Multcast Transmsson rotocol for Multmeda Servces n Smart Home Jeong-Jun Suh, Shan Guo Quan, Jong-Ku ar, Jong Hyu ar 3 and Young Yong Km Department of Electrcal and Electronc Engneerng, Yonse nversty 34 Shnchon-dong, Seodaemun-gu, Seoul, Korea e-mal : {un, sgquan, y}@yonse.ac.r Hyunda Motor Company 77- Jangdu-dong, Hwasung-s, Gyeongg-do, Korea e-mal : de999@hyunda-motor.com 3 &D Insttute, HAWHA S&C Co., Ltd. Hanwha Blg. #, Jangyo-Dong, Jung-Gu, Seoul, Korea e-mal : hyus00@hanwha.co.r Abstract Common nformaton can be sent to multple users w Broadcast/Multcast servces. In smart home, multmeda servces can be provded to users, who want to receve multmeda data rough moble phones. Most of multmeda applcatons are mplemented usng e streamng technque from contents provder to subscrbers. It s mportant to manage e bufferng amount remaned n e user sde n order to support e adequate qualty of vdeo contents and mnmze complants of subscrbers due to buffer starvaton. In e wreless envronment, t s dffcult to provde several subscrbers w multcast traffc stably, snce e data rate whch users can receve vares accordng to e channel state of each user. Therefore, we propose and dscuss an effcent usage togeer w MI (Multcast In ncast porton as well as MIM (Multcast In Multcast porton n order to mze e aggregate roughput of users and mnmze e starvaton duraton degradng QoS supported to users. We consder e multcast parttonng problem applyng lst parttonng algorm and provde e analytcal foundaton at e proposed scheme can be evaluated. Smulaton results show at proposed model can provde enhanced performance and allowed streamng qualty. Keyword: buffer synchronzaton, channel stuaton, multcast partton. Introducton sers can utlze multmeda servces convenently rough er moble phones n smart home. Broadcast/Multcast servce to transmt e same nformaton to subscrbng users s beng standardzed n 3G []. In Broadcast/Multcast meod, a base staton perodcally broadcasts nformaton to e subscrbng users n e allocated porton of resource. The proft of Broadcast/Multcast servce s to save fnte transmsson resources because a base staton concurrently transmts pacets w a group rate whch s mnmum data rate allowed to e subscrbng users. Especally n CDMA x EV-DO system, a base staton perodcally broadcasts nformaton to e subscrbng users n e allocated porton of resource. It s essental to have a suffcent amount of bufferng as to ensure streamng qualty snce most of multmeda applcatons are mplemented to streamng technque. Streamng of multmeda applcatons over a wreless envronment s a dffcult problem. Ths s due to e strngent QoS requrements and e unrelablty of wreless ln [3]. In addton, such a perodc transmsson w a group rate n Broadcast/Multcast can cause several user termnals to endure undesrable streamng qualty owng to buffer starvaton because an allowed data rate accordng to channel stuaton s changeable due to tme varyng wreless channel characterstcs. Therefore, t s necessary at dverse requrements of user termnals are reflected n multcast transmsson rate rough an adequate parttonng. In s paper, we deal w e multcast partton problem for smart home applcatons. At at tme, e herarchcal codng meod wll acheve e buffer synchronzaton due to dfferent transmsson rates [4]. In addton, we should consder e polcy whch a few user termnals w bad channel stuaton contnually can be releved from. Therefore, we propose e meod whch apples to an addtonal transmsson n uncast porton togeer w perodc multcast delvery. Such a supplementary transmsson can be accomplshed by usage of spared uncast porton except e porton for multcast transmsson. The part of s paper was presented n []. Broadcast/Multcast servces are servces ntended to provde flexble and effcent mechansms to send common nformaton to multple users. The motvaton for ese servces les n e need to utlze e most effcent meod of delvery for e use of ar nterface and networ resources when sendng e same nformaton to multple users. The type of nformaton to be sent s not lmted to text messages, but could be any type of data, ncludng multmeda (e.g voce and streamng servces at mght be sent to moble statons. The nformaton shall orgnate va SMS or pacet data technques. Broadcast servces shall be delvered va e most effectve and effcent Ths wor was supported by HIS(Hybrd Intellgent Servce latform proect, funded by MOCIE(Mnstry of Commerce, Industry and Energy, Korea. (006508 3
technque (SMS or pacet data meods based on e nformaton to be sent. Multcast nformaton shall be delvered n e most effcent and effectve technque (.e. SMS or pacet data regardless of e orgnaton source. Ths means at f a multcast s orgnated va SMS, but e best delvery s pacet data due to content en ere needs to be nter-worng between SMS and pacet data. The servces shall allow for bo moble termnated and moble orgnaton for multcast and moble termnated only for broadcast. The servce needs to be defned n such a way as to not lmt e number of users n a user group, or e number of user groups avalable. Multcast user groups are generally consdered to be a closed group at a member eer explctly has to subscrbe to e servce (publc multcast group by sendng a request to e admnstrator, by some web nterfaces, or oer mechansm. A prvate multcast group s restrcted to membershp explctly by e admnstrator manually addng members. Broadcast can also be dvded nto publc and prvate groups. A publc broadcast group s used for sendng geographc specfc nformaton. All devces n e specfc geographc area at have broadcast capablty are n e publc group and wll receve s nformaton. Examples of broadcast nformaton for s group type are emergency weaer alerts are traffc condtons. rvate broadcast groups are targeted to sendng specfc nformaton to a specfc group of devces n a specfc area. Examples of s servce would be locaton based advertsng. A user elects to receve specfc advertsements when ey are at e mall, but not at oer tmes. [8] The remanng part of s paper s composed as follows. In secton, we descrbe e proposed protocol. In secton 3, we show e smulaton results about proposed scheme. In addton, we dscuss e performance of proposed scheme and examne t from varous angles. Secton 4 concludes s paper.. Descrpton of rotocol. The ole of Multcast artton and Herarchcal Codng We assume at e number of optmal partton s two and ese two parttons have tme slot of same porton (left and rght. At at tme, e decson of a partton pont appled to proposed scheme comples w e lst partton algorm [5]. Ths algorm offers e optmal partton meod to mze e aggregate roughput of group users. In e wreless envronment, f e users whch belong to e left partton move to e rght partton or vce versa, ere wll happen e synchronzaton problem due to e dfferent amount of bufferng. Multcast transmsson by dfferent data rate can not avod resultng n e synchronzaton problem. Therefore, we apply e herarchcal codng meod [4] n order to solve s problem. Herarchcal codng has been deal to multcast transmsson over lns w dfferent bandwds and channel envronment.. Descrpton of MIM and MI rotocol A base staton transmts pacets to subscrbng users perodcally durng e multcast part ( T M as shown n Fg.. We assume at a broadcast duraton conssts of one hundred slots and s duraton s called a All of user termnals can receve pacets for T n a frame M ( T S. Especally we call e transmsson n e T M nterval MIM (Multcast In Multcast porton and e transmsson n e T nterval MI (Multcast In ncast porton. The data rate whch e base staton can transmt n every frame s called a group rate. The group rate means e mnmum rate whch e subscrbng users can receve. Fg.. Frame structure of Broadcast/Multcast system The avalable rate of each user termnal changes accordng to channel envronment contnuously. Alough ere s only one user termnal whch has a bad channel stuaton n a group, e termnals whch have good channel stuaton relatvely can not help recevng pacets to e low data rate. In case of lve streamng contents, constant streamng rate at user sde should be sustaned and a suffcent amount of bufferng should be ensured n order to acheve guaranteed servce qualty. However, subsequent bufferng degradaton of group users due to a user termnal whch has a bad channel stuaton wll mae e servce qualty of multmeda applcaton fall off and frequent buffer starvaton wll cause e user termnals to rrtate. Therefore, we ntend to apply multcast partton meod so as to prevent some user termnals from recevng pacets to undesrable data rate. Frst of all, a base staton can fnd e user termnal at wll go rough a buffer starvaton n a current frame before transmttng pacets. The base staton preferentally determnes e user termnal w MI servce as mentoned above snce e base staton nows nformaton about e data rate at each user can receve and streamng rate at e user sde. Once, e base staton determnes several user termnals whch wll receve MI servce n e current frame only. However, e goal of our scheme prevously prevents some recevers from gong rough buffer starvaton n a current frame as well as n a next If e buffer starvaton s held n e next frame as well as n e current frame, streamng qualty at e user's dsplay wll be largely damaged contnuously. Our scheme taes nto account s case and en ncludes e channel stuaton n e next The decson of MI servce appled to our proposal gets acheved by e foundaton shown as followng Table. The MI decson of each user termnal s performed n case ere exsts e possblty at buffer starvaton wll be occurred n a current frame and n e multcast porton of a next Therefore, e decson pont s shown as follows. B ( t TM = [ B ( t currenttm L stm ] ( = B( t TM T s E( next TM L TM s < 0 ( 4
Table. Descrpton of arameters for MI Decson (t B s current L E ( next buffer occupancy of user n a current tme t streamng rate at e user sde group rate whch can receve n a current frame e left porton of multcast transmsson e rght porton of multcast transmsson expectaton value of group rate whch can receve n a next frame e amount for e decson of MI servce ( user In e next secton, we wll deal w e model for obtanng E ( next..3 Channel Estmaton Meod In e prevous secton, we estmated e channel stuaton of each user termnal n a next frame n order to obtan E ( next of (. We estmate e channel stuaton of a next frame usng Marov model w eleven states. elated model s shown as Fg. [6] and parameters are shown n Table, 3, and 4 [6]. Fg.. State transton dagram of HD system Table. Transmsson ate Supported n HD System ate (Kbps 3 4 5 6 7 8 9 0 38.4 76.8 0.6 53.6 04.8 307. 64.4 9.6 8.8 843. 457.6 Table 3. State Transton ate 3 4 5 6 7 8 9 0 ate 0.995 0.99 0.0 0.99 0.40 0.75 0.005 0.99 0.08 0.0 3 4 5 6 7 8 9 0 ate 0.0 0.5 0.0 0.53 0.5 0.995 0.0 0.74 0.0.0 Table 4. robablty whch a rocess emans n Each State 3 4 5 6 7 8 9 0 ate 0.00 0.0 0.04 0.08 0.5 0.5 0.8 0.094 0.3 0.06 0.005 We assume at e state dstrbuton of each user termnal n a current frame s } {,, LL}. {,,, LL The followng notatons can be defned n Table 5. Table 5. Descrpton of arameters for an Analytcal Model mn mn mn mn, e number of users w mnmum rate n e left porton e number of users w (mnmum rate one state n e left porton e number of users w mum rate n e left porton e number of users w mnmum rate n e rght porton e number of users w (mnmum rate one state n e rght porton state transton rate for upward and downward respectvely transmsson rate whch can receve n a state probablty at e users reman n a state porton of a specfc slot duraton In ese state dstrbutons, e left porton of multcast transmsson s parttoned n user termnals up to rate and e rght porton s parttoned n user termnals after rate. If ere s user n e left porton or e rght porton of current frame, e group rate n a next frame can be summarzed as follows. Left orton of current frame (a The probablty at a group rate s = { { mn mn mn ( } mn ( q } mn mn n a next mn mn p (b The probablty at a group rate s n a next mn mn mn mn = p ( q mn mn (c The probablty at a group rate s n a next q mn mn (3 (4 mn mn = ( (5 5
(d The probablty at a group rate s n a next = ( p (6 ght orton of current frame (a The probablty at a group rate s n a next = mn mn mn mn ( p ( q mn mn ( p (7 (b The probablty at a group rate s n a next mn mn = { ( } mn mn mn mn ( p [{ ( q mn mn { ( p }] mn mn ( q mn mn mn [{ ( q } { mn (c The probablty at a group rate s ( mn } ( p }] (8 n a next mn mn = (9 q mn Therefore, e expectaton value about a group rate n a next frame can be represented as follows. When user exsts n e left porton of current frame, E ( next =. (0 When user exsts n e rght porton of current frame, E( next =. ( We can obtan E value of equaton (. If we ( next assume at user termnals are provded w MI servce n a current frame and lned up orderly, e user termnal can receve e amount of data expressed n equaton (. = where r T ( As above (, alough MI servce s provded w e specfc user termnal, e user termnals whch data rate s hgher an at of e target user also can receve same data. Ths result s due to e broadcastng characterstc of wreless transmsson. Whle, value means e rato whch would assgn tme slots to user. In order to mnmze e average starvaton duraton of all actve user termnals, e followng condton (3 should be needed. Maxmze = where = r T (3 W s observaton, we may mprove e streamng qualty for all actve users f we desgn value carefully. We wll show some examples w dfferent value and dscuss each example va smulaton..4 Synchronous Mode In s secton, we deal w e buffer synchronzaton problem. The modfed protocol s proposed assumng not to apply e herarchcal codng meod. Fg. 3 shows at MI servce slots are dvded nto prorty MI slots ( T and shared MI slots ( T S and t also represents a modfed frame structure whch prorty MI porton ( T s added to. The role of T s to resolve e problem about buffer synchronzaton accordng to e movement between e left partton and e rght partton. The prorty MI porton s preferentally served f ere s e user whch needs e buffer synchronzaton. Oerwse, e prorty MI porton s converted nto e shared MI porton. In detal, e descrpton of modfed protocol s as follows. Case The users whch belong to left partton move to e rght partton n a current Fg. 4 shows e bufferng amount of e users whch exst each partton and e followng procedure descrbes e protocol behavor of case. ( The bufferng amount of user whch moves from e left partton to e rght partton. ( The bufferng amount of user whch exsts n e rght porton orgnally. (3 The bufferng amounts of ( and ( are synchronzed by e addtonal MI servce on e T porton. (4 If e bufferng amount of ( can not reach at of ( due to e lmted resource, e users of ( wll go 6
rough starvaton duraton for e unsolved bufferng porton. Case The users whch belong to rght partton move to e left partton n a current Fg. 5 shows e bufferng amount of e users whch exst each partton and e followng procedure descrbes e protocol behavor of case. ( The bufferng amount of user whch exsts n e left porton orgnally. ( The bufferng amount of user whch moves from e rght partton to e left partton. (3 The user of ( can not be served untl e bufferng amounts of each user are synchronzed. (4 If e users of bo ( and ( are not synchronzed at e end of frame, e remaned bufferng amount of ( should be dropped. As above e descrpton, e addton of prorty MI porton presents e possblty whch can solve e buffer synchronzaton problem. However, t s nevtable at e resource effcency s degraded due to e addtonal resource consumpton. Fg. 3. Frame structure of synchronzed broadcast/multcast system Broadcastng Servce rovder st Group nd Group slot BS Tme st Group nd Group Fg. 6. Smulaton scenaro of proposed scheme A base staton provdes total M moble users w lve broadcastng contents. Contents nformaton from servce provder s transmtted to a base staton w a Constant Bt ate (CB. A base staton transmts nformaton to subscrbng users w multcast group rate parttoned to two n e allocated porton of a We apply to e proposed algorm related to eep pace w MI meod. The assumptons of our smulaton are as follows. The default value of multcast porton s 30%. acet sze s fxed to 04 bt and transmsson rates supported to user termnals are fully rates. We assume at F envronment s aylegh fadng channel and pacet loss s gnored owng to usng e perfect error correcton. Traffc type of contents provder s CB source supported n S-, snce lve broadcastng contents are transmtted by usage of streamng technque. Streamng rate at e user sde s 00 Kbps. Start up delay s set to average 600 ms. 3. Smulaton of roposed rotocol ( The Varaton of Starvaton Duraton accordng to Multcast orton The frst smulaton s e case at 40 users receve e only MIM servce. Fg. 7 shows at starvaton duraton s decreased accordng to an ncrease of multcast porton. Fg. 7 also shows at e multcast servce of 45% porton barely can accomplsh e starvaton duraton less an 6%. These results show at e only MIM servce can not support good streamng qualty. Fg. 4. sage of prorty MI porton (L Fg. 5. Behavor for e buffer synchronzaton ( L 3. Smulaton esults We carry out smulaton to evaluate e performance of HD system applyng to e proposed algorm. We mae use of S- (etwor Smulator verson [7] at s popular for networ performance evaluaton as smulaton tool. Smulaton scenaro s shown as Fg. 6. Fg. 7. Starvaton duraton accordng to e varaton of multcast porton ( erformance comparson accordng to partton types We carry out smulaton to compare a fxed partton 7
mode w a movable partton mode. (See Fg. 8 In e fxed partton mode, e rght group rate s fxed to 64.4 Kbps and n e movable partton mode, t s determned from e lst partton algorm as above mentoned n secton. Smulaton results show at ere s no dfference between two partton meods accordng to e ncrease of e number of user termnal. Fg. 8. Starvaton duraton accordng to two partton meods (3 The effect of MIM servce togeer w MI servce In s secton, we dscuss a result when MI servce s used togeer w MIM servce. In order to compare MIM servce only w MIM togeer w MI servce farly, we appled to e equal amount of transmsson slots (45%. In MI servce, e channel predcton meod mentoned n e secton has not used. Fg. 9 shows at MIM servce togeer w MI servce has better performance an MIM servce only about 3-4%. wll be also served to user termnals at have better channel stuaton an selected user. = The umber of MI Servce Weght Allocaton The slot dstrbuton by s polcy s determned from value for MI servce decson. The larger value of each user has, e more tme slots allocated to each user are. Therefore, f amount of buffer s relatvely scarce, more MI servce wll be provded. Fg. 0 shows at weght allocaton polcy has e best performance of ree polces. However, e lmted transmsson resources do not affect e performance of ree polces (less an %. = Specfc Allocaton W s allocaton polcy, a base staton selects only one user termnal whch has e smallest value of all actve users and performs MI servce to e selected user. = arg Mn{,, 3,..., Fg. 0 shows at specfc allocaton polcy has not worse performance an oer polces. } Fg. 9. Effect of MIM and MI servce (4 erformance comparson accordng to e dfferent value n e MI servce In s secton, we dscuss e performance varaton accordng to allocaton polces n e MI servce. Fg. 0 shows e performance comparson about ree polces. Far Allocaton Far allocaton polcy equally provdes e MI servce users determned by a base staton w MI tme slots. Such a far allocaton does not tae nto account each user's buffer amount or channel envronments. W e characterstc of a wreless transmsson, e data pacets transmtted by tme slot allocated to each MI servce user Fg. 0. Starvaton duraton accordng to allocaton polces (5 Effect of channel predcton meod In s secton, we dscuss e effect of e channel predcton meod mentoned n e prevous secton. The channel envronment of a next frame s predcted by Marov modelng and e estmated channel nformaton provdes e foundaton at a base staton determnes MI servce users n a current Intutvely, such as e protocol can prevent user termnals w a bad channel from gong rough contnual buffer starvaton. Fg. shows e effect of channel estmaton. In Fg., MI servce decson by e channel estmaton of a next frame results n e enhanced performance about.5%. The combnaton of MI servce w channel estmaton meod maes e average starvaton duraton dropped 8
down as half an at of only MIM servce. Fg.. Effect of channel estmaton 3. Smulaton of Synchronous Mode In s secton, we carry out e smulaton n order to examne e performance varaton accordng to e addtonal MI slots n case e herarchcal codng s not used. Smulaton parameters are same as ose of settng up n e prevous secton. The frst smulaton result s about e relaton MIM porton and prorty MI porton. In Fg., each percentage of MIM porton and prorty MI porton means e upper bound whch can be served by e base staton. In oer words, each user may not consume e allocated slots f ey are not needed n e current Smulaton results show at e MIM porton of 30% and e prorty MI porton of 0% have e best performance. Fg. 3. Starvaton duraton accordng to prorty MI servce porton 4. Concluson In s paper, we suggest e new multcast transmsson protocol for smart home applcatons. Multmeda servces can be provded to users rough moble phones n smart home. The exchange of multmeda servces n smart home has been ncreased, and e useful multcast transmsson protocol s needed. Especally, we show at e proposed protocol s sutable for lve broadcastng applcatons to use streamng technque and to requre multcast transmsson. In e proposed protocol, we apply to lst partton algorm to frame based multcast partton and enhance e streamng qualty n case of a few subscrbers. In addton, MIM & MI servce s supplementary meod for smple perodc transmsson va multcast partton. We also show at MIM & MI servce has e enhanced performance n comparson w only MIM servce. Whle, we prove at e meods for slot allocaton n MI servce do not affect e streamng QoS n terms of buffer starvaton. In addton, we also execute e channel modelng of e next frame usng Marov model and analyze e expectaton probablty va bnomal concepts. The applcaton of channel predcton va Marov model leads to e reduced buffer starvaton about % and ndcates e possblty to be able to estmate e channel stuaton n e future. Fnally, we ntroduce e modfed protocol at taes nto account buffer synchronzaton problem. In e modfed protocol, we can obtan e trade-off relaton between MIM servce porton and prorty MI porton. Also we can fnd e adequate allocaton of prorty MI porton. eferences Fg.. Starvaton duraton accordng to each servce porton In e next smulaton, we deal w e streamng performance accordng to e varaton of prorty MI porton. In s smulaton, e porton of MIM servce s fxed to 30%. The rse of prorty MI servce porton can prevent some users from gong rough starvaton duraton caused by e mperfect buffer synchronzaton. However, Fg. 3 shows at e enhanced performance can not be obtaned largely n case of e prorty MI porton allocated over 0%. [] Jeong-Jun Suh, Shan Guo Quan, Jong-Ku ar, Jong Hyu ar, and Young Yong Km, "Multcast artton roblem for Smart Home Applcatons," Internatonal Worshop on Smart Home (IWSH 006, Korea, ovember 006. [] CDMA000 Hgh ate Broadcast-Multcast acet Data Ar Interface Specfcaton, 3G C.S0054 Verson 0.9, July 4, 003. 9
[3] F. H.. Ftze, and M. esslen, "A refetchng rotocol for Contnuous Meda Streamng n Wreless Envronments," IEEE Journal on Selected Areas n Communcatons, vol. 9, no. 0, pp. 05-08, Oct. 00. [4] George C. olyzos, Herarchcal Codng of Images and Contnuous Meda for Transmsson over acet- Swtchng etwors. CSD Techncal eport CS93-73, January 993. [5] ung-hung Gau, Z. J. Haas, and B. Krshnamachar, "On Multcast Flow Control for Heterogeneous ecevers," IEEE/ACM Transactons on etworng, vol. 0, no., pp. 86-0, Feb. 00. [6]. Bender,. Blac, M. Grob,. adovan,. Sndhushayana, and A. Vterb, CDMA/HD: A Bandwd-Effcent Hgh-Speed Wreless Data Servce for omadc sers, IEEE Communcatons Magazne, vol. 38, no. 7, pp. 70-77, July 000. [7] S- Web Ste, http://www.s.edu/nsnam/ns/ [8] Broadcast/Multcast Servces - Stage, 3G S.0030 Verson.0, August, 00. Auors Jeong-Jun Suh eceved a B.S. degree n control & nstrumentaton engneerng from Chung-Ang nversty, Seoul, Korea, n 999, and M.S. degree n electrcal and electronc engneerng from Yonse nversty, Seoul, Korea, n 00. He s currently a h.d. canddate n e Department of Electrcal and Electronc Engneerng, Yonse nversty, Seoul, Korea. Hs research nterests nclude satellte ATM networs and energy management n wreless sensor networs. Shan Guo Quan eceved e B.S. degree n Department of Electronc Engneerng and Computer Scence from Yanban nversty of Scence and Technology, Yan, Chna, n 998 and e M.S. degree n Department of Control and Instrumentaton Engneerng from Kangwon atonal nversty, Chuncheon, Korea, n 00. He s currently worng toward e h.d. degree n Department of Electrcal and Electronc Engneerng, Yonse nversty, Seoul, Korea. Hs current research nterests nclude wreless multmeda communcaton, moblty management, cross-layer wreless sensor networ desgn and nterworng over hybrd wreless networs. Jong-Ku ar eceved a B.S. degree n electroncs engneerng from Yonse nversty, Seoul, Korea, n 00, and M.S. degree n Department of Electrcal and Electronc Engneerng from Yonse nversty, Seoul, Korea, n 004. He s currently a researcher n Hyunda Motor Company. Hs research nterests nclude MAC protocol n multmeda satellte networs and wreless streamng. Jong Hyu ar Has been a research scentst of e &D Insttute, Hanwha S&C Co., Ltd., Korea snce 00, and currently has become a senor researcher and manager of proects related to bhome / smart home. He has been experenced many nternatonal conferences and worshops char or C member : Char of IC 07, CI'07, SSD 07, TST 07, ISM 07, IMIS 07, IWSH 06/07 and C member of AIA 07, GC 07, C 07, bsafe 07, SmarTel 07, CES 07, MS 07, etc. He s e founder of Internatonal Conference on Multmeda and bqutous Engneerng (ME, and Internatonal Conference on Intellgent ervasve Computng (IC. In addton, he s a member of e Tas Force n e IEEE IC and Assocate Edtor of e IJSH. He s also n edtoral board of Internatonal Journal of Intellgent Technology. He s nterested n ubqutous computng / home networ applcaton and securty, multmeda servce and securty, access control, etc. Young Yong Km eceved e B.S. and M.S. degrees n Department of Electroncs Engneerng from e Seoul atonal nversty, Seoul, Korea, n 99 and 993, respectvely. He receved e h.d. degree n e Department of Electrcal and Computer Engneerng at nversty of Texas, Austn, 999. From 998 to 000, he was w e Wreless etwor Management esearch Group at Telcorda Technologes (formerly Bell Communcaton esearch, ed Ban, J, as a esearch Scentst. In 000, he oned Yonse nversty, Seoul, Korea, where he s currently an Assocate rofessor n e Department of Electrcal and Electronc Engneerng. Hs nterests nclude wreless channel modelng, performance evaluaton of moble networs, and desgn and analyss of e multmeda CDMA system. 30