IJIRST Interntionl Journl for Innovtive Reserch in Science & Technology olume Issue 3 August 5 ISSN (online): 349-6 A Novel High Resolution Algorithm for obile Detection & Cpcity Implementtion llikrjun Hiremth Dr. Srik Tle. Tech Student Associte Processor Deprtment of Digitl Electronics & Communiction System Deprtment of Digitl Electronics & Communiction System isvesvry Technologicl University, CPGSB, IAT, uddenhlli, Chikkbllpur- 56 isvesvry Technologicl University, CPGSB, IAT, uddenhlli, Chikkbllpur- 56 Abstrct In recent yers Smrt Antenn is fst rising res of communiction systems. Smrt ntenns re especilly ggressive of their pplictions in relted fields such s Fourth Genertion (4G) technologies nd these re best suitble for multi crrier modultions. In our project contins min utilize of smrt ntenns in obile communiction tht upgrdes the cpbility of the mobile devices s well s enhnced the cellulr system for exmple fst nd high bit rte, multi-use interference nd spce division multiplexing technique, enlrge the rnge, deduction of errors becuse of multi-pth fding long with nother enormous benefit is the higher security. In multiuser IO system, the optiml bit rte is very much dynmic plus chnges from one pcket to the some other neighboring pcket. This ignores trditionl bit rte dpttion lgorithm mechnism, which depends on recent history to clculte the excellent bit rte for the upcoming pckets. In this project Turbo Rte dpttion scheme for ulti User ultiple Input ultiple Output ANs is recommended. Turbo Rte Adpttion method djusts the bit rte bsed on per pcket bsis if every client lerns vribles. It s Signl to Noise Rtio when it trnsmits lone to the AP (ccess point), nd the direction besides which its signl is received t the AP. A Turbo Rte client then clrifies its pckets with these two vribles nd to llow different clients to choose the best bit rte s well s trnsmits simultneously to the Access Point. Keywords: U IO, ISO, Rte Adption, SDA, Detection Algorithm I. INTRODUCTION In recent yers, considerble exmine hs been occurring in rry ntenns which re sufficiently brillint to recognize required signl s well s interference signl. Presently, the role of Smrt Antenns in mobile communictions to enhnce the bility of corresponding chnnels hs reignited investigte nd growth in this extremely exhilrting re. Such dvnce is smrt ntenn nd the kind of numerous ccesses is bsed on SDA technique. A. Spce Division ultiple Access Sptil division multiple ccess method (SDA) is differ from frequency division multiple ccess method (FDA) s well s time division multiple ccess (TDA) method. SDA scheme in uses methods by which signls re differentited t the bse sttion derived from their origin plce in spce. SDA is combintion with FDA, TDA or CDA (Code division multiple ccess) keeping in mind the end gol to offer the recent with the extr cpcity to investigte the sptil property of the signls. The SDA method is tking into ccount the ide tht rriving signl from different sources reches unlike ntenns in n rry t distinctive times becuse of their sptil distributions. This dely pplies to differing from single or numerous clients in one region to other regions. This system permits efficient trnsmission strting with one cell then onto the next without disturbing concurrent trnsmission. Fig. demonstrtes the essentil SDA system. For Ex, GS (globl system for mobile communiction) / GPRS (Generl pcket rdio service) permits just single client t once to brodcst or cquire in frequency bnd to the bse sttion. In the event tht GS/GPRS is utilized together with Spce Division ultiple Access, multiple synchronous trnsmissions cn occurs t sme frequency bnd in like mnner increses the cpcity of the formtion. Since the Fig., there re two bems of the similr frequency f t sme time t, yet there re in distinctive directions. Therefore the cpcity of the system is improved by fctor of two. With Sptil Division ultiple Access system severl mobiles cn shre the sme frequency. All rights reserved by www.ijirst.org 99
Fig. : SDA scheme Fig. : SDA Concepts The rriving signls t the bse sttion cn be divided from the bse sttion receiver s long s their ngulr prtition is greter thn trnsmit or receiving bem widths. Therefore, the Sptil Dimension cn be exploited s Hybrid ultiple Access technique supplementing FDA, TDA nd CDA. This methodology is typiclly indicted s SDA nd it empowers different clients inside of the sme rdio cell to be suited on the similr frequency nd sme time opening s represented in Fig. II. PROBE FORUATION In most recent couple of yers significnt development in the dvncement of wireless ccessing technologies like internet services s well s cellulr systems. As result of them it is predicted tht lter on tremendous dvnce in trffic will be experienced for individul nd mobile communiction. This is becuse of two certinties first one is n increment in the number of clients nd second one is the strting of higher bit rte informtion services. This turns mjor issue to resolve for the network provider. Their current some negtive fctors in the rdition-environment which dds to the limited cpcity nd nother negtive fctor re co-chnnel interference which is brought on by n increment in the number of clients. The some different impirments dd to the degrdtion in system performnce nd cpcity re multipth fding, increse the dely occurred by signl reflection from buildings nd ountins nd clients trveling in vehicles. Remote ocl Are Networks re hving two ptterns. Initilly number of ntenns on n ccess period is expnding consistently with ordinry Access Points tody, hving few ntenns. Second is on tht point is prolifertion of little Wi-Fi devices for exmple sensors, gme consoles nd smrt phones, which ber less Form Fctor nd strict power limittions therefore normlly utilize n one nd only ntenn. These ptterns cuse multiple ntenn-aps to expend huge division of its time communicting with only one ntenn client. Accordingly, wireless-locl re networks won t crry the gretest number of concurrent trnsmissions empowered by their infr-structure. To perceive this reserchers hve recommended to utilize of U IO ANs, where multiple number of single-ntenn clients simultneously communicte with multiple ntenn ccess points. Reserchers showed tht decoding of such simultneous communictions is chievble both in the uplink s well s downlink. So fr in ny cse, investigte on multiuser IO Wireless ANs hs not tended to the bit rte selection issue or problem nd essentilly cknowledged tht the All rights reserved by www.ijirst.org
trnsmitters recognize the best bit rte. This sttement is pplicble long the downlink where there is however single trnsmitter, the Access point, nd subsequently the difficulty cn be decresed to stndrd 8.n rte dpttion technique. The uplink stte is quite different. It hs multiple numbers of concur renting trnsmitters tht ggregtely need to choose the optiml bit rtes to their ccess point. III. SYSTE ARCHITECTURE The Fig.3 shows the system rchitecture of the project. In our project we derived mthemticlly these bellowed fctors. Using these reltions, we find out direction detection of the Third genertion (3G) method nd Fourth genertion (4G) method. ooking t the design of 3G method nd 4G methods, there is slight bit different between the equtions bsed on direction detection methods. stly, compre between the ISO (3 G) nd IO (4 G) methods nd lstly conclude tht IO method is better thn ISO. A. Arry nifold ector Amount of Time Dely n electromgnetic wve undergoes when it hits the ntenn elements is computed using Arry nifold ector B. Arry Correltion trix Computtion The Arry Correltion mtrix gives the informtion bout the correltion between the Noise nd the signl elements C. Direction Computtion for 3G The Direction Computtion for 3G is computed utilizing the following eqution Fig. 3: System Architecture of Rte Adpttion Project P ISO S ( ) Here, S θ presents the Steering ector relted with the direction θ, R denotes Arry Correltion trix. And denotes rry elements In ISO Direction estimtion, n rrngement of steering vectors S θ is relted with different directions θ is frequently mentioned to s the Arry nifold. Prcticlly, it is clculted t the session of rry clibrtion. Since utilizing both rry mnifold nd n estimtion of the Arry Correltion trix, power of spectrum P B (θ) is to be computed. Directions of the rditing sources tken s peks in power spectrum. H RS All rights reserved by www.ijirst.org
D. Eigen lue Determintion for Arry Correltion mtrix The Eigen lues for the Arry Correltion trix is obtined by finding out the roots of the chrcteristic eqution E. Determining Noise Eigen lues The Eigen lues re sorted in scending order nd the first - eigen vlue corresponds to the noise. is number of ntenn elements nd is number of sources F. Determine Noise Eigen ectors The Noise eigenvectors re found out by solving the following eqution for multiple noise eigenvlues. G. Direction Detection using 4G Rte Adpttion The noise subspce Eigenvectors re orthogonl to the rry of Steering ectors t the Angles of Arrivl,,..... Due to H H this orthogonl sitution, cn demonstrte tht the Eucliden distnce d ) ) for ll ngle of rrivl,,...... Inserting this distnce expression in the denomintor occur shrp peks t the ngles of rrivl. The IO pseudo spectrum is given by P IO H H ) ) Where, ( ) is steering vector of n ngle nd denotes x mtrix consisting of noise eigen vectors. H. Comprison of 3G nd 4G Rte Adpttion Direction Detection ) ess ntenn elements. Closely Spced Sources For this Number of Antenn Elements re chosen s 8 nd Sources re present t 5 degree nd one more source is present t 9 degree. The sources re sid to be closely spced if the ngle of seprtion between the sources is t lest 4 degree ) ore Antenn Elements Closely Spced Sources For this Number of Antenn Elements re chosen s nd Sources re present t 5 degree nd one more source is present t 9 degree. The sources re sid to be closely spced if the ngle of seprtion between the sources is t lest 4 degree 3) ess Antenn elements Widely Spced Sources For this Number of Antenn Elements re chosen s 8 nd Sources re present t 3 degree nd one more source is present t 45 degree. The sources re sid to be widely spced if the ngle of seprtion between the sources is t lest 5 degree 4) ore Antenn elements & Widely Spced Sources For this Number of Antenn Elements re chosen s nd Sources re present t 3 degree nd one more source is present t 45 degree. The sources re sid to be widely spced if the ngle of seprtion between the sources is t lest 5 degree 5) esure Cpcity The Cpcity is computed using the following eqution log( sin ( ) SNRorig ) Crtio( ) log( SNR ) direction SNR org orig of source SNR of the originl signl I. ISO ETHOD OF DETECTION In this technique [] rectngulr window of uniform weighting is pplied to the time series dt to be nlyzed. The 3G Cognitive method is lso defined s Ordinry Bemforming ethod (OB). This method clcultes the men power PB (θ) by steering the rry in θ direction. The power spectrum in 3G ethod Cognitive method is presented s H S RS PB ( ) () Where, S θ presents the steering vector connected with the direction θ, nd R denotes the rry correltion mtrix nd is the number of rry elements In DETECTION evlution, rrngement of steering vectors S θ is connected with vriety of direction θ is often referred to s the rry mnifold. Prcticlly it my be mesured t the time of rry clibrtion. From the rry mnifold nd n estimte of the rry correltion mtrix P B (θ) is computed. Peks in P B (θ) re then tken s the directions of the rditing sources. All rights reserved by www.ijirst.org
All rights reserved by www.ijirst.org 3 themticl Anlysis of 3G method cognitive method A. The steering vector A for n ntenn rry comprising of elements is clculted by using eq (). ) The signl mplitude vector s is column vector of order x is given by ) s Where,,,., re mplitude of sources () The Hermitin Trnspose s H of signl vector S given by 3) H s (3) The Signl Correltion trix (P) given by 4) ] [ H ss E P (4) Substitute the vlue of s nd s H from equtions cn obtin eqution p (5) The signl vector S is x digonl mtrix comprises only digonl elements of mtrix P is given by 5) S (6) The Signl subspce is x mtrix is given by 6) H s SA A R (7) The Noise subspce is x mtrix is given by 7) I R n (8) Where, is vrince of noise nd I denotes identity mtrix The rry correltion mtrix is given by 8) R S R n R (9) Where, S R is signl subspce nd n R is noise subspce Compute the steering vector by using eqution () 9) The vlue of the power spectrum is found by using eqution (4) ) This power spectrum is computed by keeping rry correltion mtrix constnt nd vrying θ in the rnge π/<θ+.< π/.. RATE ADAPTION DETECTION ETHOD Rte Adpt Detection ethod is n cronym which stnds for Hnn Quenon. 4G Rte Adption ethod is used to give unbised estimtion of the quntity of signls nd its Angles of Arrivls furthermore strength of the wveforms. 4G RATE ADAPTATION ETHOD gurntees the presumptions tht the interference in every chnnel is uncorrelted by creting the Noise Correltion trix digonl. The incident signls re correlted by mking non digonl Signl Correltion trix. By the by, under high signl correltion the trditionl 4G RATE ADAPTATION ethod lgorithm breks down nd other methods re to be implemented to correct this wekness. We must relize tht the number of signls is then the quntity of signl Eigenvlues nd Eigenvectors is nd the quntity of noise Eigenvlues nd Eigenvectors re - (rry elements denoted s ). Becuse 4G RATE ADAPTATION ETHOD chieves the Noise Eigenvector Subspce, it is occsionlly defined s Subspce ethod. The Eigenvlues nd Eigenvectors of the correltion mtrix R re initite. Eigenvectors re relted with the signls nd Eigenvectors connected with the noise re liented. The Eigenvectors re ssocite with the littlest Eigenvlues re used to find the power spectrum. For uncorrelted signls the smllest-eigenvlues re equl to the rince of the noise. The ( ) dimensionl Subspce spnned by the Noise Eigenvectors is defined by
e e e... e where e 3 i is i th eigenvlue. () The Noise Subspce Eigenvectors re,,..... Orthogonl to the rry of Steering ectors t the Angles of H H Arrivl. Since, this orthogonl condition, it cn tell tht the Eucliden distnce d ) ) for ech Angle of Arrivl,,..... Substituting this distnce formul in the denomintor side which cretes shrp peks t the Angles of Arrivl. The 4G RATE ADAPTATION ETHOD Pseudo Spectrum is defined by P () USIC H H ) ) Where ( ) is the Steering ector of n ngle ngle nd E N is the x - mtrix comprising of Noise Eigenvectors. A. Simultion ethodology of 4G RATE ADAPTATION ETHOD Algorithm ) The Steering ector A for n ntenn rry consisting elements is clculted by using below eqution A e e id sin id ( e e id sin id sin id ( )sin )sin id ( )sin ( ) The next steps re sme in previous chpters re tken, it mence to derive the equtions up to 8 th step (eqution 9 ) 9) After deriving Arry correltion mtrix then to find the Eigen vlues of n Arry Correltion mtrix by performing the Eigen vlues decomposition method. The Eigen vlues of x Arry Correltion mtrix (R ) re estblish by performing the Chrcteristic eq is known by R I The result to the eq gives Eigenvlues,..., The Eigenvector for prticulr Eigenvlue R n n,. is estblish by solve the eq followed by Here n denotes x mtrix consisting number of unknown vribles. Expnd the eq in mtrix nottion we get R, R, R, R, R, R, R R R,,, ultiply the mtrices with set of simultneous eqs s chrcterized in re gotten R R... R R R,,, R R,,,... R... R e,,, e While there re unknown vribles, hence here contin simultneous eqs cn be solved to get vlues forms Eigenvector mtrix. ) The noise Eigen vlues re selected from ll Eigen vlues ) The power is clculted by using P () HQ H H ) ) E N is the noise Eigen vectors,..., () (),.. Those All rights reserved by www.ijirst.org 4
I. RESUTS A. ISO ethod Fig. 4: shows X xis s ngle (deg) nd y xis s power (db) for Detection of obile User using ISO ETHOD (UTIUSER) In the bove Fig.4 shows the opertion of multi user ISO method in which X xis is rnging of ntenn, where ngle of ntenn is vries -9 degree to +9 degrees. The bove exmple we tke 3 mobile users.the bove non uniform rry in which consists ntenn elements re the number of inputs sy in exmple ntenn elements re 8. In multiuser ISO method the number of E wves is equls to number of mobile users. Hence we tke three different mplitudes of three different E wve strikes t ntenn t bse sttion sy first is t mplitude is, second is nd third is t 5. Similrly, we enter three different directions from where E wve strikes t the ntenn rry sy some exmples degree, 45 degree nd 6 degrees. The bove figures we only identify the mobile user direction t degrees but we cn t distinguish t 45 degrees nd t 6 degrees. Therefore incse of multiple mobile user in ISO method is not cpble of detecting the mobile users significntly. B. IO ethod Fig. 5:(A) shows X xis s Number of Eigen lues nd y xis s Eigen ector gnitude for Number of Eigen vlues v/s Eigen ector gnitude. Fig. 5:(B) shows X xis s ngle (deg) nd y xis s power (db) for Detection the obile User using IO ETHOD Similrly, in multiuser IO we tke number of three mobile users nd number of eight ntenn elements. In multiuser IO method the number of E wves is equls to number of mobile users. Hence we tke three different mplitudes of three different E wve strikes t ntenn t bse sttion sy first is t mplitude is, second is nd third is t 5. Similrly, we enter three different directions from where E wve strikes t the ntenn rry sy some exmples degree, 45 degree nd 6 degrees. In multi user IO method we cn esily find out the direction of mobile user, s sw in figure. We enters the direction is t degree, 45 degree nd 6 degrees hence we seen in figure pek signl is t round t, 45 nd 6 degrees which is cler nd All rights reserved by www.ijirst.org 5
lso it is out of disturbnce s compred to ISO method. This indictes tht IO hs the cpbility to detect the mobile directions in best wy C. Comprison between ISO nd IO We need to clculte the cpcity of IO nd we need to see how the SNR is getting incresed nd we need to see how the cpcity of IO behving with respect to the ngle nd we need to see wht is the mount of error produced by ISO or IO lgorithm. So tht we cn execute by the turbo rte mechnism. Here Fig.6A indictes originl SNR is on X xis nd projected SNR is Y xis. If you see in the grph when originl SNR is increse s well s projected SNR is n increse. If the originl SNR is nd your projected SNR is +.4 tht it indictes. So in IO method smllest gin resolutions men.4 will lso mesured, but in ISO it is difficult. Fig.6B indictes the ngle between clients verses reduction rtio in SNR s see in the grph. The ngle between two client s increses the reduction in the SNR will lso increse. Fig.6C is one of the importnt grphs, ccording to IO the SNR of mobile device in cse IO is vry from 5 to 6 db in this rnge. We consider vrious SNR rnges such s 5 db,8 db, db, 4 db, 7 db, db, 3dB, 6 db. We kept the ngle between to 9 degrees, becuse we considering uniform liner rry nd IO presented t both trnsmitter nd the receiver. Since we consider uniform liner rry the vribility will vry from -9 degree to +9 degree. But in our grph shows us considering only positive ngles to 9 degrees nd for this fixed SNR is clculte w.r.t cpcity. If the ngle increses the cpcity lso increses which is seeing in grph. Fig.6D indictes SNR OF 5 db re treting s low SNR mobile, 7 db re treting s medium SNR mobile nd 6 db treting s high SNR mobile. For CDF (cumultive distributed function) is plotted in the rnge between to. It clssifies or normlizes the vlue of cpcity between to. When we divide the cpcity by the mximum vlue this becomes CDF grph. If you seen mobile with low SNR is lso cpcity increses, medium SNR is lso cpcity increses similrly high SNR is lso cpcity is increses hs the ngle between two clients increses. It indictes tht whether it hs low SNR, if the ngle between two mobile users is further then the IO doesn t bothered bout SNR, it cn extrct the signl. Therefore if IO cpcity is high then it cn send more pckets. But in ISO ll 3 mobile users hve sme pek, but only one mobile cn send the pcket not for ll mobiles. Rte dpter will tke both the ngle s well s rte dption nd it will send more pckets towrds the users nd no pckets towrds other users. Fig.6E shows plotting ctul direction verses estimted directions, in the grph some cses vlues re overlps nd in some cses vritions re present but estimted direction is high. FiG.6F shows plotting ctul direction verses estimted directions. In the IO method fewer vritions re present compred to ISO method hence in the grph it shows lmost overlpped in vlues. Fig. 6A: shows x xis s Originl SNR nd y xis s Projected SNR. Fig. 6B: shows x xis s ngle between clients [degree] nd y xis s SNR reduction [db]. Fig.6C: shows x xis s ngle between clients [degree] nd y xis s Cpcity Rtio. Fig.6D: shows x xis s ngle between clients [degree] nd y xis s CDF Distribution. Fig.6E: shows x xis s oction Index nd y xis s Estimted\Actul Direction using ISO. Fig.6F: shows x xis s oction Index nd y xis s Estimted\Actul Direction using IO. Fig.6G: shows x xis s oction Index nd y xis s Error in Estimtion [degree] using ISO. Fig.6H: shows x xis s oction Index nd y xis s Error in Estimtion [degree] using IO Fig.6G nd Fig.6H shows plotting grph loction detection verses estimted error of ISO nd IO methods respectively. The X xis shows loction index nd Y xis shows estimtion error. All rights reserved by www.ijirst.org 6
D. Comprison of detection of mobile users between ISO nd IO In order to compre IO nd ISO we need to consider four following functions. Fig. 7A: shows X xis s ngle (deg) nd y xis s power(db) for detecting the mobile user using ISO,IO lgorithms. Fig.7B: & b shows X xis s ngle(deg) nd y xis s power(db) for Detecting The obile User using ISO,IO lgorithms. Fig.7C: & b shows X xis s ngle(deg) nd y xis s power(db) for Detecting The obile User using ISO,IO lgorithms Fig.7D: & b shows X xis s ngle(deg) nd y xis s power(db) for Detecting The obile User using ISO,IO lgorithms ) ess ntenn elements. Widely Spced Sources For this lesser number of Antenn Elements is chosen s 8 nd widely spced nd enters the mplitude of the first E wve is nd direction is degree similrly mplitude of second E wve is tken s nd direction is 45 degree nd lstly third E wve mplitude is 3 nd its direction is t 65 degree. The sources re sid to be widely spced if the ngle of seprtion between the sources is t lest 5 degree. If we see in the figure ISO lgorithm is not cpble of detecting multiple mobiles t the sme time, where s IO system is cpble of detecting multiple mobiles t sme time. It hs produce unique distinguish peks t ech mobile user directions t, 45, 65 degrees. ) ore Antenn Elements Widely Spced Sources For this Number of Antenn Elements re chosen s nd Sources re present t 3 degree nd one more source is present t 45 degree. The sources re sid to be widely spced if the ngle of seprtion between the sources is t lest 5 degree Here both IO nd ISO cn detect the mobiles very efficiently. For the cse number of ntenn elements re high, both re performing in very good wy, but IO produces nrrower pek compres to ISO. The mount of disturbnce in terms of power is high in cse of ISO s compred to IO 3) ess Antenn elements Closely Spced Sources For this number of Antenn Elements re chosen s 8 nd number of mobile users re 3. Sources re present t degree nd 4 nd one more is 7 degree. The sources re sid to be closely spced if the ngle of seprtion between the sources is t lest 4 degree 4) ore Antenn elements & closely Spced Sources For this, number of Antenn Elements is chosen s nd number of mobile users re 3 nd Sources re present t degree nd one more source is present t 4 degree nd nother one is 8 degree. The sources re sid to be closely spced if the ngle of seprtion between the sources is t lest 4 degree. The bove figures shows both the IO nd ISO lgorithms re sme pttern. As compre to IO, in ISO system the mount of power is lrger. Hence in IO system the mount of power wsted is less. The conclusion regrding in four cses is the IO method performs the best, but ISO is little bit good. II. CONCUSION Smrt Antenn is technology tht gives wireless-networks with fewer ltency nd higher dt bit rtes. To rise the cpcity of the chnnels, novel technologies like multiple inputs nd multiple outputs (IO) cn be pplied, longside the smrt ntenns to give greter cpcity of the chnnels. In this pper introduces Turbo Rte dption method, distributed rte dption protocol for multi user IO ANs. The Rte dpttion lgorithm ws divided into ISO detection, IO Noise Subspce Detection nd the lgorithms were compred with respect to Bis nd Resolution. IO obile detection hs lwys better s compred All rights reserved by www.ijirst.org 7
to ISO obile detection. The crrying out demonstrtes tht enbling ultiuser-io with trditionl rte dpttion lessens the throughput in mny occsions, while enbling Turbo Rte in ultiuser-io rises the network throughput respectively. REFERCES [] Cisco Aironet 36 Series, http://www.cisco.com/en/us/ products/ps983/index.html, 3. [] K. Tn, H. iu, J. Fng, W. Wng, J. Zhng,. Chen, nd G.. oelker, SA: Enbling Prcticl Sptil ultiple Access in Wireless AN, Proc. AC obicom, 9. [3] E. Aryfr, N. Annd, T. Slonidis, nd E.W. Knightly, Design nd Experimentl Evlution of ulti-user Bemforming in Wireless ANs, Proc. AC obicom,. [4] H. Yu,. Zhong, A. Sbhrwl, nd D. Ko, Bemforming on obile Devices: A First Study, Proc. AC obicom,. [5] D. Tse nd P. ishwnth, Fundmentls of Wireless Communictions.Cmbridge Univ., 5. [6] D. Hlperin, W. Hu, A. Sheth, nd D. Wetherll, Predictble 8. Pcket Delivery from Wireless Chnnel esurements, Proc. AC Specil Interest Group on Dt Comm.,. [7] H. Rhul, F. Edlt, D. Ktbi, nd C. Sodini, Frequency-Awre Rte Adpttion nd AC Protocols, Proc. AC obicom, 9. [8] J. Cmp nd E. Knightly, odultion Rte Adpttion in Urbn nd ehiculr Environments: Cross-yer Implementtion nd Experimentl Evlution, Proc. AC obicom, 8. [9] K.C.-J. in, N. Kushmn, nd D. Ktbi, ZipTx: HrnessingPrtil Pckets in 8. Networks, Proc. AC obicom, 8. [] J. Bicket, Bit-Rte Selection in Wireless Networks, PhD disserttion, sschusetts Inst. of Technology, 5. []. utukuru, H. Blkrishnn, nd K. Jmieson, Cross-yer Wireless Bit Rte Adpttion, Proc. AC Specil Interest Group on Dt Comm., 9. [] S. Gollkot, S.D. Perli, nd D. Ktbi, Interference Alignment nd Cncelltion, Proc. AC Specil Interest Group on Dt Comm., 9. [3] S. Gollkot, F. Adib, D. Ktbi, nd S. Seshn, Clering the RF Smog: king 8.n Robust to Cross-technology Interference, Proc. AC Specil Interest Group on Dt Comm.,. 46 IEEE TRANSACTIONS ON OBIE COPUTING, O. 3, NO., JANUARY 4 [4] K.C.-J. in, S. Gollkot, nd D. Ktbi, Rndom Access Heterogeneous IO Networks, Proc. AC Specil Interest Group on Dt Comm.,. [5] G. Strng, iner Algebr nd Its Applictions, second ed. Acdemic, 98. [6] Ettus Inc., Universl Softwre Rdio Peripherl, http://ettus.com, 3. [7] G. Ren, H. Zhng, nd Y. Chng, SNR Estimtion Algorithm Bsed on the Premble for OFD Systems in Frequency Selective Chnnels, IEEE Trns. Comm., vol. 57, no. 8, pp. 3-34, Aug.9. [8]. Guillud, D. Slock, nd R. Knopp, A Prcticl ethod for Wireless Chnnel Reciprocity Exploittion through Reltive Clibrtion, Proc. Eighth Int l Symp. Signl Processing nd Its Applictions, 5. [9] System Description nd Operting Principles for High Throughput Enhncements to 8., Technicl Report IEEE 8.-4/ 87r, 4. [] R. Hndel nd. Huber, Integrted Brodbnd Networks; AnIntroduction to AT-Bsed Networks. Addison-Wesley ongmn, 99 All rights reserved by www.ijirst.org 8