cloud service-aware location update in mobile cloud computing

8
Published in IET Communications Received on 15th December 2013 Accepted on 27th January 2014 doi: 10.1049/iet-com.2013.1142 ISSN 1751-8628 Cloud service-aware location update in mobile cloud computing Qi Qi 1,2 , Jianxin Liao 1,2 , Yufei Cao 2 1 State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, Peoples Republic of China 2 EBUPT Information Technology Co., Ltd., Beijing 100191, Peoples Republic of China E-mail: [email protected] Abstract: Mobile devices are becoming the primary platforms for many users who always roam around when accessing the cloud computing services. From this, the cloud computing is integrated into the mobile environment by introducing a new paradigm, mobile cloud computing. In the context of mobile computing, the battery life of mobile device is limited, and it is important to balance the mobility performance and energy consumption. Fortunately, cloud services provide both opportunities and challenges for mobility management. Taking the activities of cloud services accessing into consideration, the authors propose a service-aware location update mechanism, which can detect the presence and location of the mobile device without traditional periodic registration update. Analytic model and simulation are developed to investigate the new mechanism. The results demonstrate that the service-aware location update management can reduce the location update times and handoff signalling, which can efciently save power consumption for mobile devices. 1 Introduction As mobile devices are becoming the primary platforms for many users who always roam around and access the cloud computing applications, the mobile computing and cloud computing have emerged as a new computing paradigm. Fernando et al. [1] summarise three mobile cloud computing concepts: (i) mobile device acts like a thin client connecting to the remote server through wireless network; (ii) mobile devices act both the resource providers and consumers and make up a peer-to-peer network; and (iii) mobile device ofoads its workload to a local cloudletcomprised of several multi-core computers with connectivity to the remote cloud servers. By referring the above concepts, we discuss the rst concept and consider the operation of mobile cloud computing through the core network Internet protocol (IP) multimedia subsystem (IMS). IMS is an overlay-architecture proposed by a large group of standardisation organisation such as 3rd generation partnership project (3GPP), 3rd generation partnership project 2 (3GPP2), European telecommunications standards institute (ETSI) and telecommunications and internet converged services and protocols for advanced networking (TISPAN), which combines heterogeneous access networks with IP-based services and targets to offer real-time IP multimedia applications over mobile networks [2]. On one hand, mobile cloud computing benets from the IMS in means of increasing the system efciency through the desired management mechanisms. Mobile cloud computing is used in the highly heterogeneous networks. Different mobile devices access to the cloud service through different access technologies such as wideband code division multiple access, general packet radio service, (worldwide interoperability for microwave access), and wireless local area networks. As a result, an issue of how to handle the wireless connectivity while satisfying the requirements of always-on connectivity, on-demand scalability of wireless connectivity and the energy efciency of mobile devices rises [3]. Fortunately, IMS provides service invocation, mobility management, quality of service (QoS) guarantee and security mechanism for mobile cloud services. The real-time telecommunication services in the cloud computing environment, which require signicantly higher QoS levels and mobility management, can be provided by the IMS even during the access network change. On the other hand, the traditional telecommunication operators and service providers use cloud computing to realise the elastic resource utilisation, especially for services shared by a large amount of users such as telecommunication service voice over IP (VoIP), location, conferencing and messaging services, Internet service image processing, natural language processing and multimedia search and sensor data applications [4]. To cope with high peak situations, telecommunication service infrastructures are usually over-provisioned, sometimes by an order of magnitude, and that leads to low resource utilisation rate during typical times. Hence, hosting IMS application servers in the cloud enables pay-per-use cost models for traditional telecommunication service providers and reduces the risk of bad investments. Fig. 1 shows the services in mobile cloud computing environment with a Cloud Service Monitor which is proposed to aware the mobile users service behaviours. www.ietdl.org IET Commun., 2014, Vol. 8, Iss. 8, pp. 14171424 doi: 10.1049/iet-com.2013.1142 1417 & The Institution of Engineering and Technology 2014

Upload: qi

Post on 24-Feb-2017

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Cloud service-aware location update in mobile cloud computing

www.ietdl.org

IE

d

Published in IET CommunicationsReceived on 15th December 2013Accepted on 27th January 2014doi: 10.1049/iet-com.2013.1142

T Commun., 2014, Vol. 8, Iss. 8, pp. 1417–1424oi: 10.1049/iet-com.2013.1142

ISSN 1751-8628

Cloud service-aware location update in mobilecloud computingQi Qi1,2, Jianxin Liao1,2, Yufei Cao2

1State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications,

Beijing 100876, People’s Republic of China2EBUPT Information Technology Co., Ltd., Beijing 100191, People’s Republic of China

E-mail: [email protected]

Abstract:Mobile devices are becoming the primary platforms for many users who always roam around when accessing the cloudcomputing services. From this, the cloud computing is integrated into the mobile environment by introducing a new paradigm,mobile cloud computing. In the context of mobile computing, the battery life of mobile device is limited, and it is important tobalance the mobility performance and energy consumption. Fortunately, cloud services provide both opportunities and challengesfor mobility management. Taking the activities of cloud services accessing into consideration, the authors propose a service-awarelocation update mechanism, which can detect the presence and location of the mobile device without traditional periodicregistration update. Analytic model and simulation are developed to investigate the new mechanism. The results demonstratethat the service-aware location update management can reduce the location update times and handoff signalling, which canefficiently save power consumption for mobile devices.

1 Introduction

As mobile devices are becoming the primary platforms formany users who always roam around and access the cloudcomputing applications, the mobile computing and cloudcomputing have emerged as a new computing paradigm.Fernando et al. [1] summarise three mobile cloud computingconcepts: (i) mobile device acts like a thin client connectingto the remote server through wireless network; (ii) mobiledevices act both the resource providers and consumers andmake up a peer-to-peer network; and (iii) mobile deviceoffloads its workload to a local ‘cloudlet’ comprised ofseveral multi-core computers with connectivity to the remotecloud servers. By referring the above concepts, we discussthe first concept and consider the operation of mobile cloudcomputing through the core network Internet protocol (IP)multimedia subsystem (IMS). IMS is an overlay-architectureproposed by a large group of standardisation organisationsuch as 3rd generation partnership project (3GPP), 3rdgeneration partnership project 2 (3GPP2), Europeantelecommunications standards institute (ETSI) andtelecommunications and internet converged services andprotocols for advanced networking (TISPAN), whichcombines heterogeneous access networks with IP-basedservices and targets to offer real-time IP multimediaapplications over mobile networks [2].On one hand, mobile cloud computing benefits from the

IMS in means of increasing the system efficiency throughthe desired management mechanisms. Mobile cloudcomputing is used in the highly heterogeneous networks.Different mobile devices access to the cloud service through

different access technologies such as wideband code divisionmultiple access, general packet radio service, (worldwideinteroperability for microwave access), and wireless localarea networks. As a result, an issue of how to handle thewireless connectivity while satisfying the requirements ofalways-on connectivity, on-demand scalability of wirelessconnectivity and the energy efficiency of mobile devicesrises [3]. Fortunately, IMS provides service invocation,mobility management, quality of service (QoS) guaranteeand security mechanism for mobile cloud services. Thereal-time telecommunication services in the cloud computingenvironment, which require significantly higher QoS levelsand mobility management, can be provided by the IMS evenduring the access network change. On the other hand, thetraditional telecommunication operators and serviceproviders use cloud computing to realise the elastic resourceutilisation, especially for services shared by a large amountof users such as telecommunication service voice over IP(VoIP), location, conferencing and messaging services,Internet service image processing, natural languageprocessing and multimedia search and sensor dataapplications [4]. To cope with high peak situations,telecommunication service infrastructures are usuallyover-provisioned, sometimes by an order of magnitude, andthat leads to low resource utilisation rate during typicaltimes. Hence, hosting IMS application servers in the cloudenables pay-per-use cost models for traditionaltelecommunication service providers and reduces the risk ofbad investments. Fig. 1 shows the services in mobile cloudcomputing environment with a Cloud Service Monitor whichis proposed to aware the mobile user’s service behaviours.

1417& The Institution of Engineering and Technology 2014

Page 2: Cloud service-aware location update in mobile cloud computing

Fig. 1 Mobile cloud computing

www.ietdl.org

Mobile user may move from one access network to anotheror change its access technology. However, the energycontained in its battery for a mobile device is finite; hence,one of the key issues encountered in a mobile cloud is thedesign of a power saved mobility management that supportsuser mobility effectively. In this paper, we focus on thenew features that were introduced by cloud services to themobile computing environment and propose a completesolution to reduce energy consumption during theirmovement.The remainder of this paper is structured as follows.

Section 2 surveys the related work of mobile cloudcomputing and mobility management. Section 3 introducesthe cloud service-aware location update solution to reduceenergy consumption with original contribution presented indetail. Section 4 models the proposed service-awarelocation update mechanisms. Section 5 analyses theperformance of the new mechanism. Section 6 concludesthis paper.

2 Related work

The new concept of mobile cloud computing cannot besimply illustrated as merging mobile computing and cloudcomputing technologies which introduces many technologychallenges.As the battery lifetime of mobile devices is limited, and

energy awareness for utilising the cloud resource to reduceits energy consumption is a priority of designingapplications on mobile phones. Rao et al. [5] propose anew approach for full-text keyword searches to retrievecontent matching input keywords. For low energy, thekeywords search is split into two subsets, such that onesubset is processed by the mobile phone and another by theremote server. For the same purpose, Liu and Shu [6]propose a mobile cloud system framework which canautomatically perceive network conditions and smartlyschedule data transmissions for different popularapplications, for example, social networking services (SNS)and cloud storage for various multimedia content according

1418& The Institution of Engineering and Technology 2014

to such conditions. This framework can lead to prolongedbattery life and enhance user experience in mobile devices.Ryu et al. [7] firstly work on the handoff mechanism for

mobile cloud computing. The probability of predictivemode failure that distinguishes the predictive mode and thereactive mode of fast handovers for mobile IPv6 (FMIPv6)are considered to optimise the handoff control in IP layer.The proposed scheme supports seamless handoff for variouswireless technologies in cloud computing and reduces theoverhead of traditional FMIPv6 protocol. Additionally,Choi et al. [8] propose a seamless connection handoffmechanism between different access networks in transportlayer. By setting a checkpoint, this mechanism clearscommunication channel before switching the socket ofconnection from the old network to the new network. Bythis way, it guarantees that there are no transit data on thecommunication channel at the moment of connectionhandoff. Vrat et al. [9] propose a model to enhance themobility services in cloud by utilising the concept ofhierarchical mobile IPv6 (HMIPv6) in coexistent network.It reduces the burden on existing IPv4 addresses andenhances mobility as a service in cloud computing. Tunceret al. [10] presents a virtual mobility domains on theirfloating cloud tiered Internetworking model. The virtualmobility domains support both macro and micro mobilityby leveraging a new tiered addressing, a network cloudconcept and a unique packet forwarding scheme. However,non-IP addressing and classic routing protocols is notsuitable for the mobile device in current mobile cloudcomputing environment.For second generation (2G) and 3G networks, a normal

location area update caused by a mobile device changing itslocation and a periodic location area update schemes areadopted in universal mobile telecommunications system(UMTS) and global system for mobile communications(GSM) networks, and dynamic periodic location areaupdate is proposed in [11]. Based on this, the dynamic andoptimal combination of periodic and normal location areaupdate are provided in [12–14] with comparison to the costof location update among these mechanisms.As mentioned above, the existing research on mobile cloud

computing for reducing the power consumption is offloadingapplication in mobile device to remote server. The locationupdate is indeed a forgotten procedure for mobile devices,which can be optimised for power saving. And the users’service behaviours in cloud computing environment such asservice accessing have not been considered into locationupdate.

3 Cloud service-aware location update

3.1 Motivation

IMS obtains the location and state information of mobiledevices through registration process. A binding is createdby the serving-call session control function (S-CSCF) basedon initial registration between the public user identify andthe IP address of mobile device. The S-CSCF and proxy-call session control function (P-CSCF) keep user’sregistration status including the timer that is indicated bythe ‘expires’ parameter in ‘contact’ header of REGISTERmessage. The subsequent re-registrations are performedperiodically to update the binding and refresh the timer [2].To refresh the exiting registration or notify the network that

the capability of mobile device has been changed, the mobiledevice should initiate re-registration process. (i) The

IET Commun., 2014, Vol. 8, Iss. 8, pp. 1417–1424doi: 10.1049/iet-com.2013.1142

Page 3: Cloud service-aware location update in mobile cloud computing

Fig. 2 Publish signalling for location aware by cloud services

www.ietdl.org

re-registration is triggered by two situations: periodicre-registration (PRR) is used to periodically report the‘active’ state of mobile device to the network for avoidingexpiry of the agreed registration timer between mobiledevice and network and (ii) re-registration for changecapabilities (CCRR) is initiated when the capabilities of themobile device have changed. Typically it is the case that, asthe mobile device moves and connects to another networkentry point, registration procedure is performed once moreto update registration timer in IMS. The goal of PRR is todetect if the mobile device is still registered to the network(the network will initiate de-registration procedure as theregistration timer is timeout), whereas the goal of CCRR isto inform the network when and where the locations andcapabilities of mobile device have been changed. Theformer is triggered by the timer periodically and the latter istriggered when capability parameters change. Both the PRRand CCRR processes need to refresh the user’s registrationtimer so that the mobile device can be located or it caninitiate a new session.For the IMS, if the network receives a session invited by

the mobile device or the mobile device can response asession successfully, the IMS can determine the mobiledevice is still in the registration status. Different fromtraditional telecommunication network, the powerful mobilephones in the mobile cloud computing environment accesskinds of Internet services, such as weather, instantmessaging, social network, Internet TV, online gaming,document processing and other remote web applications.The behaviours of accessing the cloud services also indicatethat the mobile device is in the registration status. Thereforetaking use of the session establishment and the cloudservice accessing to refresh the registration timer can reducethe PRR times, which can save the power of the mobiledevice caused by the periodically re-registration.

3.2 Signalling procedure of cloud service-awarelocation update

Besides session initiation protocol (SIP), the cloud servicesmay use other application layer protocol, such as HTTP.However, the S-CSCF and P-CSCF can only process SIPmessage, so the IMS cannot obtain the cloud serviceaccessing events directly. By taking use of the event whenthe mobile device accesses the cloud service, the CloudService Monitor perceiving the user’s behaviour during itsonline time and obtain the information of mobile’s devicesstatus. The SUBSCRIBE/PUBLISH method is used tonotify IMS as the active state of the mobile device, depictedin Fig. 2. The S-CSCF and P-CSCF subscribe the status ofregistered mobile devices from the Cloud Service Monitor.For avoiding the large network signalling traffic between

the Cloud Service Monitor and the S-CSCF and reducingthe load of the cloud services, the Cloud Service Monitorcaptures the user’s behaviour from the cloud services in aninterval, and publishes its information to the S-CSCF andP-CSCF.

3.3 Algorithm of cloud service-aware locationupdate

We define user’s behaviour checkpoint as an action to informthe IMS whether the mobile is attached or not. Thecheckpoint event includes a session establishment request,or CCRR event, or the publish message from the CloudService Monitor. Such an event results in checkpoint action:

IET Commun., 2014, Vol. 8, Iss. 8, pp. 1417–1424doi: 10.1049/iet-com.2013.1142

(1) Incoming or outgoing session (IOS) event: IMS receivesthe session initiated by the mobile device or the mobiledevice receives a session request from the IMS andresponses successfully, such as a VoIP session setuprequest, or session request for a streaming service in thecloud.(2) Cloud service access (CSA) event: the Cloud ServiceMonitor publishes the access events to the S-CSCF andP-CSCF when the mobile device accesses the cloud service.For example, a user may logon the twitter and watch hisfriends’ content.(3) CCRR event: when the capabilities of the mobile devicehave changed, CCRR event occurs.(4) PRR event: the PRR to the IMS network.

The cloud service-aware location update algorithm is asfollows (see Fig. 3).When the mobile device is disconnected from the network

because of abnormal reasons, such as battery removal,subscriber moving out of the service area, and so on, theIMS detects abnormal detach of the mobile device in one ofthe following two cases, depicted in Fig. 4a. (i)The mobiledevice disconnected at t2 and the next re-registration fortimer expires (t3) occurs before the arrival of the incomingsession (t4). In this case, the network can detect expirationof the timer and considers the mobile device detached. Thenext incoming session will not be delivered. (ii) The nextincoming session occurs (t8) before the timer expires (t9),that is, the timer update is not timely. In this case, the IMSattempts to deliver the next incoming session to the mobilebut fails. After failure of the session setup, the IMSconsiders that the mobile device is detached and willdisable future session terminations and registered timer.Then the session establishment, the cloud service accessing

and the capability changing event are taking into updating thetimer. When these events occur, the location is updatedpassively (t2) and the PRR event can be put off (t4).

1419& The Institution of Engineering and Technology 2014

Page 4: Cloud service-aware location update in mobile cloud computing

Fig. 3 Algorithm of cloud service-aware location update

Fig. 4 Timer sequence of location update and mobile disconnected detecting

www.ietdl.org

However, when there are no above events, the PRR recovers(t6 and t7). From Fig. 4b, we can see the putting off thelocation update of re-registration has no effect of the IMSdetecting disconnected mobile devices.

4 Analytical model

In this section, a formula will be derived for analysing thepower cost of location update. The PRR and CCRR event issending REGISTER message to IMS, and IOS event is

1420& The Institution of Engineering and Technology 2014

sending or receiving the INVITE message. The energyconsumed in these processes is dependent on the usedmodulation and coding scheme and the exact number ofsub-frames required for the process. However, for thepurpose of analysis, we consider this power cost to bedirectly proportional to the size of the messages exchanged[15]. Thus, the power cost of the mobile device forprocessing update date point event e is denoted by ce = L(message size). The CSA event is accomplished by CloudService Monitor notifying the S-CSCF when the mobile

IET Commun., 2014, Vol. 8, Iss. 8, pp. 1417–1424doi: 10.1049/iet-com.2013.1142

Page 5: Cloud service-aware location update in mobile cloud computing

www.ietdl.org

device accesses the cloud service. In this procedure, themobile device only needs to update its timer by theapplication within its operation system. The energyconsumption of this procedure is far less than sending orreceiving an SIP message.Assume the arrivals of IOS, CCRR and CSA events follow

Poisson distribution and independent with each other. WhenIMS receives the update event, it refreshes the timer, whichis called update check point. As the sum of several Poissondistribution is still a Poisson distribution, the arrival rate ofupdate check point is

l =∑

e={IOS, CCRR, CSA}

le (1)

For each update check point, the number of a certain event isλe/λ. Then, the power cost of the events for one update checkpoint is

Powere = ce le/l( )

, where e = IOS, CCRR, CSA{ } (2)

IMS update event only contains CCRR, which means theupdate check point is receiving CCRR. In Fig. 5, the timeinterval of update check point follows exponentialdistribution with a mean of λCCRR, and its probabilitydensity function is:f (lCCRR) = lCCRRe

−lCCRRt. AssumeN IMSPRR is the number of PRR event times during two update

check points for IMS location update. Astp, 1 = tp, 2 = ... = tp, n = tp, (n+1) = tp, t = N IMS

PRRtp + t∗.Then, we can deduce the probability of N IMS

PRR = n duringtwo update check points as

P N IMSPRR = n

( ) = P(ntp , t , (n+ 1)tp)

=∫(n+1)tp

ntp

lCCRRe−lCCRRt dt =e−lCCRRntp (1− e−lCCRRntp )

(3)

And the expectation of N IMSPRR is

E N IMSPRR

( ) = ∑1n=1

nP N IMSPRR = n

( )

=∑1n=1

e−lCCRRntp 1− elCCRRntp( ) = 1

elCCRRtp − 1

(4)

Hence, the power cost of PRR event is

PowerPRR = cPRRE N IMSPRR

( )(5)

The total power cost of the IMS location update and session

Fig. 5 Number of PRR event during two update check points

IET Commun., 2014, Vol. 8, Iss. 8, pp. 1417–1424doi: 10.1049/iet-com.2013.1142

establishment is

TPowerIMS = PowerCCRR + PowerIOS + PowerPRR

= cCCRRlCCRRl

+ cIOSlIOSl

+ cPRR1

elCCRRtp − 1

(6)

In the service-aware location update algorithm, the locationupdate events include CCRR events, IOS and CSA events.The update check point is the occurrence of CCRR, IOS orCSA events. The time interval of update check point tfollows exponential distribution with a mean of λ, and itsprobability density function is: f (λ) = λe−λt.Taking use of the same method of modelling the standard

IMS location update, we can also obtain the probability ofNCloudPRR = n and the expectation of NCloud

PRR during two updatecheck points in cloud service-aware location update

P NCloudPRR = n

( ) = e−lntp 1− e−lntp( )

(7)

E NCloudPRR

( ) = 1

eltp − 1(8)

Then, power cost of PRR event for service-aware locationupdate is

PowerPRR = cPRRE NCloudPRR

( )(9)

As the CSA event only needs to update the timer in mobiledevice, the total location update power cost forservice-aware location update includes the cost of CCRR,IOS and PRR events

TPowerCloud = PowerCCRR + PowerIOS + PowerPRR

= cCCRRlCCRRl

+ cIOSlIOSl

+ cPRR1

eltp − 1(10)

5 Simulation and performance analyses

5.1 Simulation

To verify our analytic model, we have developed a Java-basedtest bed, depicted in Fig. 6 which includes severaldiscrete-event simulators, the IMS network entitiesincluding P-CSCF, interrogating-call session controlfunction (I-CSCF), S-CSCF and home subscriber server(HSS), cloud services and a simple mobile device module.A timer is contained in the mobile device, P-CSCF andS-CSCF. The discrete-event simulators simulate themobility behaviours of mobile device by generating usermovement, session initiation and cloud service accessingevents, which all follows exponential distribution. IMSnetwork entities, Cloud Service Monitor and the mobile

1421& The Institution of Engineering and Technology 2014

Page 6: Cloud service-aware location update in mobile cloud computing

Fig. 6 Simulation environment

Table 1 Parameter values in simulation

Parameters Values

REGISTER 225 bytesINVITE 810 bytes200OK 100 bytesACK 60 bytes

www.ietdl.org

agent are developed based on the open source code SIP stackSIPp [16], which realised the basic transaction process. Weuse the OpenStack to provide three different vitalisedmachines, and develop the online chatting, video ondemand (VoD) and cloud storage on the cloud. Finally, thewireless link layer is simulated by several queues whichcarry some background traffic and communicate withmobile device with the specified packet loss rate.The power cost of an SIP message is set to be the length of

the message. For example, the power cost for transmitting are-registration message (225 bytes) is 225 W, where W is aconstant of proportionality for the power consumed. Thepower cost for a mobile device update its timer cloud

Fig. 7 Simulation and analytical value of PRR event number with λIOS

1422& The Institution of Engineering and Technology 2014

service is set to 10 W. Table 1 lists the parameters used inour simulations.From the analysis in Section 3, the power cost for location

update is decided by the number of periodic registration forrefreshing the timer. Figs. 7 and 8 depict the number ofperiodical registration from simulation and analytical model.With different session arrival rates (λIOS) and timer lengths(tp), the simulation and analytical results are almost thesame. The errors are because of the number of randomgenerated discrete-events. For example, if several morehandoff events generated during the simulation period, thesimulation result is bigger than the analytical result, and viceversa. As the error rate is under 1%, these experiments haveverified that the analytic model is consistent with thesimulation results. From Figs. 7 and 8, we can also see thatthe number of periodic registration for cloud service-awarelocation update is less than that for the IMS location update.By introducing the session establishment and cloud serviceaccessing events, cloud service-aware location updatereduces the number of periodic update events, but does notimpact the location update and mobile device’s re-registration.Fig. 9 is the comparing result of IMS and cloud

service-aware location update with various arrival rate of

IET Commun., 2014, Vol. 8, Iss. 8, pp. 1417–1424doi: 10.1049/iet-com.2013.1142

Page 7: Cloud service-aware location update in mobile cloud computing

Fig. 8 Simulation and analytical value of PRR event number with tp

Fig. 9 Power cost for IMS and cloud service-aware locationupdate with λCCRRtp = 1, λIOS = 0.5, λCSA = 0.5

Fig. 10 Total cost for IMS and cloud service-aware locationupdate with λIOStp = 1 and λCCRR = 0

Fig. 11 Power cost between IMS and cloud service-aware locationupdate with λCSAtp = 1 and λCCRR = 0.5

www.ietdl.org

CCRR event. We can see that introducing the IOS and CSAevents to help update the mobile device’s location, thepower cost for cloud service-aware location update withdynamic location update is obviously smaller than that forIMS location update with periodic location update. Whenthe arrival rate of CCRR event increases, the power cost forboth the IMS and cloud service-aware location updatedecreases. The reason is the CCRR event can refresh thetimer in IMS, which puts off the PRR event. Then, within aperiod of time, the number of PRR event decreases and theconsumed energy of the mobile device is saved.Fig. 10 depicts the power cost for the IMS and cloud

service-aware location update under the condition ofdifferent IOS event arrival rates. When the IOS eventarrival rate increases, both the power cost for IMS andcloud service-aware location update increase. This isbecause the energy for processing the session incoming andterminating increases, which is contained in the totalcalculated power cost. In Fig. 10, the power costs for cloudservice-aware location update are all lower than that forIMS location update. And by the increasing of the arrivalrate of cloud service accessing, the power cost for cloudservice-aware location update is decreasing. The reason is

IET Commun., 2014, Vol. 8, Iss. 8, pp. 1417–1424doi: 10.1049/iet-com.2013.1142

1423& The Institution of Engineering and Technology 2014

Page 8: Cloud service-aware location update in mobile cloud computing

Fig. 12 Power cost between IMS and cloud service-aware locationupdate with tpλCCRR = λIOS = λCSA = 0.3

www.ietdl.org

by perceiving the mobile user’s behaviour and taking use ofCSA event, the timer in IMS can be refreshed withoutperiodic registration update.Fig. 11 shows the difference of energy consumed between

periodic location update in IMS and dynamic location updatein new mechanism. The difference value increases when thearrival rate of CSA event increasing. The reason is cloudservice-aware location update takes use of the cloud serviceaccessing event to refresh timer, which reduces the numberof PRR events and its power consumption. When λCCA = 0and λIOS = 0, there is no session or cloud service event forupdate, and the timer refreshing in IMS is only based onthe CCRR and PRR events. In this case, the IMS locationupdate has the same performance with cloud service-awarelocation update, and the power consumptions of the mobiledevice of the two mechanisms are same. When the arrivalrate of IOS event grows, the difference value becomes high.Since the cloud service-aware location update can refreshtimer without the CCRR and PRR events, but the IMSlocation update needs to use the PRR events, which resultsthat the power cost for cloud service-aware location updateis much less than that for IMS location update.Fig. 12 depicts the total power cost with various register

timer values. Along with timer value increasing, the updateinterval becomes large, then the times of periodic updatedecreases, which results in the power cost for both the IMSand cloud service-aware location update decreasing.However in this case, the power cost for cloudservice-aware location update is smaller than that for IMSlocation update.

6 Conclusion

In this paper, we focus on the mobile cloud computing wherethe mobile device acts like a thin client connecting to the

1424& The Institution of Engineering and Technology 2014

remote server through wireless network. IMS provides anIP-based control plan for the mobile cloud computingenvironment, especially QoS guarantee and mobilitymanagement. However, considering the features of cloudcomputing services and the limited battery life of themobile devices, IMS location update is not efficient for theenergy consumption. By perceiving the cloud service thatthe mobile users accessing, we propose an optimisedlocation update techniques that support user mobilityeffectively. The service-aware location update detects thepresence and location of the mobile device by the mobileusers’ behaviours such as access network change, sessionestablishment and cloud service accessing. The reduction ofperiodic registration makes energy consumption of mobiledevices for location updates less. The evaluation resultsdemonstrate that the service-aware location update minimisesunnecessary energy consumption for the location update.

7 References

1 Fernando, N., Loke, S.W., Rahayu, W.: ‘Mobile cloud computing: asurvey’, Future Gener. Comput. Syst., 2013, 29, pp. 84–106

2 3GPP TS 23.228, V.8.2.0. IP Multimedia System (IMS), 20073 Dinh, H.T., Lee, C., Niyato, D., Wang, P.: ‘A survey of mobile cloud

computing architecture, applications, and approaches’. WirelessCommunications and Mobile Computing, 2011, DOI: 10.1002/wcm.1203

4 Bellavista, P., Carella, G., Foschini, L., Magedanz, T., Schreiner, F.,Campowsky, K.: ‘QoS-aware elastic cloud brokering for IMSinfrastructures’. IEEE ISCC, 2012, pp. 157–160

5 Rao, W., Zhao, K., Lagerspetz, E., Hui, P., Tarkoma, S.: ‘Energy-awarekeyword search on mobile phones’. Sigcomm Workshop on MobileCloud Computing, 2012, pp. 59–64

6 Liu, F., Shu, P.: ‘eTime: energy-efficient mobile cloud computing forrich-media applications’. IEEE COMSOCMMTC E-Letter, 2013, vol. 8

7 Ryu, S., Lee, K., Mun, Y.: ‘Optimized fast handover scheme in mobileIPv6 networks to support mobile users for cloud computing’, J.Supercomput., 2012, 59, pp. 658–675

8 Choi, M., Park, J., Jeong, Y.-S.: ‘Mobile cloud computing frameworkfor a pervasive and ubiquitous environment’, J. Supercomput., 2013,64, pp. 331–356

9 Vrat, A., Sachan, M., Dinker, A.G., Arora, D., Vaish, A., Venkatesan, S.:‘Performance analysis of enhanced mobility model in cloud computing’.Recent Trends in Information Technology (ICRTIT), 2011, pp. 638–643

10 Tuncer, H., Nozaki, Y., Shenoy, N.: ‘Virtual mobility domains – amobility architecture for the future Internet’. IEEE ICC, 2012,pp. 2774–2779

11 Lin, Y.-B., Lee, P.-C., Chlamtac, I.: ‘Dynamic periodic location areaupdate in mobile networks’, IEEE Trans. Veh. Technol., 2002, 51,pp. 1494–1501

12 Xiao, Y., Chen, H.: ‘Optimal periodic location area update for mobiletelecommunications networks’, IEEE Trans. Wirel. Commun., 2006, 5,pp. 930–937

13 Xiao, Y., Chen, H.: ‘Periodic location area update schemes for UMTS3G mobile networks: optimality and comparison’. IEEE Int. Conf.Communications, 2006, vol. 3, pp. 955–960

14 Xiao, Y., Chen, H.: ‘An analytical model of the ODPLAU scheme fortelecommunication networks’. IEEE Wireless Communications andNetworking Conf. (WCNC), 2007, pp. 3194–3198

15 Kim, R.Y., Kalle, R.K., Das, D.: ‘Joint paging area and location updateoptimization for IEEE 802.16 m idle mode’, Comput. Netw., 2011, 55,pp. 3744–3758

16 SIPp [OL]. Available at http://www.sipp.sourceforge.net/, October 2011

IET Commun., 2014, Vol. 8, Iss. 8, pp. 1417–1424doi: 10.1049/iet-com.2013.1142