Effectiveness and Accuracy of Wireless Positioning Systems

Transcription

1 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban Effectveness an Accurac of Wreless ostonng Sstems SEBASTIAN FUICU *, ARIUS ARCU *, BOGAN STRATULAT *, ANANIA GIRBAN ** * Computer Scence an Engneerng epartment ** athematcs epartment oltehnca Unverst of Tmsoara Tmsoara, B. V. arvan, No. ROANIA sebastan.fucu@cs.upt.ro, mmarcu@cs.upt.ro, bogan.stratulat@cs.upt.ro, anana.grban@gmal.com Abstract: - Localzaton or postonng s an mportant aspect of moble applcatons n orer to acheve contetaware applcatons. The man goal for localzaton process s estmatng the poston of a moble evce n ts envronment base on a set of sensors wth known postons. oern moble evces prouce nowaas, an almost all smartphones an As, have one or more wreless communcaton nterfaces n orer to communcate wth other evces. WLAN nfrastructure s a wel accepte an mplemente communcaton stanar n man noor envronment, therefore man bulngs are alrea equppe wth IEEE 8. WLAN nfrastructure access ponts. Wreless aapters the moern moble evces are equppe wth can montor rao power strength of the emttng sources nearb. Base on the receve rao sgnal strength an estmaton of stance between the evce an power source can be compute. In ths paper we want to escrbe an conclue our work regarng the possblt to esgn a smple an energ effcent postonng soluton to be mplemente on moble evces wth lmte resources. Ke-Wors: - wreless postonng sstems, rao sgnal strength, energ effcenc, trlateraton Introucton One of the most sgnfcant elements of contetawareness n ubqutous envronments s moble evce localzaton. To obtan the accurate locaton nformaton for noor envronments s a challengng problem; therefore there have been a number of attempts to esgn sstems for noor localzaton usng fferent wreless sensng technques. Sgnal strength (SS approaches are eas to mplement n the estng technologes n moble evces. We want to nvestgate the postonng methos base on the trlateraton scheme wth a smpl an clearl mathematcal moel. Localzaton or postonng s an mportant aspect of moble applcatons n orer to acheve contetaware applcatons. The man goal for localzaton process s estmatng the poston of a moble evce n ts envronment base on a set of sensors wth known postons. oern moble evces prouce nowaas, an almost all smartphones an As, have one or more wreless communcaton nterfaces, lke bluetooth or WF, n orer to communcate wth nfrastructure computng sstems or other moble evces. WLAN nfrastructure s a wel accepte an mplemente communcaton stanar n man noor envronment, therefore man bulngs are alrea equppe wth IEEE 8. WLAN nfrastructure access ponts []. Wreless aapters the moern moble evces are equppe wth can montor rao power strength of the emttng sources nearb. Base on the receve rao sgnal strength an estmaton of stance between the evce an power source can be compute. If a moble evce can montor at least three WLAN power sgnal sources wth a-pror known postons, an estmaton of the moble evce poston can be acheve. Ths approach has the avantage that no costl harware nstallatons are necessar nse the bulng, nor atonal sensors to be attache wth the moble evces []. an moble applcatons woul beneft from beng able to use WLAN for communcaton as well as postonng []. etermnng the phscal locaton of moble actve noor noes, can be one of the man ssues of a new class of applcatons an poston-epenent servces n ubqutous sstems. For eample onl localzaton applcatons for ubqutous sstems wll have a growth of value (8 tmes n 5 ears from bllon US n 5 up to 8.5 bllon US n []. Hence we can sa that we shall face wth a pronounce evelopment of the number of applcatons that run on such evces an use fferent postonng technques. ISSN: Issue 9, Volume 8, September 9

2 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban The goal of ths paper s to evaluate the possblt to mplement a smple wreless postonng sstems usng onl trlateraton. The man requrement n evelopng such a software soluton s low computatonal power resources usage an the secon one s to have mnmum mpact on power consumpton. Ths paper s structures as follows: n net secton the relate works on postonng sstems are presente. The propose mathematcal moel varants are escrbe n secton. The framework archtecture for wreless postonng s escrbe brefl n secton 4. The epermental test cases an ther results are presente n secton 5. Fnall we conclue the paper n secton 6. Relate Work In the lterature there are man approaches for esgnng methos for postonng sstems, whch are fferent n terms of stance measurement technques an mathematcal moels. There are a few fferent localzaton methos that can be use for the postonng proceure. These methos are ve nto three maor categores base on the envronment n whch the nformaton s sprea: noor, outoor or me. ostonng sstems consst of algorthms an methos to estmate the poston of an unknown target, an are classfe base on [4]: - the sgnal tpes: o nfrare, o ultrasoun, o ultra-weban, an o rao frequenc - the sgnal metrcs o global locaton sstems - outoor methos o cellular locaton sstems - me methos o noor locaton sstems. The most popular outoor localzaton sstem s known as GS (global postonng sstem. The GS recever calculates ts current poston (longtue, lattue, elevaton usng a trlateraton technque. The stance s compute base on the tme ela between the transmsson an recepton of the encrpte rao sgnal ssue b the satelltes. In comparson wth these sstems, the noor sstem poses atonal challenges. In our work we aress the noor postonng problem. fferent technques est for estmatng the poston of a moble evce n a wreless network []: - cell base - wth ths localzaton technque the poston of a evce s smpl locate aroun the poston of the access pont where the evce s connecte. - sgnal propagaton tme base - the poston s eterme usng the tme of arrval of the receve sgnal from several access ponts (s to the target evce. - sgnal angle of arrval - the poston of a evce coul be estmate base on the measure angle of arrval of receve sgnal from surrounng s. - recever sgnal strength base - the poston of a target evce s estmate usng the receve sgnal strength from surrounng s. Cell base postonng metho s ver smple an consers that a moble evce s locate near the known poston of the ths evce s connecte to. espte ts smplct ths metho s not largel use because of the lack of accurac. The propagaton tme can be rectl translate nto stance, base on the known sgnal propagaton spee [5]. Base on sgnal propagaton tme from surrounng s to the target evce, ts poston can be estmate. These technques offer ncrease accurac but the requre precse clock snchronzaton an more epensve nfrastructure. There are two mportant methos n ths categor: ToA (Tme of Arrval [6] an ToA (Tme fference of Arrval [7]. The most mportant parameter for accurate noor postonng sstems s the tme of arrval TOA of the rect Lne of Sght path [8]. Therefore an accurate estmaton of TOA from receve communcaton sgnals s requre. Angle measurements can be use for noor postonng usng AoA (Angle of Arrval of the receve sgnal or oa (rrecton of Arrval [9, ]. For ths technque atonal harware s also neee n orer to measure the angle of ncence of the receve sgnal. The AOA requres antenna arras at each noe whch ncrease the complet of the estng sstem, as well as, performng worse n multpath envronment [7]. The fourth solutons class for noor postonng sstem s base on receve rao sgnal strength of the surrounng s. Ths postonng metho uses the estng WLAN nfrastructure an oes not nee an etra harware attache to the target evce. The network car n the moble evce contnuousl measures the RSSI (Rao Sgnal Strength Incator of the s n ts neghbor. Ths nformaton s avalable ue to beacon broacast multple tmes per secon b ever []. Base ISSN: Issue 9, Volume 8, September 9

3 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban on RSSI the moble evce can estmate the stance from ts poston to the t measure the RSSI. Usng at least three s wth well known postons, a moble evce can estmate ts poston base on measure RSSI from these s. RSS base WS (Wreless ostonng Sstems were stue a lot n the last ears. There are at least three locaton algorthms n ths class of WS: - fngerprntng (RSS patterns [,, ] - smulaton (lnes of constant RSS [, ] - trlateraton (Euclan stance etracte from RSS [4] The fngerprntng metho s the most common use for WS. An estmate of the target evce poston s obtane from the RSSI measurements an usng a rao sgnal propagaton moel nse the bulng []. The propagaton moel can be obtane usng a pror RSS measurements n fferent locatons n the bulng. These measurements compose a RSS map, calle fngerprnt, are store n a atabase an can further be use to estmate the poston of a target evce. Ths metho s ve n two phases []: ntal calbraton phase where the RSS map s acheve an the postonng phase, the measure RSS values from surrounng s are compare to the ones store n the atabase n orer to estmate the current poston of the evce. The savantages of ths approach are the atabase generaton an mantenance requrements []. urng the WS lfetme, ever change n the nfrastructure requres the bul of the RSS map. The secon class of WS methos uses smulaton algorthms to bul the propagaton moels of the rao sgnal nse the bulng. The smulaton starts from the bulng map an postons of s on the map. urng the smulaton for ever the lnes of constant RSS, calle sobars, are compute. After the smulaton the map of RSS sobars for ever s obtane. Ths map can be further use b a moble evce to estmate ts poston nse the bulng. The rawbacks of ths metho are the complet of the propagaton moel smulator an the computaton resources neee to run the smulaton algorthms. A network-base localzaton metho s presente n [5], whch uses the rao propagaton sgnal strength that covers a plane, where three snffers are place n orer to lsten to a sngle sgnal strength emtte b the moble an to automatcall generate an estmate sgnal strength map. Also, the algorthm s able to establsh the moble s poston b browsng ths table. Unlke the prevousl escrbe metho a clentbase localzaton soluton s presente n [6], usng rao-frequenc (RF as well, n whch the man element s recorng an processng the sgnals receve from several base statons that are place b overlappng coverage n the plane. The trangulaton s acheve b usng two methos: the frst one requres measurng the sgnals an creatng a sgnal strength map SS- an searchng the best sgnal strength measurements; the secon metho requres the usage of a smple propagaton moel n orer to estmate the SS-. The authors of [] evelope an teste a software framework calle pos for noor postonng base on WLAN fngerprntng the can b use to bul locaton base applcatons. It conssts of an effcent, freel confgurable framework, whch s sutable for multple applcaton archtectures. The RSS measurements are performe on the moble evce, the computaton an vsualzaton of poston can be run on an nfrastructure server or on the evce tself. Base on ther tests the obtane an accurac of better than meters for an area of 5 m covere b 7 s. The authors of [] evelope a metho to entf ntersectons of lnes of constant RSSI values, calle solnes, of several s wthn nterpolate rao maps base on trangulaton. Base on ther tests an average evaton of meters was obtane for m ehbton room equpe wth 4 s. The am of the work presente n paper [7] s to stu the possbltes that the WLAN's offer to noor localzaton an to mplement an applcaton that wll help to localze a evce equppe wth a WLAN's car supportng the IEEE 8.lb protocol. The applcaton evelope b the authors calle Wlb uses a atabase wth sgnal strength values store an a mathematcal moel n orer to estmate the poston of a user wth a wreless evce nto a lbrar of ther Unverst. The thr RSS metho s base on trlateraton [4]. Ths approach s smple, eas to mplement an nees fewer resources to compute the poston. Ths metho s approprate for a moble evce where the energ an computatonal resources are lmte, even the accurac of the metho s n the best case comparable wth the other methos. The savantage of all RSS methos s the ranom evaton from mean receve sgnal strength cause b shaowng an small scale ISSN: Issue 9, Volume 8, September 9

4 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban channel effect [7]. We conser base on our tests, that the accuraces obtane b other postonng applcatons are the best that can be acheve an n the real lfe usage scenaros the postonng accurac wll ecrease sgnfcantl ue to: changes n the envronment, people movng aroun n the measure envronment, moble evce spee an orentaton, s rao sgnal transmtter s fluctuatons, etc. ower consumpton an energ effcenc are an mportant aspect for moble applcatons n general an for local postonng applcaton n specal. Therefore one of the topcs we nvestgate n ths paper s the energ effcenc of WS applcatons. The power consumpton problem of moble sstems s n general a ver comple one an remane of nterest for qute a long tme [8, 9]. In ths paper efne a software eecuton framework for moble sstems n orer to characterze the power consumpton profle of fferent tpes of moble applcatons [8]. Our goal s to nvestgate how trlateraton can be use n noor postonng, to estmate ts accurac an ts neee resources for the moble evces. We target postonng n large rooms wth fferent obects nse the room an people movng nse the room. We also want to be realstc an prove also the problems such a postonng sstem mplementaton wll face urng ts eplotaton. Ths paper covers our work n last three ears partall presente n some conference papers [4,,, an ]. athematcal moel The trlateraton assumes the estence of at least three access ponts (s wth known postons. The stance between each access pont ( an the moble evce ( have to be etermne b computatons. For ever stance a crcle can be rawn. In the eal case, these three crcles must ntersect n a sngle pont whch s actuall the poston of our moble evce (. The metho chosen for obtanng the stance between an S s base on the sgnal strength (SS measurement []. The propagaton of rao waves s nfluence b three factors: free space loss, attenuaton b the obects on the propagaton path, an the sgnal s scatterng. In the absence of obstacles, the moel for propagaton s free space loss whch can be epresse for the eal sotropc antenna as n (: t ( 4π ( 4π f r λ c ( t sgnal power at the emtter r sgnal power at the recever propagaton stance between emtter an recever λ - carrer wavelength f carrer frequenc c spee of lght The eal stuaton, when there s no obstacle, we conser the b-mensonal case when all s an the moble evce are n the same plan. If no perturbatons nterfere along the measurng, at moment t, receves a sgnal wth power from the transmtter. The stance between an coul be calculate wth the formula (: t c 4π f r ( In the real worl, measurng SS s mae wth some errors, because the SS cannot be etermne wth ver hgh precson, even n the case of obstacle absence. If we take nto account the attenuaton nuce b some obstacles the problem coul be more complcate. Hence the ntersecton of the three crcles s not a sngle pont. For three access ponts the mamum number of ponts s s, because ever two crcles can generate two ponts of ntersecton. The eal stuaton, when there s no obstacle, we conser a b-mensonal case. If no perturbatons nterfere along the measurng, at the t moment, receve, an from the transmtters,, respectvel. So, t s at: t r c 4π f from the transmtter, t r c 4π f from the transmtter, t r c 4π f from the transmtter, ( (4 (5 ISSN: Issue 9, Volume 8, September 9

5 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban The recever s at the ntersecton of the crcles havng the centre n, of raus, because n ever pont of a crcle the same power s ntercepte. In eal case, the ntersecton of the crcles s one pont onl, whch s (Fg.. : ( + ( ( + ( ( + ( (6 ρ ρ (, ρ(, (, ρ(, (8 ρ (, beng the enst of the envronment n the pont (, In the three-mensonal case, n the eal varant when there are no perturbatons n nterceptng the sgnal, an the sgnal oes not meet an obstacle, the recever s the ntersecton of the spheres wth the centers n, an of raant,, an respectvel, gven b the (, (4, (5 relatons. Fg. oston of n eal case If we accept that the sgnal reaches the recever nfluence b some perturbatons, then the ntersecton of the crcles s no longer a crcle but a oman as n the Fg.. Fg. The case Fg. The real case oman can be estmate to one pont conserng ths pont ts centre of mass an t can be consere that the appromate poston of. : ( + ( ( + ( ( + ( The coornates of the centre of mass are: (7 In the real case, where perturbatons appear at the ntercepton of the sgnal, the ntersecton of the four spheres wll be a oman whch also s reuce to ts centre of mass, as t s shown n: ( + ( + ( z z : ( + ( + ( z z ( + ( + ( z z (9 z ρ ρ ρ ρ zρ ρ (,, z (,, z (,, z (,, z (,, z (,, z z z z z z z ( ISSN: Issue 9, Volume 8, September 9

6 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban (,, z ρ beng the enst of the envronment n the pont (,, z Fg. 4 oman for the case Ths centre of mass appromates the poston of. In orer to resolve ths problem of trlateraton we propose two geometrc methos. etho I In the eal case, the soluton s generate b the followng sstem of equatons []: ( + ( : ( + ( ( + ( ( Where, (, are the moble noe s (, coornates, are the s coornates an s the stance between S an. But, as we mentone earler, n a real stuaton, the ntersecton of the three crcles generates not a sngle pont, but a set of ponts. The problem s how to choose between those man noes. We propose a pure geometrc metho, whch s ver smple to mplement an computatonal effcent. The frst conton s for the s not to be on the same as. Ths conton s gven b the net equaton: ( Our soluton requres makng pars wth ever two s, havng as a result the net relaton: ( ( r r < ( + ( < ( r + r (, Where (, an are the coornates r r of the two s. We conser an the rauses of the two crcles. The raus s the same wth the stance between an S, stance calculate base on the sgnal strength sense b e moble evce. A. If both nequaltes n relaton ( are true, ths means that we have two ponts of ntersecton. The coornates of these ponts are gven b the followng sstem of equatons (4: ( + ( ( + ( (, (, r r (4 (5 Fg. 5 Intersecton ponts of the three crcles In case each par of generates two ponts of ntersecton, t results a total number of s ponts, as n the Fg. 5. B. If the rght nequalt of relaton ( s false, then we have: ( + ( ( r + r (6 Ths means that the two crcles are not ntersecte. In ths case we conser a sngle pont ISSN: Issue 9, Volume 8, September 9

7 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban stuate on the as efne b the centers of the two crcles. The coornates of ths pont are: r k r + k + k + k, + k (7 C. If the left nequalt of relaton ( s not satsfe, then we have: ( + ( < ( r r (8 Ths means that one crcle s contane b the other one. As n the prevous case the pont s stuate on the as efne b the centers of the two crcles. ( r k r + k + k Q, + k + k r + Q ( + ( ( + ( ( r + r ( ( q + ( + ( r ( q + ( + ( (9 The cases B an C generate a sngle pont, but n case A we have two ponts. In orer to choose between them, we appl the net computaton. For each of them we make the sum of stances between the noe an ever other noe obtane through one of the three cases: A, B or C. S S N ( k + ( k k N ( k + ( k k ( If S < S then we choose noe, else we. Thus, we generate a set of choose the noe (, ponts:. The selecte ponts n ths wa form a polgon an the poston of S wll be the pont, appromate wth the center of gravt of ths polgon. The relatons are: N N ( etho II The secon metho mposes the same general conton [4], namel n the zone where moble evce s actng to be at least three access ponts. If there are more then three access ponts a new conton arses: the access ponts nee to create a conve polgon lke n fgure 6. In ths fgure we eemplf a case wth 4 access ponts. Fg. 6 Access ponts surrounng the moble evce If no perturbatons nterfere along the measurng, at the t moment, receve power from the transmtters respectvel an stances between an coul be calculate wth the formula (. The algorthm of ths metho requres groupng ever two receve RSSI measures therefore we ma wrte: ( + ( ( + ( ( ISSN: Issue 9, Volume 8, September 9

8 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban Solvng ths sstem, we wll obtan two solutons: (, an (, (4 From these two solutons we shall choose the pont that s at the same se of the lne etermne b an +, wth another (for eample -. We can choose an other (not or + because the are the vertces of a conve polgon. Ths pont wll verf the relaton: > The selecte ponts n ths wa form a polgon. The poston of pont wll be appromate b the center of gravt of ths polgon. ( n n ( + n 4 Software framework archtecture The general archtecture of the power-aware locaton framework presente n fgure 7 has a laere structure an each laer s ve n several nepenent moules [4]. The lower laer of the framework applcaton wll use the operatng sstem s rvers of fferent phscal components took nto account n the optmzng process of energ consumpton (the processor, the batter, wreless chpset, man-boar chpset, the memor an locaton algorthm nput measures (WLAN rver. The kernel of the eecuton framework takes the avalable measures through the montorng rvers, calculates the energ consumpton of the current runnng applcatons an proves a consumpton state to the locaton estmaton core. The locaton algorthm presente before s mplemente n the locaton core moule. Ths moule s confgure base on applcaton s requrements an power consumpton state an ts output s prove through a publc nterface to the applcaton level. Applcatons usng the locaton framework nterface we call locaton-aware applcatons. n Fg. 7 ostonng framework archtecture The problem n reang the ata from the network car on the movng obect, was, that t has to be rea nepenent of the moel of the wreless network car. ore etale presentaton of software archtecture an RSSI reang from the software s presente n [4]. 5 Test be an test cases eperments We nvestgate n our work how trlateraton can be use n large rooms postonng an whch are the best results some can obtan wth ths metho. Our tests use a large laborator m 5.5m (Fg. 8 an a number of access ponts or wreless routers from fferent provers or tpes. We set up a number of test cases n orer to evaluate our postonng soluton, to nvestgate an problem the soluton wll face n real lfe envronment an to etermne whch the man factors that nuce these problems are. For ever test case we run a number of eperments we base our conclusons. 5. Test case In the frst test case we nvestgate how the tpe an evce nfluences the accurac of postonng algorthm. We use 6 tpes of s from two provers: Lnk an Lnkss (WL-G7, I-64, WL-, W55-AG, W54-G, an WRT54-GS an we use also fferent evces on the same tpe. We run the testng algorthm a large number of tmes for fferent parameters: test uraton, samplng tmes, number of samples average for one poston computaton. The same test was run then a number of tmes for the same evce an the for fferent evces. We place a number of access ponts an we measure the RSSI n fferent well known postons (Fg. 8. urng these test case ISSN: Issue 9, Volume 8, September 9

9 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban eperments we selecte a number of goo evces whch are well sute to be use n the postonng sstem an also we calbrate the tests parameters. For ever poston a mnutes capture was save. urng tests were mae two reangs of RSSI per secon an we average the last secons measurements ( samples for ever poston estmaton. In the test publshe b other authors the use smaller amount of tmes for ther tests (e.g. 5 secons n []. We conser that the valaton tests shoul last a larger amount of tme because there are man propagaton phenomena we shoul account an for smaller test tmes we can obtan onl a partcular result. - Lnkss W55AG - mw, 6m, mn - Lnkss W55AG - 5mW, 6m, mn - Lnkss W54G - mw, 6m, mn - Lnk L64-8mW, 6m, mn - Lnk WL-G7 6mW, 6m, mn On goo evce we selecte for the sstem s presente n Fg. 9 (RSSI varatons an Fg. (stance estmatons. RSSI [Bm] W55AG Fg. 9 RSSI for a goo stance [m] Fg. 8 Test scene an frst test case testng_algorthm( test_tme, sample_tme, average_no begn select_evce_params(transm_power start_tmer(tmer whle( tmer < test_tme begn for each ap n surrounngs begn rea_rss(rss average_rss(rss, average_no compute_stance(average_rss recor_ata(stance en compute_poston(poston recor_ata(poston sleep(sample_tme en stop_tmer(tmer en Even f between the access ponts an the moble noe are no obstacles, the values of RSSI are ver oscllatng. These varatons are cause b a couple of elements of whch we can menton the evce tpe an reflectons of the sgnal n an noor envronment. The oscllatons of sample RSSI epen on the evce therefore we selecte a number of goo evces. To see f the fluctuatons are generate b the evces we conucte a set of measurements usng eferent tpe of evces: stance [m] W55AG Fg. stance estmaton for two goo s L64 We can observe that small changes n RSSI epresse n Bm nuce large varatons n stance estmaton. In Fg. we plot the hstograms for the stance estmatons of L64 an W55AG s shown n Fg.. The stance between an target evce was 6 meters an we obtane an average error of meters for W55AG an stanar evaton of.75; an average error meters for L64 wth a stanar evaton of.5. ISSN: Issue 9, Volume 8, September 9

10 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban Frequenc [,5 [5,5.5 [5.5,6 [6,6.5 [6.5,7 [7,7.5 [7.5,8 stance [m] Fg. stance estmatons hstogram L64 W55AG stance [m] W55AG L64 One ba eample for a evce whch cannot be emploe b a postonng sstem s presente n Fg. wth the average error of 6 meters. stance [m] Fg. stance estmaton for a nos W54G 5. Test case To eamne the mpact of the stance between the access pont an the moble evce over the RSSI we mae the followng measurements n the testbe escrbe n Fg.. stance [m] stance [m] W55AG L64 W55AG L64 Fg. 4 Estmate stance from one to two s We observe an ncrease n measurement stanar evaton when the moble evce moves awa from the ,, - - Fg. Test scene for secon test case - -4 wlg7- ol. (wlg7- Test setup:. Lnkss W55AG - mw, 9 m, an lnk L64-8mW, m, mnutes. Lnkss W55AG - mw, 6 m, an lnk L64-8mW, 6 m, mnutes. Lnkss W55AG - mw, m, an lnk L64-8mW, 9 m, mnutes -5-6 Fg. 5 Relaton between RSSI an stance for Lnkss W55AG We place a number of access ponts an we measure the SS n fferent well known postons. For smplct we place a moble evce between ISSN: Issue 9, Volume 8, September 9

11 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban two s, from meter to meter. For ever poston a mnutes capture was save. The relaton between the receve sgnal strength an the stance between emtter an recever for some evces are presente n Fg Test case Usng three s the poston of a moble evce nse the room can be estmate Fg. 6. The test uses: - Lnkss W55AG - mw - lnk L64-8mW - lnk WL-G7-8mW [m] Fg. 6 oston estmaton from s RSSI RSSI RSSI an the presence of people has nfluence on the RSSI. Ths coul be a problem because the WS are ntene to be use when the users are nse the room. Therefore a WS shoul conser the a characterstcs of the test when people are present an move n the room. 5.5 Test case 5 Runnng the framework applcaton on the moble evce has nfluence on the tme the batter wll scharge. Fg. 6 shows the batter schargng capact when no applcatons are eecute on a laptop an when the framework applcaton s eecute. A mnmum consumpton fference of 5% was obtane n our tests. An nterestng aspect s that relate to the le state power consumpton wth WLAN chpset swtche off, WLAN chpset set on an WLAN communcaton show an ncrease n power consumpton as presente n Fg. 8. We can state that wreless communcaton has three states of power consumpton observe at applcaton level: WLAN-OFF power state; WLAN-ILE power state; WLAN-CO power state Test case 4 The tests presente b other authors for ther wreless postonng sstems were taken wthout people movng aroun n the bulng. We conser that when there are people nse the room the nfluence the accurac of the postonng process. We let the sstem runnng for one a an we obtane the graphc n Fg [mw] WLAN off WLAN on, no comm WLAN on, WS WLAN on, comm Fg. 8 WS power consumpton AY NIGHT Fg. 7 a an nght postonng accurac -a-4 -a-4 fferences coul be observe n receve sgnal power between the two evces, between a an nght an between sgnal level an sgnal sperson. urng the a, the room was use b the stuents 6 Conclusons We presente n ths paper an applcaton framework for wreless postonng sstem. The epermental results wth the propose tests conuct us to the concluson that our soluton s eas to mplement, power effcent an sutable for the concept of contet-awareness n ubqutous envronments. In eal contons we obtane comparable postonng accuraces wth other solutons of aroun - meters. In the eal case we carefull selecte the proper s, we assume that no users are nse the room to ntrouce perturbatons an the target evce was statc. ISSN: Issue 9, Volume 8, September 9

12 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban In case we conser real lfe usage scenaros, there are a lot of aspects whch pose serous problems to an wreless postonng sstem: - evce changes: n case one n the sstem s change (poston, evce equpment, evce tpe, orentaton, etc., the postonng sstem shoul be recalbrate. - Envronment changes: ang, removng or changng obects postons nse the room ma nfluence the accurac of the postonng sstem an n some cases nee recalbraton. - oble evce changes: when target evce s changng ts poston, the accurac of the postonng ecrease ue to: evce orentaton, evce spee, etc. - User enst changes: when the test scene s use b other people, the nfluence the propagaton moel an further the accurac of the poston estmaton. ACKNOWLEGEENT Ths work was supporte b the grant 68/9..9 from the Romanan nstr of Eucaton CNCSIS. References: [] U. Grosmann, C. Rohrg, S. Hakoban, T. omn, an. alhaus, WLAN noor postonng base on Euclan stance an nterpolaton, roceengs of the 8th Wreless Technologes Kongress, pp. 96-5, ortmun, German, 6. [] G. Retscher, E. oser,. Vreevel an. Heberlng, erformance an Accurac Test of the WLAN Inoor ostonng Sstem pos, roc. of the r Workshop on ostonng, navgaton an Communcaton (WNC 6, pp. 7-4, 6. [] lan cgrath, GS hansets to ntegrate GS, EETmes, ec. 7. [4] H. Lm, L. Kung, J.C. Hou, H. Luo, Zero- Confguraton, Robust Inoor Localzaton: Theor an Epermentaton, INFOCO 6. 5th IEEE Internatonal Conference on Computer Communcatons. roceengs, pp. -, 6. [5]. Broton, J. Lfton, J. araso, Localzng a Sensor Network va Collaboratve rocessng of Global Stmul, roceengs of the Secon European Workshop on Wreless Sensor Networks, pp. -, 5. [6] A.A. Al an A.S. Omar, Tme of Arrval Estmaton for WLAN Inoor ostonng Sstems usng atr encl Super Resoluton Algorthm, roceengs of the n Workshop on ostonng, Navgaton an Communcaton, WNC 5, pp. -, 5. [7] C. Zhang,.J. Kuhn,. Branon, A.E. Fath,. ahfouz, Hgh Resoluton Uwb Inoor Localzaton Sstem Operatng At 8 an 4 GHz Utlzng Tme fference Of Arrval Approach, Natonal Rao Scence eetng, 6. [8] Xnrong L, Kaveh ahlavan, Super- Resoluton TOA Estmaton Wth verst for Inoor Geolocaton, IEEE Transactons on Wreless Communcatons vol. No.., pp. 4-4, 4. [9] J.C. Chen, K. Yao, R.E. Huson, Source localzaton an beamformng, IEEE Sgnal rocessng agazne, Vol. 9, No.,, pp. -9. [] N.B. rantha, A. Chakrabort, H. Balakrshnan, The Crcket Locaton-Support Sstem, roceengs of the 6th AC Internatonal Conference on oble Computng an Networkng (AC OBICO,, pp.-4. [] B. L, Y. Wang, H.K. Lee, A.G. empster, C. Rzos, A new metho for elng a atabase of locaton fngerprnts n WLAN, IEE roceengs Communcatons, Vol. 5, No. 5, 5, pp [] B.L, J. Salter, A.G. empster. C. Rzos, Inoor postonng technques base on Wreless LAN, Frst IEEE Int. Conf. on Wreless Broaban & Ultra Weban Communcatons, 6. [] U. Grossmann,. Schauch, S. Hakoban, The accurac of algorthms for WLAN noor postonng an the stanarzaton of sgnal recepton for fferent moble evces, Internatonal Journal of Computng, vol. 6, no., pp. -9, 7. [4] S. Fucu,. arcu, B. Stratulat, I. stratulat, A. Grban, An open, low power framework for WLAN noor postonng sstem, th WSEAS CSCC ultconference, Int. Conf. on Computers, -5 Jul 9. [5] J. Ymng, S. Baz, S. ane,. Agrawal, ARIANE: A namc Inoor Sgnal ap Constructon an Localzaton Sstem, roceengs of the 4 th Internatonal Conference on oble Sstems, Applcatons an Servces, 6, pp ISSN: Issue 9, Volume 8, September 9

13 WSEAS TRANSACTIONS on COUTERS Sebastan Fucu, arus arcu, Bogan Stratulat, Anana Grban [6]. Bahl, V.N. amanabhan, RAAR: An In- Bulng RF-Base User Locaton an Trackng Sstem, INFOCO. Nneteenth Annual Jont Conference of the IEEE Computer an Communcatons Socetes. roceengs. IEEE, Vol.,, pp [7] A. Karakos, I. Kours, Wreless postonng n a unverst lbrar, WSEAS Transactons on Crcuts an Sstems. Vol. 4, no. 7, pp Jul 5. [8]. arcu,. Tuor, S. Fucu, S. Copl-Crsan, F. atcu,. cea, ower Effcenc Stu of ult-threang Applcatons for ult-core oble Sstems, WSEAS Transactons on Computers, Issue, Volume 7, ecember 8. [9]. arcu,. Vlautu, H. olovan, croprocessor Thermal Characterzaton usng Thermal Benchmark Software, WSEAS TRANSACTIONS on COUTERS, Issue, Volume 5, ISSN 9-75, pp. 68-6, November 6. []. arcu, S. Fucu, A. Grban, an. opa, Epermental Test Cases for Wreless ostonng Sstem, IEEE Internatonal Conference on "Computer as a Tool", EUROCON 7, Warsaw, olan, Sep. 7, pp []. arcu, S. Fucu, A. Grban, Local Wreless ostonng Sstem, 4th Internatonal Smposum on Apple Computatonal Intellgence an Informatcs, 7. SACI '7, Tmsoara, Romana, a. 7, pp []. arcu, S. Fucu, A. Grban, A mathematcal moel for Wreless LAN noor postonng sstem, The th Internatonal Conference of TENSOR Socet on fferental Geometr & Its Applcatons an athematcal Founatons of Informaton Scences & Its Applcatons, TENSOR 8. [] J.C. Chen, K. Yao, R.E. Huson, Source localzaton an beamformng, IEEE Sgnal rocessng agazne, Vol. 9, No.,, pp. -9 ISSN: Issue 9, Volume 8, September 9