if an RSU communicates with a near- time, and desire to find other possible charging by vehicle instead of one far away. Therefore, modes. Wireless power transfer technologies are efficient task scheduling is important to serve promising to overcome this drawback and allow vehicular communication requirements by mak- EVs to charge in a wireless manner even while ing the RSU always communicate with a nearby moving. The energy conversion efficiency of RF vehicle. In addition, the overall energy con- energy transfer can reach up to 85 percent prac- 88 IEEE Wireless Communications December 2019 NING_LAYOUT.indd 88 12/19/19 3:09 PM  Definition Description Wind turbines Solar panels r v Electric power prices of an RSU and an EV at St , St time t, respectively The amount of energy sold to an EV and an Pi, Pr RSU, respectively RSU communication ranges d m Deployment and maintain costs for an RSU, Ci , Ci respectively Smart grid g g The electric power price of smart grid after t0 St+t0, St–t1 time and before t1 time, respectively Production & Management Charging efficiency for RSUs and smart grid, a, b Information services respectively Ur, Uv Utilties of an RSU and an EV, respectively TABLE 1. Main notations. Electric supplies Charging facilities tically. Currently, the design of wireless power transfer systems has drawn great attention. An efficient wireless power transfer system is con- structed in , including four coils for wireless FIGURE 1. A communication sketch for green IoV systems. charging. A source coil and two transmitter coils are installed to create and boost an electromag- netic field on the power supply equipment. An ing EVs, and minimize their costs by purchasing induction coil is installed in the EV to receive elec- energy from EVs when their energy is in short- tric power from the electromagnetic field. One age; on the other hand, EVs prefer to maximize key challenge for wireless power transfer is how their cost savings when purchasing energy from to improve the charging efficiency for EVs. RSUs, and maximize their benefits by selling ener- gy to RSUs. To resolve the contradiction of utility Green IoV Framework between RSUs and EVs as well as balance their To the best of our knowledge, the study on joint energy, an energy-harvesting strategy, leveraging optimization of energy management between bat- a three-stage Stackerberg game, is designed based tery-enabled RSUs and EVs is still vacant. With the on V2I communications. objective of both satisfying the required service requirements of EVs and reaching Nash equilib- System Model rium on the benefits of both EVs and RSUs, we The vehicular communication requirement design an intelligent energy-harvesting framework, flow arriving at an RSU is considered to follow named IEAF, based on V2I communications. a Poisson process. The RSU can be viewed as a server processing requirements from EVs, Framework Overview and modeled as an M/G/1/n queueing system. As shown in Fig. 1, solar panels and wind tur- The queueing and processing time for a service bines are two major renewable energy sourc- requirement from EVs can be obtained by queue- es to generate energy for smart grid. RSUs are ing theory. In addition, each RSU is equipped with equipped with wind turbines, and generally go a wind turbine to provide energy. Therefore, the through three stages in one battery cycle, that is, harvested energy via wind turbines can be deter- electric powers at high level, mid level and low mined by the wind speed, air density, and circular level, respectively. When they are in a high-level turbine cross-section. The harvested power of situation, the energy generation speed of wind the RF energy transfer can be obtained based on turbines is higher than the energy consumption RF energy propagation models. The main speed of RSUs. Under this circumstance, RSUs notations of our framework are summarized in have sufficient electricity and can sell redundant Table 1. energy to passing EVs through RF energy trans- When the electric power of an RSU is in fer technology. In other words, EVs can not only high-level, it has sufficient electricity to support its obtain electric power from charging facilities, but operations. When a traffic requirement arrives, it is also purchase electric power from RSUs on their stored in the waiting queue of RSUs for processing routes. When RSUs are in mid-level and low-level regardless of the electric power consumption. In situations, the energy generation speed of wind addition, RSUs can sell redundant electric currency turbines is lower than the energy consumption to passing EVs through RF energy transfer. There- speed of RSUs, and they can purchase energy fore, RSUs’ utilities in this stage can be leveraged from nearby EVs to support their operations. Both to minimize their maintenance costs, equivalent in high-level and mid-level situations, RSUs serve to maximizing their benefits, that is, Ur = Str  Pi all the requirements of EVs. When the electricity – Cid – Cim. Herein, Str is the RSU’s energy price at of RSUs is in a low-level situation, they will only time t. The amount of energy sold to an EV is Pi, serve the requirements of EVs. which should be less than the remaining energy On one hand, RSUs intend to maximize their provided by the RSU’s battery. Variable Cid is the benefits through selling redundant energy to pass- deployment cost equally apportioned for a deal, IEEE Wireless Communications December 2019 89 NING_LAYOUT.indd 89 12/19/19 3:09 PM  Stage I Stage II Stage III Pw Pw Pw Uploading requests Price Str Charging Uploading Charging based on a Charging efficiency α requests efficiency α successful efficiency α Uploading transaction requests Acquisition Total sales Pr Acquisition Total sales Total sales Pr αPr αPr Pi Acquisition αPi Price Stv Price Stv FIGURE 2. Interactions based on V2I communications: An RSU sells energy to an EV in Stage I; the RSU purchases energy from an EV in Stage II; an EV provides energy to enable the RSU to process its service requests in Stage III. while Cim is the maintenance cost for the deal. The IntellIgent energy hArvestIng unit deployment cost can be calculated by the According to the analysis above, we notice that total deployment cost divided by the battery life the utilities of RSUs and EVs always conflict with of RSUs. The deployment cost for a deal can be each other. In order to balance their benefits, we computed by the unit deployment cost times the propose an intelligent energy-harvesting strategy duration from the last deal to this deal in the cur- based on a three-stage Stackerberg game, and rent stage. Similarly, the maintenance cost for the their interaction is shown in Fig. 2. The corre- deal can be obtained. sponding process can be described as follows: For an EV, it can purchase energy from an RSU. Stage I illustrates the electric power of an RSU Its utility function is to maximize cost saving, that in high-level, and the RSU is the leader in the Stack- g g is, Uv = (S t+t0  Pi  ab – Str  Pi, where S t+t0 is erberg game that offers a price to EVs. If the price the charging price set by the smart grid when the is lower than that of the smart grid, EVs can pur- EV prepares to charge after time interval t0. The chase electric power from the RSU. We can obtain charging efficiencies of an EV charging from an the corresponding Nash equilibrium condition A ≤ d m g RSU and the smart gird are denoted by a and b, Str* ≤ B, where A = (Ci + Ci )/Pi and B = aS t+t0/b. respectively. Stage II shows the electric power of the RSU in When the electric power of an RSU is in mid-level, in which an EV plays the role of leader mid-level, it stops selling its electric currency and in the Stackerberg game and offers a price to the g begins to purchase electric power from passing RSU. If price Stv* satisfies D ≤ Stv* ≤ E, where D = S t– d m EVs to store energy for service management. t1/b and E = (Cj + Cj )/Pr, a Nash equilibrium can Under this situation, it serves all the requirements be reached, and the RSU buys the electric power submitted by the passing EVs and purchases ener- from the EV. gy from them if possible. The corresponding utility Stage III is in a low-level situation, and the RSU d m is Ur = Stv  Pr – Cj – Cj , where Stv is the price for merely serves EVs that can provide enough energy electric power sold by an EV at time t, and Pr is the to process their requirements. Therefore, when an amount of electricity sold to the RSU. The installa- EV has a request to process, it should make a trade- d tion cost for this deal is denoted by Cj , and can be off between its benefit and the response delay. If a calculated by the accumulation of the installation request is delay-tolerant, the EV can choose other cost for each unit based on the time for generating RSUs to process its requests later if its benefit can- m Pr energy by wind turbines. Similarly, C j is the not be satisfied. In this situation, the Nash equilibri- maintenance cost for this transaction. um condition is the same as that in Stage II if a deal For an EV, if its electric power is sufficient to can be made between the EV and the RSU. If the support the corresponding operations until arriving request is delay-sensitive, the EV needs to satisfy at its destination, it can sell an amount of energy to the request regardless of its benefits. Therefore, the RSUs. Therefore, the objective of the EV becomes Nash equilibrium condition is Stv* ≤ E. g to maximize its benefits: Uv = Stv  Pr – S t–t1  Pr/b, g where S t–t1 is the price of the electric power when PerforMAnce evAluAtIon the EV purchases from a charging facility of smart We utilize a real-world trajectory of taxies in grid. Shanghai, China, from April 1, 2015 to April 30, When the electric power of an RSU is in 2015 for performance evaluation. We randomly low-level, it lacks energy. In order to maintain its place RSUs in Jingan district and obtain the aver- fundamental functions, the RSU merely serves the age performance by the Monte Carlo method. requirements from EVs, which can supply enough We obtain the information of wind speeds in energy to process the uploaded requirements. The Shanghai from , and model the hourly energy utility of the RSU is the same as that in the mid-level generated by winds according to Gaussian dis- situation. For the EV, whether its requirements can tribution similar to. We also set the average be served or not depends on the energy supply. If packet size as 867.4 bytes, the energy consump- not enough energy can be supplied for the RSU to tion per bit as 2.92  10 –6) J, and the energy process its requirements, another RSU needs to be capacity of an RSU battery as 262 kJ as in. selected, causing a large delay for the response of We consider the signal-to-interference-plus-noise its requirements. Therefore, the utility of EV in this ratio (SINR)-based channel model, where the stage is to minimize the response delay for the cor- cross gain is related to the Euclidean distance and responding requirements. channel fluctuations. 90 IEEE Wireless Communications December 2019 NING_LAYOUT.indd 90 12/19/19 3:09 PM  The packet blocking probabilities of our solu- tion and the sleep-based solution are demon- strated in Fig. 3a. Packet blocking probability is defined as the number of requirements dropped by the RSU when its waiting queue is full com- pared to the number of total requirements uploaded by EVs. The sleep-based solution is an energy-harvesting scheme based on wind turbines and allows RSUs to sleep at points when the energy generation speed is lower than the energy consumption speed. We can observe that the packet blocking probability of our scheme is lower than that of the sleep-based solution, especially during the peak time, that is, 8:00–9:00 and 17:00–18:00. The reason is that our solution allows RSUs to purchase energy from EVs when (a) they have insufficient energy, while RSUs go into sleep cycles when they are lacking energy in the sleep-based solution. Average residual energy of RSUs is compared in Fig. 3b. We can observe that the performance of our scheme is better since it can balance ener- gy between EVs and RSUs. However, RSUs are forced to sleep when they have insufficient energy in the sleep-based solution. During peak time, the sleep-based solution allows the RSU to go to sleep while our algorithm serves passing EVs with its best efforts. Consequently, more requirements can be satisfied by our algorithm, and more energy is con- sumed compared to the sleep-based solution. Figure 4 shows the benefits of RSUs and EVs. The benefit of RSUs is defined as the total earn- ing minus the total maintenance and deployment (b) costs during a time period. The benefit of EVs illustrates the total cost savings plus the total FIGURE 3. Performance evaluation: a) packet blocking probability; b) residual earning. The normalized benefits of vehicles and energy. RSUs are unified values, reflecting the gaps of their achieved results. We notice that the benefit of RSUs is high when the time is between 12:00 Currently, many researchers have focused on and 15:00, since the wind energy is sufficient encoding compression cost and reducing energy and the service demands of EVs are low. RSUs consumption to support energy-constrained devic- can harvest sufficient energy and sell redundant es. However, the encryption and compression energy to EVs. technologies always increase the computational For EVs, their benefits are high when the energy overhead and consume a lot of energy. Therefore, of RSUs is insufficient, since EVs can sell their ener- low-complexity cryptographic algorithms need to gy to RSUs with the purpose of supporting their be designed to reduce the encryption overhead normal operations. In addition, when the energy and energy consumption of devices. In addi- generation speed is higher than the energy con- tion, a trade-off is required between the algorithm sumption speed between 13:00 and 15:00, RSUs complexity and security to satisfy various network can sell energy to EVs at a low price, increasing requirements. vehicular cost savings. However, the sleep-based solution cannot increase benefits of RSUs and EVs, Dynamic Power Transfer since it does not consider an energy-harvesting For a stationary charging system, drivers general- scheme between EVs and RSUs. ly park their cars at a charging facility and leave. With the increasing number of EVs and the long Research Challenges and Open Issues time consumed for charging, dynamic power Many challenges and open issues still exist for transfer is a promising technology to cope with energy management in green IoV systems. In the the above-mentioned problems. It allows EVs following, we discuss them from several aspects. to charge during their travels regardless of their movements. Currently, dynamic power transfer Energy-Aware Security Protocol faces the following challenges: Security protocols are necessary for communi- How to develop high-efficiency power transfer cations in IoV systems to protect individual pri- technologies to improve the transmission effi- vacy. Generally, a high-level security mechanism ciency, such as innovative system designs and requires high computational complexity and circuits, deserves to be well studied. consumes much energy, which challenges ener- How to reduce the cost for the establishment of gy-constrained devices. Therefore, energy-efficient the dynamic power transfer system also needs security protocols call for investigation. Howev- to be investigated, because the high cost may er, the design of energy-aware security protocols pose a constraint on the popularization among for green IoV systems is still in its initial phase. drivers. IEEE Wireless Communications December 2019 91 NING_LAYOUT.indd 91 12/19/19 3:09 PM  and residential regions. V2V energy swapping is a feasible solution to relieve heavy loads on the power grid with the help of a connected aggre- gator. Therefore, efficient energy swapping strat- egies are necessary to guarantee the energy utilization. For example, how to stimulate EVs to participate in the process of energy supply to bal- ance power demands at aggregators needs to be further investigated. In addition, online ener- gy management protocols are desired to make a rapid match among EVs for supply and demand. Conclusion This article first emphasizes research significance of energy management in green IoV communication (a) systems and presents several promising research aspects for energy management in IoVs. Then an intelligent energy-harvesting framework based on V2I communications is constructed to both sat- isfy the required services of EVs and reach Nash equilibrium on the utilities of RSUs and EVs. Perfor- mance evaluation illustrates that our framework can both meet the service demands of EVs and largely increase the benefits of both EVs and RSUs. Finally, we discuss several existing research challenges and open issues to provide a guideline for further work. Acknowledgments The work is supported by the National Nature Sci- ence Foundation of China under Grant 61971084, Grant 61632014, and Grant 61802159, the China Postdoctoral Science Foundation under Grant (b) 2018T110210, the National Natural Science Foun- FIGURE 4. Performance evaluation: a) normalized RSU benefit; b) normalized dation of Chongqing under Grant cstc2019jcyj-msx- vehicular benefit. mX0208, and the Fundamental Research Funds for the Central Universities under Grant DUT19JC18. How to reduce the mobility impact of EVs during References the charging cycle is challenging, because C. M. Martinez et al., “Energy Management in Plug-in Hybrid Electric Vehicles: Recent Progress and a Connected speeds and driving directions of EVs mainly Vehicles Perspective.,” IEEE Trans. Vehic. Tech., vol. 66, no. depend on traffic conditions and driving habits. 6, 2017, pp. 4534–49. How to seamlessly switch among different R. Atallah et al., “Energy Harvesting in Vehicular Networks: A charging facilities without drivers’ interactions is Contemporary Survey,” IEEE Wireless Commun., vol. 23, no. 2, 2016, pp. 70–77. also challenging. Z. Ning et al., “Mobile Edge Computing-Enabled Internet of Vehicles: Toward Energy-Efficient Scheduling, IEEE Network, Green Edge Computing vol. 33, no. 5, Sept./Oct. 2019, pp. 198-205.. A significant challenge caused by connections J. He et al., “Delay Minimization for Data Dissemination in Large-Scale VANETs with Buses and Taxis,” IEEE Trans. and communications between end terminals and Mobile Computing, vol. 15, no. 8, 2016, pp. 1939–50. cloud servers is the energy efficiency of green Z. Ning et al., “Green and Sustainable Cloud of Things: IoV systems. One promising solution is to sched- Enabling Collaborative Edge Computing,” IEEE Commun. ule tasks among several servers to minimize the Mag., vol. 57, no. 1, Jan. 2019, pp. 72–78. R. Atallah et al., “A Reinforcement Learning Technique energy consumption in the data center. Another for Optimizing Downlink Scheduling in an Energy-Limited choice is to jointly optimize radio and computa- Vehicular Network,” IEEE Trans. Vehic. Tech., vol. 66, no. 6, tional resources. However, most current solutions 2017, pp. 4592–4601. cannot be directly applied in vehicular networks A. Muhtar et al., “Greening Vehicular Networks with Stand- alone Wind Powered RSUs: A Performance Case Study,” due to the highly dynamic network topology. IEEE ICC, 2013, pp. 4437–42. Therefore, energy-efficient edge computing or Y. Cao et al., “A Cost-Efficient Communication fog computing in vehicular networks is desired. Framework for Battery-Switch-Based Electric Vehicle For instance, how to offload traffic from a remote Charging,” IEEE Commun. Mag., vol. 55, no. 5, May 2017, pp. 162–69. cloud to nearby fog nodes by exploiting V2I com- C. Chen et al., “Cooperative Optimization of Electric Vehi- munications has a significant impact on energy cles in Microgrids Considering Across-Time-and-Space Ener- management in IoV systems. In addition, how gy Transmission,” IEEE Trans. Industrial Electronics, vol. 66, to utilize free resources of vehicles to form fog no. 2, 2019, pp. 1532–42. S. C. Moon et al., “Wireless Power Transfer System with an nodes and construct an energy-efficient model is Asymmetric Four-Coil Resonator for Electric Vehicle Battery also challenging. Chargers,” IEEE Trans. Power Electronics, vol. 31, no. 10, 2016, pp. 6844–54. V2V Energy Swapping Strategy X. Lu et al., “Wireless Networks with RF Energy Harvesting: A Contemporary Survey,” IEEE Commun. Surveys & Tutorials, Due to the increasing number of EVs and their vol. 17, no. 2, 2015, pp. 757–89. charging demands, the power grid may be over- http://data.cma.cn/data/cdcdetail/dataCode/A.0029.0001 loaded during peak times, especially in business _310.html. 92 IEEE Wireless Communications December 2019 NING_LAYOUT.indd 92 12/19/19 3:09 PM  W. Kumar et al., “Energy and QoS Evaluation for a V2R Lei Wang ([email protected]) is currently a full professor in Network,” Proc. IEEE GLOBECOM, 2011, pp. 1–5. the School of Software, Dalian University of Technology. He X. Wang et al., “Privacy-Preserving Content Dissemination received his B.S., M.S., and Ph.D. from Tianjin University, China. for Vehicular Social Networks: Challenges and Solutions,” He was a member of technical staff with Bell Labs Research IEEE Commun. Surveys & Tutorials, vol. 21, no. 2, 2019, pp. China (2001–2004), a senior researcher with Samsung, South 1314–45. Korea (2004–2006), and a research associate at Washington X. Wang et al., “Offloading in Internet of Vehicles: A Fog-En- State University (2007–2008). His research interests involve abled Real-Time Traffic Management System,” IEEE Trans. wireless ad hoc networks, social networks, and network security. Industrial Informatics, vol. 14, no. 10, 2018, pp. 4568–78. He is a co-corresponding author of this article. Biographies Lei Guo ([email protected]) received his Ph.D. degree from the University of Electronic Science and Technology of China, X iaojie W ang received her M.S. degree from Northeastern University, China, in 2011, and received her Ph.D. degree from Chengdu, in 2006. He is currently a full professor at Chongqing Dalian University of Technology, China, in 2019. From 2011 University of Posts and Telecommunications, China. He has to 2015, she was a software engineer at NeuSoft Corporation, authored or coauthored more than 200 technical papers in China. She is a research associate at Lanzhou University. Her international journals and conferences. He is an Editor for sev- research interests are vehicular networks, edge computing, and eral international journals. His current research interests include resource management. She has published over 30 scientific communication networks, optical communications, and wireless papers in the above areas. communications. He is a co-corresponding author of this article. Z haolong N ing [M’14, SM’18] ([email protected]) Bin Hu [M’10-SM’15] ([email protected]) is currently a professor received his M.S. and Ph.D. degrees from Northeastern Univer- at Lanzhou University, an adjunct professor at Tsinghua Univer- sity in 2011 and 2014, respectively. He is an associate professor sity, China, and a guest professor at ETH Zurich, Switzerland. at Dalian University of Technology and an adjunct professor at He is an IET Fellow, a member-at-large of ACM China, Vice Lanzhou University, China. He has published over 100 scientific President of the International Society for Social Neuroscience papers in international journals and conferences. His research (China Committee), and so on. He has published more than 100 interests include mobile edge computing, vehicular networks, papers in peer reviewed journals, conferences, and book chap- and network optimization. He is a co-corresponding author of ters. He is a co-corresponding author of this article. this article. Xinyu Wu ([email protected]) is now a professor at Shenzhen Xiping Hu ([email protected]) is currenlty a professor at Lanzhou Institutes of Advanced Technology, and director of the Center University. He has over 70 papers published and presented in for Intelligent Bionic. He received his B.E. and M.E. degrees prestigious journals and conferences. He has served as the lead from the Department of Automation, University of Science and Guest Editor of IEEE Transactions on Automation Science and Technology of China in 2001 and 2004, respectively. His Ph.D. Engineering, WCMC, and so on. His research areas consist of degree was awarded from the Chinese University of Hong Kong mobile cyber-physical systems, crowdsensing, and social net- in 2008. He has published over 180 papers and two mono- works. He holds a Ph.D. from the University of British Columbia, graphs. His research interests include computer vision, robotics, Vancouver, Canada. He is a co-corresponding author of this and intelligent systems. He is a co-corresponding author of this article. article. IEEE Wireless Communications December 2019 93 NING_LAYOUT.indd 93 12/19/19 3:09 PM