Signaling Cost Analysis for Handoff Decision Algorithms in Femtocell Networks

Transcription

1 Signaling Cos Analysis for Hoff Decision Algorihms in Femocell Neworks Wahida Nasrin Jiang Xie The Universiy of Norh Carolina a Charloe Charloe, NC {wnasrin, Linda.Xie}uncc.edu Absrac Femocells are deployed o provide good indoor coverage o offload daa raffic from macrocell neworks. Unnecessary hoffs (HOs), ping-pong effecs, cell uilizaion are imporan performance merics for evaluaing he qualiy of connecions daa offloading in femocell neworks. Though significan research has been conduced on HO decision algorihms o reduce unnecessary HOs ping-pong effecs, only a few of hese sudies consider unnecessary HOs cell uilizaion ogeher. Moreover, all of he exising HO decision algorihms add exra signaling overhead o he HO procedure. Therefore, i is imporan o analyze he HO signaling cos of exising HO decision algorihms. In his paper, we propose an analyical model o sudy he HO signaling cos of differen HOs in open-access femocell neworks. In addiion, we propose HO decision algorihms ha can reduce unnecessary HOs wihou increasing HO signaling coss sacrificing cell uilizaion. Simulaion resuls show significan performance improvemen as compared o he exising HO decision algorihms. To he bes of our knowledge, his is he firs analyical model ha can be applied o all exising HOs available in open-access femocell neworks. I. INTRODUCTION Femocells are inroduced as a promising soluion o improve indoor coverage o offload daa raffic from cellular neworks (i.e., macrocells) [1]. Femocells are low-powered, shor-ranged, low-cos indoor base saions (BSs), which are deployed managed by users. Femocells can offload raffic from cellular neworks based on heir access modes he availabiliy of user equipmen (UE) wihin heir coverage area. There are hree access modes available in femocell neworks: closed, open, hybrid. These modes form hree ypes of hoffs (HOs): macro-o-femo, femo-o-macro, femo-o-femo. Only a limied number of regisered UEs can access he femocell in closed-access mode. This access mode suppors macro-o-femo femo-o-macro HOs. Any UEs ha are wihin he coverage area of an open-access femocell can access i. Besides he wo HOs available in closed-access, open-access mode also suppors users in performing HOs from a femocell o anoher femocell (femo-o-femo). On he oher h, hybrid access mode suppors all ypes of users HOs. HO plays an imporan role during daa offloading in a femocell-deployed macrocell nework. Due o indoor unplanned deploymen of femocells, unnecessary HOs pingpong effecs may happen frequenly, which severely degrades he qualiy of connecions user experience. On he oher h, offloading in femocells requires a high cell uilizaion. This work was suppored in par by he US Naional Science Foundaion (NSF) under Gran No. CNS Therefore, i is necessary o design an HO decision algorihm ha can reduce unnecessary HOs improve cell uilizaion a he same ime. Mos of he exising works propose HO decision algorihms ha reduce unnecessary HOs pingpong effecs [2] [4]. A few of hem have considered cell uilizaions while designing an HO decision algorihm [5], [6]. All of hese algorihms have used differen parameers echniques o design HO decision algorihms [2]. This pracice adds exra HO signaling overhead a he core nework increases he HO signaling cos. Limied research has been conduced o analyze he HO signaling cos in femocell neworks [7], [8]. However, hese works do no analyze he signaling cos of differen ypes of HOs in open-access modes. Moreover, a comparison of he HO signaling cos for exising HO decision algorihms is necessary. In his paper, we propose an analyical model o sudy he HO signaling cos of differen HOs in open-access femocell neworks. In addiion, we exend our work [6] o propose a arge cell selecion mehod o propose HO decision algorihms ha can reduce unnecessary HOs wihou increasing HO signaling coss sacrificing he femocell uilizaion. Finally, we compare he HO signaling cos of exising HO decision algorihms in femocell neworks. The res of he paper is organized as follows. Relaed work conribuions are inroduced in Secion II. In Secion III, he proposed arge cell selecion mehod HO decision algorihms are described. The analyical model for HO signaling coss are presened in Secion IV. Performance evaluaion is given in Secion V, followed by he conclusions in Secion VI. II. RELATED WORK AND CONTRIBUTIONS A. Relaed Work Though a number of papers on HO decision algorihms are available in he lieraure [2], only a few of he exising works consider he HO signaling cos in femocell neworks [7] [9]. An archiecure for LTE femocell neworks which inroduces an inermediae node (HeNB GW) is presened in [7]. Two mehods for mobiliy managemen are proposed in his paper. In he firs mehod, he HeNB GW acs as a mobiliy anchor o conrol HOs among femocells, i works as a relay in he second mehod. An analyical model o evaluae compare HO signaling coss of hese wo mehods are described here. However, no all HO scenarios are considered. In [9], an HO decision algorihm based on users speed raffic ypes is discussed. The signaling procedure for boh macro-o-femo /17/$ IEEE

2 femo-o-macro HOs are also presened in his paper. A simple HO decision algorihm for he macro-o-femo HO scenario ha considers users speed is discussed in [8]. The HO signaling cos is also analyzed for his HO scenario, he proposed algorihm is compared o a radiional HO decision algorihm. Boh of he papers [7], [8] analyze he HO signaling cos for a simple general scenario. For example, a femo-o-femo HO scenario is considered in [7], a macroo-femo HO scenario is discussed in [8]. However, he res of he HO scenarios in he open-access mode have no been considered. B. Conribuions In his paper, we propose HO decision algorihms an analyical model for he open-access femocell neworks. The conribuions of his paper are summarized as: We exend our work [6] for open-access femocell neworks propose a arge cell selecion algorihm, along wih HO decision algorihms for macro-o-femo, femoo-femo, femo-o-macro HO scenarios. We propose an analyical model of he oal HO signaling cos for all ypes of HOs in open-access femocell neworks. We analyze compare he oal HO signaling cos of differen exising HO decision algorihms ha are designed o reduce unnecessary HOs. We use a realisic simulaion scenario o evaluae he performance of hese exising algorihms in erms of oal HO signaling cos femocell uilizaion. III. PROPOSED HO DECISION ALGORITHM In his secion, he proposed arge cell selecion mehod HO decision algorihms are presened. The proposed algorihms work in wo phases: iniializaion uilizaion. The iniializaion phase is used o build a locaion-hisory daabase, his daabase is used in he uilizaion phase o adap he hyseresis margin (HM). The noaions used in our algorihms are lised in Table I. RSSI RSSI m RSSI sf RSSI f RSSI min Th HM ad MME FGW MBS Th spd UE spd TABLE I NOTATIONS USEDINTHEALGORITHMS Received Signal Srengh Indicaor Received Signal Srengh Indicaor for he macrocell Received Signal Srengh Indicaor for he serving femocell Received Signal Srengh Indicaor for he arge femocell Minimum received signal srengh indicaor for he macrocell Threshold for he femocell Adapive hyseresis margin Mobiliy managemen eniy Femo gaeway Macro-base saion Threshold for users speed Users speed Despie he access policies, he requiremens for a macro-ofemo a femo-o-macro HO are he same for boh closedaccess open-access femocell neworks. The proposed HO decision algorihms for hese HOs are given in Algorihm 1 Algorihm 2. We have modified he HO decision algorihms o make hem applicable for open-access femocell neworks by considering he users speed. The calculaion of Th, RSSI min, HM ad is shown in [6]. The mos imporan issue during a femo-o-femo HO in open-access mode is how o selec a proper arge femocell o perform an HO. We propose o use a locaion-hisory daabase ha can sore he ID of he arge femocell along wih he locaion fingerprin. In he iniializaion phase, when he locaion-hisory daabase is buil, each ime a UE sends a measuremen repor o he serving BS, i deermines he arge cell based on he maximum RSSI forwards his informaion, along wih he measuremen repor, o he MME/FGW. This forwarded message conains a lis of neighboring cell IDs (wih RSSI > RSSI min ), heir corresponding RSSIs, a arge cell ID. The MME/FGW sores his informaion in he daabase unil i is full. The daabase building updaing algorihm is shown in [6]. In he uilizaion phase, his daabase is used o adap he HM o selec he arge cell during a femo-o-femo HO scenario. In his manner, he serving BS can reduce he delay of selecing a arge cell each ime an HO reques arrives. The proposed HO decision algorihm for a femo-o-femo HO scenario is shown in Algorihm 3. Algorihm 1: Macro-o-femo HO decision algorihm if UE spd <Th spd hen if RSSI f >Thor RSSI m < RSSI min, RSSI f > RSSI m + HM ad hen HO o femocell; Say in macrocell; Say in macrocell; End; Algorihm 2: Femo-o-macro HO decision algorihm if UE spd >Th spd hen HO o macrocell; if RSSI sf <Thor RSSI m > RSSI min, RSSI sf + HM ad < RSSI m hen HO o macrocell; Say in he femocell; End; Algorihm 3: Femo-o-femo HO decision algorihm if UE spd >Th spd hen HO o he macrocell; if RSSI sf <Thor RSSI f >Th, RSSI sf + HM ad < RSSI f hen HO o he arge femocell; Say in he serving femocell; End; IV. ANALYTICAL MODEL FOR THE HO SIGNALING COST In open-access femocell neworks, a macro-o-femo HO happens when an acive UE moves owards a femocell bound-

3 ary, a femo-o-macro HO happens when an acive UE moves ou of he boundary. There are four evens ha occur in open-access femocell neworks, including hese HOs. All of hese evens are shown in Fig. 1. (Fig. 2(d)) represens an acive femocell UE ha moves ino he coverage area of anoher femocell, he daa session ends while wihin he area. Therefore, only a femo-o-femo HO happens in his case. On he oher h, a UE becomes acive wihin a femocell coverage area, hen i performs an HO o a femocell (Fig. 2(e)). If he probabiliies of hese evens are P r4, P r5, P r6, P r7, P r8, hen we can calculae probabiliies of all hree ypes of HOs in open access femocell neworks as: macro o femo = P r1 + P r2 + P r4, (1) femo o macro = P r1 + P r3 + P r6, (2) femo o femo = P r4 + P r5 + P r6 + P r7 + P r8. (3) Fig. 1. Timming diagrams for mobiliy evens in marco-femo HOs. In he iming diagram (Fig. 1), we assume ha a UE may become acive a any momen (A 1 ), he daa session arrival rae (λ) is a Poisson process. On he oher h, A 2 is he momen when he daa session ends. In addiion, he momen when a UE eners he range of a femocell is indicaed by 1, he momen when i leaves he range of a femocell is indicaed by 2. The firs even shown in Fig. 1(a) represens ha an acive UE moves ino a femocell coverage area moves ou while i is sill acive. Boh macro-o-femo femo-o-macro HOs happen in his case. In he second even (Fig. 1(b)), an acive UE moves ino he femocell coverage area, he daa session ends before i moves ou of he area. A macro-o-femo HO happens in his case. During he hird even (Fig. 1(c)), a UE becomes acive wihin a femocell coverage area moves ou of he area while sill acive. In his case, he UE performs a femo-o-macro HO. In he fourh even (Fig. 1(d)), since he UE becomes acive, he daa session ends wihin he femocell coverage area, no HOs happen. The probabiliies of hese evens are P r1, P r2, P r3. Besides hese four evens, he open-access mode has five addiional mobiliy evens ha can cause differen HO scenarios. These five evens are shown in Fig. 2. In he iming diagram, he even in Fig. 2(a) represens an acive UE ha moves ino a femocell coverage area from a macrocell hen moves ino anoher femocell coverage area. Boh macro-o-femo femo-o-femo HOs happen in his case. The even in Fig. 2(b) represens an acive UE ha moves beween femocells while i is sill acive, a femo-o-femo HO happens in his case. An acive femocell UE performs an HO o anoher femocell when i moves ino he nex femocell area o a macrocell when i moves ou of femocells (Fig. 2(c)). Femo-o-femo femo-o-macro HOs happen in his case. The nex even Fig. 2. Timming diagrams for mobiliy evens in femo-femo HOs. Since boh he session duraion (T D ) he duraion ha a UE says wihin a femocell coverage area (T R )are exponenially disribued, we can calculae he probabiliies of evens in Fig. 2 he same way as he evens in Fig. 1. Therefore, we can infer P r4 = P r5 = P r6 = P r1, P r7 = P r2, P r8 = P r3. Using hese values in (1), (2), (3), we obain macro o femo =2P r1 + P r2, (4) femo o macro =2P r1 + P r3, (5) femo o femo =3P r1 + P r2 + P r3. (6) In he iming diagrams, T D T R are independen rom variables. T D denoes he session duraion which is exponenially disribued wih mean 1/η, he probabiliy densiy

4 funcion of his session duraion is f TD () =ηe η. Similarly, T R is he duraion of a UE being wihin he coverage area of a femocell which is exponenially disribued wih mean 1/μ, he probabiliy densiy funcion of his duraion of say is f TR () =μe μ. T DR T Rr in he iming diagram follow he memoryless propery of he residence imes, T D T R, respecively. In addiion, he probabiliy densiy funcion of T DR is f DR (which is exponenially disribued wih mean 1/η) he probabiliy densiy funcion of T Rr is f Rr (which is exponenially disribued wih mean 1/μ). Now, we can calculae P r1, P r2, P r3 as: P r1 = P (A 1 < 1 <A 1 + T D ) P (T DR >T R ), (7) P r2 = P (A 1 < 1 <A 1 + T D ) P (T DR T D ), (8) P r3 = P ( 1 <A 1 < 1 + T R ) P (T D T Rr ). (9) Here, (7) ensures ha he session sars before he UE eners he femocell coverage area, he UE leaves he area before he session ends. Similarly, (8) indicaes ha he session sars before he UE eners he femocell coverage area ends before i leaves he area. (9) ensures ha a session sars afer he UE eners he femocell coverage area ends afer i leaves he area. Using he Laplace ransform, we have P r1 = P r2 = (1 λe λ f TD (y)dyd λe λ f TD (x)dxd (1 P r3 = λe λ f Rr ()d μe μx f DR ()dxd), (1) ηe η f DR (y)dyd), (11) ηe ηy f Rr ()dyd. (12) Afer solving (1), (11), (12), we can find he probabiliy of he hree evens as: λμ P r1 = (λ + η) 2 (μ + η), (13) P r2 = λη (λ + η) 2 (μ + η), (14) λμ 2 P r3 = (λ + η) 2 (μ + η). (15) Finally, hese hree probabiliies can be calculaed from (13), (14), (15). Then, he oal HO signaling cos of macroo-femo, femo-o-macro, femo-o-femo HOs in openaccess femocell neworks are C open macro femo = macro femo ( Tj i + P i ), (16) C open femo macro = femo macro ( T i j + P i ), (17) C open femo femo = femo femo ( T i j + P i ). (18) Here, Tj i is he delivering cos of an HO message beween node i j, P i is he processing cos of a message a node i, he erms in he brackes are he signaling cos of a successful HO. The macro-o-femo he femo-o-macro HO signaling procedures are given in [6]. By analyzing he HO signaling procedure, we ge Tj i P i for a macroo-femo HO as: (T i j ) macro femo =3TUE MBS +6TFGW+5T FBS MME (19) (Pi ) macro femo = P UE + P FBS + P FGW +2P MME. (2) Similarly, we ge Tj i P i for a femo-o-macro HO as: (T i j ) femo macro =3TUE FBS +5T FBS FGW +5TMBS MME, FGW+3T MME (21) (Pi ) femo macro = P UE + P FBS +2P FGW + P MME. (22) In addiion, we presen he HO signaling procedure for a femo-o-femo HO in Fig. 3. Now, we ge Tj i P i for a femo-o-femo HO as: (T i j ) femo femo =3TUE FBS +1TFGW FBS +2TFGW MME, (23) FGW +5TMBS MME, (Pi ) femo femo = P UE + P FBS +2P FGW + P MME. (24) Noaions for differen coss heir values are given in Table II [7], [8], [1] [12]. Fig. 3. Femo-o-femo HO signaling procedures for he proposed HO decision algorihm. V. PERFORMANCE EVALUATION In his secion, we evaluae he performance of he proposed HO decision algorihm in erms of he oal HO signaling cos, he rae of unnecessary HOs, femocell uilizaion for open-access femocell neworks. The rae of unnecessary HOs represens he probabiliy ha a UE emporarily hs over o he arge cell hs over back o he serving cell. On he oher h, he femocell uilizaion represens he probabiliy

5 ha a UE says conneced o he femocell while wihin he coverage. Furhermore, we compare our proposed algorihm wih five exising algorihms: 1) RSS TH: HO decisions based on an RSSI-based hreshold; 2) RSS TH HM: HO decisions based on boh an RSSI-based hreshold a fixed HM; 3) RSS ADHM: HM adaps based on he formula from [13], which is HM = max{hm max (1 1 d R ) 4 ;}. Here, R is he radius of he femocell, d is he disance beween he femo-base saion (FBS) a UE; 4) SINR ADHM: adapive HM is calculaed from HM = max{hm max (1 1 SINR ac SINR min SINR min SINRmax ) 4 ;} [3]; 5) RSS Speed: hresholds for boh he users speed RSSI are used o make HO decisions; 6) Proposed: locaion-hisory daabase is used o adap HM based on he indoor locaion of a UE [6]; 7) Proposed Speed: hreshold for users speed adapive HM based on he locaion-hisory daabase are used. TABLE II HO SIGNALING COST PARAMETERS TUE FBS Transmission cos beween a UE an FBS 2 TFBS FGW Transmission cos beween an FBS an FGW 2 TFGW MME Trnsmission cos beween an FGW an MME 4 TMBS MME Transmission cos beween an MBS an MME 4 TUE MBS Transmission cos beween a UE an MBS 2 P UE Processing ime a UE 4 P FBS Processing cos a FBS 3 P FGW Processing cos a FGW 2 P MME Processing cos a MME 4 We use NeLogo 5..5 [14] o simulae he indoor environmen for open-access femocell neworks. We design a singlefloored wo bedroom aparmen wih an FBS, which has he capaciy o suppor en users. The aparmen is surrounded by six neighboring FBSs, all hese FBSs are wihin he coverage area of a macrocell. Thiry users all FBSs are placed in a rom manner. These users follow a modified version of he Rom Waypoin mobiliy model, hey have a probabiliy of.7 o ener exi he aparmen. The mobiliy model is modified in a way ha he users use he door only o go in/ou of he aparmen, none of hem cross he walls. The Okumura-Haa propagaion model is used for he macrocell nework, he ITU-R P indoor pahloss model [15] is used for he femocell nework. The parameers used in our simulaion are lised in Table III [6], [16], [17] TABLE III SIMULATION PARAMETERS Macrocell ransmission power, P m 45 dbm Radius of macrocell 1.2 km Femocell ransmission power, P f 1 dbm Radius of femocell 15 m Size of daabase, d s 3 Users speed o 1 km/hr Threshold, Th -45 db Wall peneraion loss 5dB Oudoor peneraion loss 2dB-1dB RSSI min -75 db Th spd 5 km/hr HM max 5dB A. Toal HO Signaling Cos The performance of he oal HO signaling cos for all ypes of HOs in open-access femocell neworks are given in Fig. 4. The signaling cos of an HO is deermined by considering all ransmission coss processing coss during an HO. I is calculaed for an exponenial session duraion (mean 1/η =3), a residence ime (mean 1/μ =1), he session arrival rae λ (.1 o.34). Then, he oal HO signaling cos is calculaed by muliplying he signaling cos of an HO by he rae of HOs, which also includes he rae of unnecessary HOs. From he figures, we can observe ha he oal HO signaling cos increases wih he addiion of HO decision crieria. We can also observe ha using he users speed reduces he oal HO signaling cos by reducing unnecessary HOs of high speed users. The exising algorihms ha adap HMs have he highes oal HO signaling cos. These exising algorihms adap HMs eiher based on he disance beween he BS he UE or he SINR received a he UE side. As a resul, he UE has o noify he serving BS frequenly, which creaes addiional signaling cos. Moreover, hese mehods canno eliminae he number of unnecessary HOs. On he oher h, we can observe ha our proposed algorihms show beer resuls in he open-access mode. B. Rae of Unnecessary HOs When a UE performs wo consecuive HOs from a BS o anoher BS wihin a specific ime limi, we consider i as an unnecessary HO. Then, he rae of unnecessary HOs is couned as he accumulaed number of unnecessary HOs divided by he oal number of HOs. The rae of unnecessary HOs in openaccess femocell neworks are given in Fig. 5. The simulaion resuls shows ha he proposed algorihms has a lower unnecessary HO rae han he compared algorihms, which is desirable in order o provide beer performance in femocell neworks. By observing resuls of boh oal HO signaling cos rae of unnecessary HOs, we can infer ha hough he proposed HO decision algorihm wihou considering users speed has almos he same HO signaling cos as he RSS-based algorihm, i eliminaes more unnecessary HOs rae han all he compared algorihms. Fig. 5. Comparison of he rae of unnecessary HOs for open-access femocell neworks.

6 (a) Macro-o-femo HOs (b) Femo-o-macro HOs (c) Femo-o-femo HOs Fig. 4. Comparison of he oal HO signaling cos in open access femocell neworks. C. Femocell Uilizaion Mos of he ime, a proper HO decision algorihm for reducing unnecessary HOs may also reduce he uilizaion of femocells. On he oher h, a good femocell uilizaion indicaes a high raffic offload, offloading is imporan for femocell neworks. Therefore, we simulae he femocell uilizaion for open-access mode. These resuls, as shown in Fig. 6, represens good femocell uilizaion for our proposed algorihms. Fig. 6. Femocell uilizaion for differen HO decision algorihms for openaccess femocell neworks. VI. CONCLUSION In his paper, we proposed an analyical model o evaluae he hoff signaling cos of macro-o-femo, femo-ofemo, femo-o-macro hoffs in open-access femocell neworks. In addiion, we also proposed a arge cell selecion mehod hoff decision algorihms for open-access femocell neworks. The proposed algorihms are compared o five exising hoff decision algorihms wih respec o he oal hoff signaling coss femocell uilizaions. Simulaion resuls show ha our proposed algorihms can significanly reduce he oal hoff signaling coss wihou sacrificing femocell uilizaion as compared o he exising algorihms. REFERENCES [1] J. G. Andrews, H. Claussen, M. Dohler, S. Rangan, M. C. Reed, Femocells: Pas, presen, fuure, IEEE Journal on Seleced Areas in Communicaions, vol. 3, no. 3, pp , 212. [2] D. Xenakis, N. Passan, L. Merakos, C. Verikoukis, Mobiliy managemen for femocells in LTE-Advanced: Key aspecs survey of hover decision algorihms, IEEE Communicaions Surveys Tuorials, vol. 16, no. 1, pp , Firs Quarer 214. [3] Z. Becvar P. Mach, Adapive hyseresis margin for hover in femocell neworks, in Proc. Inernaional Conference on Wireless Mobile Communicaions (ICWMC), 21, pp [4] W. Nasrin J. Xie, A mobiliy managemen scheme o reduce he impac of channel heerogeneiy in cogniive radio femocell neworks, in Proc. IEEE Inernaional Conference on Sensing, Communicaion, Neworking (SECON), 216, pp [5] J.-M. Moon, J. Jung, S. Lee, A. Nigam, S. Ryoo, On he radeoff beween hover failure small cell uilizaion in heerogeneous neworks, in Proc. IEEE Inernaional Conference on Communicaion Workshop (ICCW), 215, pp [6] W. Nasrin J. Xie, A self-adapive hoff decision algorihm for densely deployed closed-group femocell neworks, in Proc. IEEE Inernaional Conference on Sensing, Communicaion, Neworking (SECON), 215, pp [7] L. Wang, Y. Zhang, Z. Wei, Mobiliy managemen schemes a radio nework layer for LTE femocells, in Proc. IEEE Vehicular Technology Conference, 29, pp [8] H. Zhang, W. Ma, W. Li, W. Zheng, X. Wen, C. Jiang, Signalling cos evaluaion of hover managemen schemes in LTE-advanced femocell, in Proc. IEEE Vehicular Technology Conference (VTC Spring), 211, pp [9] A. Ulvan, R. Besak, M. Ulvan, The sudy of hover procedure in LTE-based femocell nework, in Proc. IEEE Wireless Mobile Neworking Conference (WMNC), 21, pp [1] J. McNair, T. Tugcu, W. Wang, J. L. Xie, A survey of crosslayer performance enhancemens for mobile IP neworks, Compuer Neworks, vol. 49, no. 2, pp , 25. [11] U. Narayanan J. Xie, Signaling cos analysis of hoffs in a mixed IPv4/IPv6 mobile environmen, in Proc. IEEE Global Telecommunicaions Conference (GLOBECOM), 27, pp [12] J. Xie U. Narayanan, Performance analysis of mobiliy suppor in ipv4/ipv6 mixed wireless neworks, IEEE Transacions on Vehicular Technology, vol. 59, no. 2, pp , 21. [13] S. Lal D. K. Panwar, Coverage analysis of hoff algorihm wih adapive hyseresis margin, in Proc. Inernaional Conference on Informaion Technology (ICIT), 27, pp [14] S. Tisue U. Wilensky, NeLogo: A simple environmen for modeling complexiy, in Proc. Inernaional Conference on Complex Sysems, 24, pp [15] P. Series, Propagaion daa predicion mehods for he planning of indoor radiocommunicaion sysems radio local area neworks in he frequency range 9 MHz o 1 GHz, ITU-R Recs, 212. [16] J. Xie, User independen paging scheme for mobile IP, Wireless Neworks, vol. 12, no. 2, pp , 26. [17] W. Nasrin J. Xie, Effecs of heerogeneous frequency changes in cogniive radio femocell neworks, in Proc. IEEE Global Communicaions Conference (GLOBECOM), 216, pp. 1 6.