Research Article Approaches to Addressing Service Selection ...

Research ArticleApproaches to Addressing Service Selection Ties in Ad HocMobile Cloud Computing

Ayotuyi Tosin Akinola and Matthew Olusegun Adigun

Department of Computer Science Centre of Excellence at University of Zululand Private Bag X1001 OngoyeKwaDlangezwa 3886 South Africa

Correspondence should be addressed to Ayotuyi Tosin Akinola ruthertosingmailcom

Received 1 September 2017 Revised 6 November 2017 Accepted 9 November 2017 Published 15 February 2018

Academic Editor Youyun Xu

Copyright copy 2018 Ayotuyi Tosin Akinola and Matthew Olusegun Adigun -is is an open access article distributed under theCreative CommonsAttribution License which permits unrestricted use distribution and reproduction in anymedium providedthe original work is properly cited

-e ad hoc mobile cloud (AMC) allows mobile devices to connect together through a wireless connection or any othermeans and send a request for web services from one to another within the mobile cloud However one of the majorchallenges in the AMC is the occurrence of dissatisfaction experienced by the users -is is because there are many serviceswith similar functionalities but varying nonfunctional properties Moreover another resultant cause of user dissatis-faction being coupled with runtime redundancy is the attainment of similar quality computations during service selectionoften referred to as ldquoservice selection tiesrdquo In an attempt to address this challenge service selection mechanisms for theAMC were developed in this work -is includes the use of selected quality of service properties coupled with userfeedback data to determine the most suitable service -ese mechanisms were evaluated using the experimental method-e evaluation of the mechanisms mainly focused on the metrics that evaluate the satisfaction of usersrsquo interest via thequantitative evaluation -e experiments affirmed that the use of the shortest distance can help to break selection tiesbetween potential servicing nodes Also a continuous use of updated and unlimited range of usersrsquo assessments enhancesan optimal service selection

1 Introduction

Service-oriented computing (SOC) is an emanating in-terdisciplinary paradigm for realizing a distributed com-putation thereby changing the traditional pattern ofincurring the cost of access to unwanted software appli-cations thus depicting a better service delivery on variousservice request provisioning schemes [1] -e softwareapplications that are made available in the SOC are pri-marily a product of fundamental resources which areservices -e services are self-describing and computationalresources that enhance an automatic rapid and low-costprovisioning of software applications in a distributedenvironment Due to the explosive growth of variousmobile applications the Internet serves as a linkingmedium between the mobile clients and the serviceproviders -erefore with the view to achieving agility

and flexibility between provider and client interactionthe SOC has been generally adopted -e SOC assumedbusiness functions as modular packages in the form ofservices that run on any service-oriented architecture(SOA) [2]

Moreover the cloud computing paradigm is an exampleof the SOA that serves as an information technology ser-vicing model where computing services are delivered ondemand to customers over a network in a self-servicefashion independent of devices and locations [3] -usthe resources required to provide the requisite quality ofservice levels are shared dynamically scalable rapidlyprovisioned and virtualized with minimal service providerinteraction Users pay for the service as an operating expensewithout incurring any significant initial capital expenditurewith the cloud services employing a metering system thatdivides the computing resources into appropriate blocks

HindawiJournal of Computer Networks and CommunicationsVolume 2018 Article ID 4505290 17 pageshttpsdoiorg10115520184505290

-ere are three basic processes that are involved in theprovisioning of services in any service-oriented platformswhich include service discovery service selection andservice composition [2 4 5] By the discovery of serviceswe refer to the process of identifying potentially availableweb services which can carry out a particular taskHowever among the available web services there wouldcertainly be a particular service in the midst of the dis-covered web services which best satisfies the request of thepresent service user at the time of request based on thequality of service that was specified -is process is termedldquoweb service selectionrdquo -e composition of web servicesinvolves the integration of two or more web servicestogether to implement interconnected tasks For examplethis interconnected task can be a web service whichsearches for a flight schedule as well as another that locatesa meter taxi or a hotel reservation -is kind of request isbeing carried out via a composite web service from theprovisioning platform

-ese three concepts are very important in any serviceprovisioning platform and thus become one of the centralfocus to research scholars -erefore this calls for ade-quate mechanisms to be deployed to enable these pro-cesses to run smoothly in the course of providing optimalservices on the aforementioned platforms Addressingthese three concepts poses various challenges in themodern service provisioning platforms today especiallythe concern with the enormous release of web services thatare performing similar functionalities In the same veinwhen mobile devices were connected to the cloud plat-forms for service usage it becomes imperative that op-timal service selection is addressed to avoid the dissatisfactionwith service responses to the intended users A typicalexample of such a mobile service consumption scenario isthe mobile cloud computing paradigm -is paradigmenables the use of mobile devices to consume resident webservices within the cloud computing architecture [6ndash9]-e dissatisfaction becomes very tasking in a situationwhere the users eventually become the platform for ser-vice provisioning in the case of the AMC system whereinmobile users are constantly in a move while consumingthe web service [10]

Several parastatals such as military and other forms oforganizations which are involved in emergency opera-tions have been adopting the AMC technology towardsaddressing the pressing need for total avoidance ofinterconnectivity issues both in less-developed areas asa result of bad infrastructure and even in developed placesdue to natural disasters etc [11] For example in anemergency operation which utilizes an AMC setup thenearest mobile nodes with the required service must bethe chosen node for the operation so that the task at handwill be carried out successfully and promptly within thelimited allotted time frame Moreover the occurrence ofservice selection ties during the course of selection es-pecially in a range of different service locations within themobile devices needs to be addressed as this has a lot tospeak to the measure of satisfaction that the users ex-perience within such a system [12 13]

A contemporary project that is currently under develop-ment which looks into most of the aforementioned issues isGUIISET (GRID-based utility infrastructure-infrastructure lessfor SMME-enabled technology) from the Centre of Excellenceof the University of Zululand in South Africa It is an ad-vancement on the previous GUISET project through the in-corporation of this infrastructure less (ad hoc model) platforminto the system [14ndash17] -e goal of this incorporation is toenable the developed platform to be able to render servicesto mobile devices especially in the context of m-Healthm-Learning and m-Commerce

Several works have been done which ranges from GUI-SETrsquos implementation [18ndash20] to its performance evaluation[17 21 22] and security [15 16 23] as well as the Pricingstrategy to be used on the platform [14] However little re-search has been conducted in relation to providing an op-timally satisfactory service especially in the newly integratedad hoc model system In addition issues relating to similarcomputation scores (selection ties) which often inform un-satisfactory service usage have not been fully addressed inmost of the articles to the best of our knowledge Hence thisarticle contributes to the existing knowledge via providingsolutions to ldquoservice selection tiesrdquo in both mobile nodes andweb service provisioning to various patronizing customers viadeployable selection mechanisms Moreover service responsedelay is also addressed to reduce the time taken to delivera selected service to intended users

-e remainder of this article is as organized as followsSection 2 reviews various service selection approaches indifferent platforms to bring out the gap that is missing whichis very important on the AMC -e section also reviews theearlier attempts to address the issue of service selection invarious platforms Section 3 discusses the QoS propertyresolution strategy while Section 4 enumerates the modes ofselection occurrence in the AMC Section 5 addresses theselection ties in AMC mobile nodes using the multidynamicdistance-based approach where a typical surgical emergencyservice is used as a scenario Section 6 discusses the feedback-based selection approach for optimal web service selection inan AMC while Section 7 explains how the performanceevaluation was conducted with some discussions Section 8explains the conclusion and future work

2 Literature Review

-is section elucidates the methodologies of the serviceselection process with a view to discussing various methodsthat can be deployed for selection of services in general fromany service-oriented computing platform Moreover theselection approaches that are often deployed in an ad hocenvironment were also discussed Hence we shall beexplaining service selection methodologies and service se-lection approaches in the AMC In addition this section alsoattempts to review other approaches to optimal selection inthe AMC

21 Service Selection Methodologies -e selection method-ologies are the various methods that are adopted for selection

2 Journal of Computer Networks and Communications

services within a service-oriented architecture platform -emethodologies used in service selection are generally classifiedinto three according to the work of Swarnamugi [24]

-ese are explained below

(1) Functional-based method -is method selects theappropriate service based on the retrieval of a func-tional description of a service from the registries andthen certifies that the description and requirement ofthe interfaces match each other-ere is always a needto convert the web service into the semantic web toenhance the easy description of web service func-tionality -e semantic web service selection wasimplemented in the work of Klusch and Kapahnkeusing the hybrid version of the SAWSDLmatchmakercalled SAWSDL-MX [2 25] -e drawbacks of thefunctional service selection approachmake the serviceprovider seek for a more informative alternative thatcould differentiate their products from others in re-lation to performance Moreover considering theAMC environment the use of the highly logical al-gorithm and complex ontological data storage gen-erate a challenge for memory-restrained mobiledevices therefore the approach is considered un-suitable for deployment

(2) Nonfunctional-based method It is a common expe-rience in the service provisioning environment forservices to provide similar functionalities with dif-ferent nonfunctional properties Hence such servicescan only be differentiated by considering their non-functional attributes which can either be quality ofservice or context-based [24] In [26] the authorproposed the architecture for a web service selectionusing the QoS-based approach in a service pro-visioning environment where the service user searchesthe service registry (UDDI) for the list of all servicesthat address the concerned request-e service brokerhere assists in differentiating the various servicesin the registry -us the service registry becomesa complex task for an AMC-based environment Xinand others in [27] considered a framework thatcombines QoS attributes with user preferences ofa group of consumers to propose an algorithm anda mobile service selection model to solve the selectionchallenge But the selection was based on a group ofcollective users which is not individually tailored asexpected in an AMC environment -e work ofAmoretti and others in [28] proposed a reputation-based selection framework which emphasizes on anintra- and inter-SOP (service-oriented peers) moduleinteraction -e framework contains a componentcalled SAFE that ensures that a mobile peer computesthe reputation of a provider based on the previousexperience -e SAFE component is assigned the taskof ldquovoting strategiesrdquo to ensure that proper record ofthe aggregated reputation influences selection de-cisions However this work was silent about the sit-uation where web service attains similar computationaggregates (otherwise called selection ties) thereby

leaving behind gap to fill Furthermore the studyconducted by Akingbesote and others [18] proposeda quality of service aware Multilevel Ranking Model(MLRANK) for selecting an optimal web service incloud computing-e study addressed the occurrenceof ties within a number of services that are available inthe UDDI -e study highlights the challenge ofselecting an optimal web service when there are tieswith the used criterion where performance alterna-tives have the same score -e study achieved optimalselection by comparing the service consumerrsquos QoSpreference with the web service QoS offerings -eprovider offering that best fits the QoS preference istaken as the optimal web service -e study usednondeterministic QoS metrics and concentrated onvarious information services to test the performanceof the proposed model However this study seriallyconsiders each of the selected qualities one after theother and checks which one is higher than the other tomake a selection -is approach is not reliable in thecontext of the AMC -is is because decisions areexpected to be as fast as possible In addition the worknever considered the possibility of the consumershaving relatively equal priority for the specified QoSproperties which makes the selection yardstick lessefficient Keidl and Kemper in [29] proposed a contextprocessing framework to influence context-aware webservice provisioning However the framework uses aninsufficient number of context parameters amongwhich are location and client addresses -e locationspecified by the SOAP message body of the webservice is fixed thus it is not suitable for a dynamicenvironment like the AMC system Since the mobiledevice keeps changing location the use of fixed lo-cation as one of the parameters will be irrelevant toservice selection Keskes proposed the combination ofcontext and QoS for service selection and consideredthe effectiveness and consistency of the method inrelation to using an increasing number of service pa-rameters more than six [30] But the use of the on-tological context to decipher the information bringsabout the issue of ambiguity if deployed to memory-constrained mobile devices for service selection

(2) User-based selection method -e measure of thetrustworthiness of a particular web service is termedldquothe Reputationrdquo -is measure mainly depends onthe end usersrsquo experience of using the particularservice Various end users may have dissimilar viewsabout the same web service However the reputationis expressed as an average ranking that is given toa web service by the end users thus deriving a range ofranking from these end users -is system determinesthe QoS of a service provider through calculation ofthe difference between the published service providervalue and the user feedback response A higher dif-ference value typically shows a lower QoS rating forthat particular service provider or web service -eservice users are allowed to provide a feedback

Journal of Computer Networks and Communications 3

through the use of a pair of keys [31] -e usersauthenticate with the aid of these keys and they areallowed to update the QoS criteria based on theirexperience However the provider would not alwaysbe available to affirm and see to the proper updating ofthe QoS properties -e high possibility is that boththe user and consumer would likely not be available ata similar time therefore resulting into overloading ofthe system through unconsumed user requests

22 AMC Service Selection Approaches -e approaches toservice selection in the AMC require more criteria to be takencare of because of its inherent nature -is is because in anAMC there might be the need for the selection of the nearestmobile node to provide a particular service and at some othertimes it could be the web service that is resident within themobile node that needs to be shared among the AMC mobiledevices Considering the influx of mobile devices within anAMC there could be a sudden increase in the number ofconnecting mobile devices which invariably increases thenumber of service requests thus we itemize the followingconcern with respect to AMC selection approaches (1) anyapproach that must be used should cater for sudden in-crement in service requests [32] (2) -e approach mustaddress the dynamic nature of the mobile environment and(3) the approach should eradicate the arbitrary selection ofmobile nodes and web services during the occurrence ofservice selection ties which is a major challenge in serviceprovisioning in the AMC environment [18 33]

According to [34 35] service selection in the ad hocmobileenvironment is carried out using the following approaches (1)hop-based (2) QoS-based and (3) integratedhybrid-based-e hop count has been used for time-constrained mobileservice selection in wireless mesh networks and sensor net-works especially where the response time is very pivotal such asin critical and life-saving situations [36 37] -is systemconsiders time as a very important issue and therefore looks orsearches for the closest node to the requestor -is has theadvantages of (i) a lower risk of downlink with the least dis-tances apart (ii) a lower rate of battery consumption with thereduced distances between the nodes and (iii) lesser possi-bilities for the two nodes to be out of reach of each other [38]

-e integrated approach combines the properties fromthe routehop count with the QoS approach to arrive at theoptimal solution for a particular request Due to the natureof the request that a service user specified there might bea need to prefer particular properties more than the otherFor example time constraint might be very imperative fora user in need of urgent medical attention or user whoneeded to catch up quickly on a flight whereas for anotheruser that is requesting for an educationally related servicemight not be as more urgent Mobile devices enable the useof various services on transit through a wireless connectioneither to the cloud server or within themselves -us theintegrated (hop count QoS properties and feedback re-sponse) selection approach seeks to address usersrsquo satis-faction by ensuring that the optimal service is being selectedat every request

3 QoS Property Resolution Strategy

-e impact of quality of service cannot be overemphasized inthe process of service selection It serves as a buildingfoundation upon which other service approaches can beintegrated to enhance optimal service selection in anyservice-oriented provisioning -e selection of service isbuilt by mapping the requests onto a web service among theavailable ones -e mapping process helps in locating theactual web service -e aggregation of the QoS for serviceselection is based on the individual preferences with regardto QoS properties -e following two steps assist in ensuringthat this process is properly carried out

31 -e Scaling Process -is study uses a simple scaling ornormalization technique -is concept of scaling ensureseven distribution of QoS properties since these propertiesdiffer in value from each other with hardly any comparabilityamong them -e range of values derived fully expresses theorder and level of competitiveness among the available webservices [39] Ss set of ad hoc mobile services of1113864s1 s2 s3 sj1113865 with a set of qualities 1113864Qq1113865 being1113864q1 q2 q3 qm1113865 -is implies that when a sample of theAMC web service is selected such as si it will have a cor-responding set of qualities such as 1113864qi1 qi2 qi3 qij1113865 Nowassume that the value of the minimum and maximum ithQoS properties within all the available services are Qmin

j andQmax

j For example the general quality of some QoSproperties decreases with an increase in the value of theproperty such as service cost and response time thus byscaling we try to minimize the property as much as possibleusing the normalization equation (1) according to [40] as

Vj

Qmaxj minusQij

Qmaxj minusQmin

j

if Qmaxj minusQmin

j ne 0

1 if Qmaxj minusQmin

j 0

⎧⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎩

(1)

In a similar manner other services whose QoS propertiesincrease with an increasing value of such a property such asavailability and reliability are all normalized using the ex-pression in (2) according to [38] as

Vj

Qij minusQminj

Qmaxj minusQmin

j

if Qmaxj minusQmin

j ne 0

1 if Qmaxj minusQmin

j 0

⎧⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎩

(2)

An interval of the range of [0 1] is produced from (1) and(2) where ldquo1rdquo represents the best QoS property while ldquo0rdquorepresents the least QoS property thereby reflecting anyform of changes that occur to the general quality of any ofthe services whether it decreases or increases -e algorithmfor normalization process is as shown in Algorithm 1

From Algorithm 1 a new matrix V was generated

V Vij 1le ile n 1le jlem1113872 1113873 (3)

4 Journal of Computer Networks and Communications

32 Weighted Summation Assignment -ere is a need toassign weight to the respective normalized value of eachof the service users so that the preferences and the im-portance of a particular quality can be expressed by theuser when requesting a service -erefore treatingthe pool of services as a sample space with coordinatesthat are having origin as a vector the services can berepresented as

Srarr

V11113954V1 + V2 1113954V2 + V3 1113954V3 + middot middot middot + Vm

1113954Vm (4)

where the service is represented by Srarr

and V1 1113954Vj bothrepresented normalized QoS properties as well as theunit vector along the axis of the normalized propertySupposing the mean quality of service is given by Smeanand if Vmean represents the mean value of the ith QoSproperty then

Srarr

mean V1mean1113954V1 + V2mean

1113954V2

+ V3mean1113954V3 + middot middot middot + Vmmean

1113954Vm(5)

Considering the mean value of the normalized qualitieswe have the summation of the means ranging towards 05 as1ge Vj ge 0 hence we can rewrite the expression to be

1113954Smean 1m

radic 1113954V1 + 1113954V2 + 1113954V3 + middot middot middot + 1113954Vj1113872 1113873 (6)

Carrying out the dot product of the ith service of Si

rarr

as it projects over the mean Srarr

mean then we have theexpression

Si

rarrmiddot Srarr

mean 1m

radic V1 + V2 + V3 + middot middot middot + Vj1113872 1113873 (7)

Hence the summation of quality of service propertiescan be expressed as

Summation of quality properties 1m

radic summ

i1Vj (8)

Assuming the user request for a special preferencein any of the qualities of service that are highlightedthe summation of the preference in terms of weightassignments could be expressed for the ith QoS propertythus

Weighted summation assignment 1m

radic summ

i1VjlowastWi (9)

-e output generated from this normalization andweighted summation assignment often results in several webservices and service nodes with similar aggregate scores [18]-is challenge shows that there is a need for a mechanismthat helps to inform optimal service selection throughbreaking the occurrence of selection ties in the heteroge-neous web service provisioning

InputA set of qualities of a Web Service S(t) s1 s2 smthat each describes a service

OutputA matrix of normalized QoS parameters

Step 0 Initialization create m by 5 matrix P-us having

for(i 0 iltm i + +)dofor(j 0 jlt n J + +)doif(qf[j]eq0)

if(diffqos[j] 0)

v[i][j]larr((qmax[j]minusp[i][j])diffqos[j])

elsev[i][j]

endifelse if(qf[j]eq1)

if(diffqos[j] 0)v[i][j] larr ((p[i][j]minus qmin[j])diffqos[j])

elsev[i][j]larr1

endifreturn

V (Vij 1le ile n 1le jlem )

endifendif

ALGORITHM 1 QoS property normalization algorithm

Journal of Computer Networks and Communications 5

4 Modes of Selection Ties in the AMC

Selection ties occur in a situation where there are two or moreldquoobjectsrdquo that qualify to attend to a particular request froma service consumer [11 18] In the case of the AMC a mobilenode can be a service provider where it helps to forwarda trac via the shortest route mechanism to the tracdestination In addition the resident web services within themobile nodes in an AMC can attain similar computationscores via the processes discussed in Section 3 e literaturecurrently selects any of the mobile nodes or web services thatattain similar computation scores to address the service re-quest which in turn is usually not the optimal option isinvariably results in some level of dissatisfaction on the pathof service consumers when at other times a close friend wasable to experience a better service delivery than them due torandom selection at ldquotiesrdquo occurrence

For example let us consider a case where teachers andstudents formed an AMC within the school premises inwhich they all consume similar educational resources eexperiences of the teachers and the students will virtually bedierent from one another in situations where there aresimilar computation scores in the selection processerefore to avoid this kind of situation this article ad-dresses two major modes of ldquoselection tiesrdquo occurrence inthe AMC

(A) e mobile node selection ties(B) e web service selection ties

is article oers substantiated mechanisms with respectto these two modes towards the selection of an optimalservice provider to various service requests that are issued inthe context of the AMC environment

5 Multidynamic Distance-Based Approach

e term multidynamic distance in this context refers to theuse of QoS properties alongside with distances apart amongthe mobile nodes otherwise referred to as the shortest routein the course of service provisioning e approach here isdeplorable in an emergency state especially in a typicalm-Health situation as shown in Figure 1 e SUnodes is theprincipal service users

e tracs are expected to take the shortest route to thedestination while bearing in mind that the intended nodeshave the capacity to oer the required services e hypo-thetical quality of service parameters is as shown in Table 1containing the prices response time reliability and avail-ability of the mobile nodes e last column in the table alsodepicts the summation of route distances apart for theintended trac Suppose that a service user (SU15) in Figure1 is issuing a request for a surgical emergency service (SEx)

5

7

2

4

5

2

5

5

4

5

5

6

4

9

11

8

6

8

67

3

10

SE10

SE1

SE6

SE5

SE18

SE17

Su16

Su13

Su14

SE11

SE7

SE9

SE2

SE12

SE3

SE4

SE20

SE19

Su15

SE8

Figure 1 m-Health scenario routing trac from the source to the sink

6 Journal of Computer Networks and Communications

from any mobile nodes (all SEx) who can render such a serviceWe implemented the multidynamic algorithm that wasdepicted in Algorithm 2We used the normalization algorithmto analyze the considered qualities of service (price responsetime reliability and availability) which generated the nor-malization results that were depicted in Table 2 -ese outputsare the results that are fed into the weight assignment stage todetermine the interest of the service consumer who requests forthe service of the mobile nodes

Algorithm 2 calls for the computation of the normali-zation and weight assignment and the calculation of ag-gregate scores-e occurrence of ties further initiates the callto Algorithm 3 which invariably differentiates the serviceswith similar scores -e ldquoSortestrootrdquo uses the dynamicprogramming algorithm to effect the choice of the shortestroute that needs to be calculated in the process It considersonly the mobile nodes with the same computation or ag-gregate scores as shown in Table 3 and Figure 2 -e al-gorithm selects the mobile nodes among those with similarcomputation and further selects the one closest to therequesting node for user consumption

51 Selection of Optimal SE Service -e selection of thesurgical emergency service (SE) that meets the requestedQoS is selected by first providing ranges of weights thatdepict the amount of priority given to each QoS whichranges from 0 to 1 -is weight represents the degree ofimportance associated with a specific QoS property and theyare fractions whose sum must be equal to 1 For example ifa consumerrsquos topmost priority is on price for surgical emer-gency service then the higher weight is given to price -eother properties take lesser fractions respectively till the leastof them-e service that provides the best utility level based on

Table 1 Quality of service properties

Services PropertiesSE services Price (R) Response time (ms) Reliability () Availability () Route (m)SE1 82 200 69 75 29SE2 70 3600 202 43 12SE3 75 3300 365 50 11SE4 74 31010 316 41 12SE5 81 2365 357 48 14SE6 80 210 403 60 25SE7 82 208 67 72 12SE9 82 210 48 78 17SE10 73 250 35 45 20SE12 67 233 60 88 7SE17 55 2145 55 48 26SE18 78 305 67 53 23SE19 76 300 82 64 10SE20 80 295 45 81 16

Set parameters (p rt r)Input consumers required QoS

Call normalizationSet weights

Calculate scoreif Ss(Score)gt 1 then

call DPPrint Ss(Sortestroot)

elseprint Ss(score)

end ifend

ALGORITHM 2 Multidynamic algorithm

v vertex and n nodeInput (v n)Shortestrootlarr 0Array Shortpernodelarr 0Shortestrootlarr 0Shortrootestlarr 0RepeatFor i vminus 1 to vFor j 1 to nFor k 1 to edgeno

Shortpernode[k] search min cost u vNext kSort Shortpernode[1minus k] in descending

ifshortrootestgt Shortpernode [1]

thenshortrootest Shortpernode [1]

endifNext jShortestrootlarr Shortestroot + shortrootestv vminus 1

Next iUntil v 1

ALGORITHM 3 Shortest route DP algorithm

Journal of Computer Networks and Communications 7

price becomes the chosen one provided that it is the only onee weighted summation assignment generates the aggregatescore as given in (9)

When the number of ad hoc surgical emergency servicesthat have the maximum score is greater than one then weconsider the route to each of the destination We thenapplied the dynamic programming as shown in Algorithm 3is is done by describing our data elements needed in theform of cities and distances We then formulate this as anoptimization problem and the minimum route is de-termined based on the minimum cost tour

e idea in this section has been used in our previouspaper [11] but we considered a larger number of mobilenodes in this experiment thus this solution becomes one ofthe solution approaches under the present article e de-tailed comparison among the mobile nodes with aggregatescores was depicted in Figure 2 Other approaches withrespect to this particular scenario select any of the mobilenodes among those that attain the request of the service userbut a better utility level is guaranteed with this mechanism inplace for mobile service selection

6 Feedback-Based Approach for Web Services

e user feedback has been an approach that is generally usedfor improving the performance of a system is is due to thefact that it gives an updated and continuous informationabout the performance of an application (web services)with a view to making informed decisions about the be-haviour of such an application In this section we deployedthe use of user feedbacks to enhance better service selectionthat eradicates the occurrence of web service ties duringdynamic service selection Again citing the case of teachersand students who team up to form a group as an AMC thatexists within a school environment where both participantsshare educational materials the challenge that is often ex-perienced is that we want the system to be able to render theoptimal web service that guarantees the highest level of

satisfaction each time to the service user us we proposea middleware engine that collects the usersrsquo view afterconsuming web services and use it to range the performanceof the resident web services on the system to enhance anoptimal selection in future requests e diagram in Figure 3shows the interrelated components for eective service se-lection within an AMC

e main purpose of introducing the feedback mecha-nism is to eradicate the possibility of service selection tieswithin the AMCe setup mechanism majorly contains therater the feedback composer and the database wherein the

Table 2 Normalization of SE services

Services PropertiesSE services Price (R) Response time (ms) Reliability () Availability () Route (m)SE1 08181 02373 06175 09290 29SE2 02727 0064 00802 07828 12SE3 05 02523 03843 06829 11SE4 04545 03763 02929 08942 12SE5 07727 11064 09533 05620 14SE6 07272 1 04552 07296 25SE7 08181 09688 03694 03827 12SE9 08181 1 08504 09284 17SE10 04328 04270 03894 0530 20SE12 045 08639 08693 07829 7SE17 08012 04390 05728 07328 26SE18 06013 1 03874 04038 23SE19 0520 02089 08429 07829 10SE20 03720 07823 07832 05390 16

Surgical emergency services

29

12

7

16

0

5

10

15

20

25

30

35

Rout

es an

d ag

greg

ate s

core

s

Route (m)Aggregate scores

SE1 SE2 SE12 SE20

078280783 07827 07829

Figure 2 Shortest route service selection in critical situations

Table 3 Shortest route-based selection

Services PropertiesSE services Route (m) Aggregate scoresSE1 29 07830SE2 12 07828SE12 7 07827SE20 16 07829

8 Journal of Computer Networks and Communications

feedback records are kept Most of the previous steps such asthe normalization and the weight assignment are alreadyassumed to have been carried out in this stage We alsoassumed that various users provide the system with truthfulresponses e rating process follows two major steps whichare the collection of the user feedback and the prediction ofthe appropriate services for the service requestor

61 Collection of User Feedback e inyenux of the AMC withvarious forms of web services ultimately results in the oc-currence of similar ties within the environment When thesimilar ties occur the instruction is to select any out of thoseones to carry out the task and in the absence of such aninstruction the system goes into a temporary redundancystate in which indecision sets in Eradicating the challenge ofsimilar ties in service selection will enhance prompt responseto users through overcoming the possibility of occurrence ofredundancy states [41] e user feedback system uses thecredibility level system to collect process and rank theresponses from the users us brvbarve categories of credibilitytrust on web services were assigned namely the mostcredible (l5) more credible (l4) credible (l3) less credible(l2) and the least credible (l1) e associated strength toeach user response can be normalized in

sum5

i1li 1 (10)

Using this debrvbarnition the researcher assigned the value of02 to express the least credibility value l1 which shows a lack oftrust regarding the selected service Other values starting from04 to 06 are assigned to l2 and l3 respectively where they bothtypify a low credibility truste last sets of values of 08 and 10which represent l4 and l5 express reliable trust from the serviceusers Every rating selection made by the user is computed bybrvbarnding the average or mean of the feedback values selectedis is used to update the system for the latest rating regarding

a particular service is credibility trust utilized in the pro-posed system is coupledwith a recommendation technique thatwill assist to make the right decision for the service user [41]e recommendation technique that is adopted is an item-based collaborative systemwhich is a subset of the collaborativebrvbarltering approach is recommendation system is importantin this context for two major reasons

(1) It identibrvbares the best of the peer of web servicesamong those attaining similar aggregate scores

(2) It helps to compute brvbarnal selection based on theservices used by the requesting user provided theuser has invoked services in the system before

e item-based collaborative approach helps addressesthese two highlighted points thereby helping to carry out theprediction creation tasks on web services with the similaraggregate scores

62 Prediction Creation is section of the service rec-ommendation predicts the best web services based on theprevious ratings recorded by the system from earlier serviceuserse prediction proposed uses the binomial probabilitydensity distribution In this approach the following vari-ables were used as expressed below

Let k be the context of service selection such asm-Learning

Xk is the userrsquos rating for a service in context kPk is a userrsquos preference for a service in context ke feedback composer rates each service to select the

optimal service for users according to the following densityfor random variables Xk for which xk is a typical instance

Xk sim1nBinomial n Pk( ) (11)

where n no of mobile devices (nodes)

E Xk( ) Pk

Var Xk( ) 1nPk 1minusPk( )

(12)

We deployed the user rating into the probability pre-diction for selection By expressing the ranges over thebinomial distribution the summary was explained in twoways

(1) When the feedback is very low the mean of therating distribution Xk should correspond to a verylow value For example if the credibility from theuser based on quality experienced is l2 then thefeedback composer normalizes the probability toassign E(Xk) 02

(2) Contrariwise when the feedback is high E(Xk) isalso expected to be high us the feedback com-poser chooses E(Xk) 1 when the response is high

is study implements the following variation on themean Xk thus

Mobile users

Smart phoneLaptop computer

PDA

SQLitedatabase

Feedbackcomposer

Rater

Userinterface

(Browser)

Ratings Request Response

Rating update

Feedbackrecords

Selection algorithm

QoS metrics

Selection middleware engine

Figure 3 Rating mechanism in AMC service selection

Journal of Computer Networks and Communications 9

E Xk( ) μk Pk + 2 qk minus12

( ) qk minusPk( ) (13)

Every service user draws a rating from the mean Xk forevery service that has been rated ese sets of all usersrsquoratings are the inputs to the composer for proper selection totake place e result of (13) produces series of mean dis-tributions with each of the provided usersrsquo ratings thusaligning the results into a format of a binomial distributioncurve is is because it continually brvbarnds the mean of theratings submitted by the users which always tends to reachthe maximum value of ratings as given by a typical binomialdistribution diagram curve e ratings maintain their valueif similar values are provided by the new users who make nodierence from the earlier value in the feedback record Butwhenever the value of the new ratings is more than theprevious record the rating increases us the feedbackrating falls and rises along the binomial distributioncurve e peak of the dome-shaped top of the binomialcurve corresponds to the least rating from the users andsuch services are not often predicted to users for usebecause of the low value or rating which shows that theservice performs poorly within the specibrvbared servicefunctionality e yenattened edge tending towards 10shows the best services to be predicted to the user for useand the higher it moves from the dome-shaped curvetowards the yenattened end the better the web services thatare predicted for use in terms of QoS

63 Environmental Specication To test this proposed ap-proach this work deploys the use of the SunWireless Toolkit25 Beta Version J2ME from the SunMicrosystems packagesto test the behaviour of the selection mechanisme SQLitedatabase is shown outside of the physical connection setupaccording to Figure 4 to depict the three-layer networkmodel setup (consumer layer middle layer and databaselayer) however the DB was actually embedded within theserver machine e emulator interface design is developedto run on the J2ME-enabled devices such that it can be easilydeployed on real smartphone mobile devices that supportcommunication through the Hypertext Transfer Protocolinterconnections

e implementation of the server is accomplished bythe use of J2EE-compliant servers such as GlassbrvbarshApplication Server 312 e server-side components areconbrvbargured to run on any server that conforms to the J2EEspecibrvbarcation e server handles messages from clientsthrough the use of WMA Bridge API which enhancesmessage interaction during client service invocation fromthe server e server contains the web component whichruns within the servlet container and also uses the JAXPto communicate with the external data sources e ex-ternal data sources originate as information from theXML database which is a collection of mobile webservices

e whole setup was tested using a mobile laptoprunning on Windows 7 Professional Edition as an oper-ating system e laptop had an Intel Core(TM) i7-4500U

processor with a processing speed of 240 GHz and 800 GBRAM e application consumed 104MB of the hard drivestorage which is favourable to the mobile computing en-vironment due to low memory consumption Figure 4 alsocontains the mobile emulators developed from the Net-beans IDE together with the Apache JMeter load generatore emulator shows the typical interface of the nature ofservice requests which is similar to requests generated bya typical Apache JMeter for the purpose of testing theperformance of the proposed selection model Figure 5shows a typical emulator interface for service specibrvbarcationand query interface e Apache JMeter allows the settingof network parameters like latency and jitters during theexperiments erefore this setup mimics the AMC serviceprovisioning environment for testing the performance ofthe deployed mechanism

7 Performance Evaluation

e performance of the selection mechanism on serviceselection ties was tested with a series of experiments toensure that the concerned challenge is being addressedis article brvbarrst evaluated the selection system using certainparameters such as throughput and service availability tocheck up that the system is operating normally undera good condition We later conducted experiments todetermine the eect of user feedback on the selectionprocess that involves selection ties and as well use a case ofa real-life scenario where service selection ties occur to seehow tiesrsquo breaking of services might occur before oeringthe service to the users

71ServiceAvailabilityRatesandshyroughputPerformance ereare many options for quantifying the rate of serviceavailability within interconnected systems One of thepopular options is the use of statistical analyses [42ndash44]ese approaches are more feasible in a very stable en-vironment where mobility is not occurring too oftenus the constant dynamic nature of the AMC will not

NetbeanIDE

Mobileemulators

ApacheJMeter

SQLiteDB

Serviceclients

Database layer

httprequests

Middle layer

Figure 4 Mechanism simulation setup

10 Journal of Computer Networks and Communications

favour these approaches However in this experiment weassigned some weight indicators to certain parameters(fundamental resource availability ie HTTP authori-zation manager response assertion HTTP responsedefault and Hypertext Markup Language (HTML) linkparser) in the Apache load generator which is indi-cated at runtime by assigning values ranging from0 through 1 for each of the parameters Based on theseparameters suppose Pxy represents the weight of setparameters Sxy where Sxy sums up the indicator valuesIxy and given that

sumxa

x1Ixy 1 (14)

then the service availability of the AMC is therefore cal-culated using the expression

Ava suma

x1sumb

y1PxySxy (15)

As the number of mobile nodes in the AMC is in-creased there is a relative increase in the number of re-quests and it invariably has the corresponding impact onthe percentage number of service availability rates asshown in Figure 6 From the graph shown it can bededuced that as the number of service provider nodesincreases the service availability climbed steeply at theinitial stage but increased more at 20 nodes (ie as thenodes doubled) is can be interpreted to mean that allthings being equal the greater the number of serviceproviders the more the number of services available forconsumption within the system It was noticed that as thenumber of nodes increases within the AMC their rate ofservice availability was not so signibrvbarcant in incrementdue to relatively similar web services which also performsimilar nonfunctional qualities is can be assumed to bethe result of similar recurring service requests which existwithin the cloud e central server PC is stabilized andthe feedback records are properly aligned to enhance

relatively optimal selection without excessive delay of thequeries

In a similar manner the throughput measures the rateof successful service request delivery over the AMCcommunication channel It determines the time it takesto perform a transaction (E) over a number of sessionse throughput was measured through the JMeter sim-ulator that was integrated via the Netbeans IDE toolSince the proposed AMC system consists of connectingmobile devices the experiment is vital to access theperformance of the service selection mechanism Tocompute the throughput of the system the followingexpression is used

1ksumk

c0ET (16)

where the variable c represents the request index and kis the number of requests at a particular session Figure 7shows the results of the throughput for the loadgenerator

e number of mobile nodes represented the numberof available mobile devices within the system at the timethe experiment was conducted while the number of si-multaneously issued service requests showed the numberof users that were making an HTTP request during theexperiment e throughput of a system clearly showedthe number of completed service deliveries to therequesting service consumer e results displayed onFigure 7 show that the throughput performance of theweb service selection mechanism increases as the numberof users increases e proposed selection mechanismperformed well when the test for throughput was con-ducted e linear increment in the values of thethroughput with the increasing number of node requestsperforms in a realistic way us the throughput averagelyincreases with the increasing number of requests which

Figure 5 Service specibrvbarcation and query interface

0

01

02

03

04

05

06

07

08

09

1

10 20 30 40 50 60 70

Serv

ice a

vaila

bilit

y (

)

No of service nodes

Service availability rates

Service availability rates

Figure 6 Service selection eect on web service availability

Journal of Computer Networks and Communications 11

conbrvbarrms a good performance from the system ereforehaving conbrvbarrmed the performance of the AMC systemwith respect to provisioning of service availability as wellas the rate of completed service delivery we have achieveda good platform upon which the eect of feedback can betested on the system as well as the contribution of theselection mechanism to the body of knowledge whencompared with other AMC selection approaches undersimilar conditions (occurrence of selection ties)

72 User Feedback Eect Evaluation e proof of theconcept system here was implemented through deployingthe Glassbrvbarsh Web Server 312 platform on the centralserver PC within the AMC is platform works togetherwith the SQLite database which acquires the WSDL brvbarle aswell as the QoS information provided by the variousmobile service providers e test deployed a total of 100web services for this experiment to simulate a real-worldselection challenge in terms of user dissatisfaction Toevaluate the feedback-enabled QoS selection this ex-periment evaluated the number of service selections thatcoincide between feedback-enabled and nonfeedbackservice selections which were adopted from the existingworks [45 46] e goal of this experiment is to determinehow precise the two approaches are at ensuring that themost relevant service is selected

In the brvbarrst approach the service was selected directlywithout the aid of feedback ratings of the ad hoc mobileselection mechanism (an approach that is already used incloud computing) while in the other approach the se-lection was carried out with the aid of the feedback se-lection ratings

e same values of QoS parameters are requested withthe increasing number of users is evaluation revealedthe kind of services retrieved after each service request

and recorded the level of satisfaction ranging froma 1-star to a 5-star rating Starting with the feedback-enabled approach the expression for percentage pre-cision is given as

Precision relevant web services cap retrieved web services | |

| retrieved web services |

(17)

is gives the percentage precision through calculatingthe mean of responses from the number of users at eachlevel of the service request is precision is expressed todetermine the percentage level of satisfaction derived byeach of the service users Figure 8 shows that the feedbackindicates highly satisfactory service selection at every in-cidence of service invocation when compared with selec-tion based on the quality of service alone At some pointduring selection ties the nonfeedback eect can makea selection that coincides with the optimal service at thatpoint and this accounted for both approaches achievinga similar value only at experiment number four where thenumber of requests is 16 In addition the feedback-enabledselection mechanism enhanced the selection of web ser-vices and also gave a high tendency to solve the issue ofservice selection ties thereby providing an optimal servicefor users

e user feedback eect experiment shows that the useof a continuous updated and unlimited range of usersrsquo webservice assessment only enhances the selection of optimalservices for the service users It should however be notedthat the greater the number of services that attain theselection ties the higher the eect of the feedback-enabledsystem in selecting appropriate andmore satisfying servicesfor the user In a situation where there were few numbers ofservices both approaches often though not always as-sumed the same output result e feedback rating systemhowever enhanced the selection of the highly rated service

0

20

40

60

80

100

120

4 8 12 16 20 22 24

S

yste

m p

reci

sion

No of requests

User feedback effect

Feedback enabledNonfeedback enabled

Figure 8 Selection precision based on the feedback eect

0

01

02

03

04

05

06

07

10 20 30 40 50 60 70

ro

ughp

ut (T

ps)

No of service nodes

roughput values (Tps)

roughput values (Tps)

Figure 7 Mechanism selection throughputs

12 Journal of Computer Networks and Communications

among the selection ties for the user at each point of serviceinvocation where ever it occurs

73 Feedback Mechanism on Real-Life Scenario -is ex-periment extracted data samples from the work of Al-Masriand Mahmoud in [45] -is gives us the QoS of sevendifferent web services that were chosen from the samedomain -ese web services are e-mail web services thatshare the same functionalities -e data were provided inStrikeIroncom and XMethodsnet A total number of 20web services were used in this experiment to actually see theeffect of the proposed selection mechanism at solving theoccurrence of web service ties However if more web ser-vices are deployed then the possibility of web service tiesincreases in the system

-e QoS of the 20 web services is measured by WS-QoSMan WS-QoSMan is a module that is responsible forcollecting and measuring the QoS information of webservices -e values of the QoS metrics are normalized andassigned weights accordingly as described in Section 3-e free spider simulator (FSS) tool is also an online toolthat helps to get a free report about web service actionableinsights -e feedback for over 2 weeks was collected andpresented -e last column of Table 4 also specifies thefeedback of each of the web services as derived from theFSS -e results of various web services were computed asshown in Table 4 We compare our approach with the twoother approaches used in the literature which are the webservice ranking function approach [45] and nonfeedbackoutputs from [47] -e values of the data collected undereach approach were expressed in Table 5 which are the

normalized and ranked outputs More details can befound in [48]

74 Discussions -e results from Table 5 and Figure 9showed a critical observation that both feedback andnonfeedback techniques achieved the same ordering of webservices according to QoS-based selection although dif-ferent values were obtained and the same ordering wasrealized -is confirmed that both techniques achievedsimilar results However the WsRF technique only selectsat random from the set of services that attain similar QoSscores -e idea discussed in Section 4 helps to solve thechallenge by carrying out the prioritization based on thetrusted user feedback data that were generated by the AMCweb service selection mechanism

-e e-mail validation web services of numbers 6 9and 16 (ldquoServiceObjectsrdquo ldquoByteplantrdquo and ldquoBulkE-mailVerifierrdquo resp) all achieved similar QoS computa-tion It is understood however that the normalizationtechnique found in [31 40 42] ensures that it provides thethree best web services amongst those consideredHowever this information is not enough to justify thebest of the three web services as the feedback ratingsystem shows quite clearly that the web service number 9(Byteplant) is rated above others of comparable perfor-mance -e rating of the Byteplant e-mail validation wasseen to be relatively constant over a two-week period andmaintained a rating of 100 from different service users-is final choice of the Byteplant over the other two is notonly confirmed by the high range of computation selectedbased on the normalization during QoS computation but

Table 4 QoS metrics for various available e-mail verification web services

Web services Response time (ms) -roughput (bs) Availability () Cost (rands) Reliability () Feedback ()XMLLogic 720 6 85 12 87 59XWebservices 1100 174 81 1 79 49StrikeIron Emails 710 12 98 1 96 86CDYNE 912 11 90 2 91 65Webservicex 910 4 87 0 83 89ServiceObjects 1232 9 99 5 99 94StrikeIron Address 391 10 96 7 94 70Kickbox 428 5 86 8 60 67Byteplant 601 8 70 4 75 100QuickEmail 205 35 95 5 89 62TowerData 504 9 75 1 77 79Leadspend 832 8 81 3 83 76Briteverify 911 3 80 51 78 26Mailbox 604 10 89 5 87 69Emailanswers 505 7 85 6 95 29BulkEmailVerifier 600 5 90 4 88 96BulkEmailChecker 195 35 85 0 89 55Emailtor 220 6 87 8 75 28Verifalia 950 5 90 3 86 38Xverify 350 8 96 4 94 56

Journal of Computer Networks and Communications 13

Table 5 Results of WsRF non-FB and FB QoS metric computation

ID Web services WsRF computation Rank Nonfeedback-enabled computation Feedback-enabled computation Rank1 XMLLogic 36638 13 47648 068 132 XWebservices 32166 15 43176 064 163 StrikeIron Emails 46103 4 57113 087 54 CDYNE 39246 11 50256 072 115 Webservicex 42679 5 53689 089 46 ServiceObjects 46700 1 57705 090 37 StrikeIron Address 41955 7 52965 079 88 Kickbox 40428 10 51438 074 109 Byteplant 46700 1 57704 097 110 QuickEmail 39205 12 50215 069 1211 TowerData 42504 6 53514 083 612 Leadspend 40832 8 51842 081 713 Briteverify 20911 20 31921 058 2014 Mailbox 40604 9 51614 075 915 Emailanswers 30505 18 41617 062 1816 BulkEmailVeribrvbarer 46700 1 57706 092 217 BulkEmailChecker 32195 16 43205 065 1518 Emailtor 23020 19 34031 060 1919 Verifalia 30950 17 41961 063 1720 Xverify 35077 14 46087 066 14

36638

32166

46103

3924642679

467

41955 40428

467

3920542504 40832

20911

40604

30505

467

32195

2302

309535077

47648

43176

57113

5025653689

57705

52965 51438

57704

5021553514

51842

31921

51614

41617

57706

43205

34031

4196146087

068 064087 072 089 09 079 074

097069 083 081

058 075 062092

065 06 063 066

0

1

2

3

4

5

6

7

QoS

-bas

ed se

lect

ions

Web services

WsRFNon-FB enabledFB enabled

XML

logi

c

X W

ebse

rvic

es

Strik

elron

emai

ls

CDYN

E

Web

serv

icex

Serv

ice o

bjec

ts

Strik

elron

addr

ess

Kick

box

Byte

pla

nt

Qui

ck em

ail

Tow

er d

ata

Lead

spen

d

Brite

ver

ify

Mai

l box

Emai

l ans

wer

s

Buli

emai

l ver

ifier

Bulk

emai

l che

cker

Emai

ltor

Verif

alia

Xver

ify

WsRF versus non-FB enabled versus FB enabled

Figure 9 e comparison of WsRF non-FB and FB-enabled web service selection

14 Journal of Computer Networks and Communications

also affirmed by the extensive feedback range of qualitymaintenance -e binomial probability distribution usedby the recommendation system predicts in such a mannerthat web services are arranged orderly even to the leastsignificant value to bring about the needed differences todifferentiate a web service from another one of similarcomputation Hence the whole experiments were focusedon improving the user satisfaction in the AMC as thiswill promote frequent patronage and regular interest inconsumption of services on the GUIISET businessplatform

Summarily multidynamic and feedback mechanismshave contributed to the knowledge base of AMC serviceselection in the following ways

(1) Outrightly removal of the possibility ofdelayredundancy as a result of selection ties

(2) Breaking of service selection ties whenever theyoccur

(3) Ensuring that an optimal service is selected in anyservice requisition

(4) Promotes prompt service delivery via the multi-dynamic distance approach

(5) Increasing the user satisfaction level in the course ofservice provisioning

8 Conclusion and Future Work

Many services in the AMC platform are with similarfunctional qualities which open up the understanding thatthe effective and efficient use of a various combination ofnonfunctional QoS will enhance higher user satisfactionthrough provision of optimal services We performed a se-ries of experiments within interconnected mobile networknodes as well as P2P mobile devices Our experimentsproved that distances within mobile networks can be animportant QoS tool that will enhance the optimal selectionprocess which is typically useful in the case of emergencyneeds for example natural disaster mining process andhealth monitoring systems

Moreover the user feedback effect on web services can becomputed to effect proper ordering of these services within themobile devices in such a manner that eradicates selection ties-e outcome of the experiments showed that out of sevenselection samples only one is likely to produce an optimalselection without the feedback mechanism with the calculatedprobability of 013 -us we can conclude that the use of thefeedback mechanism section of the proposed system enhancesbetter performance in usersrsquo satisfaction in comparison withexisting works such as WsRF [45] and nonfeedback [47]

Possible future works need to consider the context of thedevice and mobile environment -us inculcating the en-vironmental context as well as the device context into theselection process is a good note for the furtherance of thiswork with proper checks in place to avoid complexity issuesAnother area for future work includes the implementation ofan autoswitching system into themechanism to switch to thenew server during failure

Conflicts of Interest

-e authors declare that there are no conflicts of interestregarding the publication of this paper

Acknowledgments

-is work is based on the research support in part by theNational Research Foundation of South Africa (Grant UIDtp11062500001 (2017-2018)) -e authors also acknowledgefunds received from the industrial partners Telkom SA LtdHuawei Technologies SA (PTY) and Dynatech InformationSystems South Africa in support of this research

References

[1] R Yu X Yang J Huang Q Duan Y Ma and Y TanakaldquoQoS-aware service selection in virtualization-based cloudcomputingrdquo in Proceedings of 14th Asia-Pacific NetworkOperations and Management Symposium (APNOMS) pp 1ndash8Seoul Republic of Korea September 2012

[2] M SathyaM Swarnamugi PDhavachelvan andG SureshkumarldquoEvaluation of QoS based web-service selection techniques forservice compositionrdquo International Journal of Software Engineeringvol 1 no 5 pp 73ndash90 2012

[3] S Marston Z Li S Bandyopadhyay J Zhang andA Ghalsasi ldquoCloud computingmdashthe business perspectiverdquoDecision Support Systems vol 51 no 1 pp 176ndash189 2011

[4] G Zou Q Lu Y Chen R Huang Y Xu and Y Xiang ldquoQoS-aware dynamic composition of web services using numericaltemporal planningrdquo IEEE Transactions on Services Comput-ing vol 5 no 3 pp 1ndash14 2012

[5] B Martini F Paganelli A A Mohammed M GharbaouiA Sgambelluri and P Castoldi ldquoSDN controller for context-aware data delivery in dynamic service chainingrdquo in Pro-ceedings of 1st IEEE Conference on Network SoftwarizationSoftware-Defined Infrastructures for Networks Clouds IoTand Services NETSOFT pp 1ndash5 London UK April 2015

[6] E E Marinelli ldquoHyrax cloud computing on mobile devicesusing MapReducerdquo vol 389Pittsburgh PA USACarnegieMellon University September 2009 MS thesis -esis

[7] F AlShahwan and M Faisal ldquoMobile cloud computing forproviding complex mobile web servicesrdquo in Proceedings of2nd IEEE International Conference on Mobile Cloud Com-puting Services and Engineering pp 77ndash84 Oxford UKApril 2014

[8] N Kaushik and J Kumar ldquoA literature survey on mobilecloud computing open issues and future directionsrdquo In-ternational Journal of Engineering and Computer Sciencevol 3 no 5 pp 6165ndash6172 2014

[9] K Bahwaireth and L Tawalbeh ldquoCooperative models in cloudand mobile cloud computingrdquo in Proceedings of 23rd In-ternational Conference on Telecommunications (ICT 2016)pp 1ndash4 -essaloniki Greece May 2016

[10] G Huerta-Canepa and D Lee ldquoA virtual cloud computingprovider for mobile devicesrdquo in Proceedings of ACM Work-shop on Mobile Cloud Computing amp Services Social Networksand Beyond pp 35ndash39 San Francisco CA USA June 2010

[11] A T Akinola M O Adigun A O Akingbesote andI N Mba ldquoOptimal service selection in ad-hoc mobile marketbased on multi-dynamic decision algorithmsrdquo in Proceedingsof 2nd World Symposium on Computer Networks and In-formation Security pp 1ndash6 Hammamet Tunisia 2015

Journal of Computer Networks and Communications 15

[12] A T Akinola M O Adigun A O Akingbesote andI N Mba ldquoOptimal route service selection in ad-hoc mobileE-marketplaces with dynamic programming algorithm usingTSP approachrdquo in Proceedings of International Conference onE-Learning Engineering and Computer Softwares pp 74ndash81Kualar Lampur Malaysia 2015

[13] J Cao K Hwang K Li and A Y Zomaya ldquoOptimal mul-tiserver configuration for profit maximization in cloudcomputingrdquo IEEE Transactions on Parallel and DistributedSystems vol 24 no 6 pp 1087ndash1096 2013

[14] M E Buthelezi M O Adigun O O Ekabua and J S IyiladeldquoAccounting pricing and charging service models fora GUISET grid-based service provisioning environmentrdquo inProceedings of the 2008 International Conference one-Learning e-Business Enterprise Information Systems ande-Government (EEE 2008) pp 350ndash355 Las Vegas NV USAJuly 2008

[15] O O Ekabua andM O Adigun ldquoGUISET LogOn design andimplementation of GUISET-driven authorizationrdquo in Pro-ceedings of First International Conference on Cloud Com-puting GRIDs and Virtualization 2010 pp 1ndash6 LisbonPortugal November 2010

[16] E K Olatunji M O Adigun E Jembere J Oladosu andP Tarwire ldquoA privacy-as-a-service model for securing data inGUISET environmentrdquo in Proceedings of Southern AfricaTelecommunication Networks and Applications Conference(SATNACrsquo13) pp 143ndash148 Stellenbosch South Africa Sep-tember 2013

[17] A O Akingbesote M O Adigun S Xulu and E JembereldquoPerformance modeling of proposed GUISETmiddleware formobile healthcare services in E-marketplacesrdquo Journal ofApplied Mathematics vol 2014 Article ID 248293 9 pages2014

[18] A O Akingbesote M O Adigun J B Oladosu andE Jembere ldquoA quality of service aware multi-level strategy forselection of optimal web servicerdquo in Proceedings of IEEEInternational Conference on Adaptive Science and TechnologyICAST pp 1ndash6 Pretoria South Africa November 2013

[19] A O Akingbesote M O Adigun M A Othman andI R Ajayi ldquoDetermination of optimal service level in cloudE-marketplaces based on service offering delayrdquo in Pro-ceedings of 2014 IEEE International Conference on ComputerCommunication and Control Technology (I4CT 2014)pp 283ndash288 Langkwawi Kedah Malaysia September 2014

[20] T Shezi E Jembere and M Adigun ldquoTowards developingfailure tolerant communication framework for GUISET ser-vicesrdquo in Proceedings of the Southern Africa Telecommuni-cation Networks and Applications Conference (SATNACTrsquo11)East London International Convection Centre East LondonSouth Africa 2011

[21] A O Akingbesote M O Adigun E Jembere and J OladosuldquoModeling the cloud e-marketplaces for cost minimizationusing queuing moderdquo Australian Journal of Basic and AppliedSciences vol 8 no 4 pp 59ndash67 2014

[22] A O Akingbesote M O Adigun S S Xulu M Sanjay andI R Ajayi ldquoPerformance analysis of non-preemptive prioritywith application to cloud E-marketplacesrdquo in Proceedings ofIEEE International Conference on Adative Technolgy (ICAST)pp 1ndash6 Paris France 2014

[23] S Msane M O Adigun and O O Ekabua ldquoA reputation-based trust management model to enforce trust amongst webservices in a GUISET grid environmentrdquo in Proceedings ofInternational conference on Semantic Web abd Web Services(SWWSrsquo08) pp 230ndash235 Las Vegas NV USA July 2008

[24] M Swarnamugi ldquoTaxonomy of web service selection ap-proachesrdquo International Journal of Computer Applicationsvol 2 no 4 pp 18ndash22 2013

[25] M Klusch and P Kapahnke ldquoSemantic web service selectionwith SAWSDL-MXrdquo in Proceedings of Second InternationalWorkshop on Service Matchmaking and Resource Retrieval inthe Semantic Web (SMRR 2008) pp 1ndash15 Karlsruhe Ger-many October 2008

[26] S Susila S Vadivel and A Julka ldquoBroker architecture for webservice selection using SOAPUIrdquo in Proceedings of In-ternational Conference on Cloud Computing TechnologiesApplications and Management (ICCCTAM 2012) pp 219ndash222 Dubai UAE December 2012

[27] M Xin M Lu and R Zhang ldquoA location-based servicesselection model and algorithm with QoS constraints undermobile internet environmentrdquo in Proceedings of InternationalConference on Service Sciences pp 124ndash129 Jiangsu ChinaMay 2014

[28] M Amoretti M C Laghi A Carubelli F Zanichelli andG Conte ldquoReputation-based service selection in a peer-to-peer mobile environmentrdquo in Proceedings of IEEE In-ternational Symposium on a World of Wireless Mobile andMultimedia Networks (WOWMOM) pp 1ndash8 Newport BeachCA USA June 2008

[29] M Keidl and A Kemper ldquoTowards context-aware adaptableweb servicesrdquo in Proceedings of the 13th International WorldWide Web Conference on Alternate Track Papers amp PostersACM pp 55ndash65 New York NY USA May 2004

[30] N Keskes ldquoContext of qos in web service selectionrdquoAmericanJournal of Engineering Research vol 2 no 4 pp 120ndash1262013

[31] Y Liu A H H Ngu and L Z Zeng ldquoQoS computation andpolicing in dynamic web service selectionrdquo in Proceedings ofthe 13th International World Wide Web Conference on Al-ternate Track Papers amp Posters ACM pp 66ndash73 New YorkNY USA May 2004

[32] H Shah-Hosseini ldquoProblem solving by intelligent waterdropsrdquo in Proceedings of IEEE Congress on EvolutionaryComputation pp 3226ndash3231 Singapore September 2007

[33] A T Akinola and M O Adigun ldquoFeedback-based serviceselection in ad-hoc mobile cloud computingrdquo in Proceedings of3rd International Conference on Advances in Computing andCommunication Engineering (IEEE-ICACCE-16) pp 72ndash77Durban South Africa November 2016

[34] Z Zhang W Sun W Chen and B Peng ldquoAn integratedapproach to service selection in mobile ad hoc networksrdquoin Proceedings of 4th IEEE International Conference onWireless Communications Networking and Mobile Computing(WICOM 2008) pp 1ndash4 Dalian China October 2008

[35] A T Akinola M O Adigun and P Mudali ldquoA federated webservice selection approach for ad-hoc mobile cloud comput-ingrdquo in Proceedings of 19th Southern African Telecommunica-tion Networks and Applications Conference (SATNAC)pp 386ndash373 Frankhurt South Africa September 2016

[36] A Varshavsky B Reid and E de Lara ldquoA cross-layer ap-proach to service discovery and selection in MANETsrdquo inProceedings of IEEE International Conference onMobile Adhocand Sensor Systems Conference pp 66ndash73 Washington DCUSA November 2005

[37] R Lacuesta J Lloret S Sendra and L Pentildealver ldquoSpontaneousad hoc mobile cloud computing networkrdquo Scientific WorldJournal vol 14 no 10 pp 1ndash19 2014

[38] K Yang A Galis and H-H Chen ldquoQoS-aware service se-lection algorithms for pervasive service composition inmobile

16 Journal of Computer Networks and Communications

wireless environmentsrdquo Mobile Networks and Applicationsvol 15 no 4 pp 488ndash501 2010

[39] E Cavalcante T Batista F Lopes et al ldquoOptimizing servicesselection in a cloud multiplatform scenariordquo in Proceedings ofthe IEEE Latin America Conference on Cloud Computing andCommunications LatinCloud pp 31ndash36 Porto Alegre BrazilNovember 2012

[40] M Makhlughian S M Hashemi Y Rastegari and E PejmanldquoWeb service selection based on ranking of QoS using as-sociative classificationrdquo International Journal on Web ServiceComputing vol 3 no 1 pp 1ndash14 2012

[41] J Zhu Y Kang Z Zheng and M R Lyu ldquoA clustering-basedQoS prediction approach for web service recommendationrdquo inProceedings of 15th IEEE International Symposium on ObjectComponentService-Oriented Real-Time Distributed ComputingWorkshops pp 93ndash98 Shenzhen China April 2012

[42] L Zeng B Benatallah A NguM Dumas J Kalagnanam andH Chang ldquoQoS-aware middleware for web services com-positionrdquo IEEE Transactions on Software Engineering vol 30no 5 pp 311ndash327 2004

[43] N Kokash and V DrsquoAndrea ldquoEvaluating quality of webservices a risk-driven approachrdquo in Proceedings of BusinessInformation System Conference pp 180ndash194 Poznan PolandApril 2007

[44] O Kondratyeva N Kushik A Cavalli and N YevtushenkoldquoEvaluating quality of web services a short surveyrdquo in Pro-ceedings of IEEE 20th International Conference on Web Ser-vices pp 587ndash594 Santa Clara CA USA June 2013

[45] E Al-Masri and Q H Mahmoud ldquoQoS-based discovery andranking of web servicesrdquo in Proceedings of 16th InternationalConference on Computer Communications and Networkspp 529ndash534 Honolulu HI USA August 2007

[46] R P Singh and K K Pattanaik ldquoAn approach to compositeQoS parameter based web service selectionrdquo Procedia Com-puter Science vol 19 no 2 pp 470ndash477 2013

[47] S Reiff-Marganiec H Q Yu and M Tilly ldquoService selectionbased on non-functional propertiesrdquo Service-OrientedComputing-ICSOC 2007 Workshops pp 128ndash138 SpringerBerlin Germany 2009

[48] A Akinola ldquoQoS-based web service selection mechanism forad-hoc mobile cloud computingrdquo University of ZululandKwaDlangezwa South AfricaUniversity of Zululand 2017MSc dissertation -esis

Journal of Computer Networks and Communications 17

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom