Analysis of a Polling System Modeling QoS Differentiation in WLANs

Size: px
Start display at page:

Download "Analysis of a Polling System Modeling QoS Differentiation in WLANs"

Transcription

1 Analyss of a Pollng System Modelng QoS Dfferentaton n WLANs T.J.M. Coenen Unversty of Twente Department of Appled Mathematcs Enschede, The Netherlands t.j.m.coenen@utwente.nl J.L. van den Berg TNO Informaton and Communcaton Technology Delft, The Netherlands j.l.vandenberg@tno.nl Rchard J. Bouchere Unversty of Twente Department of Appled Mathematcs Enschede, The Netherlands r.j.bouchere@utwente.nl ABSTRACT Ths paper nvestgates a pollng system wth a random pollng scheme, a 1-lmted servce dscplne and determnstc servce requrement modelng WLANs wth QoS dfferentaton capablty. The system contans hgh and low prorty queues that are dstngushed va the probablty of beng served next. We propose a new teraton algorthm to approxmate the watng tme of customers n the hgh and low prorty queues. As shown by smulaton results, our approxmaton s accurate for lght to moderately loaded networks. Keywords: Pollng model, QoS dfferentaton, WLAN, IEEE e AMS subject classfcaton: 90B15, 90B18, 90B22, 68M20, 60K25 1. INTRODUCTION Wreless Local Area Networks (WLANs) have become wdely avalable for nternet access and there s currently a growng demand for the support of other applcatons, n partcular speech and vdeo. Specfc mechansms then need to be deployed n order to provde approprate QoS to the varous applcatons. A typcal approach to provde such QoS dfferentaton s for example by gvng a larger share of the avalable capacty to preferred users, or gvng prorty to preferred classes. Introducton of such mechansms requres nsght nto ther performance. Ths paper nvestgates the nfluence of prortzaton of the packet delay handlng at the Medum Access Control (MAC) layer n WLANs. In IEEE WLAN prortzaton appears n the support of dfferent QoS classes. These QoS classes are mplemented va dfferent settngs of MAC layer parameters, lke ther access tme, the maxmum and mnmum value for ther Correspondng author: Tom Coenen, Unversty of Twente, Postbox 217, 7500 AE, Enschede, The Netherlands, t.j.m.coenen@utwente.nl Also afflated wth: Unversty of Twente, Department of Computer Scence, Enschede, The Netherlands Permsson to make dgtal or hard copes of all or part of ths work for personal or classroom use s granted wthout fee provded that copes are not made or dstrbuted for proft or commercal advantage and that copes bear ths notce and the full ctaton on the frst page. To copy otherwse, to republsh, to post on servers or to redstrbute to lsts, requres pror specfc permsson and/or a fee. ValueTools 2008, October 21 23, 2008, Athens, GREECE. Copyrght c 2008 ICST ISBN # back-off counter or the number of consecutve packets that may be transmtted, see [10] for an overvew of IEEE e that ncorporates these mechansms. QoS provsonng for IEEE systems has been nvestgated manly va dscrete event smulatons. Analytcal models yeldng robust nsght nto system behavour are scarce. To a large extent, such models are based on the poneerng work of Banch [1], n whch a basc system wth persstent sources,.e. sources that always have packets ready to be transmtted, s modeled and analysed usng a Markov chan approach and valdated va smulaton showng excellent agreement wth actual system behavour. Extensons to nclude physcal layer detals are gven n e.g. [9],[16]. The extenson to non-persstent sources s provded n [3],[14], where a flow level model s ntroduced that s analysed usng a Processor Sharng queueng model. Comparson wth dscrete event smulaton shows that ndeed the MAC layer can be adequately modeled va the Processor Sharng mechansm. Extensons to multple traffc classes wth dfferent QoS requrements, as e.g. n e, are among others presented n [17],[18],[19]. Although the flow level modelng of [3],[14],[17],[18],[19] captures the resource sharng behavour of the MAC layer of protocols, the essental behavour at the packet level s not captured. At that level a flow conssts of a seres of packets that are transmtted one by one, where transmssons of dfferent flows are ntertwned. Especally for real tme applcatons, such as speech/telephony, the packet level s of hgh mportance. In [4], a packet level analyss for non-persstent sources s presented, extendng the Markov model of Banch to nclude the probablty of the node gong nto an empty backoff state. We take a further step to analyze the packet level by modelng the MAC layer as a pollng model where the server works off packets at dfferent queues. The essental characterstcs of the QoS aware MAC protocol are ncorporated va the frequency at whch the server vsts the dfferent nodes. In partcular, we gve the server a hgh probablty of vstng a node wth hgh prorty packets. In our pollng model, we consder two types of queues, vz. hgh and low prorty queues, each type wth a dfferent probablty of the server movng to t. Upon departure from a queue, the server randomly selects a queue accordng to these probabltes, whch mmcs the behavour of the MAC layer n systems. Note that we do not clam to accurately model the behavour of the IEEE e protocol, but analyse a mathematcally nterestng model that provdes nsght nto the effect of prortzaton such as used n

2 the IEEE MAC layer. In our model, we wll take the probablty of movng to a hgh prorty (HP) queue to be α tmes as hgh as movng to a low prorty (LP) queue. The servce tme of a packet s consdered to be determnstc as the packet szes n the system are equal for all queues and the channel speed s assumed to be constant at all tmes. As a queue s only allowed to transmt one packet when obtanng the channel, the servce dscplne s 1-lmted. Ths paper analyzes the steady state watng tme for ths 1-lmted pollng system wth random pollng. For the 1-lmted pollng model, general results are avalable n lterature. In [6] Fuhrmann and Cooper derve the well known decomposton result for queues wth server vacatons, whch s very useful for analyzng pollng models. For symmetrc queues, so wth dentcal arrval and servce rates at the queue, and a cyclc pollng order, [7] extends ths result to gve analytcal results on the average watng tme of packets n the queues. In [2], Boxma gves a pseudoconservaton law for the mean watng tme n a pollng system wth Markovan pollng, that ncludes random pollng. Ths law provdes an exact expresson for a weghted sum of the mean watng tmes at all queues, whch need not be symmetrcal. However, results for ndvdual queues cannot be derved from ths law when the network s not symmetrc. The man contrbuton of ths paper s an analyss of the steady state margnal dstrbuton of the watng tme of packets for dfferent types of queues n a 1-lmted asymmetrc pollng model. We consder the dfferent queues n the system ndvdually and model a partcular queue as a queue wth server vacatons, where these vacatons depend on the state of the other queues. To obtan the steady state watng tme dstrbuton, we propose an teraton algorthm. The algorthm computes the margnal steady state dstrbuton of the number of packets at a tagged queue, assumng a steady state at all other queues. Iteratng ths approach over the queues, for varous settngs, we obtan the steady state watng tme dstrbuton for packets at the dfferent queues. The remander of ths paper s organsed as follows. Secton 2 descrbes the queueng networks under consderaton and the analytcal approach for determnng the dstrbuton of the watng tme of customers per queue. Numercal results of the proposed algorthm are compared wth smulaton n Secton 3, and Secton 4 concludes the paper. 2. MODEL DESCRIPTION AND ANALYSIS Consder a pollng model consstng of queues Q 1,..., Q n wth fnte buffer B and a sngle server S vstng the queues. Customers arrve at a queue Q accordngtoapossonprocess wth rate λ. The servce process at the queues s determnstc wth servce tme τ and there s no swtchover tme between the queues. The routng polcy for the server s random, meanng there s a probablty p that the server moves to queue Q upon departure from queue Q j,j=1,..., n. For a hgh prorty queue, ths probablty s α tmesashghas for a low prorty queue, that s p HP = αp LP. The servce polcy s assumed to be 1-lmted, meanng at most one customer s served at each vst of the server, and customers are served FCFS at each queue. When the server reaches an empty queue, t wll mmedately proceed to the next. When all queues are empty, the server wats at the last queue to nstantly move to the frst queue that receves a customer. To ensure stablty of the system we assume that ρ = n λτ <1. In the followng, we derve expressons for the average watng tme of a packet for both types of queue. We start by consderng one hgh prorty queue surrounded by n low prorty queues. At both queues, packets arrve accordng to a Posson process. The server wll move to the HP queue α n+α wth probablty and to a certan LP queue wth probablty. We present an algorthm to approxmate the 1 n+α watng tme of a packet for both types of queue. Ths algorthm consders queues separately as served by a server wth vacatons. The length of the vacatons depends on the number of customers at the other queues. Startng wth an arbtrary dstrbuton of the number of customers at the other queues, the steady state of the number of customers n the consdered queue s determned, usng the vacaton tme dstrbuton. Ths process s terated over the dfferent types of queues repeatedly, untl convergence occurs. For specfc cases, beng that ether the HP or LP queues are saturated, meanng they always have packets ready to be transmtted, exact results are presented. Exact results are also gven for the case where all queues have equal prorty. 2.1 General case To determne the average watng tme of a packet n the queue, we consder the queues separately, as f they are n solaton. From the pont of vew of a queue, the server s ether present and servng a packet, or away whle servng an other queue. We thus can consder a queue as an M/D/1/B queue wth vacatons (c.f. [5],[11],[12]), where the absence of the server whle servng other queues are the vacatons. The length of these vacatons, whch depends on the number of customers at the other queues, nfluences the watng tme of the packets n the queue. For llustratory reasons, we frst gve the analyss for the scenaro where there are two queues, one hgh prorty and one low prorty queue, whch as we show later can be extended to any number of queues Two queues In the two queue scenaro, each queue can be consdered separately as a queue wth a server that goes on vacaton. The duraton of a vacaton now depends on the state of the other queue. We approxmate the dstrbuton of the length of the vacaton V x, gven the number of customers N y at the other queue (HP or LP) usng the followng recurson: q y B j= 1 P (V x = kτ N y = ) = (1) P (V x =(k 1)τ N y = j)p (A y = j +1), k 1 { P (V x =0 N y = ) = 1, =0 (1 q y), =1,..., B where V x s the length of the vacaton seen by the queue x, N y,q y and A y are the number of customers at queue y, the probablty of the server pollng queue y and the number of arrvng customers at queue y durng the servce tme at queue x, respectvely. Note that length of a servce perod s known to be τ due to the 1-lmted servce dscplne, hence we wll denote ths as a servce tme. The varable x can be the HP or LP queue and y s the other type of queue. The

3 vacaton length dstrbuton s then determned usng P (V x = kτ) = B P (V x = kτ N y = )P (N y = ), k 0 =0 (2) As the steady state dstrbuton P (N y = ), s not known, we start wth an arbtrary dstrbuton, for example an always empty queue. Usng ths dstrbuton, the vacaton dstrbuton for the other queue s obtaned. We derve the steady state dstrbuton of the number of customers n the queue usng the vacaton tme dstrbuton, so that by usng Lttle s law we acqure the expected watng tme of a packet. The queue under consderaton can be seen as an M/D/1/B queue wth vacatons (c.f. [5],[11],[12]). To analyze the steady state of ths queue, we frst focus on the state of the system at embedded ponts, whch are after the departure of a customer or the end of a vacaton. The probablty p n that an embedded pont s the completon of a servce and the departng customer leaves n customers behnd, and the probablty q n that an embedded pont s a vacaton termnaton wth n customers n the system are related n the followng manner n+1 p n = g n k+1 q k, n =0, 1,.., B 2 B 1 n=0 p n + p B 1 = q n = q B = B q n =1 n=0 k=1 B k=1 g C B kq k n h n k p k + h nq 0, n =0, 1,.., B 1 k=0 B 1 k=0 h C B kp k + h C Bq 0 where g j and h j denote the probablty of j customers arrvng durng a servce and vacaton tme, respectvely, gj C and h C j denote the probablty of j or more customers arrvng. As these probabltes are known, ths set of equatons can be solved, gvng the steady state dstrbuton at the end of an nterval (ether a servce or vacaton). To determne the contnuous tme steady state dstrbuton, we note that the number of tmes a departng customer leaves a certan number of customers behnd equals the number of tmes an arrvng customer fnds ths number of customers n the system. We have to take nto account, however, that an arrvng customer can fnd B customers n the system n whch case the customer s dscarded and leaves. Let P B denote the probablty that an arrvng customer fnds the system full. To evaluate ths expresson, observe that P B = ρ ρ ρ where ρ = λτ, λ = λ s the offered load and ρ s the carred load, ρ (1 b)τ = bev +(1 b)τ where EV denotes the expected vacaton tme and b denotes the probablty that an embedded pont s a vacaton termnaton pont, b = B q n. n=0 Let σ denote the multplcatve nverse of the average nterval between consecutve embedded ponts, that s σ 1 = bev +(1 b)τ then (1 b)σ P B =1. λ The queue length dstrbuton at arrval epochs, π n, n = 0,..., B s π n = P (Arrval sees n packets Arrval s accepted)(1 P B) + P (Arrval sees n packets Arrval not accepted)p B = p n(1 P B)+P B1(n = B) where { 1fn = B 1(n = B) = 0otherwse Combnng these results, we obtan (1 b)σ π n = p n, n =0, 1,..., B 1 (3) λ (1 b)σ π B =1. λ From PASTA we obtan that the contnuous tme steady state queue length dstrbuton s gven by π n, n =0,..., B. Note that (3) requres the average vacaton tme EV,and(2) the dstrbuton of the other queue to determne the vacaton tme dstrbuton. We may terate (2) and (3) to obtan an approxmaton of the steady state queue length dstrbuton. Algorthm 1. Iteraton 1. Intalze It := 1, x:= 1, y:= 2 P (N y = ) =γ, EN (0) = 0 for =0,..., B where EN (j) denotes the average queue length of queue n teraton j 2. Determne the vacaton tme dstrbuton at queue x from (2), and EV := EV x 3. Determne the queue length dstrbuton P (N x = n) = π n, n =0,..., B, from (3) and determne the average queue length EN x(it) 4. Set y := x, x := 3 y and repeat steps 2 and 3 for ths settng EN 5. If x(it) EN x(it 1) EN x(it) < 0.01 for both x =1, 2, then STOP Else y := x, x := 3 y, It := It +1GotoStep2 The algorthm approxmates n each teraton the number of customers found at the other queue to determne the vacaton tme for the tagged queue. When ths vacaton tme s underestmated, the server swtches back early

4 to the queue and starts servcng a packet at the consdered queue (when avalable), thus leavng the server busy. When however the vacaton tme s overestmated, the approach leaves the server at the other queue for too long a perod, where ths queue mght actually have become empty, thus leavng the server dle whle t could process jobs n the tagged (non-empty) queue. The presented approach hence underestmates the capacty of the server, but equally for both queues. The average queue length of all customers n the total system, whch for larger values of B approxmately can be seen as an M/D/1 queue as t s work conservng, s known and gven by ρ(2 ρ) EN total = where ρ =(λ LP + λ HP)τ, the load of the total system. The results obtaned by the teraton gve a hgher average queue length due to underestmaton of the server capacty. The queue length of each type of customer should hence be scaled down, so that the average queue length of all customers n the system s correct. Ths leads to an mproved estmaton of the average queue length of a customer per type of queue. Usng Lttle s law, we obtan the average watng tme for each type of queue. The algorthm can start wth an arbtrarly chosen steady state dstrbuton for the queue length of the HP queue. From ths, a new steady state s computed for the same queue. Startng from each ntal dstrbuton for the HP queue, Algorthm 1 converges to the steady state dstrbuton. Theorem 1 below states that ths convergence s monotone startng from ether an empty or full HP queue. We need stochastc orderng. Let X and Y be random varables wth dstrbuton F X(.) andf Y (.), respectvely. We say that X st Y ff F X(x) F Y (x) for all x 0 (c.f. [15], p.410). Theorem 1. For each ntal dstrbuton, Algorthm 1 converges monotoncally. Proof. Let X HP and X LP denote random varables for the queue length dstrbutons of the HP and LP queue after the th teraton and let Y LP and Y HP denote the random varables for the correspondng vacaton length dstrbutons. From (2) t follows that f X0 HP st X1 HP also Y0 LP st Y1 LP as a hgher queue length for the HP queue leads to a longer vacaton length for the LP queue. From (3) t follows that f Y0 LP st Y1 LP also X0 LP st X1 LP as a longer vacaton for the server of the LP queue leads to a hgher number of packets n the LP queue. Followng the same reasonng for the LP node, we have that X0 LP st X1 LP leads to Y0 HP st Y1 HP and Y0 HP st Y1 HP leads to X1 HP st X2 HP. It thus follows that X HP 0 st X HP Smlarly we have that X HP 0 st X HP as X0 HP st X1 HP. for any 1aslongasX0 HP st X1 HP. for any 1aslong From Theorem 1, an obvous approach s to start wth { P (X0 HP 1 for n =0 = n) = (4) 0 for n>0 snce t then holds that X0 HP st X for any X wth a nonnegatve dstrbuton. Let X denote the random varable followng an equlbrum dstrbuton, that s X = X+1. We then have that as X0 HP st X,alsoX HP st X,so the teraton process cannot jump past an equlbrum. In every teraton, the dstrbuton may change, movng closer towards the equlbrum dstrbuton. Smlarly, we can start wth the dstrbuton P (X HP 0 = n) = { 1 for n = B 0 for n<b where B s the maxmum number of customers n the queue. It then follows that X0 HP st X for any X, sothatafter every step we have that X HP st X as X0 HP st X. In ths case every teraton takes a step closer to the equlbrum from above. Usng Algorthm 1 startng from both (4) and (5), we fnd our approxmaton Multple queues The approach for two queues can easly be extended to multple queues of any prorty class. The vacaton length of a consdered queue then depends on the state of all the other queues, and can be computed by analogy to (2). The vacaton length dstrbuton n ths case s gven by P (V x = kτ N y = y,y x) = (6) B P (V x =(k 1)τ N z = a z,z x) q y y x z x,y a z= z,z x,y a y= y 1 P (A z = a z z) P (A y = a y y +1) P (V x =0 N y = y,y x) = q x + y, y>0 qy Here q x denotes the probablty of the server jumpng to queue x. The vacaton length dstrbuton s found usng P (V x = kτ) = (7) B P (V x = kτ N y = y,y x)p (N y = y,y x) y=0,y x where agan the steady state queue length dstrbuton of the other queues s needed. Startng agan wth a random dstrbuton for all but one queue we fnd the vacaton tme for ths tagged queue and hence the correspondng steady state queue length dstrbuton of ths queue. Ths dstrbuton can now be used for all queues of the same class and the other class can be analyzed usng the steps of the algorthm. Note that the proof of convergence remans the same, as the analyss s done for each type of queue. In the case of multple queues wth balanced load, that s wth dentcal arrval rates at the queues, the random varables X HP, X LP, Y HP and Y LP can be used for all queues of the same type as they are ndentcal. When arrval rates at the queues are dfferent, the same reasonng can be used for all separate varables X HP j, X LP j, Y HP j and Y LP j,wherethe subscrpt j denotes a specfc queue of the type HP or LP. 2.2 Specal cases For a hgh prorty queue, t may be needed that a certan average watng tme can be guaranteed. To obtan the maxmal average watng tme n a network wth one HP queue and n LP queues, we gve results for the stuaton wth saturated LP queues. To analyze the mpact of prortzng the hgh prorty queue on the low prorty queues, we compare the average watng tme at the LP queues wthout an HP queue n the system, wth the case where the HP q x (5)

5 queue s saturated. For these specal cases, exact results are avalable, whch are gven n ths secton Saturated LP queues Consder one hgh prorty queue wth Posson(λ HP)packet arrvals and n saturated low prorty queues,.e. λ LP. Let the probablty q of vstng the hgh prorty queue be q = α n + α where α denotes the factor of mportance gven to the hgh prorty queue, meanng the probablty of vstng the HP queue compared to the LP queue s α tmes as hgh. For the HP queue, the vacaton length dstrbuton s then gven by the geometrc dstrbuton P (V = kτ) =(1 q) k q as any tme the server does not jump to the HP queue, t wll servce exactly one packet at an LP queue. As the average tme between arrvals of the server s τ and the server only q serves one customer at each vst, the HP queue s stable when q>λτ. Wth the exact dstrbuton of the vacaton length known, we can use the pgf of the number of customers n the queue as gven by 3 to determne the average number of customers n the HP queue. The average watng tme then easly follows from Lttle s law Empty and saturated HP queue We now consder the case where the low prorty queues are no longer saturated, but each have an arrval process of rate λ LP and a determnstc servce tme of value τ. Let queue n + 1 be the HP queue, the conservaton law (c.f. [8]) then states that n+1 n+1 ρ EW q = ρ λβ(2) where EW q denotes the average watng tme n the queue (not ncludng servce) and ρ = λ LP τ for = 1...n and ρ n+1 = λ HPτ, so that and ρ = nρ LP + ρ HP. Astheservce tme dstrbuton s determnstc for any queue, we have that β (2) = τ 2 and the total watng tme of a customer s EW = EW q + τ. Consder the case where there are only n LP queues, so the arrval rate at the HP queue s set equal to zero. The stablty condton s that ρ = nλ LP τ<1and t mmedately follows that n n ρ LP EW q LP = ρ λlp τ 2 EW q LP = ρτ (2 ρ)τ EW LP = Now consder the case where the HP queue s saturated. We have n dentcal LP queues, and from the perspectve of the LP queues the server ncurs a swtchover tme when t vsts the HP queue. The stablty condton for ths system nλσ s that < 1, where σ denotes the mean swtchover (1 ρ) tme, as ths s the number of arrvng customers durng the average cycle tme of a queue. Let p denote the probablty of jumpng to queue and s the average tme t takes to swtch to queue. We have a pseudo-conservaton law statng that (c.f. [2]) n ρ ρ [1 λ σ p 1 ρ ]EW = n λβ(2) + σ 1 ρ n ρ p n ρ s + ρ 2σ n p s (2), where for our model we have that λ = λ LP = λ, ρ = λτ, ρ = nλτ, β (2) = τ 2, p = 1 qτ, s = n 1 q s(2) and = q(q+1)τ (1 q) 2 σ = s as all swtchover tmes are equal. Here q denotes the probablty of the server pollng the HP queue. As the LP queues are statstcally dentcal, the expresson smplfes to EW LP = nλτ 2 + qτn + q+1 2qτ 2(1 nλτ) (1 q)(1 nλτ) 2(1 q) nλqτ [1 ] (1 q)(1 nλτ) and applyng Lttle s law the average total number of customers n the queue s obtaned. Note that ths approach can easly be extended to a case wth multple hgh prorty queues, as only the probablty of the server beng on vacaton changes, so only the values of s and s (2) need to be adjusted. 3. VALIDATION In the followng we valdate our approxmaton approach by comparson wth smulaton results. For a wde varety of settngs, varyng the load of the system and the grade of prortzaton, the average watng tmes of packets at the ndvdual queues are determned. Note that the approach presented calculates the dstrbuton of the watng tme, but only the averages are used n the followng for comparson wth smulaton. Results for the scenaro wth one hgh prorty and one or two low prorty queues are consdered, together wth the specal cases. 3.1 General case Two queues Table 1 shows the average watng tme of packets n a queue computed by the algorthm compared wth smulaton results for dfferent loads of the system n the case of two queues, one HP and one LP queue. The table shows the mpact of varyng α, the relatve mportance of the HP queue compared to a LP queue. The load at the queues s balanced,.e. each queue has the same arrval rate of packets. The probablty of movng to the HP queue s q = α,whch n+α s α tmes as hgh as for the LP queue and the buffer sze s set to 15 for all cases. The mpact of the dfferentaton appears to be hgher when the load of the system ncreases. For a low load, the queues are often empty, thus makng t possble for the server to attend to packets drectly upon arrval. As the load ncreases, the queues wll be fuller and the watng tme depends more on the frequency at whch the server vsts the queues. We observe that the accuracy of the algorthm deterorates as the load of the system ncreases. For a hghly loaded system, the queues wll at tmes be fully loaded, causng arrvng packets to be lost. Ths effect s not taken nto account when usng the pseudoconservaton law to scale the obtaned results. Smulaton however shows that the mpact of ths approxmaton s lmted, as the average

6 Table 1: Average watng tme n a two node network wth balanced load Rates Smulaton Algorthm Error α λ LP λ HP LP HP LP HP LP HP number of packets n the system remans close to a system wth nfnte queues. In a smlar fashon Table 2 shows results for unbalanced arrval rates, wth the probablty q = 2 (α = 2) of vstng 3 the HP queue kept constant. For more unbalanced stuatons, the results deterorate, especally for hgher loads. For the node wth the lower arrval rate, the error made by the algorthm s bgger, as the average queue length s smaller. Comparng the mpact of ncreasng the load of the LP queue on the HP and vce versa shows that the ncrease n load of the HP queue has a bgger mpact on the average watng tme at the LP queue than ncreasng the load of the LP queue has on the HP queue. As an ncrease of the load wll cause the queue to be non-empty for a larger fracton of the tme, the mpact t has on the other queue by causng the server to go on a vacaton becomes larger. As a HP has a hgher probablty of beng vsted, ncreasng the load of ths queue has a bgger mpact than ncreasng the load at the LP queue Three queues In Table 3 we consder the scenaro wth three queues, one HP queue and two LP queues. The table shows the average watng tme of packets computed by the algorthm compared wth smulaton results for the stuaton wth balanced load. As for the stuaton wth two nodes, we observe that for hgher loads, the mpact of the prortzaton ncreases. Agan, the results deterorate as the load of the system ncreases. Comparson wth the results of Table 1 furhter shows that the mpact of prortzaton s hgher when more nodes are actve n the network. The decrease n the average watng tme of customers for the HP queue s stronger relatve to the decrease for the two node stuaton. Wth more queues present, the relatve ncrease n probablty of beng vsted s hgher when the value of α s ncreased. For example, ncreasng the value of α from 2 to 3 for both stuatons gves the followng relatve ncrease (r..): α =2 α = 3 r.. 2nodes q = 2 q = % 3 4 3nodes q = 1 q = 3 20% 2 5 For all settngs, no more than 15 teratons were needed by the algorthm wth the accuracy set n such a way that the last step gave an mprovement less that 1%. Longer runs wth hgher accuracy dd not mprove the results sgnfcantly. To run the teratons, the values of P (V x = kτ N y = ) for the two node case and P (V x = kτ N y = y,y x) for the three node case had to be computed once usng the teratons gven n (1) and (6), whch s tme consumng for large values of the buffer szes. For hghly flled buffers however, the geometrc dstrbuton can be used, as the probablty of the vacaton havng a duraton of kτ s then very close to the probablty of frst vstng k other queues before vstng the consdered queue, as the other queues wll not become empty durng the process. The tme needed for the teraton tself s very lmted, as (2) (or (7)) only encompasses the addton over all possble values of queue lenghts and (3) s a small enough system of equatons to be solved wthn seconds. 3.2 Specal cases Saturated low prorty queues For a user wth mportant traffc, the QoS dfferentaton s of hgh mportance. To get an dea of the mpact of the settngs for the dfferentaton, a worst case scenaro can be analysed to see the mnmal prortzaton that s needed to obtan a certan average watng tme for the hgh prorty packets. The worst case scenaro s when all other (low prorty) queues always have traffc to transmt. Fgure 1 shows the average watng tme of a packet n the HP queue, for dfferent values of n, the number of saturated low prorty queues n the system. The arrval rate at the HP queue s set to λ HP =0.01. The three lnes represent the results of the model for α = 2...4, the grade of prortzaton. It clearly follows from the fgure that where for a sparce network (low number of LP queues) the dfferentaton has a lmted effect and that for a dense network (hgh number of LP queues) gvng more prorty has a much bgger mpact Empty and saturated hgh prorty queue The dfferentaton between users s prmarly done to provde better performance for more mportant traffc. However, t also has to be taken nto account what the mpact s on performance of the less mportant traffc. If the prortzaton of the hgh prorty queue s too hgh, the low prorty queues mght be starved. To analyse the mpact on the low prorty queues, we compare the stuaton wthout the HP queue (or an empty HP queue) wth the stuaton that the HP queue always has packets to transmt. In the latter case, we vary the grade of prortzaton. Fgure 2 shows the average watng tme of a packet n an LP queue, for dfferent values of n, for dfferent settngs of the HP queue. The arrval rate λ LP s set to 0.01 for each of the n LP queues. In ths case the HP queue s ether absent (or empty) n whch

7 Table 2: Average watng tme n a two node network wth unbalanced load Rates Smulaton Algorthm Error λ LP λ HP LP HP LP HP LP HP Table 3: Average watng tme n a three node network wth balanced load Rates Smulaton Algorthm Error α λ LP λ HP LP HP LP HP LP HP alpha = 2 alpha = 3 alpha = 4 Empty HP alpha = 2 alpha = 3 alpha = avg watng tme avg watng tme number of LP nodes number of LP nodes Fgure 1: Average watng tme for the scenaro wth saturated LP queues Fgure 2: Average watng tmes for the scenaro wth an empty or saturated HP node case the complete network behaves as a standard M/D/1 queue where each separate queue has the same average behavour or the HP s saturated, wth dfferent values for α, the grade of prortzaton. For hgher values of α the server wll more often be processng HP packets, leavng less capacty for the LP queues. Ths shows from the fgure as the watng tme reaches hgh values already for lower values of n. When the network s sparce, we see there s already a substantal mpact of the dfferentaton on the watng tme of the low prorty packets. 4. CONCLUSION In ths paper we analyzed the mpact of QoS dfferentaton on the delay of packets for dfferent classes of queues usng a 1-lmted pollng model wth a random schedulng polcy and determnstc servce tmes, capturng the random nature of the MAC layer protocol. The model gves nsght n the effect of the parameter settngs on the QoS n a WLAN for the ndvdual classes of queues. We developed an approxmaton approach for the packet delay n a network wth hgh and low prorty queues. Comparson wth smulaton results shows that for low to moderately loaded systems, the approach works well. 5. REFERENCES [1] Banch, G., Performance analyss of the IEEE dstrbuted coordnaton functon, IEEE Journal on Selected Areas n Communcatons, vol. 18,

8 nr. 3, , [2] Boxma, O.J. and Weststrate, J.A., Watng tmes n pollng systems wth Markovan server routng, Informatk-Fachberchte, vol. 218, , [3] Cheung, S-K., van den Berg, J.L. and Bouchere, R.J., Decomposng the queue length dstrbuton of processor-sharng models nto queue lengths of permanent customer queues, Performance Evaluaton, vol. 62, , [4] Engelstad, P.E. and Østerbø, O.N., Non-Saturaton and Saturaton Analyss of IEEE e EDCA wth Starvaton Predcton, Proceedngs of ACM MSWM 05, Montreal, Canada, [5] Fuhrmann, S.W., A Note on the M/G/1 Queue wth Server Vacatons, Operatons Research, vol. 32, no. 6, , [6] Fuhrmann, S.W. and Cooper, R.B., Stochastc decompostons n the M/G/1 queue wth generalzed vacatons, Operatons Research, vol. 33, , [7] Fuhrmann, S.W., Symmetrc queues served n cyclc order, Operatons Research Letters, vol. 4, nr. 3, , [8] Groenendjk, W.P., Conservaton laws n pollng systems, Ph.D. Thess, Unversty of Utrecht, [9] Hadz-Velkov, Z. and Spasenovsk, B., Capture effect n IEEE Wreless LANs, Proceedngs of IEEE ICWLHN 01, Sngapore, [10] IEEE, IEEE Standard for Wreless LAN Medum Access Control (MAC) and Physcal Layer (PHY) specfcatons, Medum Access Control (MAC) Qualty of Servce Enhancements, IEEE std e, [11] Kramer, M., Statonary dstrbutons n a queueng system wth vacaton tmes and lmted servce, Queueng Systems, vol. 4, 57-68, [12] Lee, T.T., M/G/1/N queue wth vacaton tme and lmted servce dscplne, Performance Evaluaton, vol. 9, no. 3, , [13] Levy, H., Pollng Systems: Applcatons, Modelng, and Optmzaton, IEEE Transactons on Communcatons, vol. 38, no. 10, [14] Ltjens, R., Rojers, F., van den Berg, J.L., Bouchere, R.J. and Fleuren, M., Performance analyss of Wreless LANs: An ntegrated packet/flow level approach, Proceedngs of the 18th Internatonal Teletraffc Congress, Berln, Germany, , [15] Ross, S.M., Stochastc Processes, second edton, John Wley & Sons, New York, [16] Wu, H. et al., Performance of relable transport protocol over IEEE wreless LAN: Analyss and enhancement, Proceedngs of IEEE INFOCOM 02, New York, USA, 2002 [17] Xao, Y., Performance Analyss of Prorty Schemes for IEEE and IEEE e Wreless LANs, IEEE Transactons on Wreless Communcatons, vol. 4, no. 4, July [18] Xong, L. and Mao, G., Saturated Throughput Analyss of IEEE e Usng Two-Dmensonal Markov Chan Model, Proceedngs QShne 06, Waterloo, Canada, [19] Zhu, H. and Chlamtac, I., An Analytc Model for IEEE e EDCF Dfferental Servces, Proceedngs of IEEE ICCCN 03, Dallas, USA, 2003.

On the Exact Analysis of Bluetooth Scheduling Algorithms

On the Exact Analysis of Bluetooth Scheduling Algorithms On the Exact Analyss of Bluetooth Schedulng Algorth Gl Zussman Dept. of Electrcal Engneerng Technon IIT Hafa 3000, Israel glz@tx.technon.ac.l Ur Yechal Dept. of Statstcs and Operatons Research School of

More information

Real-Time Guarantees. Traffic Characteristics. Flow Control

Real-Time Guarantees. Traffic Characteristics. Flow Control Real-Tme Guarantees Requrements on RT communcaton protocols: delay (response s) small jtter small throughput hgh error detecton at recever (and sender) small error detecton latency no thrashng under peak

More information

Analysis of Collaborative Distributed Admission Control in x Networks

Analysis of Collaborative Distributed Admission Control in x Networks 1 Analyss of Collaboratve Dstrbuted Admsson Control n 82.11x Networks Thnh Nguyen, Member, IEEE, Ken Nguyen, Member, IEEE, Lnha He, Member, IEEE, Abstract Wth the recent surge of wreless home networks,

More information

Simulation Based Analysis of FAST TCP using OMNET++

Simulation Based Analysis of FAST TCP using OMNET++ Smulaton Based Analyss of FAST TCP usng OMNET++ Umar ul Hassan 04030038@lums.edu.pk Md Term Report CS678 Topcs n Internet Research Sprng, 2006 Introducton Internet traffc s doublng roughly every 3 months

More information

Adaptive Network Resource Management in IEEE Wireless Random Access MAC

Adaptive Network Resource Management in IEEE Wireless Random Access MAC Adaptve Network Resource Management n IEEE 802.11 Wreless Random Access MAC Hao Wang, Changcheng Huang, James Yan Department of Systems and Computer Engneerng Carleton Unversty, Ottawa, ON, Canada Abstract

More information

For instance, ; the five basic number-sets are increasingly more n A B & B A A = B (1)

For instance, ; the five basic number-sets are increasingly more n A B & B A A = B (1) Secton 1.2 Subsets and the Boolean operatons on sets If every element of the set A s an element of the set B, we say that A s a subset of B, or that A s contaned n B, or that B contans A, and we wrte A

More information

Cluster Analysis of Electrical Behavior

Cluster Analysis of Electrical Behavior Journal of Computer and Communcatons, 205, 3, 88-93 Publshed Onlne May 205 n ScRes. http://www.scrp.org/ournal/cc http://dx.do.org/0.4236/cc.205.350 Cluster Analyss of Electrcal Behavor Ln Lu Ln Lu, School

More information

Scheduling and queue management. DigiComm II

Scheduling and queue management. DigiComm II Schedulng and queue management Tradtonal queung behavour n routers Data transfer: datagrams: ndvdual packets no recognton of flows connectonless: no sgnallng Forwardng: based on per-datagram forwardng

More information

Analysis of QoS in WLAN

Analysis of QoS in WLAN Analyss of QoS n WLAN PAAL E. ENGELSTAD AND OLAV N. ØSTERBØ Paal E. Engelstad s Research Scentst n Telenor R&D An analytcal model s proposed to descrbe the prorty schemes of the EDCA mechansm of the IEEE

More information

A Binarization Algorithm specialized on Document Images and Photos

A Binarization Algorithm specialized on Document Images and Photos A Bnarzaton Algorthm specalzed on Document mages and Photos Ergna Kavalleratou Dept. of nformaton and Communcaton Systems Engneerng Unversty of the Aegean kavalleratou@aegean.gr Abstract n ths paper, a

More information

Performance Analysis of Markov Modulated 1-Persistent CSMA/CA Protocols with Exponential Backoff Scheduling

Performance Analysis of Markov Modulated 1-Persistent CSMA/CA Protocols with Exponential Backoff Scheduling Performance Analyss of Markov Modulated -Persstent CSMA/CA Protocols wth ponental Backoff Schedulng Pu ng Wong, Dongje Yn, and Tony T. Lee, Abstract. Ths paper proposes a Markovan model of -persstent CSMA/CA

More information

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization Problem efntons and Evaluaton Crtera for Computatonal Expensve Optmzaton B. Lu 1, Q. Chen and Q. Zhang 3, J. J. Lang 4, P. N. Suganthan, B. Y. Qu 6 1 epartment of Computng, Glyndwr Unversty, UK Faclty

More information

Performance Evaluation of IEEE e based on ON-OFF Traffic Model I. Papapanagiotou PhD. Student

Performance Evaluation of IEEE e based on ON-OFF Traffic Model I. Papapanagiotou PhD. Student erformance Evaluaton of IEEE 82.e based on ON-OFF Traffc Model I. apapanagotou hd. Student Wreless Telecommuncaton Laboratory Unversty of atras 265 4 atras Greece papapanag@upnet.gr J.S. Vardakas hd. Student

More information

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour 6.854 Advanced Algorthms Petar Maymounkov Problem Set 11 (November 23, 2005) Wth: Benjamn Rossman, Oren Wemann, and Pouya Kheradpour Problem 1. We reduce vertex cover to MAX-SAT wth weghts, such that the

More information

ANALYTICAL MODEL AND PERFORMANCE ANALYSIS OF A NETWORK INTERFACE CARD. Abstract

ANALYTICAL MODEL AND PERFORMANCE ANALYSIS OF A NETWORK INTERFACE CARD. Abstract ANALYTICAL MODEL AND PERFORMANCE ANALYSIS OF A NETWORK INTERFACE CARD Naveen Cherukur 1, Gokul B. Kandraju 2, Natarajan Gautam 3, and Anand Svasubramanam 4 Abstract One of the key concerns for practtoners

More information

A Sub-Critical Deficit Round-Robin Scheduler

A Sub-Critical Deficit Round-Robin Scheduler A Sub-Crtcal Defct ound-obn Scheduler Anton Kos, Sašo Tomažč Unversty of Ljubljana, Faculty of Electrcal Engneerng, Ljubljana, Slovena E-mal: anton.kos@fe.un-lj.s Abstract - A scheduler s an essental element

More information

Load-Balanced Anycast Routing

Load-Balanced Anycast Routing Load-Balanced Anycast Routng Chng-Yu Ln, Jung-Hua Lo, and Sy-Yen Kuo Department of Electrcal Engneerng atonal Tawan Unversty, Tape, Tawan sykuo@cc.ee.ntu.edu.tw Abstract For fault-tolerance and load-balance

More information

RAP. Speed/RAP/CODA. Real-time Systems. Modeling the sensor networks. Real-time Systems. Modeling the sensor networks. Real-time systems:

RAP. Speed/RAP/CODA. Real-time Systems. Modeling the sensor networks. Real-time Systems. Modeling the sensor networks. Real-time systems: Speed/RAP/CODA Presented by Octav Chpara Real-tme Systems Many wreless sensor network applcatons requre real-tme support Survellance and trackng Border patrol Fre fghtng Real-tme systems: Hard real-tme:

More information

UBICC Publishers 2008 Ubiquitous Computing and Communication Journal

UBICC Publishers 2008 Ubiquitous Computing and Communication Journal UBICC Journal Ubqutous Computng and Communcaton Journal 008 Volume 3. 008-04-30. ISSN 99-844 Specal Issue on Moble Adhoc Networks UBICC ublshers 008 Ubqutous Computng and Communcaton Journal Edted by Usman

More information

Interclass Collision Protection for IEEE e Wireless LANs

Interclass Collision Protection for IEEE e Wireless LANs Interclass Collson Protecton for IEEE 82.e Wreless LANs Woon Sun Cho, Chae Y. Lee Dstrbuted Coordnaton Functon (DCF) n IEEE 82. and Enhanced Dstrbuted Channel Access (EDCA) n IEEE 82.e are contenton-based

More information

Video Proxy System for a Large-scale VOD System (DINA)

Video Proxy System for a Large-scale VOD System (DINA) Vdeo Proxy System for a Large-scale VOD System (DINA) KWUN-CHUNG CHAN #, KWOK-WAI CHEUNG *# #Department of Informaton Engneerng *Centre of Innovaton and Technology The Chnese Unversty of Hong Kong SHATIN,

More information

Wishing you all a Total Quality New Year!

Wishing you all a Total Quality New Year! Total Qualty Management and Sx Sgma Post Graduate Program 214-15 Sesson 4 Vnay Kumar Kalakband Assstant Professor Operatons & Systems Area 1 Wshng you all a Total Qualty New Year! Hope you acheve Sx sgma

More information

Parallelism for Nested Loops with Non-uniform and Flow Dependences

Parallelism for Nested Loops with Non-uniform and Flow Dependences Parallelsm for Nested Loops wth Non-unform and Flow Dependences Sam-Jn Jeong Dept. of Informaton & Communcaton Engneerng, Cheonan Unversty, 5, Anseo-dong, Cheonan, Chungnam, 330-80, Korea. seong@cheonan.ac.kr

More information

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields 17 th European Symposum on Computer Aded Process Engneerng ESCAPE17 V. Plesu and P.S. Agach (Edtors) 2007 Elsever B.V. All rghts reserved. 1 A mathematcal programmng approach to the analyss, desgn and

More information

Voice capacity of IEEE b WLANs

Voice capacity of IEEE b WLANs Voce capacty of IEEE 82.b WLANs D. S. Amanatads, V. Vtsas, A. Mantsars 2, I. Mavrds 2, P. Chatzmsos and A.C. Boucouvalas 3 Abstract-There s a tremendous growth n the deployment and usage of Wreless Local

More information

Lobachevsky State University of Nizhni Novgorod. Polyhedron. Quick Start Guide

Lobachevsky State University of Nizhni Novgorod. Polyhedron. Quick Start Guide Lobachevsky State Unversty of Nzhn Novgorod Polyhedron Quck Start Gude Nzhn Novgorod 2016 Contents Specfcaton of Polyhedron software... 3 Theoretcal background... 4 1. Interface of Polyhedron... 6 1.1.

More information

Fast Computation of Shortest Path for Visiting Segments in the Plane

Fast Computation of Shortest Path for Visiting Segments in the Plane Send Orders for Reprnts to reprnts@benthamscence.ae 4 The Open Cybernetcs & Systemcs Journal, 04, 8, 4-9 Open Access Fast Computaton of Shortest Path for Vstng Segments n the Plane Ljuan Wang,, Bo Jang

More information

Scheduling Remote Access to Scientific Instruments in Cyberinfrastructure for Education and Research

Scheduling Remote Access to Scientific Instruments in Cyberinfrastructure for Education and Research Schedulng Remote Access to Scentfc Instruments n Cybernfrastructure for Educaton and Research Je Yn 1, Junwe Cao 2,3,*, Yuexuan Wang 4, Lanchen Lu 1,3 and Cheng Wu 1,3 1 Natonal CIMS Engneerng and Research

More information

with `ook-ahead for Broadcast WDM Networks TR May 14, 1996 Abstract

with `ook-ahead for Broadcast WDM Networks TR May 14, 1996 Abstract HPeR-`: A Hgh Performance Reservaton Protocol wth `ook-ahead for Broadcast WDM Networks Vjay Svaraman George N. Rouskas TR-96-06 May 14, 1996 Abstract We consder the problem of coordnatng access to the

More information

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS ARPN Journal of Engneerng and Appled Scences 006-017 Asan Research Publshng Network (ARPN). All rghts reserved. NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS Igor Grgoryev, Svetlana

More information

Non-Split Restrained Dominating Set of an Interval Graph Using an Algorithm

Non-Split Restrained Dominating Set of an Interval Graph Using an Algorithm Internatonal Journal of Advancements n Research & Technology, Volume, Issue, July- ISS - on-splt Restraned Domnatng Set of an Interval Graph Usng an Algorthm ABSTRACT Dr.A.Sudhakaraah *, E. Gnana Deepka,

More information

X- Chart Using ANOM Approach

X- Chart Using ANOM Approach ISSN 1684-8403 Journal of Statstcs Volume 17, 010, pp. 3-3 Abstract X- Chart Usng ANOM Approach Gullapall Chakravarth 1 and Chaluvad Venkateswara Rao Control lmts for ndvdual measurements (X) chart are

More information

Channel 0. Channel 1 Channel 2. Channel 3 Channel 4. Channel 5 Channel 6 Channel 7

Channel 0. Channel 1 Channel 2. Channel 3 Channel 4. Channel 5 Channel 6 Channel 7 Optmzed Regonal Cachng for On-Demand Data Delvery Derek L. Eager Mchael C. Ferrs Mary K. Vernon Unversty of Saskatchewan Unversty of Wsconsn Madson Saskatoon, SK Canada S7N 5A9 Madson, WI 5376 eager@cs.usask.ca

More information

A fair buffer allocation scheme

A fair buffer allocation scheme A far buffer allocaton scheme Juha Henanen and Kalev Klkk Telecom Fnland P.O. Box 228, SF-330 Tampere, Fnland E-mal: juha.henanen@tele.f Abstract An approprate servce for data traffc n ATM networks requres

More information

Derivation of Three Queue Nodes Discrete-Time Analytical Model Based on DRED Algorithm

Derivation of Three Queue Nodes Discrete-Time Analytical Model Based on DRED Algorithm Seventh Internatonal Conference on Informaton Technology Dervaton of Three Queue Nodes Dscrete-Tme Analytcal Model Based on DRED Algorthm Jafar Ababneh, Hussen Abdel-Jaber, 3 Fad Thabtah, 3 Wael Had, EmranBadarneh

More information

Analysis of Continuous Beams in General

Analysis of Continuous Beams in General Analyss of Contnuous Beams n General Contnuous beams consdered here are prsmatc, rgdly connected to each beam segment and supported at varous ponts along the beam. onts are selected at ponts of support,

More information

SLAM Summer School 2006 Practical 2: SLAM using Monocular Vision

SLAM Summer School 2006 Practical 2: SLAM using Monocular Vision SLAM Summer School 2006 Practcal 2: SLAM usng Monocular Vson Javer Cvera, Unversty of Zaragoza Andrew J. Davson, Imperal College London J.M.M Montel, Unversty of Zaragoza. josemar@unzar.es, jcvera@unzar.es,

More information

DESIGNING TRANSMISSION SCHEDULES FOR WIRELESS AD HOC NETWORKS TO MAXIMIZE NETWORK THROUGHPUT

DESIGNING TRANSMISSION SCHEDULES FOR WIRELESS AD HOC NETWORKS TO MAXIMIZE NETWORK THROUGHPUT DESIGNING TRANSMISSION SCHEDULES FOR WIRELESS AD HOC NETWORKS TO MAXIMIZE NETWORK THROUGHPUT Bran J. Wolf, Joseph L. Hammond, and Harlan B. Russell Dept. of Electrcal and Computer Engneerng, Clemson Unversty,

More information

The Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique

The Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique //00 :0 AM Outlne and Readng The Greedy Method The Greedy Method Technque (secton.) Fractonal Knapsack Problem (secton..) Task Schedulng (secton..) Mnmum Spannng Trees (secton.) Change Money Problem Greedy

More information

An Optimal Algorithm for Prufer Codes *

An Optimal Algorithm for Prufer Codes * J. Software Engneerng & Applcatons, 2009, 2: 111-115 do:10.4236/jsea.2009.22016 Publshed Onlne July 2009 (www.scrp.org/journal/jsea) An Optmal Algorthm for Prufer Codes * Xaodong Wang 1, 2, Le Wang 3,

More information

On Achieving Fairness in the Joint Allocation of Buffer and Bandwidth Resources: Principles and Algorithms

On Achieving Fairness in the Joint Allocation of Buffer and Bandwidth Resources: Principles and Algorithms On Achevng Farness n the Jont Allocaton of Buffer and Bandwdth Resources: Prncples and Algorthms Yunka Zhou and Harsh Sethu (correspondng author) Abstract Farness n network traffc management can mprove

More information

Efficient QoS Provisioning at the MAC Layer in Heterogeneous Wireless Sensor Networks

Efficient QoS Provisioning at the MAC Layer in Heterogeneous Wireless Sensor Networks Effcent QoS Provsonng at the MAC Layer n Heterogeneous Wreless Sensor Networks M.Soul a,, A.Bouabdallah a, A.E.Kamal b a UMR CNRS 7253 HeuDaSyC, Unversté de Technologe de Compègne, Compègne Cedex F-625,

More information

Modelling a Queuing System for a Virtual Agricultural Call Center

Modelling a Queuing System for a Virtual Agricultural Call Center 25-28 July 2005, Vla Real, Portugal Modellng a Queung System for a Vrtual Agrcultural Call Center İnc Şentarlı, a, Arf Orçun Sakarya b a, Çankaya Unversty, Department of Management,06550, Balgat, Ankara,

More information

Compiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz

Compiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz Compler Desgn Sprng 2014 Regster Allocaton Sample Exercses and Solutons Prof. Pedro C. Dnz USC / Informaton Scences Insttute 4676 Admralty Way, Sute 1001 Marna del Rey, Calforna 90292 pedro@s.edu Regster

More information

Performance analysis of distributed cluster-based MAC protocol for multiuser MIMO wireless networks

Performance analysis of distributed cluster-based MAC protocol for multiuser MIMO wireless networks RESEARCH Open Access Performance analyss of dstrbuted cluster-based MAC protocol for multuser MIMO wreless networks Azadeh Ettefagh *, Marc Kuhn, Celal Eşl and Armn Wttneben Abstract It s known that multuser

More information

S1 Note. Basis functions.

S1 Note. Basis functions. S1 Note. Bass functons. Contents Types of bass functons...1 The Fourer bass...2 B-splne bass...3 Power and type I error rates wth dfferent numbers of bass functons...4 Table S1. Smulaton results of type

More information

Routing in Degree-constrained FSO Mesh Networks

Routing in Degree-constrained FSO Mesh Networks Internatonal Journal of Hybrd Informaton Technology Vol., No., Aprl, 009 Routng n Degree-constraned FSO Mesh Networks Zpng Hu, Pramode Verma, and James Sluss Jr. School of Electrcal & Computer Engneerng

More information

CS 534: Computer Vision Model Fitting

CS 534: Computer Vision Model Fitting CS 534: Computer Vson Model Fttng Sprng 004 Ahmed Elgammal Dept of Computer Scence CS 534 Model Fttng - 1 Outlnes Model fttng s mportant Least-squares fttng Maxmum lkelhood estmaton MAP estmaton Robust

More information

Helsinki University Of Technology, Systems Analysis Laboratory Mat Independent research projects in applied mathematics (3 cr)

Helsinki University Of Technology, Systems Analysis Laboratory Mat Independent research projects in applied mathematics (3 cr) Helsnk Unversty Of Technology, Systems Analyss Laboratory Mat-2.08 Independent research projects n appled mathematcs (3 cr) "! #$&% Antt Laukkanen 506 R ajlaukka@cc.hut.f 2 Introducton...3 2 Multattrbute

More information

Extension and Comparison of QoS-Enabled Wi-Fi Models in the Presence of Errors

Extension and Comparison of QoS-Enabled Wi-Fi Models in the Presence of Errors Extenson and Comparson of QoS-Enabled W-F Models n the Presence of Errors Ioanns Papapanagotou, Georgos S. Paschos*, Stavros A. Kotsopoulos** and Mchael Devetskots Electrcal and Computer Engneerng, North

More information

USER CLASS BASED QoS DIFFERENTIATION IN e WLAN

USER CLASS BASED QoS DIFFERENTIATION IN e WLAN USER CLASS BASED QoS DIFFERENTIATION IN 802.11e WLAN Amt Kejrwal, Ratan Guha School of Computer Scence Unversty of Central Florda Orlando, FL-32816 USA E-mal: {kejrwal, guha}@cs.ucf.edu Manak Chatterjee

More information

Analytic Evaluation of Quality of Service for On-Demand Data Delivery

Analytic Evaluation of Quality of Service for On-Demand Data Delivery Analytc Evaluaton of Qualty of Servce for On-Demand Data Delvery Hongfe Guo Haonan Tan ( guo@cs.wsc.edu) (haonan@cs.wsc.edu) Abstract Qualty of servce (QoS) measured as balkng probablty and average watng

More information

Virtual Memory. Background. No. 10. Virtual Memory: concept. Logical Memory Space (review) Demand Paging(1) Virtual Memory

Virtual Memory. Background. No. 10. Virtual Memory: concept. Logical Memory Space (review) Demand Paging(1) Virtual Memory Background EECS. Operatng System Fundamentals No. Vrtual Memory Prof. Hu Jang Department of Electrcal Engneerng and Computer Scence, York Unversty Memory-management methods normally requres the entre process

More information

Mathematics 256 a course in differential equations for engineering students

Mathematics 256 a course in differential equations for engineering students Mathematcs 56 a course n dfferental equatons for engneerng students Chapter 5. More effcent methods of numercal soluton Euler s method s qute neffcent. Because the error s essentally proportonal to the

More information

Reducing Frame Rate for Object Tracking

Reducing Frame Rate for Object Tracking Reducng Frame Rate for Object Trackng Pavel Korshunov 1 and We Tsang Oo 2 1 Natonal Unversty of Sngapore, Sngapore 11977, pavelkor@comp.nus.edu.sg 2 Natonal Unversty of Sngapore, Sngapore 11977, oowt@comp.nus.edu.sg

More information

Optimization of Local Routing for Connected Nodes with Single Output Ports - Part I: Theory

Optimization of Local Routing for Connected Nodes with Single Output Ports - Part I: Theory U J.T. (: 33- (pr. 0 Optmzaton of Local Routng for Connected odes wth Sngle Output Ports - Part I: Theory Dobr tanassov Batovsk Faculty of Scence and Technology ssumpton Unversty Bangkok Thaland E-mal:

More information

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Proceedngs of the Wnter Smulaton Conference M E Kuhl, N M Steger, F B Armstrong, and J A Jones, eds A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Mark W Brantley Chun-Hung

More information

A SIMULATION ANALYSIS OF AGGREGATION STRATEGIES IN A WF 2 Q+ SCHEDULERS NETWORK

A SIMULATION ANALYSIS OF AGGREGATION STRATEGIES IN A WF 2 Q+ SCHEDULERS NETWORK A SIMULATION ANALYSIS OF AGGREGATION STRATEGIES IN A WF 2 Q+ SCHEDULERS NETWORK R. G. Garroppo, S. Gordano, S. Nccoln, F. Russo {r.garroppo, s.gordano, s.nccoln, f.russo}@et.unp.t Department of Informaton

More information

CHAPTER 2 PROPOSED IMPROVED PARTICLE SWARM OPTIMIZATION

CHAPTER 2 PROPOSED IMPROVED PARTICLE SWARM OPTIMIZATION 24 CHAPTER 2 PROPOSED IMPROVED PARTICLE SWARM OPTIMIZATION The present chapter proposes an IPSO approach for multprocessor task schedulng problem wth two classfcatons, namely, statc ndependent tasks and

More information

VISUAL SELECTION OF SURFACE FEATURES DURING THEIR GEOMETRIC SIMULATION WITH THE HELP OF COMPUTER TECHNOLOGIES

VISUAL SELECTION OF SURFACE FEATURES DURING THEIR GEOMETRIC SIMULATION WITH THE HELP OF COMPUTER TECHNOLOGIES UbCC 2011, Volume 6, 5002981-x manuscrpts OPEN ACCES UbCC Journal ISSN 1992-8424 www.ubcc.org VISUAL SELECTION OF SURFACE FEATURES DURING THEIR GEOMETRIC SIMULATION WITH THE HELP OF COMPUTER TECHNOLOGIES

More information

Type-2 Fuzzy Non-uniform Rational B-spline Model with Type-2 Fuzzy Data

Type-2 Fuzzy Non-uniform Rational B-spline Model with Type-2 Fuzzy Data Malaysan Journal of Mathematcal Scences 11(S) Aprl : 35 46 (2017) Specal Issue: The 2nd Internatonal Conference and Workshop on Mathematcal Analyss (ICWOMA 2016) MALAYSIAN JOURNAL OF MATHEMATICAL SCIENCES

More information

Efficient Content Distribution in Wireless P2P Networks

Efficient Content Distribution in Wireless P2P Networks Effcent Content Dstrbuton n Wreless P2P Networs Qong Sun, Vctor O. K. L, and Ka-Cheong Leung Department of Electrcal and Electronc Engneerng The Unversty of Hong Kong Pofulam Road, Hong Kong, Chna {oansun,

More information

FAHP and Modified GRA Based Network Selection in Heterogeneous Wireless Networks

FAHP and Modified GRA Based Network Selection in Heterogeneous Wireless Networks 2017 2nd Internatonal Semnar on Appled Physcs, Optoelectroncs and Photoncs (APOP 2017) ISBN: 978-1-60595-522-3 FAHP and Modfed GRA Based Network Selecton n Heterogeneous Wreless Networks Xaohan DU, Zhqng

More information

Priority-Based Scheduling Algorithm for Downlink Traffics in IEEE Networks

Priority-Based Scheduling Algorithm for Downlink Traffics in IEEE Networks Prorty-Based Schedulng Algorthm for Downlnk Traffcs n IEEE 80.6 Networks Ja-Mng Lang, Jen-Jee Chen, You-Chun Wang, Yu-Chee Tseng, and Bao-Shuh P. Ln Department of Computer Scence Natonal Chao-Tung Unversty,

More information

Avoiding congestion through dynamic load control

Avoiding congestion through dynamic load control Avodng congeston through dynamc load control Vasl Hnatyshn, Adarshpal S. Seth Department of Computer and Informaton Scences, Unversty of Delaware, Newark, DE 976 ABSTRACT The current best effort approach

More information

Improvement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration

Improvement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration Improvement of Spatal Resoluton Usng BlockMatchng Based Moton Estmaton and Frame Integraton Danya Suga and Takayuk Hamamoto Graduate School of Engneerng, Tokyo Unversty of Scence, 6-3-1, Nuku, Katsuska-ku,

More information

Learning the Kernel Parameters in Kernel Minimum Distance Classifier

Learning the Kernel Parameters in Kernel Minimum Distance Classifier Learnng the Kernel Parameters n Kernel Mnmum Dstance Classfer Daoqang Zhang 1,, Songcan Chen and Zh-Hua Zhou 1* 1 Natonal Laboratory for Novel Software Technology Nanjng Unversty, Nanjng 193, Chna Department

More information

Efficient Distributed File System (EDFS)

Efficient Distributed File System (EDFS) Effcent Dstrbuted Fle System (EDFS) (Sem-Centralzed) Debessay(Debsh) Fesehaye, Rahul Malk & Klara Naherstedt Unversty of Illnos-Urbana Champagn Contents Problem Statement, Related Work, EDFS Desgn Rate

More information

A Fair Access Mechanism Based on TXOP in IEEE e Wireless Networks

A Fair Access Mechanism Based on TXOP in IEEE e Wireless Networks 11 Internatonal Journal of Communcaton Networks and Informaton Securty (IJCNIS) Vol. 8, No. 1, Aprl 16 A Far Access Mechansm Based on TXOP n IEEE 8.11e Wreless Networks Marjan Yazdan 1, Maryam Kamal, Neda

More information

Comparisons of Packet Scheduling Algorithms for Fair Service among Connections on the Internet

Comparisons of Packet Scheduling Algorithms for Fair Service among Connections on the Internet Comparsons of Packet Schedulng Algorthms for Far Servce among Connectons on the Internet Go Hasegawa, Takahro Matsuo, Masayuk Murata and Hdeo Myahara Department of Infomatcs and Mathematcal Scence Graduate

More information

Available online at ScienceDirect. Procedia Computer Science 103 (2017 )

Available online at  ScienceDirect. Procedia Computer Science 103 (2017 ) Avalable onlne at www.scencedrect.com ScenceDrect Proceda Computer Scence 03 (207 ) 562 568 XIIth Internatonal Symposum «Intellgent Systems», INTELS 6, 5-7 October 206, Moscow, Russa Retral queueng systems

More information

Fibre-Optic AWG-based Real-Time Networks

Fibre-Optic AWG-based Real-Time Networks Fbre-Optc AWG-based Real-Tme Networks Krstna Kunert, Annette Böhm, Magnus Jonsson, School of Informaton Scence, Computer and Electrcal Engneerng, Halmstad Unversty {Magnus.Jonsson, Krstna.Kunert}@de.hh.se

More information

A KIND OF ROUTING MODEL IN PEER-TO-PEER NETWORK BASED ON SUCCESSFUL ACCESSING RATE

A KIND OF ROUTING MODEL IN PEER-TO-PEER NETWORK BASED ON SUCCESSFUL ACCESSING RATE A KIND OF ROUTING MODEL IN PEER-TO-PEER NETWORK BASED ON SUCCESSFUL ACCESSING RATE 1 TAO LIU, 2 JI-JUN XU 1 College of Informaton Scence and Technology, Zhengzhou Normal Unversty, Chna 2 School of Mathematcs

More information

Channel-Quality Dependent Earliest Deadline Due Fair Scheduling Schemes for Wireless Multimedia Networks

Channel-Quality Dependent Earliest Deadline Due Fair Scheduling Schemes for Wireless Multimedia Networks Channel-Qualty Dependent Earlest Deadlne Due Far Schedulng Schemes for Wreless Multmeda Networks Ahmed K. F. Khattab Khaled M. F. Elsayed ahmedkhattab@eng.cu.edu.eg khaled@eee.org Department of Electroncs

More information

Achievable Bandwidth Estimation for Stations in Multi-Rate IEEE WLAN Cells

Achievable Bandwidth Estimation for Stations in Multi-Rate IEEE WLAN Cells Achevable Bandwdth Estmaton for Statons n Mult-Rate IEEE 802. WLAN Cells Eduard Garca, Davd Vamonte, Rafael Vdal and Josep Paradells Wreless Networks Group - echncal Unversty of Catalona (UPC) {eduardg,

More information

Performance Analysis of a Reconfigurable Shared Memory Multiprocessor System for Embedded Applications

Performance Analysis of a Reconfigurable Shared Memory Multiprocessor System for Embedded Applications J. ICT Res. Appl., Vol. 7, No. 1, 213, 15-35 15 Performance Analyss of a Reconfgurable Shared Memory Multprocessor System for Embedded Applcatons Darcy Cook 1 & Ken Ferens 2 1 JCA Electroncs, 118 Kng Edward

More information

Computer Communications

Computer Communications Computer Communcatons 3 (22) 3 48 Contents lsts avalable at ScVerse ScenceDrect Computer Communcatons journal homepage: www.elsever.com/locate/comcom On the queueng behavor of nter-flow asynchronous network

More information

Load Balancing for Hex-Cell Interconnection Network

Load Balancing for Hex-Cell Interconnection Network Int. J. Communcatons, Network and System Scences,,, - Publshed Onlne Aprl n ScRes. http://www.scrp.org/journal/jcns http://dx.do.org/./jcns.. Load Balancng for Hex-Cell Interconnecton Network Saher Manaseer,

More information

Biostatistics 615/815

Biostatistics 615/815 The E-M Algorthm Bostatstcs 615/815 Lecture 17 Last Lecture: The Smplex Method General method for optmzaton Makes few assumptons about functon Crawls towards mnmum Some recommendatons Multple startng ponts

More information

Quantifying Performance Models

Quantifying Performance Models Quantfyng Performance Models Prof. Danel A. Menascé Department of Computer Scence George Mason Unversty www.cs.gmu.edu/faculty/menasce.html 1 Copyrght Notce Most of the fgures n ths set of sldes come from

More information

WITH rapid improvements of wireless technologies,

WITH rapid improvements of wireless technologies, JOURNAL OF SYSTEMS ARCHITECTURE, SPECIAL ISSUE: HIGHLY-RELIABLE CPS, VOL. 00, NO. 0, MONTH 013 1 Adaptve GTS Allocaton n IEEE 80.15.4 for Real-Tme Wreless Sensor Networks Feng Xa, Ruonan Hao, Je L, Naxue

More information

Fast Retransmission of Real-Time Traffic in HIPERLAN/2 Systems

Fast Retransmission of Real-Time Traffic in HIPERLAN/2 Systems Fast Retransmsson of Real-Tme Traffc n HIPERLAN/ Systems José A Afonso and Joaqum E Neves Department of Industral Electroncs Unversty of Mnho, Campus de Azurém 4800-058 Gumarães, Portugal {joseafonso,

More information

Virtual Machine Migration based on Trust Measurement of Computer Node

Virtual Machine Migration based on Trust Measurement of Computer Node Appled Mechancs and Materals Onlne: 2014-04-04 ISSN: 1662-7482, Vols. 536-537, pp 678-682 do:10.4028/www.scentfc.net/amm.536-537.678 2014 Trans Tech Publcatons, Swtzerland Vrtual Machne Mgraton based on

More information

CMPS 10 Introduction to Computer Science Lecture Notes

CMPS 10 Introduction to Computer Science Lecture Notes CPS 0 Introducton to Computer Scence Lecture Notes Chapter : Algorthm Desgn How should we present algorthms? Natural languages lke Englsh, Spansh, or French whch are rch n nterpretaton and meanng are not

More information

Term Weighting Classification System Using the Chi-square Statistic for the Classification Subtask at NTCIR-6 Patent Retrieval Task

Term Weighting Classification System Using the Chi-square Statistic for the Classification Subtask at NTCIR-6 Patent Retrieval Task Proceedngs of NTCIR-6 Workshop Meetng, May 15-18, 2007, Tokyo, Japan Term Weghtng Classfcaton System Usng the Ch-square Statstc for the Classfcaton Subtask at NTCIR-6 Patent Retreval Task Kotaro Hashmoto

More information

Network Coding as a Dynamical System

Network Coding as a Dynamical System Network Codng as a Dynamcal System Narayan B. Mandayam IEEE Dstngushed Lecture (jont work wth Dan Zhang and a Su) Department of Electrcal and Computer Engneerng Rutgers Unversty Outlne. Introducton 2.

More information

Delay Variation Optimized Traffic Allocation Based on Network Calculus for Multi-path Routing in Wireless Mesh Networks

Delay Variation Optimized Traffic Allocation Based on Network Calculus for Multi-path Routing in Wireless Mesh Networks Appl. Math. Inf. Sc. 7, No. 2L, 467-474 2013) 467 Appled Mathematcs & Informaton Scences An Internatonal Journal http://dx.do.org/10.12785/ams/072l13 Delay Varaton Optmzed Traffc Allocaton Based on Network

More information

Query Clustering Using a Hybrid Query Similarity Measure

Query Clustering Using a Hybrid Query Similarity Measure Query clusterng usng a hybrd query smlarty measure Fu. L., Goh, D.H., & Foo, S. (2004). WSEAS Transacton on Computers, 3(3), 700-705. Query Clusterng Usng a Hybrd Query Smlarty Measure Ln Fu, Don Hoe-Lan

More information

SRB: Shared Running Buffers in Proxy to Exploit Memory Locality of Multiple Streaming Media Sessions

SRB: Shared Running Buffers in Proxy to Exploit Memory Locality of Multiple Streaming Media Sessions SRB: Shared Runnng Buffers n Proxy to Explot Memory Localty of Multple Streamng Meda Sessons Songqng Chen,BoShen, Yong Yan, Sujoy Basu, and Xaodong Zhang Department of Computer Scence Moble and Meda System

More information

Related-Mode Attacks on CTR Encryption Mode

Related-Mode Attacks on CTR Encryption Mode Internatonal Journal of Network Securty, Vol.4, No.3, PP.282 287, May 2007 282 Related-Mode Attacks on CTR Encrypton Mode Dayn Wang, Dongda Ln, and Wenlng Wu (Correspondng author: Dayn Wang) Key Laboratory

More information

Evaluation of Parallel Processing Systems through Queuing Model

Evaluation of Parallel Processing Systems through Queuing Model ISSN 2278-309 Vkas Shnde, Internatonal Journal of Advanced Volume Trends 4, n Computer No.2, March Scence - and Aprl Engneerng, 205 4(2), March - Aprl 205, 36-43 Internatonal Journal of Advanced Trends

More information

2x x l. Module 3: Element Properties Lecture 4: Lagrange and Serendipity Elements

2x x l. Module 3: Element Properties Lecture 4: Lagrange and Serendipity Elements Module 3: Element Propertes Lecture : Lagrange and Serendpty Elements 5 In last lecture note, the nterpolaton functons are derved on the bass of assumed polynomal from Pascal s trangle for the fled varable.

More information

Constructing Minimum Connected Dominating Set: Algorithmic approach

Constructing Minimum Connected Dominating Set: Algorithmic approach Constructng Mnmum Connected Domnatng Set: Algorthmc approach G.N. Puroht and Usha Sharma Centre for Mathematcal Scences, Banasthal Unversty, Rajasthan 304022 usha.sharma94@yahoo.com Abstract: Connected

More information

Assignment # 2. Farrukh Jabeen Algorithms 510 Assignment #2 Due Date: June 15, 2009.

Assignment # 2. Farrukh Jabeen Algorithms 510 Assignment #2 Due Date: June 15, 2009. Farrukh Jabeen Algorthms 51 Assgnment #2 Due Date: June 15, 29. Assgnment # 2 Chapter 3 Dscrete Fourer Transforms Implement the FFT for the DFT. Descrbed n sectons 3.1 and 3.2. Delverables: 1. Concse descrpton

More information

Shared Running Buffer Based Proxy Caching of Streaming Sessions

Shared Running Buffer Based Proxy Caching of Streaming Sessions Shared Runnng Buffer Based Proxy Cachng of Streamng Sessons Songqng Chen, Bo Shen, Yong Yan, Sujoy Basu Moble and Meda Systems Laboratory HP Laboratores Palo Alto HPL-23-47 March th, 23* E-mal: sqchen@cs.wm.edu,

More information

Subspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points;

Subspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points; Subspace clusterng Clusterng Fundamental to all clusterng technques s the choce of dstance measure between data ponts; D q ( ) ( ) 2 x x = x x, j k = 1 k jk Squared Eucldean dstance Assumpton: All features

More information

Efficient Broadcast Disks Program Construction in Asymmetric Communication Environments

Efficient Broadcast Disks Program Construction in Asymmetric Communication Environments Effcent Broadcast Dsks Program Constructon n Asymmetrc Communcaton Envronments Eleftheros Takas, Stefanos Ougaroglou, Petros copoltds Department of Informatcs, Arstotle Unversty of Thessalonk Box 888,

More information

An Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices

An Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices Internatonal Mathematcal Forum, Vol 7, 2012, no 52, 2549-2554 An Applcaton of the Dulmage-Mendelsohn Decomposton to Sparse Null Space Bases of Full Row Rank Matrces Mostafa Khorramzadeh Department of Mathematcal

More information

Determining the Optimal Bandwidth Based on Multi-criterion Fusion

Determining the Optimal Bandwidth Based on Multi-criterion Fusion Proceedngs of 01 4th Internatonal Conference on Machne Learnng and Computng IPCSIT vol. 5 (01) (01) IACSIT Press, Sngapore Determnng the Optmal Bandwdth Based on Mult-crteron Fuson Ha-L Lang 1+, Xan-Mn

More information

CS 268: Lecture 8 Router Support for Congestion Control

CS 268: Lecture 8 Router Support for Congestion Control CS 268: Lecture 8 Router Support for Congeston Control Ion Stoca Computer Scence Dvson Department of Electrcal Engneerng and Computer Scences Unversty of Calforna, Berkeley Berkeley, CA 9472-1776 Router

More information