The 4th Interntionl Symposium on Wireless Personl Multimedi Communictions WPMC'), Temporl Secondry Access Opportunities for WLAN in Rdr Bnds Miurel Tercero, Ki Won Sung, nd Jens Znder Wireless@KTH, Royl Institute of Technology KTH), SE-64 4 Kist, Sweden Emil: {mitv, sungkw, enz}@kth.se Abstrct In this pper we quntify the temporl opportunities for secondry ccess in rdr bnds. Secondry users re ssumed to be WLANs which opportunisticlly ccess the rdr spectrum. Ech WLAN device employs dynmic frequency selection DFS) s mechnism to protect the rdr from interference. We lso consider n dvnced interference protection mechnism, which is termed temporl DFS. It exploits the temporl vrition of interference power due to the rottion of rdr ntenn. It is observed tht the probbility of ccessing the rdr bnd is significntly higher when the temporl DFS is used compred to the conventionl DFS. As consequence, more WLANs utilize the rdr spectrum when the temporl DFS mechnism is considered. This shows tht hving better knowledge of the primry user ctivity cn bring bout the incresed opportunity of secondry spectrum ccess, nd thus improve the spectrum utiliztion. Index Terms Secondry spectrum ccess, rdr spectrum, ggregte interference, temporl opportunity, temporl DFS. I. INTRODUCTION The useful rdio spectrum is fully llocted to vrious systems such s TV brodcsting, stellite, rdio detection nd rnging system rdr), nd mobile communictions. However, it does not necessrily men tht the spectrum is efficiently used. Mesurement results indicte tht the llocted spectrum is mostly being under-utilized [], [2]. The spectrum utiliztion cn be improved by using dynmic spectrum ccess DSA) mechnism bsed on hierrchicl ccess strtegy, i.e. secondry spectrum ccess, which is envisioned by cognitive rdio [3]. It llows secondry users to ccess the spectrum tht hs lredy been ssigned to primry users if they do not cuse hrmful interference to the primry users. Secondry spectrum ccess hs lredy been implemented in the 5GHz frequency bnd 55-535 MHz nd 547-5725 MHz). These bnds re primrily ssigned for rdrs nd WLAN devices re secondry users tht opportunisticlly ccess the spectrum. WLANs protect the rdr by implementing n interference protection mechnism clled dynmic frequency selection DFS) specified in the stndrd IEEE 82.h [4]. The opportunity of secondry ccess cn be found in sptil nd temporl dimension. The sptil opportunity comes from the ttenution of rdio signl from the secondry trnsmitter, wheres the temporl opportunity exploits the ctivity or This is lso termed s rdio locl re network RLAN) nd wireless ccess system WAS) in the literture. mobility of the primry user [5]. The DFS mechnism minly relies on the sptil opportunity. The temporl spect of secondry ccess in the rdr spectrum hs not been studied well. Thus, it is of gret interest to exmine how much the secondry ccess to rdr will benefit from the temporl opportunity. A typicl rdr hs n ntenn with shrp min bem, which is rotting in regulr mnner. This property is expected to crete temporl opportunity to the secondry users. In [6], the uthors proposed becon signl from the rdr tht helps WLANs ccess the spectrum temporlly while the min bem of rdr ntenn does not fce it. However, there is lck of quntifiction of the ggregte interference nd the spectrum opportunity. In this pper, we quntittively ssess the temporl opportunities of the secondry ccess under the ssumption tht the ntenn pttern nd the rottion of the rdr re perfectly known to the WLAN devices. We consider n dvnced interference protection scheme tht utilizes the informtion of rdr rottion, which is termed temporl DFS. The performnce of the temporl DFS is compred with benchmrk cse where WLANs implement the conventionl DFS. In the performnce evlution, we consider multiple WLANs tht re heterogeneously distributed over lrge re. In essence, our im is: to quntify benefit of exploiting the temporl spect of rdr in the trnsmission opportunity of secondry users. The primry user must be protected from hrmful interference in order for WLANs to ccess the spectrum. Therefore, it is importnt to develop n ccurte model to describe the ggregte interference coming from the trnsmission of multiple WLANs in order to ssess the temporl opportunities for WLANs in the rdr bnd. In [7], we proposed mthemticl model of the ggregte interference tht considers n interference protection scheme resembling DFS. The model ws pplied to prcticl secondry ccess scenrios in [8], [9]. Since the work in [7] hs limittion tht the secondry users re ssumed to be uniformly distributed, it hs been extended to ccount for the heterogeneity of secondry user distribution in [], where res with denser secondry user intensities hot zones) re modeled s nnulus sectors. In this study, we further extend our model in [] to obtin the probbility distribution of the ggregte interference under the temporl DFS. The rest of the pper is orgnized s follows: Section II detils the system model nd explins the concept of temporl 26
The 4th Interntionl Symposium on Wireless Personl Multimedi Communictions WPMC'), θ H r H 2Δr H RB -> Primry receiver -> Secondry trnsmi er Fig.. Representtion of the scenrio when secondry users re heterogeneously distributed. DFS. Section III presents the nlytic model for the ggregte interference. Section IV shows the numericl results obtined from mthemticl nlysis. Finlly, we close with the conclusion in Section V. II. SYSTEM MODEL A. Distribution of the secondry users We ssume tht multiple secondry users re plced in lrge circulr re. The distribution of secondry users is heterogeneous in the sense tht there re regions of concentrtion, clled s hot zones. Detiled concept of hot zone cn be found in []. The secondry users my interfere with rdr which is locted t the center of the circle. Fig. illustrtes the considered system model. Let us consider n nnulus sector with inner rdius R, outer rdius R 2, nd centrl ngle θ c. Assume tht the secondry users re uniformly distributed inside the nnulus sector. Then, define r,θ ) s the loction of n rbitrry secondry user, where the rndom vrible RV) r is the distnce from the primry user to the user, nd the RV θ is its ngle. The pdf of the loction is given by: 2y f r,θ y, θ) = R2 2 R2 )θ, R <y R 2, θ θ c. c ) The nnulus sector cn represent both the bckground re nd the hot zone depicted in Fig. since circle is specil cse of the nnulus sector. For the cse of the bckground re, the following prmeters re pplied: R =, R 2 = R B, nd θ c =2π. As for the hot zone, the prmeters re R = r H Δ rh, R 2 = r H +Δ rh, nd θ c = θ H. Let us define N T s the totl number of secondry users in the circle. By denoting the number of secondry users in the bckground re nd the hot zone re by N B nd N H, respectively, we hve N T = N B + N H.Letρ B nd ρ H be the densities of secondry users in the bckground nd the hot zone, respectively ρ H ρ B ). B. Interference Protection Mechnism The protection of rdr is usully expressed in terms of the interference to noise rtio INR). We consider the minimum tolerble INR of 9 db. Let us define the mximum ggregte interference power tht the rdr cn tolerte s A thr.the INR of 9 db corresponds to A thr = 9 dbm with the rdr prmeters shown in Tble I. The ITU recommendtion M.638 [] describes the clcultion of mximum ggregte interference for different rdrs. Let us denote the ggregte interference the rdr receives from multiple ctive WLANs by I.IfI is greter thn A thr, the rdr will experience degrdtion in performnce. Therefore, regultory constrint β is defined s the mximum permitted probbility of interference such tht: Pr[I >A thr ] β. 2) We define ξ s the interference power tht the rdr would receive from WLAN if it were to trnsmit from the position r,θ ). It is given by ξ = G r θ )G w P t Lr )X, 3) where G r θ ) denotes the rdr s ntenn gin dependent on the position of the secondry user. G w is the WLAN s ntenn gin, P t denotes the effective trnsmit power of the secondry user, X is rndom vrible modeling shdow fding effect, nd Lr ) is the distnce dependent pth loss model defined s: Lr )=Cr α, 4) where C nd α re the pth loss constnt nd exponent, respectively. We employ simplified ntenn pttern model for rdr s shown below: { G mx G r θ )= r, if <θ θ MB, G min r, otherwise. where G mx r nd G min r re the mximum nd minimum of the ntenn gin, nd θ MB is the min bem width of the rdr. Thus, ξ cn hve two vlues in given loction depending on the rdr ntenn gin. Let ξ mx be ξ when the WLAN fces rdr min bem nd let ξ min denote ξ otherwise. The ctul trnsmission of user is regulted by n interference protection mechnism. Let I be the ctul interference from WLAN. We consider n instntneous moment when N T secondry users in the system desire to trnsmit. Then, the ggregte interference I is 5) N T I = I. 6) = The DFS relies on spectrum sensing of individul WLAN device such tht it decides whether it cn trnsmit or not bsed on the received rdr pulse power. Note tht the rdr receiver is generlly collocted with the trnsmitter. Thus, the estimtion of the received pulse power is equivlent to the estimtion of interference tht the WLAN will generte 27
The 4th Interntionl Symposium on Wireless Personl Multimedi Communictions WPMC'), 5 ) ON OFF 5 b) 5 ) ON 5 OFF b) 4 4 4 HOT ZONE 4 HOT ZONE 3 3 3 3 2 2 2 2 2 2 2 2 3 3 3 3 4 4 4 4 5 5 5 5 5 5 5 5 5 5 5 5 Fig. 2. Representtion of the exclusive region when secondry users re homogeneously distributed without shdow fding. ) represents when WLANs use DFS s interference protection mechnism nd b) when temporl DFS is used. Fig. 3. Representtion of the exclusive region when secondry users re heterogeneously distributed without shdow fding. ) represents when WLANs use DFS s interference protection mechnism nd b) when temporl DFS is used. to the rdr. Let I thr be the interference threshold used by ech individul WLAN to decide if it cn trnsmit. In the conventionl DFS mechnism, the decision is mde bsed on the mximum received pulse power during the sensing period. This mens tht I is shown s { ξ, if ξ mx I = I thr, 7), otherwise. Though the WLAN ctully induces interference power of ξ, the DFS relies on ξ mx for the decision of trnsmission. Thus, the conventionl DFS results in conservtive decision when the WLAN is not fcing the min bem of the rdr. If the WLAN is wre of the rdr ntenn nd rottion pttern, it cn mke refined decision bsed on the instntneous interference rther thn worst cse. Therefore, I under the temporl DFS is given by { ξ, if ξ I = I thr, 8), otherwise. Fig. 2 nd Fig. 3 illustrte how the number of ctive WLANs is incresed by using the temporl DFS scheme when the secondry users re homogeneously or heterogeneously distributed. Shdow fding is not considered in the figures in order to clerly show the difference between the four cses. It is observed tht the temporl DFS drmticlly increses the size of re where WLANs cn trnsmit, nd thus provides higher opportunity to the secondry users. Note tht we ssume tht the secondry users hve the perfect knowledge of temporl rdr chrcteristics. In prcticl environment, it my be difficult for the WLAN devices to know the exct ntenn pttern nd rottion of the rdr. Thus, this result cn be viewed s the mximum chievble gin of the temporl DFS. In order to compute the re where WLANs cn not trnsmit, let us define Er nd Er 2 s the re of the exclusive region when DFS nd temporl DFS re used, respectively. These re given by Er = π I thr G mx r G w P t C ) 2/α, 9) Er 2 = θ MB 2π Er )+ θ ) 2/α MB I thr 2 G min. r G w P t C ) Notice tht shdow fding is not considered. A mthemticl model of ggregte interference in the presence of shdow fding will be discussed in the next section. III. PROBABILITY DISTRIBUTION OF AGGREGATE INTERFERENCE A. Interference from n rbitrry secondry user We extend the model in [] to ccount for the temporl spect of rdr. The probbility distribution function pdf) of ξ is given s where f ξ z) =hz,r 2,θ ) hz,r,θ ), ) hz,y,θ )= Ωθ )z 2 α +erf ln z Gr θ )Gw P t Ly) 2σ 2 X ) 2σ X 2 α. 2) In 2), σx db denotes the stndrd devition of the shdowing in db scle, nd the constnt Ωθ ) is given by ) 2 α Ωθ )= exp R2 2 R2 )α 2σX 2 G r θ )G w P t C /α 2). 3) When I thr is pplied to user, it stops trnsmission if ξ exceeds I thr s depicted in 7) nd 8). This mens tht portion of secondry users hve zero trnsmission power. Tht portion of users is given s F ξ I thr ) where F ξ ) denotes the cumultive distribution function CDF) of ξ. Thus, the pdf of I is s follows: 28
The 4th Interntionl Symposium on Wireless Personl Multimedi Communictions WPMC'), F ξ I thr ), if z = f I z) = f ξ z), <z I thr 4), otherwise. The derived f I z) cn be directly pplied to both the bckground nd the hot zone secondry users by dusting the prmeters R, R 2, nd θ c s discussed in II-A. Note lso tht the interference threshold I thr is ssumed to be predetermined to stisfy the condition 2). B. pdf of the ggregte interference Let us define I B nd Ii H s the interference from the secondry user in the bckground nd the secondry user i in the hot zone, respectively. The pdf of I B nd Ii H re given by 4). Since we ssume tht N B secondry users re uniformly distributed within the bckground re nd N H uniform secondry users re within the hot zone, the totl interference received t the primry user from ll the secondry users I T is computed s the sum of the ggregte interference from the users in the bckground I B nd the ggregte interference from the users in the hot zone I H. N B I T = I B + I H = I B + = N H i= I H i. 5) We employ cumulnt-bsed pproch to pproximte the pdf of I T. Note tht the cumulnts hve n ttrctive property tht the m th cumulnt of the sum of independent RVs is equl to the sum of the individul m th cumulnts [2]. Also, the first nd second cumulnts of RV correspond to the men nd vrince. Let k I B m) nd k I H m) denote the m th cumulnt of I B nd I H, respectively. Then, k I T m) = k I B m)+k I H m) 6) N B N H = m)+ m). k I B = From the cumulnts of the I T, the pdf of I T cn be pproximted s known distribution by employing the method of moments. We use log-norml distribution to pproximte the pdf of I T s in 7). ) lnz) μi T f I T z) = exp z 2πσ 2 2σ 2. 7) I T I T i= The prmeters μ I T nd σ 2 I of the pdf cn be obtined from T the first nd second cumulnt computtions s k I H i k I T ) = E[I T ] = exp[μ I T + σ 2 I /2], 8) T k I T 2) = Vr[I T ] = exp[σ 2 I ] ) exp[2μ T I T + σ2 I ]. 9) T IV. NUMERICAL RESULTS The prmeter vlues used to model the primry nd secondry systems re described in Tble I. The bsic propgtion loss model used in this study is the C-suburbn WINNER TABLE I SIMULATION PARAMETER VALUES Prmeters Vlues. Primry receiver Meteorologicl rdr) Frequency bnd [MHz] 56 Antenn height [meter] 3 Trnsmission power [kw] 25 Bndwidth [MHz] 4 Antenn gin [dbi] 4 Rottion per minute RPM) [6 /sec] Antenn min bem width θ MB 2 2. Secondry trnsmitters WLAN) Antenn height [meter].5 Trnsmission power [P t in W].2 Bndwidth [MHz] 2 Antenn gin [dbi] Rdius of bckground re [R B in km] 5 Shdowing stndrd devition [σx db in db] 8 Interference to noise rtio threshold [INR in db] -9 Aggregte interference threshold [A thr in dbm] -9 Probbility of mximum interference β.5 model which is proposed for 5GHz bnd by WINNER proect in [3]. The prmeters of the propgtion model in db scle re: Ld) =4.236 + 3.5225 log d[meter]). 2) The heterogeneous distribution of secondry users is modeled by considering hot zone t 5 km from the rdr with Δ rh =5 km nd θ H =. The hot zone nd bckground densities re ρ H =2/km 2 nd ρ H =/km 2, respectively. Thus, in this work we consider 8378 WLANs in totl. Fig. 4 presents the probbility of trnsmission for WLAN depending on the distnce from the rdr. Conventionl DFS is compred with the temporl DFS. It is observed the temporl DFS provides the secondry users with higher trnsmission opportunity when WLANs re not fcing the rdr min bem. This effect is significnt prticulrly when the secondry users re ner the rdr. On the other hnd, the trnsmission probbility is reduced if WLAN fces the min bem. The difference in the performnce between the conventionl DFS nd the temporl DFS is more noticeble when the secondry users re heterogeneously distributed, i.e. when there is region close to the rdr where secondry users re densely populted. Fig. 5 shows the probbility of trnsmission in the temporl domin. Let us consider WLAN who is 4 km wy from the rdr under the heterogeneous user distribution. This figure illustrtes tht the secondry user hs higher chnce to trnsmit during most of the time by using the temporl DFS. Note tht we ssume it tkes 6 seconds for the rdr to complete rottion RPM of ). During the 6 seconds, the WLAN cn trnsmit 99.45% of the time with the probbility of lmost when the temporl DFS is employed. Although 29
The 4th Interntionl Symposium on Wireless Personl Multimedi Communictions WPMC'), Probbility of trnsmission.8.6.4.2 Using Temporl DFS nd fcing rdr Using Temporl DFS nd not fcing rdr 2 4 6 8 2 4 6 8 2.8.6.4.2 Homogeneous Distribution Distnce r from the rdr [km] Heterogeneous Distribution Using Temporl DFS nd fcing rdr Using Temporl DFS nd not fcing rdr 2 4 6 8 2 4 6 8 2 Distnce r from the rdr [km] Fig. 4. Probbility of trnsmission for given WLAN t distnce r from the rdr. Probbility of trnsmission.8.6.4.2 Heterogeneous Distribution Using Temporl DFS 2 4 6 8 2 Time [sec] Fig. 5. Probbility of trnsmission s function of time for WLAN 4km wy from the rdr. the trnsmission is interrupted once every 6 seconds, it is negligible in most of mobile dt services. On the contrry, this probbility is fixed to 9% ll the time for the conventionl DFS cse. V. CONCLUSION We quntified the number of WLANs tht cn ccess the rdr spectrum s secondry users. Prticulrly, we exmined how much opportunity the secondry users cn obtin if they exploit the temporl vrition of interference due to the rottion of rdr ntenn. The WLANs re homogeneously nd heterogeneously distributed. The heterogeneous distribution considers regions with higher secondry user densities, which is termed hot zones. The number of ctive secondry users is clculted under the condition tht the ggregte interference to the rdr does not exceed the interference threshold, gurnteing no hrmful interference to the primry user. The trnsmission of ech WLAN is regulted by n interference protection mechnism, temporl DFS, which tkes into ccount the rdr ntenn pttern nd rottion. We compred the performnce of the temporl DFS with the conventionl DFS in terms of the probbility of trnsmission nd the number of ctive WLANs. It is shown tht by using the temporl DFS s the interference control mechnism WLANs cn increse the probbility of ccessing the rdr spectrum compred with the conventionl DFS mechnism. Thus, the rdr spectrum is better utilized nd the rdr is protected from interference. In this work, we ssumed tht the WLANs cn detect the rdr nd estimte the temporl vrition of propgtion loss ccurtely. The study of the miss detection, flse lrm, nd inccurte estimtion of temporl rdr chrcteristics remin s the direction for future reserch. ACKNOWLEDGMENT The reserch leding to these results hs received prtil funding from the Europen Union s Seventh Frmework Progrmme FP7/27-23 under grnt greement n 24833 QUASAR). The uthors lso would like to cknowledge the VINNOVA proect MODyS for providing prtil funding. REFERENCES [] D. Cbric, I. O Donnell, M.-W. Chen, nd R. Brodersen, Spectrum Shring Rdios, IEEE Circuits nd Systems Mgzine, vol. 6, no. 2, pp. 3 45, 26. [2] Federl Communictions Commission Spectrum Policy Tsk Force, Report of the spectrum efficency working group, Tech. Rep., Nov. 22, [Online]. Avilble: http://www.fcc.gov/sptf/reports.html. [3] J. Mitol III nd G. Q. Mguire Jr, Cognitive Rdio: Mking Softwre Rdios More Personl, IEEE Personl Communictions, vol. 6, no. 4, pp. 3 8, Aug. 999. [4] IEEE Std 82.h-23, Spectrum nd Trnsmit Power Mngement Extensions in the 5 GHz bnd in Europe, Oct. 23. [5] K. W. Sung, S.-L. Kim, nd J. Znder, temporl spectrum shring bsed on primry user ctivity prediction, IEEE Trnsctions on Wireless Communictions, vol. 9, no. 2, pp. 3848 3855, Dec. 2. [6] Z. Horváth nd D. Vrg, Chnnel lloction technique for eliminting interference cused by RLANs on meteorologicl rdrs in 5 GHz bnd, Infocommunictions Journl, vol. 64, no. 3, pp. 24 34, 29. [7] K. W. Sung, M. Tercero, nd J. Znder, Aggregte Interference in Secondry Access with Interference Protection, IEEE Communictions Letters, vol. 5, no. 6, pp. 629 63, Jun. 2. [8] M. Tercero, K. W. Sung, nd J. Znder, Impct of Aggregte Interference on Meteorologicl Rdr from Secondry Users, in Proc. IEEE WCNC, Cncun, Mr. 28 3 2. [9] K. W. Sung, E. Obregon, nd J. Znder, On the Requirements of Secondry Access to 96-25 MHz Aeronuticl Spectrum, in Proc. IEEE DySPAN, Achen, My 3 6 2. [] M. Tercero, K. W. Sung, nd J. Znder, Aggregte Interference from Secondry Users with Heterogeneous Density, in Proc. IEEE PIMRC, Toronto, Sep. 4 2. [] Rec. ITU-R M.638, Chrcteristics nd protection criteri for shring studies for rdioloction, eronuticl rdionvigtion nd meteorologicl rdrs operting in the frequency bnds between 525 nd 585 MHz, Interntionl Telecommuniction Union Std., 23. [2] M. Aluid nd H. Ynikomeroglu, A Cumulnt-Bsed Chrcteriztion of the Aggregte Interference Power in Wireless Networks, in Proc. 7st IEEE Vehiculr Technology Conference VTC), Tipei, My 6-9 2. [3] IST-4-27756 WINNER II, D..2 v.2 WINNER II Chnnel Models, [Online]. Avilble: https://www.ist-winner.org/winner2- Deliverbles/. 3