Characterizing the Downlink Transmission Power for WCDMA Networks in Multiservice Environments

Transcription

1 Charaterizing the Downink Transmission Power for WCDA etworks in utiservie Environments Chie Dou and iang-uen Huang Department of Eetria Engineering ationa Yunin University of Siene and Tehnoogy o. 23, Setion 3, University Road 64 Touiu, Yunin, TAIWA Abstrat: - This paper haraterizes the downink transmission power for WCDA networks in mutiservie environments for both urban and suburban maro es. To auate the required transmission power for the user in the e, various modeing approahes are used to investigate the orthogonaity, the path oss, and the inter-e interferene ratio of the user. The variation and sattering pattern of the required transmission power for eah servie onnetion are observed as a funtion of distane between the base station (BS) and the mobie station (S). The phenomena and their expanations presented in the paper onerning the WCDA downink power issues are important for the power ontro and admission ontro of the system. Key-Words: - WCDA, Downink, Transmission power, utiservie, aro e, Orthogonaity Introdution As it is known that the WCDA downink transmission power is interferene dependent. The interferene an be either due to own e or adjaent es. WCDA empoys orthogona odes in the downink to separate users, and without any mutipath propagation the orthogonaity remains when the base station (BS) signa is reeived by the mobie. However, if there is suffiient deay spread in the radio hanne, the mobie wi see part of the BS signa as intra-e mutipe aess interferene. The orthogonaity of one orresponds to perfety orthogona users. The impat of downink orthogonaity fator (OF) on WCDA system performane has been deepy investigated in the iterature reenty []-[4]. By utiizing the reationship between the OF and the hanne dispersion, it has been shown that the time-averaged OF an be determined from the hanne deay profie []. In the downink, different mobies aso experiene different inter-e interferene from adjaent es depending on the oation. The otherto-own-e interferene ratio an be defined as the summation of the ratio of every other e to own e base station transmission power reeived by the user. The reeived power from a e depends on the path oss from the transmitting BS to the mobie station (S). Two ommony adopted propagation modes for maro e senarios are Okumura-Hata mode [5] and COST 23 Wafish-Ikegami mode [6]. Both modes an be appied to urban and suburban areas with some modifiations of parameter setting. Sine both the orthogonaity fator and the inter-e interferene are oation dependent, that is, their vaues are varied aording to the distane between the BS and the S, the reeived power by a user moving in the e are affeted by them as we as by the path oss. This paper haraterizes the downink transmission power for WCDA networks in mutiservie environments for both urban and suburban maro es. In Setion 2, the equation of the required transmission power for eah servie onnetion to meet the target E b / is first derived. To auate the required transmission power for the S, various modeing approahes are needed to mode the orthogonaity, the path oss, and the intere interferene ratio of the S. In Setion 3, the required transmission power for a S moving away from the BS to the e edge is auated as a funtion of distane between the BS and the S. Suh auations are performed times and the resuts are potted together to observe the power variations and sattering as a funtion of distane. Here, the term power sattering signifies the deviation between different sampes of the auated transmission power at the same distane. The standard deviation of the power sattering at eah distane is omputed. The power variations and power sattering observed in Setion 3 for eah servie onnetion pertaining to different traffi asses are important for the system design, suh as power ontro and admission ontro. Finay, in Setion 4 the disussions and onusions are given.

2 Tabe Parameters and Definitions Definition Parameter W The hip rate The number of users of ass j in e m j υ The ativity fator of user i of ass j in e m R The bit rate of user i of ass j in e m P The required transmit power for user i of ass j from e m The noise spetra density of the mobie reeiver frontend rf P The tota downink transmission power of e m tota The path oss from the serving e m to user i of ass j (n) The path oss from another e n to user i of ass j in m e m α The orthogonaity fator of user i of ass j in e m The number of es in the system 2 Power odeing Approahes ) m The required ( E for user i (i,, ( m ) ) of b ass j ( j,, K) in e m (m,, ) is based on the ink quaity equation for the WCDA downink onnetion [7]: E / b W P. ( ) ( ) m m o υ R ( α ) Ptota / + rf W + Ptota / m n, n m () The parameters used in () are expained in Tabe. ( m ) Soving for gives: P ( ( E ) P m) b υ ( ) ( ) α Ptota + P m W / R n, n tota m m + rf j W (2) Assume that the tota transmit power of different es are equa. This assumption is approximatey vaid if there is a reguar grid of base stations, and the traffi distribution is uniform. The tota downink transmission power of e m an be written as P tota K j j i P ( where ( E ) b rf W m) W / R υ K j j i η D,. (3), (4) K and j η [ ( α + I )]. (5) D In (5), I j i ( n, n m m / ) is defined as the other-toown-e interferene ratio reeived at user i of ass j in e m. In the downink, this ratio depends on the user oation and is therefore different for eah user. To auate the required transmit power for eah servie onnetion using the equation expressed by (2), the three servie-independent parameters, I, and, must be modeed propery first. The modeing approahes for these parameters adopted in this paper are desribed as foows. 2. odeing the Orthogonaity Fator Assume that the impuse response of the maro e radio hanne an be modeed in disrete form as h( τ ) β δ ( τ T ), (6) where T / W is the hip time, β is the ompex ampitude of the -th mutipath omponent, and denotes the tota number of fingers. From (6), the average power deay profie (PDP) of the hanne an be expressed as 2 P( τ ) Ε[ β ] δ ( τ T ), (7) where E[ ] denotes spatia averaging over the fast fading. We wi assume Rayeigh fading 2 2 harateristis, i.e., Ε β ] β, and an exponentia [ 2 deaying PDP β aording to COST 259 [8] ' T / τ rms 2 e β, (8) ' T / τ e rms where τ rms is the RS deay spread (DS). The Greenstein mode [9] suggests that τ rms is ognorma at every base-to-mobie distane d, with a median that grows midy with d and a deibe standard deviation that is independent of d. The DS at distane d between the BS and the S an therefore be expresses as ε τ rms Td y, (9) where T is the medium vaue of τ rms at distane d Km, ε is an exponent that ies between.5-., and y is a ognorma variate. Speifiay, Y og y is a Gaussian random variabe over the terrain at distane d, having zero mean and a standard deviation, σ, that ies between 2-6 db. y An approximate formua for the OF is presented in [] for the ase where the tota reeived wideband power is muh greater than the inter-e interferene and the noise power, 2 βm, () α 2 m β, m where is the number of Rake fingers. The vaue of is seeted so that a resovabe mutipath omponents an apture ertain perentage of tota power. Ceary, at eah d the arger impies the greater OF. In this study we assume 6 and the α

3 vaue of is seeted so that 95% tota power is aptured. In genera, the OF is dereased as d inreases; however, vioations may be ourred due to the hange in the number of ative Rake fingers. Fig. The τ rms versus d for the urban e. samped OF at eah d in Figs. 3 and 4. In Fig. 3 it an be observed that for the urban e the OF deaying sharpy as d is sma (from to.5 Km). As d beomes arger (d >.5 Km), the hanging of the OF beomes sowy. In Fig. 4 we observe that the samped mean of the OF at eah d for the suburban e deaying smoothy as d inreases. Compare Fig. 3 and Fig. 4, we observe the foowing different harateristis of the OF for the urban and suburban maro es. First, the satter pattern of the OF for the urban e shown in Fig. 3 beomes onvergent as d beomes arger. However, the satter pattern for the suburban e shown in Fig. 4 is sti divergent as d beomes arger. This important observation has great infuene on the behaviors of different power variations for the users in the urban and suburban es, as we wi expain in the next setion. Seond, at the same d the OF for the suburban e is arger than that for the urban e. In genera, the OF for the suburban areas is greater than.5, and the OF for the urban areas is smaer than.3 as d >.5 Km. From (2) it is ear that the arger OF impies the smaer required transmission power for the S. Fig. 2 The τ rms versus d for the suburban e. To auate the RS DS and the OF for a S in the e with e radius.5 Km, we adopted the parameters suggested in [9] for the Greenstein mode. Toronto urban area with T.92 us and σ.9db is assumed for the urban e onditions. y For suburban maro e onditions, Toronto suburban area with T.28 us and σ 4.7dB is y assumed. For both urban and suburban areas ε.5 is used. Fig. and Fig. 2 show the τ rms (in us) versus the distane d for the urban and suburban e onditions, respetivey. At eah distane d we take sampes for the τ rms. Ceary, the vaues of the samped τ rms sattered with respet to the median ε. The urve for the ε T d T d versus the distane d is drawn in the figures as a smooth urve. Aso, the samped mean at eah d for the τ rms is potted aong with the median urve. Ceary, the RS DS for the urban e is greater than that for the suburban e. Given the τ rms auated by (9), the assoiated OF for the user an be obtained from (8) and (). Fig. 3 and Fig. 4 show the OF versus the distane d for the urban and suburban e onditions, respetivey. Corresponding to the samped τ rms at eah d auated in Figs. and 2, we aso have Orthogonaity Fator (OF) Orthogonaity Fator (OF) Fig. 3 The OF versus d for the urban e. Fig. 4 The OF versus d for the suburban e. 2.2 odeing the Path oss Two ommony adopted propagation modes for maro e senarios: Hata mode and COST 23 WI mode are used to auate the path oss for the S. The Hata mode for an urban e with base station

4 antenna height of 3 m, mobie antenna height of.5 m and arrier frequeny of 95 Hz is [5] og ( d ), where is the path oss in db. The path oss for a suburban e with an additiona area orretion of 8 db is [5] og ( d ). In the Hata mode, the path oss is determined ony by one parameter, d; however, in the COST-WI mode more ompex situations are onsidered. Speifiay, in the WImode we assume the buiding separation for the urban e is 2 m and for the suburban e is 4 m; the widths of roads for the urban e is m and for the suburban e is 2 m. For both urban and suburban es we assume the heights of buidings is 5 m and the non-ine-of-sight (OS) situations are onsidered. In addition, the road orientation with respet to the diret radio path ϕ is onsidered. Fig. 5 pots the path oss versus the distane d (varies from.75 Km to.5 Km) for both Hata and COST- WI modes. The mobie moves horizontay with initiay ϕ 9 and d.75 Km. From Fig. 5 we observe that the path oss for the urban e is arger than that for the suburban e, and the path oss auated by the COST-WI mode is arger than that auated by the Hata mode. Urban urves shown in Fig. 6 but not four urves. In Fig. 6 we notie that the other-to-own-e interferene ratio auated by the COST-WI mode is arger than that auated by the Hata mode as d <.5 Km. The urve obtained by the Hata mode inreases exponentiay, however the urve obtained by the COST-WI mode an be divided into three parts due to three different onditions of streetorientation funtion. The first part of the urve orresponds for 55 ϕ < 9 and inreases sharpy than that obtained by the Hata mode. The seond part of the urve orresponds for 35 ϕ < 55, and the third part of the urve orresponds for ϕ < 35. From (2) it is ear that the arger other-to-own-e interferene ratio impies the arger required transmission power for the S. Other-to-own-e Interferene Ratio Hata ode COST-WI ode Fig. 6 The inter-e interferene ratio versus d. Path oss [db] Distane BS-User [K m ] Suburban Hata ode COST-WI ode Fig. 5 The path oss of the S versus d. 2.3 odeing the Inter-Ce Interferene Fig. 6 pots the other-to-own-e interferene ratio versus the distane d, using the path oss obtained in Fig. 5. In this study we assume a symmetri 9 es euar mode with an omnidiretiona antenna overs eah e of hexagona shape. From Fig. 6 we observe that the urban e and the suburban e have the same other-to-own-e interferene ratio. This is beause the other-to-own-e interferene ratio is defined as the summation of individua ratios between the path osses from the serving e and from another e to the S. Athough the path oss auations for urban and suburban situations are different, the division operation between path osses eiminates this differene. Thus, there have ony two 3 Power Variations and Sattering This study onsiders three QoS asses shared on a WCDA downink, inuding speeh,, and streaming. To ompute the transmit power for eah ative user pertaining to different asses by (2), the foowing assumptions of parameters are used: rf W -3.6 dbm/hz; the hanne symbo rate for users is 2.3 Kbps, for users is 8.5 Kbps, for users is 36.8 Kbps; the target E b / for users is 5.5 db, for users is 4.5 db, for users is 6.5 db; the ativity fator for both and is.67, for is one. Fig. 7 shows the required transmission power for a user moving from the BS toward the e edge as a funtion of d. Obviousy, the mobie in the urban e required more transmission power than in the suburban e. The power variation for the suburban e is muh greater than that in the urban e. To simpify the presentation ony the resuts obtained by the Hata propagation mode are shown in here. To observe the power variation and sattering onditions for users in the e as a funtion of distane, different auations same as that obtained in Fig. 7

5 are performed and their resuts are potted together. Figs. 8(a) and 9(a) show the required transmission power versus the distane d obtained for the users in the urban e and suburban e, respetivey. Sine at eah d one hundred sampes are generated, the standard deviation of these sampes represents the degree of power sattering at that distane. Figs. 8(b) and 9(b) show the standard deviation of the required transmission power versus the distane d obtained for the users in the urban and suburban e, respetivey. Ceary, the dotted area shown in Fig. 9(a) is muh arger than that shown in Fig. 8(a). It is interested to notie that the standard deviation of the required transmission power for the user in the urban e goes downwardy with futuation as d inreases; however, the standard deviation of the required transmission power for the suburban e amost keep steadfast as d >.5 Km. The main reason that auses this differene is due to the modeing of the orthogonaity fator. In (2), the required transmission power for the user is determined by the three terms in the square braket. The ast two terms, one is reated to the inter-e interferene ratio and the other is reated to the path oss, have no satter onditions at eah distane d. Thus we know that the power sattering patterns as shown in Figs. 8(a) and 9(a) are aused by the in (2), that is the OF of α the user. Sine for the urban e, as d inreases the satter pattern of the OF beomes onvergent, that is the deviations of the samped vaues for the OF beome smaer. Hene, the standard deviation of the required transmission power for the users in the urban e goes downwardy as shown in Fig. 8(b), from about 2 dbm to 4 dbm, as d inreases. Contrariy, the satter patter of the OF for the suburban e remains divergent, that is the deviations of the samped vaues for the OF amost hod the same distribution, as d inreases. Hene, the standard deviation of the transmission power for the users in the suburban e amost keep steadfast as shown in Fig. 9(b), between dbm, as d >.5 Km. Urban Suburban Fig. 7 The required transmit power for a user. (a) The P vs. d P Fig. 8 The and σ vs. d in urban e. (b) The σ vs. d Fig. 9 The P and σ vs. d in suburban e. Fig. The P and σ vs. d in urban e. Fig. The P and σ vs. d in suburban e. Figs. and show the required transmission power and its standard deviation for the users in the urban and in the suburban e, respetivey. As the same Figs. 2 and 3 are potted for users. The trends for the standard deviation of the required transmission power for the users of different servie asses, as shown in Figs. 8(b), (b), and 2(b), are the same for the urban e, that is, a goes downwardy as d inreases. In Fig. (b), the standard deviation goes down from 5 dbm to 9 dbm, with a differene of 6 dbm. In Fig. 2(b), the standard deviation goes down from 4 dbm to 8 dbm, with a differene of 6 dbm aso. Atuay, this differene of 6 dbm has aready appeared in Fig. 8(b). For the suburban e, the trends shown in Figs. 9(b), (b), and 3(b) are aso simiar, that is, a keep steadfast as d inreases. For

6 users, the standard deviation keeps steadfast at about 7 dbm. For users, the standard deviation keeps steadfast at about 6 dbm. Compare the required transmission power and its standard deviation for the users of different servie asses, we find that in both urban and suburban e the order of their magnitudes is >>. VoI P Fig. 2 The P and σ vs. d in urban e. The ast two terms, one is reated to the inter-e interferene ratio and the other is reated to the path oss, have no satter onditions at eah distane d. Thus we know that the power sattering patterns as shown in Figs. 8(a) and 9(a) are aused by the α in (2), that is the OF of the user. This study observes many interesting phenomena onerning the WCDA downink power issues and their expanations are aso provided. For instane, the standard deviation of the required transmission power for the user in the urban e goes downwardy as d inreases, however, for the suburban e it amost keeps steadfast as d >.5 Km. The reason is that for the urban e as d inreases the satter pattern of the OF beomes onvergent, however, the satter patter of the OF for the suburban e remains divergent, that is the deviations of the samped vaues for the OF amost keep the same distribution, as d inreases. (a) The Fig. 3 The P (b) The σ vs. d and σ vs. d in suburban e. P vs. d 4 Conusions This paper haraterizes the downink transmission power for WCDA networks in mutiservie environments for both urban and suburban maro es. To satisfy the QoS requirement of eah servie onnetion, the BS must provide the required transmission power for the S ontinuousy. To auate the required transmission power for the S, various modeing approahes are adopted to mode the orthogonaity, the path oss, and the inter-e interferene ratio of the S. For a WCDA downink, the BS tota transmission power is shared by a onnetions pertaining to different servie asses. The understanding of the power harateristis for eah individua onnetion as we as the aggregation of servie onnetions is hepfu in designing the power ontro and admission ontro agorithms of the system. A system with we-designed power ontro and admission ontro an not ony ower the dropping and boking probabiities of the as but aso inrease the number of ative as in the e, thus enhane the system apaity. However, these topis are beyond the sope of this study. In (2), the required transmission power for the user is determined by the three terms in the square braket. Referenes: [].B. ehta,.j. Greenstein, T.. Wiis, and Z.Kosti, Anaysis and resuts for the orthogonaity fator in WCDA downinks, in Pro. IEEE VTC 22-Spring, pp. -4. [2] K.I. Pedersen and P.E. ogensen, The downink orthogonaity fators infuene on WCDA system performane, in Pro. IEEE VTC 22-Fa, pp [3] O. Awoniyi,.B. ehta, and.j. Greenstein, Charaterizing the orthogonaity fator in WCDA downinks, IEEE Trans. on Wireess Communiations, vo. 2, no. 4, Juy 23. [4] C. Passerini and G. Faiasea, odeing of orthogonaity fator using ray-traing preditions, IEEE Trans. on Wireess Communiations, vo. 3, no. 6, ov. 24. [5] H. Homa and A. Toskaa, (Editors), WCDA for UTS Radio Aess for Third Generation obie Communiations, revised ed., John Wiey & Sons, Chihester, Engand, 2. [6] E. Damosso and.. Correia (Editors), COST 23 - Digita mobie radio towards future generation systems, Fina Report, European Commission COST Teeom Seretariat, Brusses, Begium, 999. [7] K. Sipiä, Z.C. Honkasao, J.. Steffens, and A. Waker, Estimation of apaity and required transmission power of WCDA downink based on a downink poe equation, in Pro. IEEE VTC-Spring 2, pp [8]. Correia (Editor), COST Wireess fexibe Personaized ommuniations, Fina Report, John Wiey & Sons, ay 2. [9].J. Greenstein, V. Ereg, Y.S. Yeh, and.v. Cark, A new path-gain/deay-spread propagation mode for digita euar hannes, IEEE Trans. on Veh. Tehno., vo. 46, no. 2, pp , ay 997.