RESEARCH ARTICLE A Study of Qulity of Service Resource Alloction Schemes for Moile Helth Idris Krim nd Tr Ali Yhiy* Deprtment of Computer Science nd Engineering, University of Kurdistn Hewler, Kurdistn Region F.R. Irq *Corresponding uthor s emil: t.irhim1@ukh.edu.krd Received: 05 Novemer 2017 Accepted: 04 Decemer 2017 Aville online: 23 Decemer 2017 ABSTRACT In the perspective of helthcre, mhelthcre or moile helth cre is very importnt since it hs the ility nd the fesiility to tke cre of people s helth esier thn efore. mhelth is comintion of the ppliction on smrtphone tht monitors the ptient s helth nd the communiction technology tht trnsmits the dt of the ppliction to the predefined destintion which is the hospitl or the doctors. To investigte resource lloction mong users in the context of mhelth, different kind of scheduling lgorithms hve een used to study nd nlyze the ehvior of results tking into considertion diverse scenrios, types of pplictions nd lod of the network. Simultion results nd deep nlysis were done throughout the rticle, nd they demonstrted interesting conclusions out the comintion of ll prmeters tht were studied in this rticle. Keywords: mhelth, Qulity of Service, Scheduling, Universl Moile Telecommuniction System 1. INTRODUCTION Nowdys, mhelth hs ecome glol phenomenon nd controlled moile communiction power in helthcre. The reltion etween helthcre nd moile communiction hs ecome criticl fctor in sving humn life s well s resulting in emerging nd development of new mrkets. In the perspective of the World Helth Orgniztion, definition of helth is stte of complete physicl, mentl, nd socil welleing, not merely the sence of disese or infirmity. Helthy society comes from helthy individuls, ecuse helthy people improve the society y providing etter lifestyle, through their hrdworking, eduction, less time sence, higher concentrtion, etc. Successful helthcre systems produce helthier citizens, decrese lod on hospitls, nd therefore, increse the proportion of helthy popultion (Bjørnlnd et l., 2012). Access this rticle online DOI: 10.25079/ukhjse.v1n1y2017.pp60-66 E-ISSN: 2520-7792 Copyright 2017 Kmir, et l. Open Access journl with Cretive Commons Attriution Non-Commercil No Derivtives License 4.0 (CC BY-NC-ND 4.0). At present, most of the cell phones re smrtphones tht re used very widely ll over the world. According to the development technology, ccessing these devices hs ecome very esy. Applictions of mhelth increse continuously s it hs ecome esier for users to enefits from these pplictions compred to the older tools nd methods. The criticl spect of mhelth is tht, it is importnt to improve helthcre plns nd policies. Hence, the im is to progress nd crete tools to mnge ptient s helthcre utomticlly. The min ojectives of mhelth re chronic disese mngement, Personl Helth Record pplictions, ehviorl chnge, helth informtion, remote monitoring, etc. (Konschk nd Dvloor, 2011). There is significnt reltionship etween telecommuniction nd mhelth re. Moile network re mde so esier to look fter ptients through their moiles, notly specil cse-ptients. Universl Moile Telecommuniction System (UMTS) network hs gret impct in this re. mhelth pplictions depend on scheduling lgorithms which re used in se sttions (Node B). Bse sttions hve direct impct s one of the network elements. In the re of moile networks, vrious scheduling lgorithms cn e found, nd 60
ech one hs its importnce (Lkher nd Richriy, 2011) including; weighted fir queue (WFQ), priority queueing (PQ), custom queueing (CQ), modified round roin, nd proportionl firness. In this rticle, we review nd investigte the performnce of different scheduling lgorithms in 3G networks for mhelth ppliction under different context nd scenrios. The rticle is orgnized s follows: Section 2 descries the literture review on scheduling lgorithms in mhelth. Section 3 explins the simultion environment nd results nd section 4 concludes the rticle. 2. LITERATURE REVIEW Qulity of service (QoS) is the mechnism tht gurntees the service delivery from the source to destintion. This delivery must e cceptle in the perspective of time-frme nd should hve stility. By the QoS, the service performnce in terms of ccurcy, reliility, nd speed will e mesured. In network re, QoS provides multiple network services to multiple networks trffics. QoS uses different prmeters for different services such s dely vrition, ndwidth, pcket loss rtio, nd error rtion. QoS predicts nd mnges which prmeter is delivered erlier or vice vers. It uses circuit switching for rel-time pplictions which is more relile with system cpcity. There re two well-known services in QoS including; differentited services (DiffServ) which re chllenging to solve QoS prolems without resource estlishment reservtion. Controlling of DiffServ is in the IP heder which is clled DiffServ Code Point (DSCP). Regrdless the trffic flow stte, the DSCP wre routers perform trffic differentition for received pckets. QoS levels which hve een promised y QoS mechnisms is ppliction requirements tht re conveyed y integrted service rchitecture. For resource reservtion protocol, which is responsile for QoS requirements, signling protocol is used. This will e done long pth from the sender to the receiver (Chioriu, 2004). To gurntee the QoS in ny network, there re mny methods such s QoS provisioning, dmission control, resource lloction, nd scheduling lgorithms. In this rticle, we re more focused on scheduling lgorithms to investigte their ehvior for the different types of mhelth pplictions regrdless their types: Rel-time nd non-rel time. 2.1. Scheduling Algorithms Scheduling lgorithms designing is criticl issue in cse of multiple-user delivery in networks, nd it is importnt for incresing QoS nd network performnce. The very importnt fctor for QoS prmeters throughput, dely, jitter, trffic received, trffic sent, etc., to e delivered in high qulity using scheduling lgorithms. There re mny chllenges in the design of scheduling lgorithms in moile networks. Due to the limited ndwidth, pcket loss rte, more pcket overhed, mny scheduling lgorithms hve to mke QoS provisioning in wireless networks. In the following, review of some well-known scheduling lgorithms such s WFQ, proportionl fir, CQ, PQ, nd mximum throughput (Fth nd Leung, 2002) cn e found. 2.2. First in First Out (FIFO) It is simplest scheduling lgorithm for resource lloction policy. It works exctly like FIFO Queue. If k is k th resource lock, i is i th user, t is current time, nd T i is time instnt in cse of i th request ws ordered. Eqution 1 elow represents FIFO scheduling lgorithm (Kumr nd Krg, 2011): FIFO i.k m = t T i 2.3. Round Roin (RR) It is procedure of shring resources in firness time to ll user. Cycle time performnce will e smll in cse of smll pcket size. It consists of more thn one seprted FIFO queue. T i is the difference etween FIFO nd RR while it represents lst time tht user is served in. In wireless systems, engged resource mount cnnot e depended lonely ecuse RR is not fir in terms of throughput even in good chnnel condition. As well, it trets the different type of ppliction regrdless their type (rel-time nd non-rel time) firly which is not gurnteeing the QoS of oth types (AdulAhd, 2012). 2.4. WFQ WFQ is well-known scheduling lgorithm. It supports QoS design y shring fir ndwidth depending on given weight nd ility to void strvtion nd shre resources sed on the weight of resources. Vrile length pckets re supported y WFQ, so the higher the resource is, the higher the weight. Similr to other schedulers, it protects ech flow from the other nd gurntees of end-to-end dely will e otined if the dt flow is in the form of leky ucket pproch. Trffics in WFQ hs n expectle service. Eqution 2 represents the WFQ (Piro nd Cpozzi, 2013; El Gili nd Tl, 2016): FIFO RR i.k = i * i.k m w m w i : Is specific weight i: Is the i th numer of user which is use to correct RR userspecific metric. m RR i.k : Is specific metric in RR for i th user. (2) 61
2.5. PQ PQ hs simple wy to offer multiple services to multiple pcket clsses. Received pckets hve their own different priorities, nd ech clss will e dded in seprte queue. Regrdless the witing time of the other queues, clsses tht contin highest priority pckets, re trnsmitted first through the output entrnce efore lower priority pcket clsses. Ech queue entirely forwrds its pckets sed on FIFO queue ses lthough it is good policy of queueing tht provides differentited service. However, it hs some weknesses such s strvtion of service in pckets with lower priority nd hving extreme dely in high priority trffic flow (AdulAhd, 2012) [Figure 1]. 2.6. CQ CQ is scheduling lgorithm uses weighted round roin in certin percentge for ndwidth lloction. It provides gurntee for ll protocols with miniml level of service in the network environment. Reserved ndwidth is used during trnsmission y permitting non-system queue, this is to gurntee service level for ll trffics. CQ hs the ility to dynmiclly llocte unused ndwidth for ny required protocols. It llows firness tht is not llowed y PQ. It is le to control unpredictle queued trffic y controlling interfce of ville ndwidth. Associted output queue, counts configurle yte to specify the mount of system dt delivering efore going to the next queue. During the process of queue, pckets re sent unit numer of yte exceeds. Counting exceeded numer of yte is provided y queue list (Zkriyy nd Rhmn, 2015). 3. SIMULATION ENVIRONMENT AND RESULTS In this work, OPNET Modeler 14.5 Acdemic version hs een used to uild nd test the defined scenrios for network topology (Kudtrkr nd Sonkusre, 2014). Simulted network consists of four different scenrios including; hevy nd light pplictions s well s smll nd lrge numers of UEs. We tke into considertion the network topology s illustrted in Figure 2. More precisely, we study the sme topology with nd without QoS prmeters, i.e., with nd without scheduling lgorithms. The sis of the scenrios re two fctors, the numers of UEs nd the lod of ppliction, i.e., how much type of trffic is sent during the session regrdless the type of ppliction, this implies the quntity of trffic sent from ech UEs to the core network of the UMTS, nd these scenrios re represented y the following: Figure 1. Priority queueing digrm (El Gili nd Tl, 2016) Figure 2. Network topology without qulity of service 62
Figure 3. () Averge emil trffic received (pcket/s). () Averge emil trffic sent (pcket/s) Figure 4. () Averge in video conferencing. Trffic received (pcket/s). () Averge in video conferencing trffic sent (pcket/s) Light 1: Scenrio using light pplictions nd 6 UEs Light 2: Scenrio using light pplictions nd 12 UEs Lod 1: Scenrio using pplictions nd 6 UEs Lod 2: Scenrio using pplictions nd 12 UEs We used three type of pplictions, Emil, video conferencing, nd voice over IP (VoIP). They re corresponding to the wy tht the ptient is using to send its dt to the network nd communiction with the doctor. 3.1. Scenrios Without QoS Considertion Regrding the different scenrios without QoS considertion, results hve een collected for ech ppliction, ccording to the following metrics used for ech ppliction. We selected trffic sent/received y Emil ppliction. In the light scenrios, the verge of trffic sent nd received is lmost the sme, since there is no lod in the network nd the network is in its idel cse, ut in loded scenrios, trffic sent is higher thn trffic received s shown in Figure 3 nd 3. As for video conferencing ppliction, loded scenrios with higher numer of users hve higher trffic sent nd hve much higher vlues of results in comprison to the light scenrios. In trffic received the light scenrio with higher numer of users hs more trffic received. In trffic received 63
Figure 5. () Averge voice, trffic sent (pcket/s). () Averge voice, trffic received (pcket/s) Figure 6. Averge in server emil, trffic sent (pcket/s) light scenrios hve higher result in comprison to the loded scenrios. Results re shown in Figure 4 nd. 3.1.1. VoIP ppliction As for VoIP Appliction, the scenrio with higher numer of users hs higher vlue for the results [Figure 5 nd ]. Server emil collects sttistics of e-mil ppliction trffic. According to the light nd hevy scenrios, the server emil includes trffic sent nd trffic Received. The sent trffic from the side of the server for oth scenrios; high nd low lods re lmost the sme, this is relted to the position of the server in the network s it does not tke prt of the trffic circultion, which mens tht it is only the entity tht sends Figure 7. Averge Jitter for ll UEs voice ppliction (s) the trffic nd it is not wre out the lod of the network. This is why the results of ll scenrios re lmost the sme [Figure 6 nd ]. There re different results for Jitter regrding different UEs in the network re shown in Figure 7. The vrition of jitter does not ffect the qulity of voice of trffic from ll the nodes of the network to the voice ppliction server s the difference is negligile. In totl cell downlink throughput nd totl cell uplink, light scenrios hve much higher results. In light scenrios, the higher numer of users hs the higher result. Regrding the loded scenrios, the less numer of users hve the higher results s shown in Figure 8 nd. 64
3.2. Scenrios under QoS Considertion with Different Types of Scheduling Algorithms In this section, we use different scheduling lgorithms to schedule resources to the different UEs with the different pplictions in the cell. Three scheduling lgorithms hve een compred together, s well s they re compred to the results of the scenrios without QoS considertion. The scheduling lgorithms re; (i) WFQ, (ii) Custom Queueing, nd (iii) PQ. Different metrics of different pplictions re used including; emil, Video Conferencing, Voice s n ppliction result, nd emil Server, UMTS Cell Node B, nd UMTS Cell RNC s network elements (cell) result. According to the scheduling lgorithms, ech ppliction hs its effect, for instnce for emil pplictions est-otined result is for PQ nd the worst cse is for non QoS considertion scenrios while in voice est cse is otined for Custom Queueing, nd the worst cse is when QoS is not considered. In generl, est result hs een otined in WFQ nd worst result in the cse when QoS is not considered hs een otined [Tle 1]. 4. CONCLUSION This work cn e considered s most extensive work descriing QoS in UMTS. Even though, the scheduling lgorithms re very well known in the literture, ut it ws Figure 8. () UEs Universl Moile Telecommuniction System (UTMS) cell totl cell downlink throughput (pcket/s). () UEs UMTS cell totl cell uplink throughput (pcket/s) Tle 1. Comprtive study nd numericl results Appliction QoS prmeters With QoS Without QoS CQ PQ WFQ Emil Trffic sent (pcket/s) 0.008008356 0.011014135 0.008573049 0.006746156 Trffic received (pcket/s) 0.008580147 0.013044601 0.010662851 0.009320277 Video conferencing Trffic sent (pcket/s) 57.48079009 38.33422277 47.87740367 47.89509439 Voice Pcket end to end dely 1.123708162 1.262379656 4.012161176 3.304099513 Voice Jitter (s) 0.082319108 0.08818447 0.081607347 0.083434111 Pcket dely vrition 0.161716886 0.153191918 1.032398954 0.213906852 Trffic received (pcket/s) 36.01005062 33.85857466 47.80811591 35.9156749 Trffic sent (pcket/s) 314.0062325 314.2310879 348.283347 314.6287906 Server Emil Trffic received (pcket/s) 0.00410533 0.006262756 0.004993621 0.004690154 Trffic sent (pcket/s) 0.004474816 0.005640508 0.004231354 0.004344788 Voice ppliction Voice Jitter 0.093051975 0.092920397 0.093049294 0.093074535 UMTS cell Totl cell downlink throughput (pcket/s) 162.0383074 148.2678996 193.5982075 152.7733964 Totl cell uplink throughput (pcket/s) 184.0255733 152.5627731 215.1141614 190.3080855 RNC Totl received throughput (pcket/s) 136.1711101 132.8733612 175.8565147 134.5973343 Totl trnsmit lod (pcket/s) 127.0149248 128.3554896 170.3477872 125.9567736 QoS: Qulity of service, CQ: Custom queueing, PQ: Priority queueing, WFQ: Weighted fir queue, UMTS: Universl Moile Telecommuniction System 65
never used in UMTS comined with OPNET. The min ojective in this rticle ws studying QoS in terms of resource lloction gurntee using UMTS network. This is to fulfill mhelth requirements in term of continuously looking fter ptients. Minly, we divided the investigted solutions into two prts; First prt (QoS not considered): Four scenrios, ccording to light nd hevy pplictions with low nd high numer of users. This is to test ll QoS prmeters for ll pplictions including, throughput, Dely, Jitter, Pcket end-to-end dely, trffic received, trffic sent, etc., regrding ech QoS prmeter. All otined results re plotted nd explined. Second prt (QoS considered): Regrding QoS resource lloctions, three different scheduling lgorithms re pplied for the sme scenrio s well s result of tht scenrio without QoS, nd ll four cses re compred to ech other. REFERENCES AdulAhd, M. (2012). Modifying round roin lgorithm for process scheduling using dynmic quntum precision. Specil Issue of Interntionl Journl of Computer Applictions, 6, 5-10. Bjørnlnd, D. F., Hnæs, K., & Kinu, T. (2012). The Socio-Economic Impct of Moile Helth. Indi: Boston Consulting Group, Telenour Group. Chioriu, C. (2004). QoS in UMTS. Sweden: Helsinki University of Technology. El Gili, M., & Tl, S. (2016). The effect of queuing mechnisms first in first out (FIFO), priority queuing (PQ) nd weighted fir queuing (WFQ) on network s routers nd pplictions. Wireless Sensor Network, 8, 77-84. Fth, H., & Leung, C. (2002). An overview of scheduling lgorithms in wireless multimedi networks. IEEE Wireless Communictions, 9(5), 76-83. Konschk, C., & Dvloor, R. (2011). Moile pplictions in helthcre. Divurgent, ispce. Kudtrkr, A., & Sonkusre, R. (2014). Performnce Anlysis of Routing Protocols for Rel Time Appliction. Interntionl Journl of Advnced Reserch in Computer nd Communiction Engineering, 7, 5072-5077. Kumr, A., & Krg, A. (2011). Queuing lgorithem sed qulity of service (Qos) for scheduling environment model in wimx network with opnet modeler. USA: Glol Journls Inc. Lkher, A., & Richriy, V. (2011). A Survey Report on Scheduling Algorithms for UMTS. Piro, G., & Cpozzi, F. (2013). Downlink pcket scheduling in LTE cellulr networks: Key design issues nd survey. IEEE Communictions Surveys nd Tutorils, 15(2), 678-700. Zkriyy, I., & Rhmn, N. A. (2015). Simultion of clss-sed weighted fir queue lgorithm on n IP router using OPNET. Journl of Soft Computing nd Decision Support Systems, 2(4), 1-5. 66