The Internatonal Arab Journal of Informaton Technology, Vol. 1, o. 1, January 213 1 QoS-Based Performance and Resource Management n 3G Wreless etworks n Realstc Envronments Aymen Issa Zrekat Department of Informaton Technology, Mu tah Unversty, Jordan Abstract: The thrd generaton networks lke Unversal Moble Telecommuncaton Systems (UMTS) offers multmeda applcatons and servces that meet end-to-end qualty of servce requrements. The load factor n the uplnk s crtcal and t s one of the mportant parameters whch has a drect mpact on the resource management as well as on the cell performance. However, n ths paper, the fractonal load factor n the uplnk and the total downlnk power are derved taken nto account the mult-path propagaton n dfferent envronments. The analyss s based on changng new parameters that affect the Qualty of Servce (QoS) as well as the performance, such as: servce actvty factor, energy-to-nose rato (Eb/), nterference factor, and the non-orthogonalty factor of the codes. The mpact of these parameters on the performance and the capacty as well as the total throughput of the cell s also nvestgated. It s shown n ths paper that n addton to the above parameters, the type of the envronment has a major effect on the nose rse n the uplnk as well as the total power n the downlnk. The nvestgaton s based on dfferent types of servces,.e., voce (conversatonal: 12.2kbt/s), packet swtched servces wth rates (streamng 64 and 128kbt/s) and (nteractve 384kbt/s). Addtonally, the obtaned results n ths paper are compared wth some smlar results n the lterature. Keywords: 3G wreless networks, QoS, rado resource management, UMTS, load factor, nose rse. Receved August 6, 21; accepted January 3, 211 1. Introducton Thrd generaton networks lke Unversal Moble Telecommuncaton Systems (UMTS) n Europe wll support multmeda servces accordng to the 3 rd Generaton Partnershp Project (3GPP) specfcaton [1, 2, 3, 16],.e., the supported bt rates wll be at least 144 Kbps for rural areas, 384Kbps for urban/suburban area and 2 Mbps for ndoor/low range outdoor envronment. UMTS s referred as Wdeband Code Dvson Multple Access (WCDMA) based, because Frequency Dvson Duplex (FDD) and Tme Dvson Duplex (TDD) are appled to pared and unpared bandwdth respectvely at 2GHz band. For Core etwork (C), the tradtonal crcut swtched network wll evolve to modern packet swtched network e.g., IP-based network. As the Emergng of nternet and moble applcatons, UMTS users are capable of accessng both telecom and nternet resources. Qualty of Servce (QoS) becomes a crtcal ssue for the success of UMTS and to provde end users wth perceptve QoS, the network resources at varous nodes must be optmally utlzed. Therefore, Rado Resource Management (RRM) plays an mportant role n the provson of UMTS servces. W-CDMA access networks, such as the one consdered n UTRA-FDD proposal [2], provde an nherent flexblty to handle the provson of future 3G moble multmeda servces. 3G wll offer an optmzaton of capacty n the ar nterface by means of effcent algorthms for rado resource and QoS management. RRM entty s responsble for utlzaton of the ar nterface resources and covers power control, handover, admsson control, congeston control and packet schedulng [3, 13, 18]. These functonaltes are very mportant n the framework of 3G systems because the system reles on them to guarantee a certan target QoS, to mantan the planned coverage area and to offer hgh capacty. The QoS provsonng for multmeda traffc has ganed nterest n the lterature n recent years, as the problem arses n the context of 2.5G and 3G systems and s not present n 2G systems. Thus, Das et al. [6] has developed a general framework for QoS provsonng by combnng call admsson control, channel reservaton, bandwdth reservaton and bandwdth compacton. Dxt et al. [8] among others have dscussed the evoluton scenaros from 2G to 3G networks and the QoS network archtecture proposal by 3GPP for UMTS. In ths respect, few studes algned to the 3GPP specfcatons are avalable n the open lterature [7, 17]. For the admsson control, several old and recent schemes have been suggested for the uplnk [4, 1, 15, 22, 23] under dfferent condtons and at a lower extent for the downlnk [12]. More recently, Ho et al. [9] have bult mathematcal models for varous call admsson schemes and have proposed an effectve lnear programmng technque
2 The Internatonal Arab Journal of Informaton Technology, Vol. 1, o. 1, January 213 for searchng a better admsson control scheme. However, the presented nvestgaton n ths paper s nnovatve n the sense that the capacty analyss n both the uplnk and downlnk s derved and studed n more realstc envronments where the mult-path propagaton s ncluded n the dervaton of the fractonal load factor as well as the nose rse. Moreover, the mpact of the load factor on the system capacty s studed by changng new mportant performance parameters. The rest of the paper s organsed as follows. In secton 2, the uplnk fractonal load factor s derved, whereas, the total downlnk power s derved n secton 3. The comparatve results come n secton 4 for both the uplnk secton 4.1 and the downlnk secton 4.2, whereas the comparson wth other results comes n secton 4.3, followed by the conclusons and the future work n secton 5. 2. Uplnk Load Factor Dervaton Accordng to the cell breathng phenomenon n the thrd generaton systems lke UMTS, whch s based on W-CDMA, when the load ncreases, the coverage area of the cell shrnks as the nterference s ncreased. Therefore, the uplnk nterference margn (.e., nose rse) s needed to estmate the coverage of the cell. The followng requrement must be fulflled by a user k: P k P PG = PG k > = E b I - P + I + I I - P + ε I + I ntra k nter ntra k ntra Where, p j : where j users connected to other cells I nter j : I p : where users connected to own cell ntra s other cell-to-own cell nterference factor. PG=W/R: s the processng gan of the used servce: W: s the total bandwdth, R s the bt rate. P k : s the mnmum receved power of user k. I ntra : s the own cell nterference. I nter : s the other cell nterference. I : s the thermal or the background nose. E b / : s the requred energy per bt to nose to the user at ode B. From equaton 1, we get: I p I I ntra k ntra PG E b P k 1 (1 ) P k PG E b I ntra I Then, the mnmum receved power P k of a user k can be reformulated as follows: 1 I P k 1 I ntra 1 PG ( E / ) 1 PG ( E / ) (1) (2) for a sngle cell case and assumng that all users are usng the same servce, then the total uplnk power can be calculated as follows [11]: I up I (1 ) P ntra k k 1 1 up Where the uplnk load equaton (.e., the uplnk capacty utlzed) s defned as: 1 (1 ) up k 1 PG E b Assumng that all users have the same servce and havng the same E b / then the load n equaton 4 can be smplfed further to: 1 (1 ) up 1 PG E b The voce actvty factor, v has a clear mpact on the uplnk nose rse. Ths value s usually taken to be.5 plus.17 capacty overhead from control channels, whch means that the voce actvty factor s taken to be.67 (5% voce actvty and DPCCH (Dedcated Physcal Control Channel) overhead), whereas, ths factor for data servce s consdered to be 1% [3, 11]. Therefore, the voce actvty factor should be ncluded n the load equaton as follows: 1 (1 ) PLenv d up 1PG E b v Recall from [24], where the capacty bounds are derved n dfferent propagaton envronments usng the extended verson of Hata (.e., Extended COST- 231)[5]. PL env (d ) s the path loss whch s gven n dfferent propagaton envronments, env (.e., urban, suburban, dense-urban or rural) as a functon of the dstance, d ncludng the shadowng effects. In W- CDMA coverage and capacty are closely related. Therefore, satsfed user s the one who s able to transmt a power whch s enough to cope wth the (E b / ) at the base staton and less than the maxmum transmsson power, P k,max, even f the user s located at the edge of the cell and t s possble to get the sutable value of max n order to mantan the coverage. Then by substtutng equaton 6 n equaton 3 we get. P k,max PLenv d 1 I (1 ) 1 max 1 PG E b where, PL env (d ) s the path loss (ncludng shadowng effects) at dstance d, and max s the maxmum allowable load factor for assurng the requred coverage. (3) (4) (5) (6) (7)
QoS-Based Performance and Resource Management n 3G Wreless etworks 3 3. Downlnk Load Factor Dervaton The load factor n the downlnk follows the same steps and behavor n the uplnk wth small dfference [11], where the orthogonal factor of the codes,, must be ncluded n ths case as the codes n the downlnk are not fully orthogonal. Therefore, the downlnk fractonal load can be estmated accordng to equaton 5 above as follows: 1 ((1 ) ) PLenv d DL 1PG E b v The capacty of the uplnk s the crtcal one because n CDMA systems lke UMTS, the scarce resource s the transmsson power. However, gven the frequency mode FDD of UMTS, the power budget of the uplnk and the downlnk s ndependent. The power of the uplnk s lmted by the User Equpment (UE), whle n the downlnk t s lmted by the capabltes of ode B. In the downlnk, all users share the same ode B and therefore, the total downlnk power s shared between them, the more users, the less the power per user. Accordng to the above and from equaton 1, the DL power requred for user s gven as follows: Where: and, Where: ( I I ntra I nter ) ( E b / ) v PT PG ( PT PT ) I nter PLenv ( d ) I ntra PT PLenv ( d ) PT : s the downlnk power requred for user. PT: s the total T x power n the downlnk. : s the non-orthogonal factor of the code n the downlnk. By substtutng equatons 1 and 11 n 9, the DL power requred for user can be reformulated as follows: ( E ) I L ( d ) ( ) PT PT b v p PT PG Lp ( d ) The total DL power represents the transport channels (.e., the sum of the common channels and dedcated channels [14]. Hence, PT PCCH PT (8) (9) (1) (11) (12) (13) Where P CCH s the total power used for downlnk common channels and s the number of users, then from equaton 13, PT can be redefned as follows: From equaton 14: PT PT PCCH PT PT PCCH By substtutng the value of PT n equaton 12 nto equaton 15 n order to get the total power n the downlnk n the left hand sde, as follows: I PL ( d ) ( E ) v PCCH ( PG PL ( d ) ) PT env b env ( PG PL ( d ) ) ( E ) v ( ) env b 4. Comparatve Results 4.1. Sngle Cell Uplnk Analyss Ths secton presents a set of results n the uplnk for a conversatonal voce servce (12.2kbps AMR). However, Fgures 1 and 2 represent the uplnk nose rse for data servce (.e., PS: 64kbt/s and 128kbt/sec). Fgure 3 representng the path loss aganst the number of users n dfferent propagaton envronments. Although there are large numbers of parameters that can be consdered, the followng results wll concentrate on the nfluence of three man parameters: servce actvty factor, v, the uplnk nterference factor, ε, and energy-to-nose rato, Eb/, Fgures 4, 5 and 6 respectvely. These results are generated usng MATLAB package whch s a very powerful tool for smulaton and other mathematcal computatons. Table 1 show the parameters used n ths study unless t s specfed explctly. Requred E b / Table 1. Default system parameters [11, 21]. Parameters Moble maxmum transmsson power Thermal nose Inter-cell nterference to own cell power rato, Values 3, 4, 5, 6,7,8 db (varable of study) 125 mw -13 dbm.5,.55,.67 (varable of study) Servce actvty factor (v) of voce.67 Servce actvty factor (v) of data 1 Wave length.15 Heght of the moble Heght of the base staton WCDMA chp rate Voce AMR (adaptve mult rate) servce Packet swtched (PS) data servce on-orthogonalty factor n the DL 1.5 m 5 m (3.48 Mcps) or 384 Hz 12.2 kb/s 64, 128, 384 kb/s (16).5,.7,.9 (varable of study) Block Error Rate (BLER) (for Data =.1, for Voce =.1) (14) (15)
4 The Internatonal Arab Journal of Informaton Technology, Vol. 1, o. 1, January 213 4.1.1. Uplnk ose Rse for Packet Swtched Data Servce Assumng E b / of 1.5dB and (.e., own cell to other cells nterference) s assumed to be.5, and the data actvty factor s taken to be 1 as assumed before. Fgures 1 and 2 show two examples of the nose rse for packet swtched data servce. In Fgure 1, the bt rate s assumed to be 64kbt/s whereas n Fgure 2, t s assumed to be 128kbt/s. The nose rse of approxmately 4dB s equvalent to 5% load factor and the nose rse of 7dB s equvalent to 75% load factor. Therefore, and accordng to Fgure 1, the throughput of 175kbt/s s supported by 4dB nose rse whereas the throughput of approxmately 275kbt/s s supported by 7dB nose rse. However, n Fgure 2, the same parameters used n Fgure 1 are used here except that the bt rate s taken here to be 128kbt/s. It can be notced from Fgure 2 that double value of throughput s supported by the same nose rse; the throughput of 3kbt/s s supported by 4dB nose rse and the throughput of 55kbt/s s supported by 7dB nose rse. Fgure 1. Uplnk nose rse as a functon of uplnk data throughput (PS: 64kbt/s). On the other hand, the loss s expected to be hgher n other envronments (.e., urban, dense-urban) as the buldngs are expected to be hgh and therefore, the LOS would not be clear. As the number of users ncreases, the path loss becomes hgher, Fgure 3 explans ths stuaton. Fgure 3. Path loss aganst dstance n dfferent propagaton envronments. 4.1.3. The Effect of the Servce Actvty Factor, v on the ose Rse Fgure 4 demonstrates the effect of the servce actvty factor, v on the uplnk nose rse as well as the system capacty. The rest of the parameters are kept fxed as mentoned n the ttle part of the Fgure (.e., ε=.5, Eb/=4dB). It s clear from Fgure 4 that the servce actvty factor has a clear mpact on the uplnk nose rse. When the servce actvty factor ncreased from.5 to 1. the uplnk nose rse ncreases from approxmately 6dBm to 12dBm nose rse, where the maxmum number of actve users (.e., system capacty) acheved at ths range s 6. Ths explans the drect mpact of the servce actvty factor on the nose rse; the hgher the actvty servce factor, the hgher the nose rse and the better the capacty. Fgure 2. Uplnk nose rse as a functon of uplnk data throughput (PS: 128kbt/s). 4.1.2. Uplnk Path Loss n Dfferent Propagaton Envronments Fgure 3 shows the path loss n dfferent propagaton envronments. It vsualzes the path loss aganst the number of users. It s clear that the loss n the free space envronment s low compared wth other envronments as the Lne-Of-Sght (LOS) between the transmtter and the recever s clear wth no obstacles. Fgure 4. Uplnk nose rse aganst number of users for dfferent servce actvty factors. 4.1.4. The Effect of the Uplnk Interference Factor, on the ose Rse Fgure 5 demonstrate the effect of the nterference factor,, on the uplnk nose rse as well as the system capacty. Two parameters are kept fxed (.e., v=.67, Eb/=4dB). It s clear from Fgure 5 that the hgher
QoS-Based Performance and Resource Management n 3G Wreless etworks 5 the nterference factor, the hgher the nose rse. For example, at nterference factor, =.5, the maxmum number of 6 users can be reached at nose rse of approxmately 6.dBm, whle at hgher nterference (.e., =1.), the same number of users s reached at about 9.dBm nose rse. Ths explans the fact that the nterference factor n the uplnk has also a drect mpact on the nose rse. The hgher the nterference factor, the hgher the nose rse, whch means that the system reaches ts maxmum capacty. However, n the rest of the Fgures (.e., Fgures 1, 11, 12 and 13) urban envronment s consdered n the analyss and the servce s taken to be PS wth 384kbt/s. 4.2.1. Total Downlnk Power n Dfferent Propagaton Envronments It s clear from Fgures 7, 8 and 9 that not only the user receved power and the capablty of the base staton has an effect on the qualty of servce, but also the type of envronment and the propagaton condtons have an mpact. It s clear from Fgures 7, 8 and 9 that as we have a clear envronment (.e., free space) the total receved power at the base staton wll be hgh regardless of the bt rate. However, the maxmum power s acheved when we have PS 384kbt/s. It s around 47.5dBm n the free space and 41dBm n the dense urban area. The followng nvestgaton s assumed only n urban envronment. Fgure 5. Uplnk nose rse aganst number of users for dfferent nterference factors. 4.1.5. The Effect of Eb/ on the ose Rse It s clear from Fgure 6 that Eb/ has a clear mpact on the nose rse. As Eb/ ncreases, the nose rse s also ncreases. For example, 3 users can have servce wth 3dBm nose rse at Eb/=3dB. On the other hand, 3 users can have servce wth about 7dBm nose rse at Eb/=6dB. Of course, choosng Eb/ s a plannng ssue and t has to be decded by UMTS specfcatons. Fgure 7. Total downlnk power aganst the number of PS users n dfferent envronments. Fgure 6. Uplnk nose rse aganst number of users for dfferent Eb/ values. 4.2. Downlnk Analyss The analyss n the downlnk s based on Packet- Swtched (PS) wth rates: 64kbps, 128kbps and 384kbps. There are seven Fgures n ths analyss: Fgures 7, 8, 9, 1, 11, 12 and 13. All Fgures are generated for the total downlnk power aganst the number of actve users n dfferent propagaton envronments (.e., free space, urban, dense urban, suburban and rural) n Fgures 7, 8 and 9. Fgure 8. Total downlnk power aganst the number of PS users n dfferent envronments. Fgure 9. Total downlnk power aganst the number of PS users n dfferent envronments.
6 The Internatonal Arab Journal of Informaton Technology, Vol. 1, o. 1, January 213 4.2.2. The Effect of Interference Factor, ε The effect of the nterference factor s shown n Fgure 1. It can be seen from Fgure 1 that as the nterference factor between users ncreased, the total downlnk power wll be hgher as ncreasng the nterference between users means that each user has to ncrease hs power n order to acheve the requred Eb/ at ode B. Fgure 12. The effect of the voce actvty factor, v on the total downlnk power n urban envronment. Fgure 1. The effect of the nterference factor, ε on the total downlnk power n urban envronment. 4.2.3. The Effect of Orthogonalty Factor, α It s clear from Fgure 11 that the orthoganalty factor between the users n the downlnk has a clear mpact on the total downlnk power. Ths factor s very mportant as the codes n the downlnk are not orthogonal. However, the perfect orthogonalty beween the codes can be acheved when we have α=1, but ths s not realstc assumpton n real lfe. It can be seen from Fgure 11 that when we have perfect orthogonalty (α=1), the maxmum receved power s acheved (around 49dBm). Whle wth realstc orthogonalty factor (α=.5) only around 41dBm total receved power s acheved. 4.2.5. The Effect of Eb/ The requred Eb/ has a clear mpact on the total downlnk power. Ths effect can be seen from Fgure 13. As the requred Eb/ ncreases, the total receved power s gettng hgher. For example, 4 users are served at power 39dBm when we have Eb/=5.5dB. On the other hand, 4 users are served at power 41dBm when we have Eb/=8.8dB. Fgure 13. The effect of Eb/ on the total downlnk power n urban envronment. 4.3. Comparson wth Exstng Results The results obtaned n ths paper are compared wth some smlar results n the lterature [19, 2]. All presented results n [2] are assumed only an deal space envronment. Two Fgures are used n ths comparson (.e., Fgures 14 and 15). Fgure 14, shows the uplnk nose rse aganst the number of voce users. Fgure 11. The effect of the orthogonalty factor, α on the total downlnk power n urban envronment. 4.2.4. The Effect of the Servce Actvty Factor, v Fgure 12 shows the effect of the servce actvty factor, v on the total downlnk power. When the servce actvty factor s hgh, the total receved power reaches ts maxmum (.e., 3 users are served), t s 48dBm when we have the value of v=1., and t s 41dBm when the value of v=.5. Fgure 14. Uplnk nose rse (db).
QoS-Based Performance and Resource Management n 3G Wreless etworks 7 The author assumed n ths paper a value of 1% for the voce actvty factor, v whch s not realstc n a real network wth AMR of 12.2kbt/s, whereas the other-to-own cell nterference, s assumed to be 5%. It can be seen from Fgure 14 that the uplnk load s 2dB for 5 users n my results, whereas, t s 3dB for the same number of users n the compared results. The explanaton of ths s that when the uplnk nose rse s low, the capacty of the cell wll be better. My results become better when the number of users n the cell ncreases. The same parameters are used n Fgure 15, however, the voce actvty factor s assumed to be.67% (5% for voce actvty + dedcated control channel overhead). In Fgure 15, the total downlnk power s demonstrated. My curve shows better results than the compared one. It s clear that as the users reach the cell edge, the requred receved power by those users s ncreased whch leads to degradaton n the cell throughput. In [19], the nvestgaton s presented based on the downlnk pole capacty. Fgure 16. Total downlnk power n macro cell envronment. Fgure 17. Total downlnk power n macro cell envronment. Fgure 15. Total downlnk bower (dbm) n free space. Two Fgures are presented n ths comparson Fgures 16 and 17. In Fgure 16, the value of Eb/ s assumed to be 8, whereas the orthogonalty factor s taken to be 5% and the other-to-own cell nterference s assumed to be also 5%. Only macro cells envronment s assumed here. It can be seen from Fgure 16 that the curve for the total downlnk power gves better results compared to my results only when the number of users s low. As the number of users ncreases, the total receved power becomes better n my results than the compared results. Ths means that my model s able to mantan the downlnk power to a lower level even when the number of users s large. Consequently, the total cell throughput s also mantaned. Smlar results are demonstrated n Fgure 17, but n ths tme, the nterference factor, ε s taken to be.8. When ths factor s hgh, ths means that more users are comng to the cell from outsde, whch leads to ncrease n the total downlnk power requred by those users. 5. Conclusons and Future Work 3G wreless networks offer dfferent QoS guarantees and an optmzaton of the capacty n the ar nterface by means of effcent rado resource management algorthms. In ths paper, the uplnk and downlnk load factor n addton to the total receved power n the downlnk are derved and analyzed by means of a set of mportant parameters that have an mpact on 3G performance as well as the QoS; such as the path loss, the nterference factor, the actvty factor, the orthogonalty factor, energy to nose rato, as well as the capacty and the total throughput of the cell. The mpact of these parameters on the performance s studed n dfferent propagaton envronments wth dfferent types of servces; conversatonal, streamng and nteractve, n order to reflect the real lfe stuaton. These mportant parameters should be taken nto consderaton by the network desgners n the plannng ssues of the future of 3G networks. The obtaned results are compared wth exstng results n the lterature. The above analyss was only done for one cell. In the future, ths work could be extended to mult-cell analyss. Acknowledgment I would lke to thank Scentfc Research Support Fund (SRSF), Mnstry of Hgher Educaton and Scentfc Research, Jordan for the fnancal support.
8 The Internatonal Arab Journal of Informaton Technology, Vol. 1, o. 1, January 213 References [1] 3GPP TR 25.922 v4.1. (21-29), Rado Resource Management Strateges, Techncal Report, 21. [2] 3GPP TS 25.211, Physcal Channels and Mappng of Transport Channels onto Physcal Channels (FDD), avalable at: http://www.3gpp. org/ftp/specs/html-nfo/25211.htm, last vsted 21. [3] 3GPP TS 25.331 v4.., Rado Resource Control Protocol Specfcaton, avalable at: http://www.3gpp.org/ftp/specs/html-nfo/25331. htm, last vsted 21. [4] Began K. and Zrekat A., Interference Based CAC for Up-Lnk Traffc n UMTS etworks, n Proceedngs of World Wreless Congress, USA, pp. 298-33, 22. [5] COST: COST-231, Dgtal Moble Rado Towards Future Generaton Systems, Techncal Report, European Communtes, EUR18957, Chapter 4, Telecom Secretarat, Brussels, 1999. [6] Das S., Jayaram R., Kakan., and Sen S., A Call Admsson Control Scheme for Qualty-of- Servce (QoS) Provsonng n ext Generaton Wreless etworks, Wreless etworks, vol. 6, no. 1, pp. 17-3, 2. [7] De R., Imben D., Vgnal L., and Karlsson M., Load Control Strateges for Mxed Servces n WCDMA, n Proceedngs of the Vehcular Technology Conference, Tokyo, pp. 825-829, 2. [8] Dxt S., Guo Y., and Antonou Z., Resource Management and Qualty of Servce n Thrd Generaton Wreless etworks, IEEE Communcaton Magazne, vol. 39, no. 2, pp. 125-133, 21. [9] Ho C., Copeland A., Lea C., and Stüber L., On Call Admsson Control n DS/CDMA Cellular etworks, IEEE Transactons on Vehcular Technology, vol. 5, no. 6, pp. 1328-1343, 21. [1] Holma H. and Laakso J., Uplnk Admsson Control and Soft Capacty wth MUD n CDMA, n Proceedngs of Vehcular Technology Conference (VTC), Amsterdam, pp. 431-435, 1999. [11] Holma H. and Toskala A., WCDMA for UMTS: HSPA Evoluton and LTE, 4 th Edton, John Wley and Sons, 27. [12] Knutsson J., Butovtsch P., Persson M., and Yates D., Downlnk Admsson Control Strateges for CDMA Systems n A Manhattan Envronment, n Proceedngs of Vehcular Technology Conference, Canada, pp. 1453-1457, 1998. [13] Kuokkwee W., Othman M., Shamala S., and Arffn A., Enhanced Dynamc Bandwdth Allocaton Proportonal to Queue Length wth Threshold Value for VBR Traffc, The Internatonal Arab Journal of Informaton Technology, vol. 4, no. 2, pp. 117-124, 27. [14] Lo A., Hejenk G., and Bruma C., Performance of TCP over UMTS Common and Dedcated Channels, n Proceedngs of the 12 th IST Moble and Wreless Communcatons Summt, Portugal, pp. 138-142, 23. [15] Lu Z. and El-zark M.,, SIR Based Call Admsson Control for DS-CDMA Cellular Systems, IEEE Journal on Selected Area n Communcatons, vol. 12, no. 4, pp. 638-644, 1994. [16] Parasad R., Mohr W., and Konäuser W., (Ed.), Thrd Generaton Moble Communcaton System, Artech House, London, 2. [17] Rave W., Kohler T., Vogt J., and Fettwes G., Evaluaton of Load Control Strateges n UTRA/FDD etwork, n Proceedngs of the Vehcular Technology Conference, Greece, pp. 271-2714, 21. [18] Sallent O., Pereze-R J., and Sanchez R., Uplnk RRM for Conversatonal and Interactve Servces n UTRA-FDD, Kluwer Academc Publsher, 22. [19] Spla K., Honkasalo K., Laho-steffens J., and Wacker A., Estmaton of Capacty and Requred Transmsson Power of WCDMA Downlnk Based on a Downlnk Pole Equaton, n Proceedngs of IEEE 51 st Vehcular Technology Conference, Japan, pp. 12-15, 2. [2] Thrasvoulos G. and Trstan L., The Capacty of A WCDMA etwork: Case Study, Bechtel Telecommuncaton Techncal Journal, vol. 3, no. 1, pp. 73-78, 25. [21] UMTS etwork Coverage Plannng, avalable at: http://www.umtsworld.com/technology/coverage. htm, last vsted 21. [22] Zrekat A. and Al-began K., Soft Handover- Based CAC n UMTS Systems, n Proceedngs of IEEE Internatonal Conference n Telecommuncaton, French Polynesa, pp. 137-1312, 23. [23] Zrekat A. and Al-began K., Effcent etwork- Level Call Admsson Control Algorthm for 3G etworks, n Proceedngs of the 6 th Unted Kngdom Smulaton Socety Conference, Emmanuel College, UK, 23. [24] Zrekat A., Al-Began K., and Smth K., Comparatve Capacty/Coverage Analyss of CDMA Cell n Dfferent Propagaton Envronments, Wreless Personal Communcatons, vol. 28, no. 3, pp. 25-231, 24.
QoS-Based Performance and Resource Management n 3G Wreless etworks 9 Aymen Issa Zrekat s an Assocate Professor at the Informaton Technology Department, Mu tah Unversty, Jordan. He has obtaned hs BSc n computer scence from Yarmouk Unversty, Jordan n 199 and MSc n computatonal engneerng from Unversty of Erlangen, Germany n 2. Addtonally, he has obtaned hs PhD from Bradford Unversty, UK n 23. In January, 21, he has jont the performance modellng and engneerng research group at the Computng Department of Bradford Unversty, UK. Hs area of research s n the performance evaluaton and resource management of 3G wreless moble networks and beyond. He has more than 35 nternatonal refereed journal and conference papers n ths feld and he s responsble for revewng a set of papers n ths area of research n a very reputable Journals. He has publshed two mportant books n LAMBERT publsher. He s a member of some natonal coordnatng commttees; (such as the coordnatng commttee for the management of Queen Rana Al-Abdullah center for Educatonal Technology, also Commttee for Hgher Educaton Accredtaton Commsson, and Internatonal Organzatons. He has been apponted as an assstant professor at the Informaton Technology Department, Mutah Unversty, Jordan. In 29, he has been promoted to assocate professor n the same Unversty. He has been apponted as an assstant dean at the Faculty of Scence. Recently, he has been apponted as the charman of the Informaton Technology, Mu tah Unversty.