3-D BSS Geometric Indicator for WLAN Planning

Transcription

1 3-D BSS Geometric Indicator for WLAN Panning Aexandre Gondran, Oumaya Baaa, Aexandre Caminada and Haim Mabed University of Technoogy Befort-Montbéiard, SET Lab, Befort, France E-mai: {aexandre.gondran, oumaya.baaa, aexandre.caminada, Abstract: Optima panning of wireess networs is one of the most fundamenta probems in WLAN design. This paper discusses the organisation of BSS in WLAN extended service set panning for arge-scae depoyment. We tae advantage of a new geometric criterion to evauate the coverage quaity and appy them to WLAN panning. The obtained resuts show that using such criteria faciitates WLAN panning and improves the WLAN QoS. 1. INTRODUCTION WLAN networ panning consists in seecting a ocation for each transmitter and setting the parameters of a sites in order to suppy users a wireess access to their oca networ. The objective is to respect financia requirements and to guaranty a given Quaity of Service (QoS). There are two reevant stages in WLAN networ panning. Firsty we have to seect a set of sites from a ist of candidates that have been identified as potentia sites; for each site, we must choose the antenna pattern as we as its azimuth that indicates the main propagation direction and the emitted power of the antenna. The 4-upet (site, antenna pattern, azimuth, emitted power) is caed AP configuration. Seecting a set of AP configurations from a ist of candidate AP configurations is a ocation probem usuay caed ACP probem for Automatic Ce Panning in ceuar networ context. The second important stage is to aocate one of the avaiabe frequencies to each AP configuration in order to minimize interferences. The frequency set depends on the standard ( a, b or g) and aso on specific restriction on spectrum usage in each environment. This probem is caed AFP probem for Automatic Frequency Panning and becomes very famous in the context of ceuar GSM/GPRS/EDGE systems. Usuay these two probems are treated successivey in WLAN: the design process begins choosing antenna sites then aocating the avaiabe frequencies to the seected sites. The first studies on ACP probem were defined as a covering probem [1] [2] without in with AFP. Later, various constraints were added to the ACP probem in order to ease the AFP probem; the ACP probem became over constrained. A arge variety of constraints are described in the iterature. The most current constraint consists to add some ce-overapping to covering probem. For exampe prohibiting the seection of two cose sites [3] [4] or minimizing the overapping area between ces [5] [6] [7]. More sophisticated approach is to evauate the deviation between interfering transmitter [8] [9]. Another approach is to estimate the capacity of channe frequency reuse [10]. Ti in WLAN, some other constraints not connected to frequency aocation may aso be added to the ACP probem in order to improve the QoS ie the estimation of the throughput [9] [11], of the networ capacity [12] or the evauation of the impact of hidden nodes [7] Each constraint concerns a particuar aspect of QoS which is directy ined to interference management that is the number and ampitude of scrambing signas on carrier. This paper introduces a new constraint based on Basic Service Set (BSS) geometry to ease frequency aocation in WLAN panning. Considering a BSS composing a networ our innovative wor consists in defining a 3 dimension performance indicator of BSS shape in two contexts: used separatey for each BSS and used gobay for best server BSS inside an Extended Service Set. In the iterature, there are few wors which tae into account the ce (in ceuar networ) or BSS (in WLAN) shape as criteria and study its impact on the networ QoS. Reininger et a. defined a notion of ce connectivity [8]. Hao et a. used this concept during a pre-optimization phase of networ design [15]. Jedidi et a. introduced ce criteria measuring the distortion from the dis [13]. Mabed et a. [14] deepened those criteria and study their impact on interferences in GSM. In this paper we adapt and appy this category of criteria in WLAN context to study their impact on QoS. The environment is specific as the transmitter and the receiver are inside buiding and the signa propagation is in 3 dimensions to tae into account buiding foors. The paper is organized as foows. The second section introduces the geometric criteria and shows how we define them in WLAN panning. The third section expains and presents eary experimentation resuts. In the fourth section we appy geometric criteria to the whoe networ and show that it improves the WLAN design process. Finay, we summarize our main resuts and go further into current wor. 2.1 Definition of ce 2. GEOMETRIC CRITERIA In GSM or UMTS networs the area to be covered is defined as a 2-D grid of pixes. Then a ce reative to one antenna is a set of pixes associated to a given base station.

2 We define a ce C as a set of pixes where the signa received from the station exceeds a given quaity threshod q. C = { bi, j/ Fi, j> q, i, j} (1) Where b i,j represents the pixe of co-ordinates (i, j) and F i,j is the signa strength received at pixe b i,j from the antenna. 2.2 Reference geometric mode In a 2-D space, a pixes have 8 neighbours except pixes on space borders. In ceuar networ Jedidi et a. [13] evauate the convexity of a ce by counting for each pixe of the ce the number of pixe neighbours that beong to the same ce. For a given pixe b i,j, the function V computes this number of neighbours. Jedidi et a. defined then the geometrica quaity of a ce by the foowing criteria: Vb ( i, j) bi, j C GC ( ) = (2) 8 C The term in denominator is necessary to compare ces having different sizes. The function G vaue is between 0 and 1. The coser G is to 1, the coser the ce geometry is to a dis. Mabed et a. [14] brought some corrections to the origina geometric criteria. They ceary normaized the criteria in order to compare it with the idea ce shape which is a dis. So they repaced the denominator by a better evauation of one dis in a grid space. Knowing the ce size C they estimated the vaue of one dis by: 8 C 6 π C, taing into account that bordered pixes do not have 8 neighbours. So Mabed et a. define the function: Vb ( i, j) bi, j C G' ( C) = (3) 8 C 6 π C This criterion G is between 0 and 1. G gives a better measurement of ce distortion from the dis. This vaue equas to 1 if the ce shape is a dis and decreases as the ce geometry defers from it. This resut improves interference management so QoS in GSM networ [14]. Where b i,j, represents the pixe of co-ordinates (i, j, ) and F i,j, is the signa strength received at pixe b i,j,. Then the rea BSS (in 3-D) is defined as the union of a 2-D BSS, we get: = K C = C where K is the number of foors. = 1 It is necessary to adapt the geometric measurement to tae into account the 3-D space. We choose to eep the same definition of G for each foor as if they were independent. G (C ) is the vaue of the geometric indicator of the configuration in the foor. Vb ( i, j, ) bi, j, C G' ( C ) = (5) 8 C 6 π C We regroup foor-indicator inside a unique indicator defined by the foowing equation: = KC GWLAN ( C) = G' ( C ) (6) = 1 C Where each 2-D indicator is normaized on the basis of its percentage of coverage in 3-D. In the further section we wi evauate two networs with this performance indicator. 3. EARLY EXPERIMENTATIONS Our testbed is composed of one three foor buiding (figure 1). Each foor is a 45m square. We defined 252 candidate sites for AP instaation. For each site we can choose between 2 antenna patterns: one omnidirectiona antenna (without azimuth) and one directiona antenna with 8 possibe azimuths. The two antennas are abe to transmit the signa at 4 different powers. Then for each site we may choose between 36 configurations. So gobay we get 9072 candidate AP configurations. A WLAN consists in the instaation of some AP configurations among these 9072 candidates that is combinations. 2.3 Geometric indicator for WLAN To evauate and compare the AP configurations in WLAN context, it is necessary to adapt the formua to the specific indoor environment. The area to be covered whie depoying a WLAN is generay a buiding. Each foor is defined as a 2- D grid of pixes, so we have gobay a 3-D grid of pixes. In WLAN the same AP can cover severa foors at the same time then it is not correct to compare a WLAN BSS which is in three dimensions with 2-D ce in ceuar networ. Consequenty we defined for each foor a 2-D BSS C as a set of pixes at the same foor where the signa received from the AP exceeds a given quaity threshod q. C = { bi, j, / Fi, j, > q, i, j} (4) Figure 1 Topoogy buiding: candidate AP ocations The histogram depicted by figure 2 represents the distribution of those AP configurations according to G WLAN vaue. We see that G WLAN vaues foow a Gaussian distribution centred on 0.9 mean vaues. This geometric indicator denoted isoated geometric indicator concerns each AP configuration. We fixed the threshod vaue q at -80dBm. It is about 10dBm higher than the minimum power necessary to

3 estabish a connection given by constructors which is more reaistic. Figure 5 Topoogy buiding: AP ocation Figure 2 AP configuration distribution according to the isoated geometric indicator: G WLAN Figures 4 and 6 show representative vaues of this indicator on a 3-D buiding for 2 different AP ocation and compare them to the corresponding BSS shape. Figure 3 Topoogy buiding: AP ocation a. foor 0 b. foor 1 Figure 6 BSS of the AP configuration with G WLAN = 0.95 Figure 6 corresponds to another site configuration. Here we instaed our antenna on foor 0 (see spot on figure 5). Figure 6.a describes the coverage area of foor 0 and the coverage area of foor 1 is described by figure 6.b. Foor 2 is not covered at a. However we can ceary see that the BSS shapes are more compact than those of figure 4. We observe that the BSS of figure 6 has a greater G WLAN vaue than the BSS of figure 4. In our case study, the higher is G WLAN the more compact and convex is the BSS. This compacity is good for interference management between the BSS of the extended service set. 4. NETWORK GEOMETRY Figure 4 BSS of the AP configuration with G WLAN = 0.86 In site configuration described by figure 4 we instaed the antenna on foor 1 (see spot on figure 3). Figure 4.b represents the coverage area of the foor 1 corresponding to pixes receiving at east 80dBm signa. We see that the BSS shape is sighty dispersed. The foor 0 and foor 2 are aso covered by the same antenna but their respective areas are ess wide than that of foor 1. The BSS shape depicted by figure 4.c is aso sighty dispersed. That dispersion creates isoated pixes where the BSS overapping and neighbourhood are poory managed driving to interference probem. In WLAN, when a cient receives severa signas from different AP, he estabishes the communication with the AP from which he receives the highest signa. Such AP is caed the Best Server of the cient. For each AP we define his Best Server ce BS as the set of pixes where the signa received from the AP exceeds a given quaity threshod q and is the highest. Simiary to the definition of the simpe BSS C and C we defined for each foor a Best Server BSS BS as: / and m BS, { },, = bi, j, F > q F > F m A i j { } i, j, i, j, i, j, (7) Where Fi, j, is the signa power received on the pixe (i,j,) = K from the AP. In the same way, we define: BS = BS. = 1 Now, we evauate for each Best Server BSS the geometric indicator G WLAN (BS ). To evauate the geometric quaity of the entire WLAN networ, we proposed to compute the weighted average according to the size of each Best Server BSS i.e. the respective pixe number of each Best Server BSS.

4 BS GNetwor GWLAN ( BS ) BS = (8) To mae our tests we considered two networs each one composed of 12 AP distributed over the three-foor buiding (see figures 7 and 9). Figure 9 Topoogy buiding and AP ocation Figure 7 Topoogy buiding and AP ocation Figure 10 Networ 2 Best Server cartography (G Networ = 0.91) 5. CONCLUSION Figure 8 Networ 1 Best Server cartography (G Networ = 0.94) These networs resut from an optimization process which try to cover the buiding at a given data rate in each pixe. Figures 8 and 10 describe Best Server cartography for each networ. Each coour corresponds to a different Best Server BSS. Figure 8 describes the Best Server BSS of the first networ having a G Networ vaue of Figure 10 describes the Best Server BSS of the second networ having a G Networ vaue of Best Server BSS of the first networ are ess dispersed and so more convex than those of the second networ. We notice that the number of ce neighbours in figure 8 is smaer than that in figure 10. As a matter of fact, the Automatic Frequency Panning (AFP) probem wi be easier to sove for the first networ than for the second networ. As for ceuar systems, the G Networ indicator seems quite suitabe to evauate the AFP probem feasibiity in WLAN. Our next step is to show how this geometric performance impacts on interference performance. At the moment, a comparative study from expert on WLAN immediatey concudes on a better design invoving a better QoS management with networ 1. There is a cear identification of BSS distribution and borderines on the 3 foors. This is not the case on networ 2. WLAN panning is a crucia probem for arge-scae depoyment. In this artice we considered geometric criteria used in ceuar networs context and adapt them in WLAN context. Our resuts show that the higher is the function vaue the more convex is the BSS shape. Then we defined two geometric indicators for WLAN: an isoated one which characterizes the shape of a unique BSS and a Best Server one which characterizes the average vaue for the goba networ. In our case study, we notice that there is a reationship between both geometric indicators and BSS convexity. Using those criteria during WLAN panning seems very interesting to improve the frequency panning process. A networ having a good vaue of the G Networ has rea faciity for frequency assignment. Ongoing wor aim to vaidate those concusions on much more sampes and determine what is a good geometric vaue. More over we have to study the reationship between the isoated geometric indicator and the Best Server indicator in order to define how to use them during optimization phase of WLAN design process. REFERENCES [1] Fortune S.J., Gay D.M., Kernighan B.W., Landron O., Vaenzuea R.A., Wright M.H., WISE design of indoor wireess systems : practica computation and optimization, IEEE Computationa Science and Engineering, 2(1), pp , [2] Anderson H.R., McGeehan J.P., Optimizing microce base station ocations using simuated anneaing techniques, Vehicuar Technoogy Conference 2, pp , 1994.

5 [3] Lee Y., Kim K., Choi Y., Optimization of AP pacement and channe assignment in wireess LANs, IEEE Conference on Loca Computer Networs, [4] Rodrigues R.C., Mateus G.R., Loureiro A.A.F., On the design and capacity panning of a wireess oca area networ, IEEE / IFIP Networ Operation and Management Symposium, pp , [5] Sherai H.D., Pendyaa C.M., Rappaport T.S., Optima ocation of transmitters for micro-ceuar radio communication system design, IEEE Journa on Seected Areas in Communications 14(4), pp , [6] Mathar R., Niessen T., Optimum positioning of base stations for ceuar radio networs, Wireess Networs 6, pp , [7] Amadi E., Capone A., Cesana M., Maucei F., Optimizing WLAN Radio Coverage, IEEE Internationa Conference on Communications 2004, 1, pp , [8] P. Reininger, A. Caminada. Muticriteria design mode for ceuar networ. Annas of Operation Research, 107, 2001, pp [9] K. Jaffrès-Runser, J.-M. Gorce and S. Ubéda, "QoS constrained wireess LAN optimization within a mutiobjective framewor", in IEEE Wireess Communications Magazine, Specia Issue on 3G/4G/WLAN/ WMAN Panning and Optimization, Dec [10] Bahri A., Chamberand S., On the wireess oca area networ design probem with performance guarantees, Computer Networs 48, pp , [11] Ling X., Yeung K.L., Joint access point pacement and channe assignment for wireess LANs, IEEE Wireess Communication and Networing Conference, pp , [12] Promma C., Kabara J., Tipper D., Charnsripinyo C., Next generation wireess LAN system design, IEEE Proceedings of Micom 1, pp , [13] A. Jedidi, A. Caminada, G. Fine: 2-Objective Optimization of Ces Overap and Geometry with Evoutionary Agorithms, Lecture Notes in Computer Science, Voume 3005, 2004, p.p [14] H. Mabed, A. Caminada, Geometric criteria to improve the interference performances of ceuar networ, IEEE Vehicuar Technoogy Conference, Montrea. Sept [15] J. K. Hao, M. Vasquez. Heuristic approach for antenna positioning in ceuar networs. Journa of Heuristics, 2000, pp