Optimal and water-filling Algorithm approach for power Allocation in OFDM Based Cognitive Radio System

Transcription

1 Internatona Journa of Engneerng Research and Technoogy. ISS Voume 10, umber 1 (017) Internatona Research Pubcaton House Optma and water-fng Agorthm approach for power Aocaton n OFDM Based Cogntve Rado System Dr. A.. Jadhav Professor, Department of Eectroncs & Teecommuncaton, D. Y. Pat Coege of Engneerng & Technoogy Kasba Bawada,416006, Kohapur, Maharastra, Inda. S. R. Mujawar Department of Eectroncs & Teecommuncaton, D. Y. Pat Coege of Engneerng & Technoogy Kasba Bawada,416006, Kohapur, Maharastra, Inda. Mrs. P. S. Pse Assocate Professor, Department of Eectroncs & Teecommuncaton, D. Y. Pat Coege of Engneerng & Technoogy Kasba Bawada,416006, Kohapur, Maharastra, Inda. Abstract Technoogy pays vta roe n today s word. Growng technoogy n wreess communcaton and demand for spectrum s growng day by day. Many appcatons n wreess are sharng same medum and due to stran whch eads to ack of spectrum n gven mted band. As per the report pubshed by (FCC) Federa Communcaton Commsson the spectrum band are not utzed effcenty where as some of spectrum bands are used heavy. The potenta souton for ths probem s to aocate the spectrum dynamcay through means of cogntve rado. In ths paper water-fng and optma power aocaton agorthm has been nvestgated n smuaton envronment. Present numerca resut show that optma power aocaton agorthm can acheve hgher transmsson compare to water-fng (.e. cassca) power aocaton agorthm. The above agorthm s compared to tota transmsson power, nterference and ndvdua peak constrants. Keywords- Cogntve Rado, Prmary user, secondary user Orthogona frequency dvson mutpexng. Introducton Wth the advance deveopment n wreess communcatons, frequency spectrum has becomng a vauabe natura resource, and mtaton of the spectrum s a serous ssue. On other hand Federa Communcaton Commssons (FCC) has reported the underutzaton of eectromagnetc rado spectrum [1]. In some cases, the spectrum band s underutzed by censed users (prmary user) and uncensed users (secondary user) are not aowed to operate n censed spectrum bands. Ths eads to unbaanced spectrum utzaton. Spectra effcency can be ncreased sgnfcanty by gvng access of the frequency bands to a group of uncensed users (referred to as secondary or CR users). Cogntve rado(cr) system s unque softer-ware defned rado proposed to mprove spectrum effcency by expotng unused spectrum n dynamcay changng envronments. The cogntve rado s an nnovatve system desgn whch nvoves smarty sensng the hoes of spectrum and then determnng the transmsson characterstcs wth respectve to symbo rate, power, bandwdth, atency etc. of a group of secondary users based on the nature of the users to whom the spectrum has been censed (referred to as prmary users). A Cogntve rado (CR) network s a technoogy n whch a frequency band used by one or mutpe prmary users n a prmary network can be operated by a secondary user s network whch conssts of one or mutpe secondary users. To guaranty the quaty of servce (QoS) of the Prmary User (PU) and to maxmze the transmsson rate of the secondary users s one of the most cruca roes for a Cogntve rado system [1]. Orthogona frequency dvson mutpexng (OFDM) s a promsng tech for cogntve rado systems. Wth OFDM, the SU has the abty to smoothy fexbe f the spectra gaps eft unused by PUs. It aso can determne ts ocaton; sense spectrums of other devces, and change n frequency, adjust output power and even ater transmsson parameters and characterstcs. It can fuf the SUs communcaton needs wthout aterng the FCC rues. Cogntve rado many works on three tasks ncudng: Rado scene anayss n whch CR can detect spectrum hoes, ghty used band, nterference and Channe state estmaton, n whch CR determne the channe capacty and the state of the channe; Spectrum management, n whch CR make the spectrum sharng effcent. The CR desgn s an featurng new methods of rado desgn phosophy whch nvoves smoothy sensng the vacant spectrum and then determnng the transmsson characterstcs (e.g., symbo rate, power, bandwdth, atency) of a group of secondary users based on the behavor of the users to whom the spectrum has been censed (referred to as prmary users). Athough opportunstc spectrum access woud aow CR user to dscover and access avaabe spectrum 470

2 Internatona Journa of Engneerng Research and Technoogy. ISS Voume 10, umber 1 (017) Internatona Research Pubcaton House resources, one of the man am s to utze the avaabe spectrum resources n an effcent manner. OFDM base CR cogntve Rado:-OFDM s a mut-carrer moduaton technque that can overcome many probems that arse wth hgh bt-rate communcatons, the most serous of whch s tme dsperson. The data contanng symbo stream s spt nto severa ow rate streams, and these streams are transmtted on dfferent carrers. Because ths spttng ncreases the symbo duraton by orthogonay overappng carrers (subcarrers), mutpath echoes affect ony a sma part of the neghborng symbos. The remanng nter-symbo nterference (ISI) s removed by stretchng of the OFDM symbo wth a cycc prefx (CP). Other advantages of OFDM ncude hgh spectra effcency and robustness aganst narrowband nterference (BI). Orthogona frequency dvson mutpexng (OFDM), because of ts smooth fexbty n aocatng the spectrum, has been recognzed as a best ar nterface technoogy for CR systems []. Because of the coexstence of CR and prmary users s near to each other of bands, mutua nterference between these users s the mtng factor n order to acheve a better performance for CR systems. we propose a power oadng agorthm that maxmzed the downnk transmsson rate whe keepng the tota nterference ntroduced to dfferent PU recevers beow a threshod eve.a dspense agorthm for optma resource aocaton n OFDMA based CR systems has been proposed. When usng orthogona frequency Dvson Mutpexng n cogntve rado network, the power aocaton schemes for spectrum resources w be convenent and very fexbe. However, t become chaengng Roe to aocate power to ndvdua sub channes n the OFDM-based cogntve rado networks. Due to the above reasons, OFDM-based CR systems have more attenton and the reated resource aocaton probems have become good research topcs. Ths paper focuses on nvestgatng the research chaenges nvoved n the power aocaton for OFDM based cogntve rado system. Hence, the desgn probem s that gven an nterference threshod prescrbed by the prmary users, how much power shoud be transmtted nto each CR user s subcarrer such that the transmsson rate of the CR user s maxmzed. In ths paper optma power aocaton agorthm s proposed as compare to water-fng agorthm whch maxmze the downnk transmsson capacty of cogntve rado users whe keepng the nterference ntroduce to PU beow a specfed threshod. SYSTEM MODEL We consder a wreess system consstng of L sub-channes censed to dfferent prmary users. Each of these prmary users behaves dfferenty or has uncorreated actvty n ther band. A the subchannes are dvded nto mutpe subcarrers as shown n fgure1 and they are opportunstcay avaabe to some secondary or cogntve user whch uses the band n OFDM fashon. the tota number of subcarrers are wth M sub-channes censed to dfferent prmary users. In ths, frst, we dscuss the subcarrer groupng strategy that s used to maxmze the transmsson rate and mnmze nterference. Second, we outne n detas the system mode used n ths thess. Ths ncudes a descrpton of the transmtter and recever, the adaptve subcarrer aocaton scheme used n our work, as we as the channe mode used n the anayss. Underay Mode We consder a downnk transmsson scenaro. It s assumed that the frequency bands of bandwdth B1, B... BL have been occuped by PU1, PU... PUL. As n Fgure 3.1, SUs can occupy ether the spectrum of PUs or the adjacent spectrum of PUs. The avaabe bandwdth for CR transmsson s dvded nto subcarrers based OFDM system, and the bandwdth for each subcarrer s f H z. In the downnk transmsson scenaro, there are three nstantaneous fadng gans: between the SUs transmtter and SUs recever for the th subcarrer denoted as ss ; between the SUs transmtter and th PU recever denoted as sp ; between th Pus transmtter and SUs recever denoted as ps. We assume that these nstantaneous fadng gans are perfecty known at the SUs transmtter. Interference ntroduced to PU by SU We assume that the sgna transmtted on the subcarrer s an dea yqust puse. The power spectrum densty of the th subcarrer can be wrtten as: Where, X (f) = P T s ( snπf T s πft s ) (1) P : tota transmts power n the th subcarrer. T s : the symbo duraton. Then the nterference ntroduced to the th PU band by the th subcarrer s: I = P sp Ts d + B d B snπf Ts πf Ts df () d : dstance n frequency between the th subcarrer and the th PU band, B : represents occuped bandwdth by the th PU. Interference ntroduced to SU by PU The power spectrum densty of the PU sgna after M-fast Fourer transform (FFT) processng can be expressed as: E[I (ω) = 1 πm Fgure 1: Dstrbuton of prmary and CR users. π π X PU e jω sn ω y M sn ω y dy (3) X PU (e jω ) : power spectrum densty of the PU sgna. The PU sgna has been taken to be an eptcay ftered whte nose process wth amptude P PU. Accordng to the 471

3 Internatona Journa of Engneerng Research and Technoogy. ISS Voume 10, umber 1 (017) Internatona Research Pubcaton House nterference ntroduced to the th subcarrer by the th PU band can be wrtten as: J (P PU ) = sp d + f E I ω dω d f (4) Optma Power Loadng Agorthm Accordng to Shannon capacty formua, the transmsson rate at the th subcarrer s gven by: R P = og 1 + σ + ss p L =1 J σ = addtve whte Gaussan ose (AWG) varance. Our objectve s to maxmze the tota transmsson rate of SU s, expressng mathematcay as: ss p C = P f og 1 + σ L + =1 Subjected to: P r (5) max (6) J I (d, P ) I t a,, (7) P 0,, (8) P P T (9) P G 10 G = ndvdua peak power constrant. C = tota transmsson capacty of SU. = number of OFDM subcarrers. P r = the probabty. ow the Probabty nterference constrant n Eq. (7) can be wrtten as P r sp = K () = T s K () () P I t d +B / ( snπf T s πf Ts ) df d B / a,, (11) Snce sp s assumed to be Rayeght dstrbuted wth known sp parameterλ, the dstrbuton of corresponds to an exponenta dstrbuton wth the parameterλ. The constrant n Eq. (11) can be evauated n cosed form for the Rayegh fadng case as foows λ () I t. () P K 1 e α, (1) After some mathematca Manpuaton, Eq (1) can be wrtten as () I P K () t λ, (13) ( n 1 a ) It s cear that ths s a probem of convex optmzaton. (9) s a concave functon, I (P) s a convex functon. Accordng to convex optmzaton theory, when the tota transmsson capacty s maxmzed, the optma power of the th subcarrer s gven by:. P = θ + ω + 1 L =1 ξk σ + L =1 J ξ, v, θ, ω= Lagrange mutpers. Water-Fng Loadng Agorthm ss, 14 In water-fng agorthm, whch s optma power aocaton agorthm n conventona OFDM system, we use the tota power aocaton by unform oadng as the power constrant. the aocated power n the th subcarrer because of the th subcarrer because of the th nterference constrant s wrtten as P () = P K (),, (15) where P can be cacuated by assumng strct equaty n the th nterference constrant n Eq. (9). Usng Eq. (15), n ths equaty constrant We can wrte P = () () I P K = t λ ( n 1 a ), (16) () I t λ 1 n (1 a ) Usng Eq. (15) and (17), we can cacuate P () as I t (17) P = K λ n 1, a ow we need to cacuate power vaues P (L+1) due to the tota power constrant. In order to meet the tota power constrant, we use the standard water-fng agorthm to dstrbute tota power PT among CR subcarrers. Accordng to the waterfng agorthm wth a tota power constrant PT, the power vaues can be wrtten as P (L+1) = max 0, 1 α σ + =1 ss, (19) Where the agrange constant α can be cacuated from the foowng equton max 0, 1 α σ + L L () =1 J J () ss = P T (0) (WF) The power vaue for th subcarrer, denoted by P are obtaned usng the standard water-fng agorthm as mentoned n Eq (18) and (19) consderng the tota power constrant equa to the tota power aocated by unform oadng agorthm. The power vaues w satsfy the tota power constrant gven n Eq. (1) however t s checked that f the power vaues satsfy the nterference constrants specfed n Eq (1). If a partcuar nterference constrant s (WF) not satsfed,the power vaue n each subcarrer P s reduced such that the a nterference constrants are satsfed. Aso, f none of these nterference constrants s met strcty, 47

4 Internatona Journa of Engneerng Research and Technoogy. ISS Voume 10, umber 1 (017) Internatona Research Pubcaton House (WF) the power vaue P s ncreased unt one of these nterference constrants s met strcty. SIMULATIO RESULTS System Specaton. Performance of optma power aocaton agorthm are compared wth performance of cassca power oadng agorthms.e. water-fng power aocaton schemes that are used for conventona OFDM-based cogntve rado system parameters and network confguratons used for experments s descrbed as. Performance Parameters The four mportant performance parameters for evauaton are as: a) Power budget s the aocaton of avaabe power among the avaabe user. b) Interference: Ths term s used for nterference between Prmary User and Secondary User. c) Probabty: t s transmsson rate of CR user C s maxmzed whe the probabty that nterference to PU s s kept beow threshod d) Indvdua Peak Power: t s defned as the power constrant to CR to protect prmary user by restrctng the secondary user reman constant at average vaue of.6x10^6 (bps) because as we ncrease power budget for CR user, the nterference constrant become domnant and transmsson rate of CR user does not ncreases. In ths regon the CR system operates n an nterference mted scenaro. Power Budget (mwatt) Maxmum Transmtted Data Rate (bps) (x 10^6) Water fng Optma Tabe : Power Budget Vs transmtted data rate System parameter specfcaton. Sr no. Parameters Vaues 1 o of Sub-channes 5 o's Band wdth 1 Mhz 3 Power n db -0dBm 4 ose n db -110 db Tabe 1: System Parameters Fgure : Maxmum Transmtted data rate vs. power budge (PT) for CR user for dfferent power aocaton agorthm Transmsson rate vs. Power budget. The Fgure s a graph of maxmum transmsson rate for the CR user versus the tota power budget for varous agorthms. The vaue of I () t has been fxed to the ndvdua peak power constrant. From ths fgure and tabe, t s observed that the optma agorthm s abe to acheve hgher transmsson rate for a gven power budget than the waterfng agorthm. For optma power aocaton agorthm as ncreased n the power budget form 0.1 mw to 0.5 mw the maxmum transmtted data rate reach up to x10^6 (bps) and reman constant at average vaue of 7.7x10^6 (bps) and for water-fng power aocaton agorthm maxmum transmtted data rate reach up to x10^6 (bps) and Fgure 3: Bar-graph for Maxmum Transmtted data rate vs. power budge for dfferent power aocaton agorthm (PT) for CR user 473

5 Internatona Journa of Engneerng Research and Technoogy. ISS Voume 10, umber 1 (017) Internatona Research Pubcaton House Transmsson data rate versus nterference. Fgure 4: Transmsson data rate for the CR user versus Interference threshod for second PU The fgure 4 s graph of maxmum transmsson rate for CR user versus the nterference threshod for second PU user band. It s observed from fgure 4 and tabe 3 that the proposed optma power aocaton agorthm acheves hgher transmsson rate as compare to waterfng power aocaton agorthm. The proposed optma agorthm acheves hgher transmsson rate x10^6 bps than that of other agorthms and water-fng agorthm acheves hgher transmsson rate upto x10^6 bps. The transmsson rate versus nterference threshod curve saturates after a certan vaue of (I () t ). The reason s that athough (I () t ) s reaxed by ncreasng ts vaue, other constrants ((I (1) t ), (I (3) t ) and P T ) becomes domnant. The optma power aocaton gves better resuts because t mts the nterference by mtng power to the subcarrers near to prmary user s and aocate more power to subcarrers whch are far from prmary users thus mtng nterference. Interference (mwatt) Maxmum Transmtted Data Rate (bps) (x 10^6) Water fng Optma Tabe 3: Transmsson rate for the CR user vs nterference threshod for dfferent power aocaton agorthm. Fgure 5: Bar-graph for Maxmum Transmtted data rate vs. nterference threshod Transmsson rate versus probabty. The Fgure 6 s a graph of transmsson rate for the CR user versus probabty a. It s observe that the proposed optma power aocaton agorthm performs best over other waterfng power aocaton agorthm. It s observed from fgure 6and tabe 4 that for probabty of 0.75 the avaabe transmsson rate s x10^8 bps for optma power aocaton agorthm and t s x10^7 bps for water-fng agorthm for certan sgna power and as we ncrease the probabty to 0.95 the avaabe transmsson rate decrease to x10^6 for optma and x10^6 for water fng. Ths happens because as we go on ncreasng sgna power of secondary user to ncrease probabty the nterference toward prmary user aso ncrease so the achevabe transmsson rate of CR user decreases for a gven power budget and nterference threshods. Fgure 6: Transmsson rate for the CR user versus probabty wth nstantaneous nference ntroduced to PU 474

6 Internatona Journa of Engneerng Research and Technoogy. ISS Voume 10, umber 1 (017) Internatona Research Pubcaton House Maxmum Transmtted Data Probabty Rate (bps) Water - Optma fng caused by secondary user s, ess power s assgned to the subcarrers whch are near to prmary user s band and more power s assgned to the subcarrers whch are far from the prmary user band. Tabe 4: Transmsson rate for the CR user vs. probabty wth nterference ntroduce to the PU. Transmsson vs. ndvdua peak power Constrants. The Fgure 8 and tabe 5 s a pot of the achevabe maxmum transmsson rate for the CR user versus ndvdua power constrants for varous agorthms. It s observed that the proposed optma agorthm s abe to acheve hgher transmsson rate for a gven power budget than the waterfng power aocaton agorthm agorthm. It s observed from fgure 8 and tabe 5that as we ncrease the ndvdua peak power constrant from 0.1x10^-3 to 0.5x10^-3 for CR user, the transmsson rate of CR user for optma agorthm ncreases from.8714x10^5 to x10^5 and for water-fng t ncreases from x10^5 to x10^5 t s Because the nterference to the prmary user s under a threshod and the power constrant to CR protect prmary user by restrctng the secondary user hence the transmsson rate of CR user does ncrease as the ndvdua peak power constrant ncreases. Ths s expected as ndvdua subcarrer can be aocated more power. Fgure 7: Bar-graph of Transmsson rate for the CR user versus probabty wth nstantaneous nference ntroduced to PU COCLUSIOS The smuaton procedure s carred out for the mpementaton of the proposed system dscussed n prevousy. The smuaton anayss s many dvded nto two dfferent power aocaton agorthm namey water-fng and optma (adaptve) power oadng agorthm. the smuaton s done between four dfferent parameters name as power budget, nterference threshod,probabty to nterference ntroduced to PU and Indvdua peak constrants to each carrer Vs. Transmsson data rate of CR user(n bps) Transmsson rate vs. tota power budget. It s concuded that as the tota power budget s ncreased then the maxmum transmsson data rate for CR user s found to be ncreased. Further t s observed that the proposed optma power aocaton agorthm performance better than water-fng and abe to acheve more transmsson rate about 75%. Ths s because the power dstrbuted n Optma agorthm s ke a adder profe.e. to reduce the nterference Fgure 8: Transmsson rate for the CR user vs. ndvdua peak power Constrants Peak Power Maxmum Transmtted Data Rate (bps) (mw) Water fng Optma Tabe 5: Transmsson rate for the CR user vs. ndvdua peak power Constrants 475

7 Internatona Journa of Engneerng Research and Technoogy. ISS Voume 10, umber 1 (017) Internatona Research Pubcaton House REFERECES 1. S. Haykn, Cogntve rado: Bran Empowered Wreess Communcatons, IEEE J. Seet. Areas Communcatons, vo. 3, pp , Feb G. Bansa, P. Kagneed, and V. K. Bhargava, Jont Sensng and Power Loadng Agorthms for OFDM- Based Cogntve Rado Systems, n Proc. IEEE Wreess Communcatons and etworkng Conf., pp. 1 5, Apr Fgure 6.9: Bar-graph for Maxmum Transmtted data rate vs. ndvdua peak Constrants Transmsson data rate versus nterference. From smuaton resut t s concuded that transmsson rate for CR user ncrease for a fxed specfed nterference Threshod. It s observed that optma power aocaton agorthm performs better than water-fng agorthms and abe to acheve more transmsson rate about 80%. Ths s because optma power aocaton agorthm reduces the nterference rato between the prmary user s and secondary user s by mtng power rato to the nearest cogntve user and ncreasng power rato as the dstance between prmary and secondary s more. Transmsson rate versus probabty. It s concuded that transmsson rate goes on decreasng as probabty to the nterference goes on ncreasng. Further t s observed that at partcuar probabty the optma power aocaton agorthm performs better than water-fng power aocaton agorthm and gves more transmsson rate about 70%. Because unke other optma power aocaton agorthm, optma power aocaton reduce the nterference probabty dependng on dstance between prmary user s and secondary user s. 3. Z. Hasan, G. Bansa, E. Hossan, and V. K. Bhargava, Energy-effcent power aocaton n OFDM-based cogntve rado systems: A rsk-return mode, IEEE Trans. Wreess Communcatons, vo. 8, pp , Dec G. Bansa, Z. Hasan, J. Hossan, and V. K. Bhargava, Subcarrer and power adaptaton for mutuser OFDMbased cogntve rado systems, atona Conference on Communcatons (CC), pp. 1 5, Jan S. C. Yan, P. Y. Ren, and F. S. Lv, Power Aocaton Agorthms for OFDM Based Cogntve Rado Systems, IEEE Internatona Conference on Wreess Communcatons etworkng and Mobe Computng (WCOM), pp. 1 4, G. Bansa, J. Hossan, and V. K. Bhargava, Optma and Suboptma Power Aocaton Schemes for OFDM-based Cogntve Rado Systems, IEEE Trans. Communcatons, vo. 7, pp , ov Q. L. Q, A. Mnturn, and Y. Q. Yang, An effcent water-fng agorthm for power aocaton n OFDMbased cogntve rado systems, IEEE Internatona Conference on Systems and Informatcs (ICSAI), pp , 01. Transmsson vs. ndvdua peak power Constrants. It s concuded that transmsson rate s ncreased as ncreased n ndvdua peak power constrants, wth whch nstantaneous nference ntroduced to PU band remans beow nterference threshod. Further t s observed that proposed optma power aocaton agorthm s abe to perform better water-fng power aocaton agorthms and gves more transmsson rate about 60% because unke other optma power aocaton agorthm assgn dfferent nterference threshod to dfferent secondary subcarrer depend upon step sze.e. dstance between prmary user s and secondary user s. From a above concuson from Smuaton resuts t s observed that our proposed optma power aocaton agorthms can acheve hgher transmsson rate for CR user compared to the water-fng power aocaton agorthms 476