RESOURCE ALLOCATION IN VEHICULAR CLOUDCOMPUTING SYSTEMS WITH HETEROGENEOUS VEHICLES AND ROADSIDE UNITS
ABSTRACT
Vehicular cloud computing (VCC) system coordinates the vehicular cloud (consisting of vehicles’ computingresources) and the remote cloud properly to provide in-timeservices to users. Although pervious works had established themodels for resource allocation in the VCCsystembased on semiMarkovdecisionprocesses(SMDP),fewofthemdiscussedheterogeneityofvehiclesandinfluencesof roadsideunits(RSUs).Heterogeneousvehiclesmadebydifferentmanufacturersmaybeequippedwithdifferentamountofcomputingresources;andfurthermore,RSUcanenhancethe computingcapabilityofVCC.Therefore, this work proposes an SMDP model for VCC resourceallocation that additionally considers heterogeneous vehicles andRSUs, and an approach for finding the optimal strategy of VCCresource allocation. The two additional features significantlyelaborate the SMDP model, and demonstrate different resultsfrom the original model. Simulation shows that the resourceallocation in the VCC system can be captured by the proposedmodel, which performs well in terms of long-term expected values(consisting of consumption costs of power and time), undervarious parameter settings.
EXISTING SYSTEM:
Previous works on vehicular networks focused on coping with traffic congestion on roads or preventing emergencyconditions, e.g., watering roads and vehicle malfunction. Forinstance, the work in considered insufficient positioningcapability of GPS, and hence improved RFID to assist in positioning, to establish a VANET environment that canposition vehicles more precisely. The works in considered deployment of sensors and RSUs on roads inVANETs, with minimal deployment cost and number oftransmission hops. On mobile clouds, the work in introduced a model ofmobile devices connected to mobile clouds, and investigatedcloud storage services on mobile clouds. The work in adopted the SMDP to address the resource allocation problemin mobile cloud computing systems. The work in [20]proposed a model with mobile clouds and vehicular networksin which each vehicle can upload photos and videos of traffic conditions to the cloud, and share them with other vehicles.The mobile clouds proposed in were integrated to VANET, and the work in established a VCC system framework. The previous works on VCC systems are dividedinto five categories: security and privacy, data aggregation,energy efficiency, interoperability, and resource management.The work in investigated the problem of safety andprivacy arising from practical applications of VCC systems,and proposed some strategies for addressing the problems ofprivacy leaks of road users and authentication of high-speedmoving vehicles. The work in integrated trajectory data oftaxis in real world to establish a model that can evaluate andforecast the serviceability of mobile vehicular cloudlet. Some research focused on resource management in VCCsystems. The work in considered a cloud consisting ofRSUs to help resource management and allocation betweenRC and VC. The considered the V2V in the VCC systemto be applied in the scheduling model of resource sharing. Withthe concept of V2V in the VCC system, the work in defined a centralized VCC system that allocates computingresources of homogenous vehicles. The work in adoptedSMDP to analyze video streaming in heterogeneous cognitivevehicular networks to establish a resource allocation model topromote quality of service (QoS). The work in adopted SMDP to establish the optimal resource allocation strategy toallocate machine-type communication gateways in a softwaredefinednetworkingframeworkforinternetofthings.
PROPSED SYSTEM:
This work integrates both RSU’s and vehicles’ computing resources in the VCC system to provide services. Additionally,most previous works assumed homogeneous vehicles in V2Iand V2V frameworks. However, in reality, vehicles made bydifferent manufactures may be equipped with VEs withdifferent amount of computing resources. Therefore, this workestablishes a model for resource allocation in the VCC systemwith heterogeneous vehicles and RSUs based semi-Markovdecision processes (SMDP), which allocates eithervehicular cloud (consisting of vehicle’ computing resources)or remote clouds to handle vehicle’s service requests. The SMDP is one of the Markov decision processes (MDPs).MDPs aim to find the optimal decision-making strategy toaffect a system with a dynamic stochastic process, in order tomaximize a long-term value. MDPs are categorized into threetypes: in the discrete-time MDP, transition times of the systemmodel and the timing points to make decisions to control thesystem are fixed and known; in the continuous-time MDP,these timing points are not fixed and can be arbitrary at anycontinuous time point; in the SMDP, these timing points are not fixed either, but are arbitrary at discrete time points. In VCC systems, the times between vehicle arrivals and between vehicle departures are not fixed, i.e., the transitiontimes in the system are not fixed. In addition, optimal decisionsare made immediately for each service request from vehicles,i.e., at arbitrary discrete times. Therefore, the SMDP is appliedto model the concerned VCC system. In this VCC system,heterogeneous vehicles meet the reality but their providedresources are more uncertain; whereas RSUs can providerelatively stable and reliable computing resources. This workassumes that different types of vehicles provide differentamount of computing resources and arrive at the VCC systemwith different probability distributions, so that the SMDPmodel becomes more complex.
CONCLUSION
This work has proposed an SMDP model for resource allocation in the VCC system that considers heterogeneousvehicles and integrates V2V and V2I, to meet the practice.Introducing heterogeneity of vehicles and RSUs makes themodel becomes increasingly complex, in which much moresystem states of the VCC system need to be considered, andthe transition a states requires more computation andrestrictions. Simulation shows that the SMDP for the systemprovides a promising approach for allocating resources in thesystem. A line of the future work is to further considerheterogeneity of service requests, and their influence to theresource allocation. In addition, it would be of interest toinvestigate fairness and priority of resource allocation toheterogeneous vehicles and service requests in the system.
REFERENCES
[1] Q. Ding, B. Sun, and X Zhang, “A Traffic-light-aware Routing Protocolbased on street connectivity for urban vehicular ad hoc Networks,” IEEECommunications Letters, vol. 20, no. 8, pp. 1635-1638, 2016.
[2] M. Rebai, L. Khoukhi, H. Snoussi, and F. Hnaien, “Optimal placementin hybrid VANETs-sensors networks,” in Proceedings of WirelessAdvanced (WiAd), IEEE Press, pp. 54-57, 2012.
[3] C.-C. Lin and D.-J. Deng, “Optimal two-lane placement for hybridVANET-sensor networks,” IEEE Transactions on Industrial Electronics,vol. 62, no. 12, pp. 7883-7891, 2015.
[4] M. Gerla, “Vehicular cloud computing,” in Proceedings of 11th AnnualMediterranean Ad Hoc Networking Workshop (Med-Hoc-Net), IEEE Press, pp. 152-155, 2012.
[5] S. Bitam, A. Mellouk, and S. Zeadally, “VANET-cloud: A generic cloud computing model for vehicular ad hoc networks,” IEEE WirelessCommunications, vol. 22, no. 1, pp. 96-102, 2015.
[6] L. Gu, D. Zeng, and S. Guo, “Vehicular cloud computing: A survey,” inProceedings of IEEE Globecom Workshops, pp. 403-407, 2013.
[7] R. Hussain, J. Son, H. Eun, S. Kim, and H. Oh, “Rethinking vehicularcommunications: Merging VANET with cloud computing,” inProceedings of 4th International Conference on Cloud ComputingTechnology and Science (CloudCom), IEEE Press, pp. 606-609, 2012.
[8] E. Lee, E. K. Lee, M. Gerla, and S. Y. Oh, “Vehicular cloud networking:architecture and design principles,” IEEE Communications Magazine,vol. 52, no. 2, pp. 148-155, 2014.
[9] S. C. Ng, W. Zhang, Y. Zhang, Y. Yang, and G. Mao, “Analysis of accessand connectivity probabilities in vehicular relay networks,” IEEEJournal on Selected Areas in Communications, vol. 29, no. 1, pp. 140150,2011.
[10] M. Báguena, C. T. Calafate, J. C. Cano, and P. Manzoni, “An adaptiveanycasting solution for crowd sensing in vehicular environments,” IEEETransactions on Industrial Electronics, vol. 62, no. 12, pp. 7911-7919,2015