Determination of Medium Access Probability of Cognitive Radio under Different Fading Channels

Transcription

1 International Journal of Soft Computing and Engineering (IJSCE) ISSN: 3-37, Volume-, Iue-3, July Determination of Medium Acce Probability of Cognitive adio under Different Fading Channel ezwan Ahmad, Md. Imdadul Ilam, M.. Amin Abtract The correct deciion in detecting the preence of the primary ue i a vital requirement in cognitive radio network. Incorporation of patial fale alarm make the derivation of probability of correct deciion a difficult tak. The previou literature perform the tak only for the cae of the received ignal under ditribution of the fading channel. In thi paper, we enhance the work for three mall cale fading channel:, and Nakagami-m Fading Channel to get the real cenario of a cognitive radio network in an urban area. The impact of fading paramete and ening range on the profile of probability of correct deciion i alo invetigated to optimize the performance of the network. Index Term Cognitive radio, pectrum ening, pectrum hole, medium acce probability, patial fale alarm. I. INTODUCTION With the advent of new high data rate wirele application, a well a growth of exiting wirele ervice, demand for additional bandwidth i rapidly increaing. To be available anywhere and anytime when needed, the wirele medium ha to be exploited. Currently, pectrum allotment i operated by providing each new ervice with it own fixed frequency block. Demand for acce to pectrum ha been growing dramatically, and i likely to continue to grow in the foreeeable future. New ervice, uch a unlicened wirele Internet acce and atellite digital audio broadcating (DAB), are being launched and are quickly reaching hundred of thouand of conume. At the ame time, mot prime pectrum ha been aigned, and it i becoming increaingly difficult to find pectrum that can be made available for either new ervice or to expand exiting infratructure. Exiting command-and-control pectrum allocation defined by government regulatory agencie prohibit unlicened acce to licened pectrum, contraining them intead to everal heavily populated, interference-prone frequency band. A a reult, there exit an apparent carcity of tranmiion pectrum that i forcing a critical rethinking of Manucript received on July,. ezwan Ahmad, Department of Electronic and Communication Engineering, Eat Wet Univeity, Johurul Ilam City, Aftabnagar, Dhaka 9, Bangladeh. Md. Imdadul Ilam, Department of Computer Science and Engineering, Jahangirnagar Univeity, Dhaka, Bangladeh. M.. Amin, Department of Electronic and Communication Engineering, Eat Wet Univeity, Johurul Ilam City, Aftabnagar, Dhaka 9, Bangladeh. how wirele communication i performed. Additionally, it may be difficult to gain acce to a large contiguou block of frequency pectrum, which can be a major limitation for high data rate tranmiion. Even though much of the pectrum ha been allocated, preliminary meaurement indicate that uage of the frequency pectrum i pae both patially and temporally. A the pectrum i underutilized by the primary licene holde, there exit pectrum hole that can be ued by econdary ue, yielding efficient utilization of the pectrum and eliminating the apparent pectrum carcity. To enure that exiting ervice can continue to grow and evolve, it i important that efficient acce to and ue of the radio pectrum hould be promoted []. A a ignificant part of frequency reource are already allocated to licened ytem, upending new ytem from being introduced. However, according to many meaurement campaign, pectrum utilization in thee licened band i relatively low [], []. To face thi problem, new dynamic pectrum acce policie are propoed, that allow a pectrum owner called primary uer (PU) to leae part of the unued pectrum reource for unlicened (opportunitic) ue by the o-called econdary uer (SU). The tranceiver that can exploit uch a pectrum cenario without cauing a harmful interference to licened ue i named a cognitive radio (C) [3]-[5]. There are four main function of a C and they are a follow:. Spectrum Sening finding unued pectrum;. Spectrum Management finding the bet available pectrum to ue baed on quality of ervice criteria; 3. Spectrum Mobility allowing for the hift from one frequency to another; 4. Spectrum Sharing cheduling and haring pectrum in a fair manner. In [6], the author ha developed a theory to find expreion for the medium acce probability by including the effect of the conventional fale alarm (CFA) probability and the patial fale alarm (SFA) probability. The paper explicitly dicue the probability of correct deciion only for the cae of received ignal under ditribution of the fading channel. In the preent paper, we enhance the work of Han et. al. [6] for the following three mall cale fading channel:, and Nakagami-m fading channel to get the real cenario of a C network in an urban area. The impact of the fading paramete and the ening range on the profile of the probability of correct deciion i alo invetigated to optimize the performance of the network. The paper i organized a follow: Section I decribe the ytem model of the work, where ening conideration in C 459

2 Determination of Medium Acce Probability of Cognitive adio Under Different Fading Channel and SFA and medium acce probability i decribed. In thi ection, a brief decription i given about the application of the medium acce probability for different fading channel. Section III decribe the reult of the invetigation in detail and finally the concluion i given in Sec. IV. II. SYSTEM MODEL. Sening Conideration in C and SFA Problem The objective of conventional ening in C i to undetand the real on-off tatu of a PU inide the SU ening range. To properly determine one of the above two hypothee: PU off, acce opportunity i available (denoted by Η ) or PU on, acce opportunity i unavailable (denoted by ( H ) i obtained by analyzing the received ignal x [7]-[]: Η : x[ w[, n,,, N, () H : x[ [ w[, n,,, N, where x [ i the received ignal by the SU, [ n ] i the ignal from the PU locating at the edge of the ening region; and N i the number of ample. The ditance between the PU and the SU i denoted by r and the ening range i denoted by which enure the econdary receiver and primary receiver are not interfered by each of other relevant tranmitter []. If the primary tranmit power i S and the noie power i n, then the relation between the target primary-ignal-to-noie ratio (PSN) ~ and the ening range r i given by the following equation: ~ Sh r, () n where h(.) denote, for a given ditance, the power lo including the path-lo, hadowing, and alo multipath fading []. A we know that SU i granted to utilize the idle primary channel dynamically if no buy PU i detected inide the econdary ening range r. Another way we can ay that a SU i permitted to acce primary channel regardle of the tranmitting PU outide the ening range. In other word, a SU i granted permiion to acce the primary channel regardle of the tranmitting PU reide outide the ening range. PU r SU PU Fig. Sening rage of a SU within a cell: Shaded part of the figure i the ening rage of the SU. The ening range with repect to a SU (haded region), the cell boundary and the location of the PU within and outide the ening region i hown in Fig. ; where i the cell radiu and r i the radiu of the ening region. However, in real ituation, the cenario may be quite different. It wa hown earlier that when the ditance between the SU and the PU (denoted by r) i larger than the ening range (denoted by r ), the SU till can ene the preence of the PU with ome detection probability []. It wa hown that the detection probability doe not decreae efficiently to a lower level in term of the primary ignal from the outide of the ening range. Thi phenomenon i proved by the experiment performed in []. On the other hand, the area r increae rapidly a r increae. The larger area r mean the higher probability to meet the PU. A a reult, conidering the detection probability ( r ) and the probability to meet a PU jointly, the SU can uffer from high probability to loe acce opportunity.. SFA and Medium Acce Probability For a quantitative analyi and a practical cenario, let u conider that a PU i located at a ditance from the SU. We further conider that the PU ha a probability p to be in the tranmitting tate at any intant of time. The mall cale primary network i conidered to be a follow: for the exitence of a PU in a circular oberved window, of which the area i, where [6]. Here the pectrum ening i to properly determine the one of the two tate Η and Η, where Η denote the acce opportunity i to be available; and Η denote the acce opportunity to be unavailable. In addition, the tate Η i divided into two ub-tate: Η,- and Η,. The ub-tate Η,- denote that the PU i in the witching on tate inide the ening range and Η, denote that the PU i in the tranmitting tate and reiding at a point outide the ening range. Thu, we can write Η Η : : Η,- : x[ w[, Η, : x[ [ w[, x[ [ w[, n,,, N, n,,, N, where [ i the primary ignal when the PU i inide the econdary ening region, and it power i given by (3) Sh r, r r ; while [ i the primary ignal when the PU reide outide the econdary ening region, and it power i Sh r, r r. For the SU, the probability of acce opportunity provided by the primary network i the probability that the pectrum hole i available for the SU, denoted by P Η, and i given by [6] P Η p, r. (4) The conditional probability, P Η Η,-, denote the CFA probability, and P Η Η, denote the SFA probability. Hence, the total acce opportunity lot by ening i the um 46

3 P, that i, of P Η Η,- and Η Η, P Η Η P Η Η.,-, For Η,- and including the CFA probability, P MA P (5) N where T x[ / N i the tet tatitic with energy n detection (ED) and i the detection threhold [6]. Since P Η Η, i the probability of the PU to be detected under r, thu for Η, and including the SFA probability, P Η Η,, the medium acce probability can be written a P MA P Η P Η, Η, p p D dr, where f (r) i the probability denity function (pdf) of the PU location and i given by International Journal of Soft Computing and Engineering (IJSCE) ISSN: 3-37, Volume-, Iue-3, July (6) r / a f (r) hould give dr and p D (r) i the detection probability. Thu the total medium acce probability, P MA, i the um of P MA and P MA a given by Eq. (5) and (6): P (,- Η,- P Η Η,- P Η, P Η Η, p) P T Η p r P Η Η,-, the medium acce probability i given by the following expreion,- Η,- P Η Η,- ( p) P T Η, p dr. In the following, we briefly decribe how to calculate the probability P T Η,- and the detection probability p D (r) to evaluate the total medium acce probability P MA a given by Eq. (7). It i to be mentioned here that the upper limit of the integral in Eq. (7) i r which i derived from Eq. (4). p P Η D (7), (8).3 Medium Acce Probability in the cae of Nakagami-m,, and Ditribution of the Fading Channel In thi ection we decribe the medium acce probability a given by Eq. (7) in the cae of different ditribution of the fading channel. For the pdf f X (x) of X for a fading channel, the conventional fale alarm probability appearing in Eq. (7), viz., P Η Η,- can be calculated a P Η Η,- f X ( x) dx, (9) where i defined earlier. Thu the probability P T Η,- can be obtained a T Η P Η P,- Η,- () f X ( x) dx, A an example, for Nakagami-m Channel, the probability P T i given by the following expreion Η,- P T Η,- f N,, m d, () where m m m m,, / f N m e, () m ( m) where i the time-average power of the received ignal and m i the Nakagami fading parameter, ometime called hape factor of the Nakagami-m ditribution. The value of m range between / and. Taking the variation of the average ignal trength with ditance r a (r), we can write the detection probability P D (r) a m m m m / ( r) PD e d. (3) m ( m) For other fading cae, the appropriate pdf f X (x) hould be ued in Eq. (9) to calculate the CFA probability. III. ESULTS For each of the fading cae, we elect two et of paramete: one for the cae of the abence of any PU and the other for the cae of the preence of the PU. For Nakagami-m fading cae, we chooe the fading paramete,. and m =. For fading channel, the only parameter:.. For fading cae,. and the peak amplitude of the dominant ignal, A =. For the pdf of the ditribution,.4,., [6]. The variation of ignal trength with ditance i conidered a kr, where n i the path lo exponent and we have taken k =.3 and n =.7. The value of p i taken a.7. Figure how the variation of the medium acce probability P MA againt the probability of acce opportunity P Η taking. and r 3. The condition of the network i better for the cae when the received ignal follow the pdf. For other three cae (, and Nakagami-m), the probabilitie P MA are very cloe to each other and each of them rie exponentially. When the acce opportunity P Η >.85, the for, and Nakagami-m fading cae are almot equal. n 46

4 Determination of Medium Acce Probability of Cognitive adio Under Different Fading Channel P(H) Fig. The variation of medium acce probability P MA againt the probability of acce opportunity P Η taking =., r P(H) Fig. 5 The variation of medium acce probability P MA againt the probability of acce opportunity P Η taking =. r P(H) Fig. 3 The variation of medium acce probability P MA againt the probability of acce opportunity P Η taking =., r P(H) Fig. 6 The variation of medium acce probability P MA againt the probability of acce opportunity P Η taking =., r P(H) Fig. 4 The variation of medium acce probability P MA againt the probability of acce opportunity P Η taking =.5, r P(H) Fig. 7 The variation of medium acce probability P MA againt the probability of acce opportunity P Η taking =.5 r 4. 46

5 The impact of the threhold i eaily viualized from Fig. - 4, where the threhold i varied a =.,. and.5. A i oberved, a very little improvement i found for the cae of the pdf, but for the other three fading cae, the profile of the medium acce probability, P MA, rie. It i alo oberved from Fig. 5-7 that the profile of P MA i heavily dependent on the parameter r, the ening range. If the value.,. and.5 for the parameter are applied for r 4, the increae in P MA i found more prominent and the profile of all the curve approach to linear. IV. CONCLUSION The paper invetigate the impact of fading condition of the wirele channel, ening range and threhold value of the received ignal on the medium acce probability. The profile of the medium acce probability change from exponential to linear with the increment of the radiu of the ening region, r. The performance of the network i found to be the bet for the cae of normal fading but the ituation i very poor for other three fading condition. The impact of mall cale fading i prominent in an urban area hence bad reult of, Nakagami-m and fading reflect the realitic cenario of the wirele network, which i the actual finding of the paper. Still, we have the cope of incorporating the Walfih-Ikegami model, Okumura-Hata model and Lee Prediction model intead of the ordinary path lo exponent formula ued in the preent invetigation. Even, including the adaptive equalizer or match filter with the detector can improve the performance of the network. EFEENCES [] S. Haykin, Cognitive radio: brain-empowered wirele communication, IEEE Journal on Selected Area in Communication, vol. 3, pp. -, Feb. 5. [] D.Cabric, S.Mihra, and.w.brodeen, Implementation iue in pectrum ening for cognitive radio, in Proc. 38th Ailomar Conf. Signal,Sytem and Compute, Pacific Grove,CA, 4, pp [3] A. Ghaemi and E. S. Soua, Spectrum Sening in Cognitive adio Network: The Cooperation-Proceing Trade-Off, Wiley Wirele Commun. and Mobile Comp. Special Iue on Cognitive adio, Software-Defined adio, and Adaptive Wirele Sytem, vol. 7, no. 9, pp. 49 6, Nov. 7 [4] Jun Ma, Geoffery Ye Li, and Biing Hwang (Fred) Juang, Signal Proceing in Cognitive adio, in Priceeding of the IEEE, May 9, vol. 97, no. 5, pp [5] A Ghaemi, ES Soua, Interference aggregation in pectrum-ening cognitive wirele network, IEEE J. Select Topic Signal Proce., vol., no., pp. 4 55, 8. [6] Weijia Han, Jiandong Li, Qin Liu, and Linjing Zhao, Spatial Fale Alarm in Cognitive adio, IEEE Commun. Lett.., vol. 5, no. 5, pp. 58 5, May.. [7] Maha Derakhhani, Tho Le-Ngoc, Aggregate Interference and capacity-outage Analyi in a Cogitive adio Network, IEEE Tran. Veh. Technol., vol. 6, no,, pp. 96-7, January. [8] Maha Derakhhani, Tho Le-Ngoc, Beacon tranmitter placement on aggregate interference and capacity-outage in a cognitive radio network, in 7 nd IEEE Vehicular Technology Conference, Ottawa, Canada, Sep., pp. -5. [9] Marco Cardenu-Jaurez, Mounir Ghogho, Spectrum Sening and Throughput Trade-off in Cognitive radio under Outage contraint over Nakagami Fading, IEEE Commun. Lett., vol. 5, no., pp. -3, October. [] A. Ghaemi and E. S. Soua, Collaborative Spectrum Sening for Opportunitic Acce in Fading Environment, in Proc. IEEE t Symp. Dynamic Spectrum Acce Network, Baltimore, MD, Nov. 5, pp International Journal of Soft Computing and Engineering (IJSCE) ISSN: 3-37, Volume-, Iue-3, July [] F F. Digham, M.S. Alouini, M. Simon, On the energy detection of unknown ignal over fading channel, IEEE Tran Commun., vol. 55, no., pp , Jan. 7. [] D. Cabric, Experimental tudy of pectrum ening baed on energy detection and network cooperation, IEEE Tran. Wirele Commun., vol. 7, no., pp , Nov. 8. ezwan Khan ha completed hi B.S. in Computer Engineering degree from the American International Univeity of Bangladeh, Dhaka in 4. He i currently tudying Applied Phyic at Eat Wet Univeity for hi M.S. degree. Hi field of tudy i wirele communication and network. Md. Imdadul Ilam ha completed hi B.Sc. and M.Sc Engineering in Electrical and Electronic Engineering from Bangladeh Univeity of Engineering and Technology, Dhaka, Bangladeh in 993 and 998 repectively and ha completed hi Ph.D degree from the Department of Computer Science and Engineering, Jahangirnagar Univeity, Dhaka, Bangladeh in the field of network traffic engineering in. He i now working a a Profeor at the Department of Computer Science and Engineering, Jahangirnagar Univeity, Savar, Dhaka, Bangladeh. Previouly, he worked a an Aitant Engineer in Sheba Telecom (Pvt.) LTD (A joint venture company between Bangladeh and Malayia, for Mobile cellular and WLL), from Sept.994 to July 996. Dr Ilam ha a very good field experience in intallation of adio Bae Station and Switching Cente for WLL. Hi reearch field i network traffic, wirele communication, wavelet tranform, OFDMA, WCDMA, adaptive filter theory, ANFIS and array antenna ytem. He ha more than hundred reearch pape in national and international journal and conference proceeding. M.. Amin received hi B.S. and M.S. degree in Phyic from Jahangirnagar Univeity, Dhaka, Bangladeh in 984 and 986 repectively and hi Ph.D. degree in Plama Phyic from the Univeity of St. Andrew, U. K. in 99. He i currently working a a Profeor of Electronic and Communication Engineering at Eat Wet Univeity, Dhaka, Bangladeh. He erved a a Pot-Doctoral eearch Aociate in Electrical Engineering at the Univeity of Alberta, Canada, during He wa an Alexander von Humboldt eearch Fellow at the Max-Planck Intitute for Extraterretrial Phyic at Garching/Munich, Germany during Dr. Amin wa awarded the Commonwealth Potdoctoral Fellowhip in 997. Beide thee, he ha alo received everal award for hi reearch, including the Bangladeh Academy of Science Young Scientit Award for the year 996 and the Univeity Grant Commiion Young Scientit Award for 996. Hi current reearch field are wirele communication and network and alo nonlinear plama dynamic. He i a member of the IEEE. 463