Machine to Machine (M2M) Communications in Virtualized Vehicular Ad Hoc Networks

Machine to Machine (M2M) Communications in Virtualized Vehicular Ad Hoc Networks


With the growing interest in the use of internet of things (IoT), machine-to-machine (M2M) communications have become an important networking paradigm. In this paper, with recent advances in wireless network virtualization (WNV), we propose a novel framework for M2M communications in vehicular ad-hoc networks (VANETs) with WNV. In the proposed framework, according to different applications and quality of service (QoS) requirements of vehicles, a hypervisor enables the virtualization of the physical vehicular network, which is abstracted and sliced into multiple virtual networks. Moreover, the process of resource blocks (RBs) selection and random access in each virtual vehicular network is formulated as a partially observable Markov decision process (POMDP), which can achieve the maximum reward about transmission capacity. The optimal policy for RBs selection is derived by virtue of a dynamic programming approach. Extensive simulation results with different system parameters are presented to show the performance improvement of the proposed scheme.

Machine-to-machine (M2M) communications; vehicular ad-hoc networks (VANETs); random access; wireless network virtualization (WNV).



H.4Information Systems ApplicationsMiscellaneous

1 Introduction

In recent years, with the rapid development of internet of things (IoT) and wireless network technologies, machine-to-machine (M2M) communications have been attracting lots of attention from industry and academia [1, 2, 3, 4, 5, 6]. European Telecommunications Standards Institute (ETSI) estimated that M2M principles apply particularly well to networks where a large number of machines are used, even up to the 1.5 billion wireless devices in the future [7]. Therefore, M2M communications are taken into account to create a revolution in our future IoT world [8].

As the main distinct characteristic from traditional wireless communications, a large number of M2M devices are usually involved in most applications and scenarios based on M2M communications [9]. Meanwhile, as one of the application scenarios, the vehicular networks also have the same feature. Since the number of vehicles has increased considerably every year, the vehicular networks have to be able to sustain an increasing number of vehicles [10, 11, 12]. Moreover, for the vehicular networks, the main characteristic is not fixed in size, and the network needs to be scalable to adapt to the continuously and diversely changing network topology due to the nodes’ high mobility [13, 14]. Furthermore, during each time slot, the data transmission with M2M communications in vehicular networks is usually deemed as small-sized, while the frequency of their making data connections is higher than traditional devices since they have specific applications and functions [15], e.g., Global Positioning System (GPS) in the vehicles. Therefore, how to support more vehicles simultaneously connecting and accessing to the network is an important and inevitable issue [16].

Several research efforts have investigated M2M communications and resource allocation for vehicular networks. In [17], the authors considered a heuristic resource allocation scheme in vehicular networks, which have significantly less computational complexity than the iterative scheme and negligible performance degradation. A cooperative control policy of collection of vehicles operating is proposed by [18] in the vehicular ad-hoc network (VANET) with access-constrained fading channels. In [19], the authors proposed an optimization policy for joint resource allocation problem about resource blocks (RBs) and power in moving small cells, with the goal of enhancing the quality of service (QoS), and solved the optimization problem based on iterative resource allocation algorithm (IRAA).

With recent advances in wireless communications and networks [20, 21, 22, 23, 2, 24, 25, 26, 27, 28, 9, 29, 30, 31, 32, 33, 34, 35, 19, 36, 37, 38, 39, 40, 41, 42, 11, 43, 44, 45, 46, 47], some excellent works have been done in the vehicular networks with M2M communications. However, most existing researches ignore an important issue that the vehicles may fail to access the network if there is no enough radio resource allocated to the random access (RA) process [48]. In addition, only one class of vehicular networks is considered in most existing works. However, in practical networks, there are multiple classes of vehicles with different applications or services with M2M communications, such as military applications, emergency services, environmental monitoring, safety and collision avoidance for driverless cars, etc. [49]. Obviously, different applications have different QoS requirements, and they should be treated differently. 

In this paper, with recent advances in wireless network virtualization (WNV) [27, 50, 51, 37], a novel framework for M2M communications in vehicular networks with WNV is proposed [52]. The vehicular networks’ capacity can be improved by M2M communications to support many characteristics of modern vehicles such as cross-platform networking, autonomous monitoring, and visualization of the devices and information [14]. Meantime, using WNV and software-defined networking techniques [53, 54, 55, 41], a physical vehicular network can be sliced into multiple virtual vehicular networks, then differentiated vehicular network services can be provided based on differentiated QoS requirements [50]. Moreover, a decision-theoretic approach to optimize the random access process of M2M communications can be developed. Specifically, RA process in M2M communications is formulated as a partially observable Markov decision process (POMDP). The vehicles can obtain the maximum transmission rate from the belief state in POMDP optimization policy, which encapsulates the history of system state and access decision.

The rest of this article is organized as follows. System model is presented in Section 2. In Section 3, an optimization policy for the random access process via POMDP formulation is proposed, and the optimal RBs selection scheme will be given. Simulation results are compared and analysed in Section 4. Finally, we conclude the proposed work in Section 5 with future works.

2 System Model

In this section, we firstly describe the system model for the vehicular network with M2M communications. Following the general frameworks of WNV, there are three parts in the system model, which are described as follows.

2.1 Physical Resource in Vehicular Network

Given the system model in Figure 1, a single-cell scenario with vehicles and one eNodeB is considered in the physical vehicular network [56]. Each vehicle is equipped with a machine-type communication devices (MTCDs). In the proposed system model, can be denoted the time point that the MTCDs can access the eNodeB, where the total number of time slots is represented as , . Then, denotes the duration of a time slot, it represents as . Moreover, one time period includes time slots, from time point to . Meanwhile, RBs will be offered when the MTCDs attempt to access the vehicular network. The number of RBs in random access phase denotes , and represents the -th RB in the access phase, where . The state of each RB in one time slot can be described as idle or busy. The set is represented as the state of the -th RB, and , where and stands for the RB is idle or busy, respectively. In this paper, we assume that the number of busy (or idle) RBs in each time slot satisfies Poisson distribution [57].

Figure 1: The architecture of M2M communications in a vehicular network with WNV.

In the access phase, different transmission rates can be offered by the -th RB after the -th MTCD has accessed. Then, denotes the transmission rate achieved by the -th MTCD in time slot , it can be represented as


where denotes the bandwidth offered by the -th RB, is the transmit power consumed by the -th RB, () represents the channel gain when the -th (-th) MTCD accesses to the network, which follows Gaussian distribution with zero mean and unit variance, and represents the noise power [5].

2.2 Wireless Network Virtualization

In order to realize WNV, the hypervisor is considered as an important component and needs to be set in the vehicular network. Normally, the hypervisor can be implemented at the physical eNodeB, and it provides functions to connect physical resource and virtual eNodeB [58]. Moreover, the hypervisor takes the responsibility of virtualizing the physical eNodeB into a number of virtual eNodeBs, such as virtual devices and RBs, which can be used by different mobile virtual network operators (MVNOs) [27]. Furthermore, the hypervisor is also used to scheduling the air interface resources for WNV.

In this paper, the physical vehicular network is abstracted and sliced into multiple virtual networks by the hypervisor according to different application functions or QoS requirements. As a result, it can develop the diverse class of virtual vehicular network, and each virtual network includes MTCDs that have same or similar applications.

2.3 Virtual Networks

As can be seen in Figure 1, the physical vehicular network with M2M communications can be sliced into virtual networks by hypervisor. For the -th virtual network, vehicles equipped with MTCDs are included in this virtual network, which have the same or similar functions. Meanwhile, in the -th virtual network, RBs can be offered by virtual eNodeB to random access transmission in the initial time slot. represents the number of RBs used for random access. Besides, the obtained transmission rate in each virtual network can be denoted as , where and represent the obtained average transmission rate in the highest or lowest level virtual network, respectively.

3 Optimization Policy of Random Access via POMDP

In this section, we formulate a decision-theoretic approach via POMDP to optimize the random access in the vehicular network. After that, each tuple of POMDP is described in detail, followed by the reward and optimization objective. Then, the solution of problem formulation will be given.

3.1 POMDP Formulation

Different from Markov modelling, POMDP has a distinctive characteristic since it emphasizes “partially observable”. In the proposed scheme, the random access can be formulated as a POMDP optimization problem because RBs states cannot be directly observed by MTCDs [59]. Based on the state of RBs that encapsulates the decision and observation history, the MTCDs will take an effective action in each time slot. In what follows, the random access process of vehicles equipped with MTCDs will be formulated as a POMDP optimization policy and given its solution. For simplicity, we discuss the POMDP formulation in the -th virtual network as an example.

1) Action Space

At the beginning of time slot , according to its current belief state, the -th vehicle equipped with MTCD will access to network and determine which action to select [60]. Let be the set of all available actions, and the action taken by this MTCD in time slot is denoted as


In set , denotes that the MTCD will not access the eNodeB and select sleeping mode, and represents that the MTCD will select the -th RB to access the network.

2) State Space and Transition Probabilities

The system state space is the set of all RB states in the vehicular network, and the state in time point is represented as , where [61, 44]. Moreover, the state of the -th RB can be represented as


We assumed that each RB state in the proposed scheme is discretized. Then, the transition probability of the -th RB state from state to state is represented as


Considering Poisson distribution, Eq. (4) can be calculated by


where denotes the occurred frequency of the state , and represents the total number of states varying from to in long-term statistics.

3) Observation Space

As mentioned above, the MTCD needs to observe the RB state because it is difficult to acquire the full knowledge of each RB state [59]. Assume that some MTCDs have selected RBs to access virtual network, it means that part of RBs are in busy state during this time slot. Hence, before accessing the network and making decision, the MTCD needs to observe the RB state. Let be the observation state of the -th RB in time slot , where . can be given as


Then in time slot , the observation state can be represented as , where , and the set of all observation states can be denoted as .

Then, the probability of observation state can be denoted as


According to the action that be taken by MTCD and the observation state, the conditional probability of observation is calculated as


where and denote the probabilities of false observation, respectively.

4) Belief State

For POMDP optimization policy, belief state is deemed as an important element. Although the MTCD cannot obtain the RB state directly, it can achieve the state from its action decision and observation history encapsulated by the belief state. In other words, a probability distribution over states, is sufficient statistics for the history, which means that the optimal decision can be made based on the belief state. 

Let be the belief space, where denotes the conditional probability that the RB state is in at the beginning of time slot prior to state transition. In order to maintain a belief state, the knowledge of the system dynamics and the state transition probabilities in POMDP are considered as necessary.

According to [59], the belief state can be updated easily after each state transition to incorporate additional step information into history. In the proposed scheme, the belief state can be updated by using Bayes’ rule at the end of each time slot [59, 45, 62] [63]. It can be calculated as follows,


According to Eq. (9), the belief states have ability to capture all the history information at . As a result, the past actions and observations can be saved by constantly updating the belief state. It means that it is reasonable to make decisions based on the belief state.

5) Reward Function and Objective

As discussed above, idle RBs may be offered by eNodeB and the MTCD will choose the proper RB to access in each virtual vehicular network. In this paper, we consider the transmission rate as a reward, the maximum transmission rate offered by RB can be used for performance evaluation. Therefore, for each system state, the corresponding transmission rate in virtual network can be represented as


where denotes the transmission rate offered by the -th RB in time slot .

Then, in the proposed scheme, the optimization objective is to maximize the transmission rate that can be achieved by MTCDs. Hence, the system reward in each time slot is represented as


and the long-term total discounted reward is given as


where represents the discount factor.

The optimal policy in the proposed scheme can be represented as a set of action , , which maximises the expected long-term total discounted reward during a time period. Then, the optimal policy is given as


3.2 Solving the POMDP Problem

In this subsection, an optimal policy for selecting RB by the MTCD can be derived, which maximizes the transmission rate in vehicular networks based on the state of each RB. Meantime, the formulation problem in POMDP can be solved by a dynamic programming methodology.

In the proposed scheme, in order to obtain the maximum expected long-term total discounted reward during a time period, the optimization goal is to determine which action should be taken. Let be a value function, it represents the maximum expected reward that can be obtained from , given the belief state at the beginning of time slot. Suppose that the MTCD that attempts to access the network makes action and observes state , the reward can be accumulated starting from . The reward includes two parts: one is the immediate reward , the other is the maximum expected future reward starting from , given the belief state  [57]. Then, the value function is represented as


and after a time period, the maximum expected reward is calculated as


According to Eqs. (14) and (15), it should be found that the optimization policy of random access with M2M communications will affect the total reward in two ways: on the one hand, how to obtain the immediate reward; on the other hand, how to transform and select the belief state that determines the future reward.

4 Simulation Results and Discussions

In this section, simulation results are presented to show the performance improvement of the proposed scheme modeled by POMDP. Network scenario is described in Figure 1. In order to ensure the comparison fairness, other schemes are listed as follows: the existing scheme without observation, random selection scheme and the existing scheme with perfect knowledge [64]. For simplicity, the simulation is discussed by taking the highest class virtual vehicular network as an example.

In the simulation, we assume a single-cell scenario with one eNodeB and 50 vehicles equipped with MTCDs. In addition, 25 RBs are offered by eNodeB. The physical vehicular network is sliced into 5 virtual networks based on different applications and QoS requirements. For each virtual vehicular network, one virtual eNodeB and multiple MTCDs are set. The number of MTCDs is set in the first virtual vehicular network, and the number of MTCDs in other virtual vehicular networks is set , respectively. At beginning of each time slot, RBs allocate to each virtual eNodeB. Moreover, the probability that RB remains in the idle state, in the busy state, transits from busy to idle state or from idle to busy state is given as , , and , respectively. Besides, and are assumed equal, they range from to . Furthermore, the available transmission bandwidth in the first and other virtual networks are set MHz and MHz, respectively. The transmit power set dBm in each virtual vehicular network. Channel gains follow Gaussian distribution with zero mean and unit variance.

As can be seen in Figure 2, it illustrates the average reward improvement that is relative to the transmission capacity by the proposed scheme. Obviously, it can be seen that the proposed scheme has the higher reward than the existing scheme without observation and random selection scheme. Although the performance of the existing scheme without observation is closer to the proposed scheme, the advantages of the proposed scheme always exist in each the time period. According to POMDP optimization policy, the characteristic of observation can be shown by comparing with other algorithms. Moreover, the transmission capacity in the proposed scheme can be improved nearly compared with the random selection scheme.

Figure 2: Average reward within different time period in the heterogeneous traffic scenario.

Figure 3 shows the reward achieved by the proposed scheme with those by other two existing schemes in different number of RBs. It easily can be found from Figure 3 that when the number of RBs is , there is nearly no difference between these schemes. The reason is that the available number of RBs is less and there are not diverse decisions to make. However, with the increasing number of RBs, especially, when the number of RBs reaches , the advantage of performance in the proposed scheme is prominent. Due to the increasing number of available RBs in the virtual network, more access chances are offered to MTCDs, then the received reward is also improved. Thus, more advantages will be shown if the proposed scheme applies into the network that has large number of RBs. 

Figure 3: Average reward with different number of RBs in the heterogeneous traffic scenario.
Figure 4: Average reward with different probability of false observation in the heterogeneous traffic scenario.

Figure 4 depicts the variation of average reward with different probability of false observation. The results show that the average reward degrades with the increasing probability of false observation. When the probability of false observation is , the achieve reward in the proposed scheme will be close to the existing scheme with perfect knowledge. However, when the probability of false observation is , the performance in the proposed scheme has degraded obviously. Especially, the performance is even worse than the random selection scheme when the probability of false observation is high. The reason is that the MTCDs have to give up or falsely select RBs to access when the probability of false observation is high. As a consequence, there is lower average reward to be obtained.

5 Conclusions and Future Work

In this paper, we proposed a novel framework for M2M communications in vehicular networks with WNV. In the proposed framework, the physical resources are virtualized as virtual resources according to vehicles’ functions and QoS requirements. As a consequences, the virtual eNodeB and the vehicles equipped with MTCDs that have the same or similar functions can form a virtual vehicular network. In addition, we formulated the random access process of vehicular as a POMDP, by which MTCDs can select proper RBs to achieve the maximum transmission rate. Simulation results were presented to demonstrate that, with the proposed framework, the transmission rate achieved by MTCDs can be improved significantly. Future work is in progress to consider energy consumption and cooperative communications for vehicular networks in our framework. 


This work is jointly supported by the National Natural Science Foundation of China under Grant No. 61372089 and No. 61571021.


  1. D. T. Wiriaatmadja and K. W. Choi. Hybrid random access and data transmission protocol for machine-to-machine communications in cellular networks. IEEE Trans. Wireless Commun., 14(1):33–46, Jan. 2015.
  2. O. Arouk and A. Ksentini. Multi-channel slotted aloha optimization for machine-type-communication. In Proc. ACM MSWiM’14, pages 119–125. Montreal, QC, Sep. 2014.
  3. Duong Nguyen-Huu and Thinh Nguyen. Data synchronization via random network coding. In Proc. ACM DIVANet’14. Montreal, QC, Sep. 2014.
  4. Peyman TalebiFard and Victor C.M. Leung. A content centric approach to dissemination of information in vehicular networks. In Proc. ACM DIVANet’12, Sep. 2014.
  5. C. M. Angelopoulos, J. Buwaya, O. Evangelatos, and J. Rolim. Traversal strategies for wireless power transfer in mobile ad-hoc networks. In Proc. ACM MSWiM’15, pages 31–40. Cancun, Mexico, Nov. 2015.
  6. Wasif Masood, Johannes Klinglmayr, and Christian Bettstetter. Experimental evaluation of pulse-coupled oscillator synchronization in EEE 802.15.4 networks. In Proc. ACM DIVANet’14. Montreal, QC, Sep. 2014.
  7. R. Lu, X. Li, X. Liang, X. Shen, and X. Lin. Grs: The green, reliability, and security of emerging machine to machine communications. IEEE Comm. Mag., 49(4):28–35, Apr. 2011.
  8. A. Laya, L. Alonso, and J. Alonso-Zarate. Is the random access channel of LTE and LTE-A suitable for M2M communications? A survey of alternatives. IEEE Commun. Surveys Tutorials, 16(1):4–16, Jan. 2014.
  9. A. Rajandekar and B. Sikdar. A survey of mac layer issues and protocols for machine-to-machine communications. IEEE Internet of Things J., 2(2):175–186, Apr. 2015.
  10. K. Zheng, Q. Zheng, P. Chatzimisios, W. Xiang, and Y. Zhou. Heterogeneous vehicular networking: A survey on architecture, challenges, and solutions. IEEE Commun. Surveys Tutorials, 17(4):2377–2396, Fourth quarter 2015.
  11. M.J. Booysen, S. Zeadally, and G.-J. van Rooyen. Survey of media access control protocols for vehicular ad hoc networks. IET Communications, 5(11):1619–1631, Jul. 2011.
  12. Z. Wei, F. Richard Yu, and A. Boukeche. Trust based security enhancements for vehicular ad hoc networks. In Proc. ACM DIVANet’14, pages 103–109. Montreal, QC, Sep. 2014.
  13. W. Viriyasitavat, M. Boban, H.-M. Tsai, and A. Vasilakos. Heterogeneous vehicular networking: A survey on architecture, challenges, and solutions. IEEE Veh. Tech. Mag., 10(2):55–66, Jun. 2015.
  14. M.J. Booysen, J.S. Gilmore, S. Zeadally, and G.-J. van Rooyen. Machine-to-machine (M2M) communications in vehicular networks. KSII Trans. Internet and Informations Systems, 6(2):529–546, Feb. 2012.
  15. Z. Wei, F. Richard Yu, and A. Boukeche. Cooperative spectrum sensing with trust assistance for cognitive radio vehicular ad hoc networks. In Proc. ACM DIVANet’15, pages 27–33. Cancun, Mexico, Nov. 2015.
  16. Y. Chen, F. Richard Yu, and B. Zhou. Improving throughput in highway transportation systems by entry control and virtual queue. In Proc. ACM DIVANet’13, pages 9–14. Barcelona, Spain, Nov. 2013.
  17. S.-P. Pyun, W. Lee, and D.-H. Cho. Resource allocation for vehicle-to-infrastructure communication using directional transmission. IEEE Trans. Intelligent Transportation Systems, 17(4):1183–1188, Apr. 2016.
  18. G. Guo and L. Wang. Control over medium-constrained vehicular networks with fading channels and random access protocol: A networked systems approach. IEEE Trans. Veh. Tech., 64(8):3347–3358, Aug. 2015.
  19. S. Jangsher and Victor O.K. Li,. Resource allocation in moving small cell network. IEEE Trans. Wireless Commun., 15(7):4559–4570, Jul. 2016.
  20. L. Ma, F. Yu, V. C. M. Leung, and T. Randhawa. A new method to support UMTS/WLAN vertical handover using SCTP. IEEE Wireless Commun., 11(4):44–51, Aug. 2004.
  21. Fei Yu and V. C. M. Leung. Mobility-based predictive call admission control and bandwidth reservation in wireless cellular networks. In Proc. IEEE INFOCOM’01, Anchorage, AK, Apr. 2001.
  22. Z. Li, F. R. Yu, and M. Huang. A distributed consensus-based cooperative spectrum sensing in cognitive radios. IEEE Trans. Veh. Tech., 59(1):383–393, Jan. 2010.
  23. F. Yu and V. Krishnamurthy. Optimal joint session admission control in integrated WLAN and CDMA cellular networks with vertical handoff. IEEE Trans. Mobile Computing, 6(1):126–139, Jan. 2007.
  24. R. Xie, F. R. Yu, Hong Ji, and Yi Li. Energy-efficient resource allocation for heterogeneous cognitive radio networks with femtocells. IEEE Trans. Wireless Commun., 11(11):3910 –3920, Nov. 2012.
  25. A. Attar, H. Tang, A.V. Vasilakos, F. R. Yu, and V.C.M. Leung. A survey of security challenges in cognitive radio networks: Solutions and future research directions. Proceedings of the IEEE, 100(12):3172–3186, 2012.
  26. Yu Fei, Vincent W. S. Wong, and Victor C. M. Leung. Efficient QoS provisioning for adaptive multimedia in mobile communication networks by reinforcement learning. Mob. Netw. Appl., 11(1):101–110, Feb. 2006.
  27. C. Liang and F. R. Yu. Wireless network virtualization: A survey, some research issues and challenges. IEEE Commun. Surveys Tutorials, 17(1):358–380, Mar. 2015.
  28. Yifei Wei, F. R. Yu, and Mei Song. Distributed optimal relay selection in wireless cooperative networks with finite-state Markov channels. IEEE Trans. Veh. Tech., 59(5):2149 –2158, June 2010.
  29. Quansheng Guan, F. R. Yu, Shengming Jiang, and Gang Wei. Prediction-based topology control and routing in cognitive radio mobile ad hoc networks. IEEE Trans. Veh. Tech., 59(9):4443 –4452, Nov. 2010.
  30. S. Bu, F. R. Yu, Y. Cai, and P. Liu. When the smart grid meets energy-efficient communications: Green wireless cellular networks powered by the smart grid. IEEE Trans. Wireless Commun., 11:3014–3024, Aug. 2012.
  31. R. Xie, F. R. Yu, and H. Ji. Dynamic resource allocation for heterogeneous services in cognitive radio networks with imperfect channel sensing. IEEE Trans. Veh. Tech., 61:770–780, Feb. 2012.
  32. F. R. Yu, H. Tang, M. Huang, Z. Li, and P. C. Mason. Defense against spectrum sensing data falsification attacks in mobile ad hoc networks with cognitive radios. In Proc. IEEE Military Commun. Conf. (MILCOM)’09, Oct. 2009.
  33. F. R. Yu, Minyi Huang, and H. Tang. Biologically inspired consensus-based spectrum sensing in mobile ad hoc networks with cognitive radios. IEEE Network, 24(3):26 –30, May 2010.
  34. C. Luo, F. R. Yu, H. Ji, and V. C. M. Leung. Cross-layer design for TCP performance improvement in cognitive radio networks. IEEE Trans. Veh. Tech., 59(5):2485–2495, 2010.
  35. F. Yu, V. Krishnamurthy, and Victor C. M. Leung. Cross-layer optimal connection admission control for variable bit rate multimedia traffic in packet wireless CDMA networks. IEEE Trans. Signal Proc., 54(2):542–555, Feb. 2006.
  36. F. R. Yu, Peng Zhang, Weidong Xiao, and P. Choudhury. Communication systems for grid integration of renewable energy resources. IEEE Network, 25(5):22 –29, Sept. 2011.
  37. G. Liu, F. R. Yu, H. Ji, V. C. M. Leung, and X. Li. In-band full-duplex relaying for 5G cellular networks with wireless virtualization. IEEE Network, 29(6):54–61, Nov. 2015.
  38. Quansheng Guan, F. R. Yu, Shengming Jiang, and V.C.M. Leung. Capacity-optimized topology control for MANETs with cooperative communications. IEEE Trans. Wireless Commun., 10(7):2162 –2170, July 2011.
  39. S. Bu and F. R. Yu. Green cognitive mobile networks with small cells for multimedia communications in the smart grid environment. IEEE Trans. Veh. Tech., 63(5):2115–2126, June 2014.
  40. J. Liu, F. R. Yu, C.-H. Lung, and H. Tang. Optimal combined intrusion detection and biometric-based continuous authentication in high security mobile ad hoc networks. IEEE Trans. Wireless Commun., 8(2):806–815, 2009.
  41. Qiao Yan, F. R. Yu, Qingxiang Gong, and Jianqiang Li. Software-defined networking (SDN) and distributed denial of service (DDoS) attacks in cloud computing environments: A survey, some research issues, and challenges. IEEE Commun. Survey and Tutorials, 18(1):602–622, 2016.
  42. S. Bu, F. R. Yu, and H. Yanikomeroglu. Interference-aware energy-efficient resource allocation for heterogeneous networks with incomplete channel state information. IEEE Trans. Veh. Tech., 64(3):1036–1050, Mar. 2015.
  43. L. Zhu, F. R. Yu, B. Ning, and T. Tang. Cross-layer handoff design in MIMO-enabled WLANs for communication-based train control (CBTC) systems. IEEE J. Sel. Areas Commun., 30(4):719–728, May 2012.
  44. S. Zhang, F. R. Yu, and V.C.M. Leung. Joint connection admission control and routing in IEEE 802.16-based mesh networks. IEEE Trans. Wireless Commun., 9(4):1370 –1379, Apr. 2010.
  45. F.Yu, V. W. S. Wong, and Victor Leung. Efficient QoS provisioning for adaptive multimedia in mobile communication networks by reinforcement learning. Mobile Networks and Applications, 11(1):101–110, Feb. 2006.
  46. Shengrong Bu, F. R. Yu, and P.X. Liu. Dynamic pricing for demand-side management in the smart grid. In Proc. IEEE Online Conference on Green Communications (GreenCom)’11, pages 47–51, Sept. 2011.
  47. S. Bu and F. R. Yu. A game-theoretical scheme in the smart grid with demand-side management: Towards a smart cyber-physical power infrastructure. IEEE Trans. Emerging Topics in Computing, 1(1):22–32, June 2013.
  48. N. Zhang, G. Kang, J. Wang, Y. Guo, and F. Labeau. Resource allocation in a new random access for M2M communications. IEEE Comm. Letters, 19(5):843–846, May. 2015.
  49. J. Kim, J. Lee, J. Kim, and J. Yun. M2M service platforms: Survey, issues, and enabling technologies. IEEE Commun. Surveys Tutorials, 16(1):61–76, First quarter 2014.
  50. C. Liang, F. R. Yu, and X. Zhang. Information-centric network function virtualization over 5G mobile wireless networks. IEEE Network, 29(3):68–74, May. 2015.
  51. C. Liang and F. R. Yu. Wireless virtualization for next generation mobile cellular networks. IEEE Wireless Comm., 22(1):61–69, Feb. 2015.
  52. M. Hoffmann and M. Staufer. Network virtualization for future mobile networks: General architecture and applications. In Proc. IEEE Int. Conf. Commun. (ICC), pages 1–5. Kyoto, Japan, Jun. 2011.
  53. D. Kreutz, F.M.V. Ramos, P. Esteves Verissimo, C. Esteve Rothenberg, S. Azodolmolky, and S. Uhlig. Software-defined networking: A comprehensive survey. Proceedings of the IEEE, 103(1):14–76, Jan. 2015.
  54. Laizhong Cui, F. R. Yu, and Qiao Yan. When big data meets software-defined networking (SDN): SDN for big data and big data for SDN. IEEE Network, 30(1):58–65, Jan. 2016.
  55. Y. Cai, F. R. Yu, C. Liang, B. Sun, and Q. Yan. Software defined device-to-device (D2D) communications in virtual wireless networks with imperfect network state information (NSI). IEEE Trans. Veh. Tech., 2015. DOI:10.1109/TVT.2015.2483558.
  56. W. Lu and M. D. Renzo. Stochastic geometry modeling of cellular networks: Analysis, simulation and experimental validation. In Proc. ACM MSWiM’15, pages 179–188. Cancun, Mexico, Nov. 2015.
  57. P. Si, F. R. Yu, H. Wang, and Y. Zhang. Optimal control of partially observable markov processes over a finite horizon. China Communications, 10(8):1–17, Aug. 2013.
  58. R. Kokku, R. Mahindra, H. Zhang, and S. Rangarajan. Nvs: A substrate for virtualizing wireless resources in cellular networks. IEEE Commun. Surveys Tutorials, 20(5):1333–1346, Oct. 2012.
  59. C. Luo, G. Min, F. R. Yu, Y. Zhang, L. T. Yang, and V. Leung. Joint relay scheduling, channel access, and power allocation for green cognitive radio communications. IEEE J. Sel. Areas Commun., 33(5):922–932, May. 2015.
  60. Q. Zhao, L. Tong, A. Swami, and Y. Chen. Decentralized cognitive MAC for opportunistic spectrum access in Ad Hoc networks: A POMDP framework. IEEE J. Sel. Areas Commun., 25(3):589–600, Apr. 2007.
  61. Y. Wang, F. Richard Yu, M. Huang, A. Boukeche, and T. Chen. Securing vehicular ad hoc networks with mean field game theory. In Proc. ACM DIVANet’13, pages 55–60. Barcelona, Spain, Nov. 2013.
  62. G. Cili, H. Yanikomeroglu, and F. R. Yu. Cell switch off technique combined with coordinated multi-point (CoMP) transmission for energy efficiency in beyond-LTE cellular networks. In Proc. IEEE ICC’12, June 2012.
  63. R. D. Smallwo and E. J. Sondik. Optimal control of partially observable markov processes over a finite horizon. Operations Research, 21:1071–1088, 1973.
  64. L. Zhu, F. R. Yu, B. Ning, and T. Tang. Handoff performance improvements in mimo-enabled communication-based train control systems. IEEE Trans. Intell. Transp. Syst., 13(2):582–593, Jun. 2012.
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minumum 40 characters
Add comment
Loading ...
This is a comment super asjknd jkasnjk adsnkj
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test description