Slow Fading Channel Selection: A Restless Multi-Armed Bandit Formulation

Size: px
Start display at page:

Download "Slow Fading Channel Selection: A Restless Multi-Armed Bandit Formulation"

Transcription

1 Slow Fadng Channel Selecton: A Restless Mult-Armed Bandt Formulaton Konstantn Avrachenkov INRIA, Maestro Team BP95, Sopha Antpols, France Emal: k.avrachenkov@sopha.nra.fr Laura Cottatellucc, Lorenzo Magg Eurecom Moble Communcatons Department BP193, F Sopha Antpols, France Emal: {laura.cottatellucc,lorenzo.magg}@eurecom.fr Abstract We deal wth a mult-access wreless network n whch transmtters dynamcally select a frequency band to communcate on. The slow fadng channel attenuatons follow an autoregressve model. In the sngle user case, we formulate ths selecton problem as a restless mult-armed bandt problem and we propose two strateges to dynamcally select a band at each tme slot. Our objectve s to maxmze the SNR n the long run. Each of these strateges s close to the optmal strategy n dfferent regmes. In the general case wth several users, we formulate the problem as a stochastc game wth uncountable state space, where the objectve s the SINR. Then we propose two strateges to approxmate the best response polcy for one user when the other users strategy s fxed. I. INTRODUCTION Next generaton of wreless networks s expected to be characterzed by a hgh decentralzaton/dstrbuton of control functons among nodes to support self-organzng and selfhealng capabltes. Network devces shall be able to montor and sense the surroundngs, learn from ther montorng and smartly and dynamcally allocate resources. Ths perspectve scenaro s attractng a consderable amount of research efforts to develop learnng technques able to optmze the tradeoff between exploraton and explotaton of envronment and resources. A relevant class of learnng algorthms s the Mult- Armed Bandt (MAB) one. In the classc MAB problem there exst several arms that offer a reward when pulled (n analogy wth gamblng on bandts n casnos). Each arm s assocated wth a Markov process, and the reward of an arm s a functon of ts state. Gttns provded n [1] a dynamc allocaton procedure, then dubbed Gttns ndex, whch s optmal f the arms that are not pulled do not evolve over tme. The more general case when the arms that are not pulled keep evolvng n tme s known as Restless MAB. It was proven by Papadmtrou and Tstskls n [2] that restless MAB are PSPACE-hard n general. In [3], Whttle proposed to adopt a heurstc Lagrangan relaxaton to extend the Gttns ndex to the restless case, whch s asymptotcally optmal under certan lmtng regme [4]. In ths work, we consder a wreless network where transmtters can select a frequency band from a shared pool to communcate on. The evoluton of the slow fadng channel attenuaton assocated to each frequency band and each transmtter s a random process that can be well approxmated by an autoregressve process [5]. We assume that all such random processes are ndependent of each other. The goal of each transmtter s to maxmze ts average Sgnal to Interference and Nose Rato (SINR) n the long run. To get nsght nto ths problem, frst we focus on a sngle transmtter system to nvestgate the exploraton-explotaton trade-off for the randomness ntroduced n the system by the autoregressve channel attenuatons. Then, we consder the mult-transmtter case where the problem s further complcated by the randomness ntroduced by the autonomous band selectons of multple transmtters. For the sngle termnal case, the problem of dynamc frequency allocaton for SNR maxmzaton can be modeled as a restless Mult Armed Bandt (MAB) snce the transmtter only knows the nstantaneous attenuatons on the bands utlzed n the past and they evolve also when not utlzed. To the best of the authors knowledge, there are no avalable results on the MAB problem for autoregressve processes. We propose two heurstc frequency allocaton strateges, one called myopc and the other randomzed. When the AR processes possess smlar autocorrelaton functons, we suggest to use the myopc strategy. Instead, when there s one AR process havng a much hgher autocorrelaton, we suggest to use the randomzed strategy. In the scenaro wth multple transmtters the problem s formulated as a stochastc game wth uncountable state space. We focus on a two-user system and we assume that user 1 s oblvous of the presence of user 2 and follows a plan sngle-user myopc approach. Then we propose two strateges for user 2 to approxmate ts best response aganst user 1 s strategy. Agan, one strategy s myopc and the other s randomzed, wth respect to the SINR objectve functon. A lexcal remark. We say that we sample a frequency band when we utlze t for the communcaton n a certan tme slot. II. MODEL In Secton III we consder one transmtters and one recever, whle n Secton IV we deal wth a model wth two transmtters. Tme s dvded nto slots and, at the begnnng of a tme slot, each transmtter (or user) selects a frequency band, out of a pool of M dfferent ones, to transmt. At the recever, a sngle-user decoder per transmtter s deployed. In the two-transmtter case, when a both users access the same frequency band at tme slot t, they nterfere wth each other, and the SINR (Sgnal to Interference plus Nose Rato) for user j = 1, 2 at tme t s SINR,j [t] = P j h,j [t] 2 N 0 + q j P q h,q [t] 2

2 where P j s the transmt power of user j, h,j [t] C s the -th channel coeffcent of user j at tme t and N 0 s the varance of the addtve whte Gaussan nose at the recever. When only one user s present, the SINR defnton bols down to the classc SNR. For smplcty of notaton, henceforth we wll denote the channel attenuaton coeffcent h,j [t] 2 as g,j [t]. Let us descrbe now our channel model. In [5] t s shown that, under slow fadng condtons, the SNR (Sgnal to Nose Rato) of ndoor wreless channels can be well approxmated by an autoregressve (AR) model. Ths means that, under such condtons, we can model the channel attenuatons as g,j [t] = p,j k=1 ϕ (k),j g,j[t k] + c,j + ϵ,j [t] where ϕ,j R, {ϵ,j [t]} t s an..d. Gaussan process wth zero mean and varance σ,j 2, c,j > 0, and p,j s the order of the model. Moreover, all the channels consdered are ndependent of each other,.e. ϵ 1,j 1 [t] s ndependent of ϵ 2,j 2 [t] when ether 1 2 or j 1 j 2. We assume the AR process to be wde sense statonary (WSS),.e. the roots of the polynomal z p p k=1 ϕ(k),j zp k must le nsde the unt crcle. III. SINGLE USER: MDP FORMULATION In ths secton we consder the sngle user case. In order to smplfy the notaton, we drop the user ndex. In our study we consder an AR(1) channel attenuaton model,.e. g [t] = ϕ g [t 1] + c + ϵ [t] For ϕ < 1, the process s WSS, and the (uncondtoned) expected value of channel attenuaton g [t] at any tme nstant t can be expressed as m = E(g [t]) = c t. 1 ϕ Therefore we can say that E(SNR [t]) = P m /N 0, for all t. Straghtforward computatons show that the autocovarance functon of the channel attenuaton can be wrtten as E ( (g [t] m )(g [t n] m ) ) = ϕ n σ 2 1 ϕ 2. (1) We now llustrate the two fundamental assumptons of ths paper. Frst, the coeffcents ϕ and σ are known by the transmtter, whch mght have estmated them durng a tranng phase. Second, the transmtter, at tme t, only knows the nstantaneous attenuatons of the frequency bands utlzed up to tme t 1. Indeed, we assume that the recever estmates g,j and broadcast ths nformaton on the channel along wth an dentfer for the transmtter and the frequency band. The goal of the user s to dynamcally swtch among the channels at each tme slot n order to maxmze the expected average SNR over an nfnte horzon. Equvalently, t wants to maxmze the expected average over tme of channel attenuatons, denoted by r(π): T 1 1 } max π {r(π) = lm E π (g π(t) [t]) T T t=0 (2) where π s a dynamc samplng strateges over the channels 1,..., M. The reader should notce that a channel samplng strategy π at tme t may depend on the whole hstory of the observed channels and of the samplng decsons up to tme t. Ths class also ncludes statc strateges, that choose one channel once for all. Intutvely, when there exsts a channel wth much lower uncondtoned expected attenuaton,.e. m m k for all k, a statc selecton of the channel s the nearly optmal strategy, snce wth hgh probablty g [t] > g k [t] for all k for almost all t. In ths secton, we want to study how to dynamcally select the band on whch to transmt when, a pror, all of them are nearly equvalent,.e. there exsts m m k, for all k. At each tme slot, there s always one channel better than the others, hence we wsh to track dynamcally the evoluton over tme of the best channel. Intutvely, the samplng choce at each nstant has to be a trade-off between exploraton and explotaton. To gve a hnt, the most natural polcy, that we wll call myopc, at each tme step t ams at maxmzng the expected value of SNR[t], gven all the prevous channel observatons. On the other hand, the statcal nformaton about channels that are not used becomes more and more obsolete, therefore n some cases t mght be better to explore dfferent channels wth a randomzed strategy. We can formulate the optmzaton problem (2) as a restless Mult Armed Bandt problem (MAB for short), n whch a user at each tme nstant t selects an arm (here, frequency band) whch gves a reward (here, the SNR) and all the arms, ncludng the ones that have not been selected, evolve accordng to a certan stochastc process (here, an autoregressve process). More specfcally, we can descrbe the decson problem at hand as a Markov Decson Process (MDP) wth an uncountable set of states S or, equvalently, as a Partally Observable MDP. Let us descrbe t n detal. At tme t, we call n (t) the number of steps ago n whch channel has been last used. The attenuaton of channel at tme t condtoned on ts last observaton s a Gaussan r.v., and we denote ts mean and varance as µ (t) and ν (t), respectvely: µ (t) = E ( g [t] g [t n (t)] ) = ϕ n (t) 1 ϕ n (t) g [t n (t)] + c (3) 1 ϕ ν (t) = Var ( g [t] g [t n (t)] ) = σ 2 1 ϕ 2n (t) 1 ϕ 2 where g [t n (t)] s the attenuaton of channel durng ts last utlzaton. At tme step t, thanks to the Markov property of the AR(1) process, the whole statstcal nformaton about channel s hence contaned n (µ (t), ν (t)). We observe that µ R, whle ν s bounded between [σ 2; σ2 /(1 ϕ2 )]. The decson on whch channel to utlze at tme t hnges on the set S[t]: S[t] = {µ 1 (t), ν 1 (t), µ 2 (t), ν 2 (t),..., µ M (t), ν M (t)}. (5) By utlzng the MDP jargon [6], we call by S[t] the state of the decson problem at tme t. The state space S s the uncountable collecton of all the possble states. In each state S S, a set of actons A = {1, 2,..., M} s avalable to (4)

3 the transmtter, whch represents the collecton of channels that can be selected at tme slot t. If channel s selected, then we map the reward for the user n state S[t] to the expected channel attenuaton at tme t condtoned on the last observaton of channel tself,.e. µ (t). The state of the system at tme t + 1 evolves stochastcally, accordng to the followng Markovan rule. If channel s selected at tme t, then at tme t + 1, µ (t + 1) = ϕ Y + c, where Y N ( µ (t), ν (t) ) ν (t + 1) = σ 2. Instead f channel s not selected at tme t, µ (t + 1) = ϕ µ (t) + c ν (t + 1) = ϕ 2 ν (t) + σ 2. A. Heurstc algorthms The theory of MDP allows us to clam that there exsts an optmal statonary strategy π O for the problem (2). Unfortunately, the computaton of π O turns out to be a dffcult task. Indeed the soluton to a Markov Decson Problem wth uncountable state can only be approxmated by means of dscretzaton algorthms [6], and even n ths case the curse of dmensonalty entals that the sze of the dscretzed state space ncreases exponentally wth the number of arms. A dfferent approach would be to compute the Whttle ndex [3] of each channel, but ths approach s not guaranteed to be optmal. Hence, t becomes crucal to devse a smple polcy whose performance s reasonably close to the optmal r(π O ). In the followng we propose the most natural statonary strategy one can thnk of,.e. the myopc polcy π M that ams at maxmzng the nstantaneous expected SNR n each state. Such a polcy does not take nto account that the statstcs of the channel that have not been selected for a long perod mght become too stale. Frst, we need to ntalze the algorthm, and we choose to sample the coeffcent of each channel once. Algorthm 1. Myopc polcy π M. For 0 t M 1 select channel t,.e. π M (S[t]) = t + 1. For t M, π M (S[t]) = argmax µ (t). {1,...,M} We ntend to compare the performance of the myopc polcy wth a more sophstcated one, that we call randomzed strategy and s nspred by the Thompson samplng strategy for Bayesan Mult Armed Bandt problems [7]. We suggests to draw, n each state S[t], one realzaton of the random varable ξ = g [t] g [t n (t)], for each channel = 1,..., M. Then, the arm correspondng to the hghest realzaton of ξ s chosen. Ths procedure does not always follow the myopc rule, but wth a certan probablty explore the arms that, though possessng a lower µ, mght be optmal snce ther last observaton s too stale. Algorthm 2. Randomzed polcy π R. For 0 t M 1 select channel t,.e. π R (S[t]) = t + 1. For t M, draw a realzaton of the Gaussan varable ξ N (µ (t), ν ()) for all = 1,..., M. Select π R (S[t]) = argmax ξ. =1,...,M B. Smulatons In ths secton we show the results of some smulatons, gvng a hnt about the performance of the myopc and the randomzed polces, descrbed respectvely n Algorthm 1 and 2. Gven a statonary polcy π, we want to assess ts average reward r(π). We compare the myopc and randomzed polces wth ) the optmal polcy π O, approxmated by means of a state dscretzaton technque [6], wth ) the upper bound for the performance of any strategy, computed by selectng the channel wth the hghest coeffcent g at each tme step: π U (t) = argmax g [t], t 0 (6) =1,...,M and wth ) the statc polcy π S, that selects off-lne the arm wth the hghest expected value, and no longer swtches to other channels,.e. π S [t] = argmax m, t 0. =1,...,M Of course, the strategy π U s not applcable, snce t s not causal. In theory, ts performance s achevable only when the channels are determnstc hence perfectly predctable,.e. σ = 0 for all = 1,..., M. We now show the performance of the fve polces under scrutny, the myopc π M, the randomzed π R, the statc π S, the optmal π O, and the upper bound polcy π U, under dfferent channel condtons. Frst, we consder 3 arms, where arms 2,3 are statstcally equvalent, and ϕ 2 = ϕ 3 = 0.3, σ2 2 = σ3 2 = 1, and m 2 = m 3 = 8. Arm 1 has the same coeffcents ϕ 1 = 0.3, σ1 2 = 1 as arms 2,3. In Fgure 1 we show the performance of the fve polces when m 1 vares wthn [7; 9]. We see that, under these condtons, the myopc polcy outperforms the randomzed one snce the latter wastes too much tme n explorng arms that are not optmal. As ntuton confrms, the statc polcy π S performs as well as the myopc π M when arm 1 has the hghest expected value m 1 > m 2 = m = 3. Instead, for m 1 < m 2 = m 3, dynamcally swtchng between the arms 2,3 s benefcal wth respect to statcally selectng one of the two. As we see n Fgure 1, when all the arms are characterzed by the same uncondtoned expectaton,.e. m = 8, for = 1, 2, 3, the statc polcy π S s outperformed by both the myopc and the randomzed strateges. It s ndeed better to swtch among the channels to attempt to track the best nstantaneous channel at each tme nstant, based on the prevous observatons. Remarkably, the performance of the myopc polcy π M s close to the optmal π O. Hence, we evaluate our algorthms n a dfferent scenaro, n whch the value m s are the same for all the channels, but there exsts one channel (say, 1) whose autocovarance functon (1) decays consderably more slowly than the others. It s clear from Fgure 2 that there are lapses n whch channel 1 s by far the best, and some others n whch ts channel coeffcent g 1 plummets below the others. From Fgure 2 we observe that the myopc strategy often fals to track channel 1 when t s the best. The reason s qute ntutve: durng the lapse n whch channel 1 s the worst one, the myopc strategy does not choose t, then ts last observaton become obsolete, and consequently the predcton µ 1 (t) tends to m 1 = 10. Thus, t s hghly

4 r(π) r(π U ) upper bound r(π O ) optmal r(π M ) myopc r(π R ) randomzed r(π S ) statc arm m 1 Fgure 1. Performance of myopc and randomzed algorthm wth 3 arms (channels). Arms 2 and 3 are statstcally equvalents, wth ϕ 2 = ϕ 3 = 0.3, σ2 2 = σ2 3 = 1, and m 2 = m 3 = 8. Arm 1 has the same ϕ 1 = 0.3, σ1 2 = 1 as arms 2,3, whle the performance of the proposed algorthms are assessed when m 1 vares wthn [7; 9]. h[t] arm 1 arm 2 arm 3 π R randomzed π M myopc t Fgure 2. Channel (or arm) selecton when c = 10 for all, ϕ 1 = 0.9, ϕ 2 = ϕ 3 = 0.3, σ1 2 = 1.5, σ2 2 = σ2 3 = 0.5. The randomzed strategy succeeds n trackng the frst channel wth hgher autocorrelaton, when t s the best one. probable (and ths probablty ncreases wth M) that one of the other, suboptmal, channels, havng a fresher observaton, offers a hgher predcton. It easly follows that, for ts nherent features, the randomzed polcy s more sutable to such knd of stuatons, as results n Fgure 3 confrm. We consdered 3 arms (frequency bands). Arms 2 and 3 are statstcally equvalents, wth ϕ 2 = ϕ 3 = 0.3, σ 2 2 = σ 2 3 = 1, c 2 = c 3 = 10. Arm 1 has the same coeffcents c 1 = 10, σ 2 1 = 1 as arms 2,3, whle the performance of the proposed algorthms are assessed when the coeffcent ϕ 1 vares wthn [0.3; 0.98]. As we ntutvely explaned before, when the coeffcent ϕ 1 s suffcently hgh,.e. ϕ > 0.85, the randomzed strategy outperforms the myopc one. Notably, the myopc polcy s quas-optmal for ϕ 1 < 0.6, whle the the randomzed one s nearly optmal for ϕ 1 > 0.9. IV. MULTI USER: STOCHASTIC GAME FORMULATION In ths secton we dscuss the more general scenaro descrbed n Secton II, n whch two transmtters dynamcally select one among M channels at each tme slot. If some users choose the same channel n one tme slot, they nterfere wth each other. Therefore, the objectve functon for each user s ts SINR, and no longer ts SNR. Snce n the sngle user case the decson process can be descrbed as an MDP, then the scenaro wth two users can be formalzed as a stochastc game, also called compettve MDP [8], wth uncountable state space. In general, a stochastc game s an MDP n whch the nstantaneous rewards for each player and the transton probabltes among the states are controlled by the jont actons of the players n each state. In our case, the set of channels h 1,j,..., h M,j for player j evolve ndependently from the ones avalable to any other player k j, and the acton space for each player s stll A = {1,..., M},.e. the channel ndces to be selected at each slot. Therefore, we are allowed to formulate the game as a stochastc game n whch each user j controls ts own Markov chan on the state space S j. As n the sngle user case S j s the set of all the possble states (5). Formally, the state space of the stochastc game at hand s the Cartesan product S = S 1 S 2. Let us denote by π j a samplng strategy for user j and by π j the one for the other users. Possbly, π j, π j are randomzed polces. We defne the nstantaneous reward for user j n state S [t] S as the expected reward E ( SINR j [t] S [t], π, π ). Thus, the nteracton on the players occurs only on the nstantaneous rewards ganed n each state, through the SINR expresson. Thus we can say that our model s a rewardcoupled stochastc game. Ths model s very smlar wth the one dealt wth n [9], except that here the state space s uncountable and there are no constrants on the rewards. A. Heurstc Best Response We now propose a heurstc best response polcy for user 2. Suppose that user 1 s oblvous of the presence of user 2 and performs a myopc polcy π1 M to maxmze the expected average of channel attenuatons over tme, as n the sngle user case. On the other hand, user 2 knows the parameters of the channels, the current state, and the strategy of user 1. Thus, user 2 stll faces an MDP wth uncountable states, whch s equvalent to the stochastc game descrbed before, when user 1 fxes ts own statonary strategy. Let us gve an nsght on a possble strategy for user 2. Assume that, for user 2, channel 1 presents at tme t the hghest coeffcent g 1,2[t], but the expected SINR guaranteed by channel 2 wth suboptmal attenuaton s hgher, snce the nterference s much weaker. Then, t s n general not clear what user 2 should do. A myopc soluton would suggest to swtch to the free channel 2, but on the other hand, n such a way the nformaton about channel 1 becomes stale, and moreover channel 1 tself mght become free n a near future. Then, n analogy wth the sngle player case, we propose two strateges, one myopc and one randomzed, to approxmate the best response for user 2

5 r(π U ) upper bound r(π O ) optmal r(π M ) myopc r(π R ) randomzed r(π S ) statc arm upper bound best response r 2 (π 1 M,π 2 MS ) SINR myopc r 2 (π 1 M,π 2 RS ) SINR randomzed r(π) r(π) φ φ 1,2 Fgure 3. Performance of myopc and randomzed algorthm wth 3 arms (frequency bands). Arms 2 and 3 are statstcally equvalents, wth ϕ 2 = ϕ 3 = 0.3, σ2 2 = σ2 3 = 1, c 2 = c 3 = 10. Arm 1 has the same coeffcents c 1 = 10, σ1 2 = 1 as arms 2,3. ϕ 1 vares wthn [0.3; 0.98]. aganst a myopc polcy π1 M that user 1 mplements regardless of user 2 s behavor. We suppose the algorthms are ntalzed by samplng each channel once. Algorthm 3. SINR myopc polcy π MS 2 for user 2, aganst myopc polcy π M 1 for user 1. π2 MS (S [t], π1 M ) = argmax E(SINR,2 [t] S [t], π1 M {1,...,M}, ). Algorthm 4. Randomzed polcy π2 RS for user 2, aganst myopc polcy π1 M for user 1. Draw a realzaton of the random varable ξ = SINR,2 [t] (S [t], π1 M, ), for all = 1,..., M. Select π2 RS (S [t], π1 M ) = argmax ξ. =1,...,M About the performance of polces π MS, π RS, we can do smlar consderatons to the one made for the myopc and randomzed algorthms n the sngle user case. Let us explan the results llustrated n Fgure 4. We consdered 2 users and 2 channels. The nose varance s N 0 = 1 and P 1 = P 2 = 1. The channels for user 1 are almost determnstc,.e. σ1,1 2 = σ2,1 2 = 0.1 and ϕ 1,1 = ϕ 2,1 = 0.3, m 1,1 = 2, m 2,1 = 0.5. Thus user 1, that s unaware of the presence of user 2 and adopts a myopc polcy π1 M, selects channel 1 almost always. For user 2, σ1,2 2 = 0.8, σ2,2 2 = 0.4, m 1,2 = 8, m 2,2 = 3, ϕ 2,2 = 0.3. Hence, a statc strategy for user 2 would suggest not to collde and to select channel 2. Anyway, sometmes t s benefcal for user 2 to select channel 1 when ths s good enough. Indeed, for values of ϕ 1,2 approachng 1, the autocorrelaton of channel 1 for user 2 ncreases, and the randomzed polcy π RS succeeds n trackng channel 1 n the tme slots n whch ts coeffcent g s large enough to overwhelm the nterference caused by user 1. V. CONCLUSIONS We proposed two strateges to dynamcally select one out of a pool of M slow fadng channels, modeled as autoregressve Fgure 4. Best response strategy of user 2 aganst a myopc polcy for user 1. For user 1, σ1,1 2 = σ2 2,1 = 0.1 and ϕ 1,1 = ϕ 2,1 = 0.3, m 1,1 = 2, m 2,1 = 0.5. For user 2, σ1,2 2 = 0.8, σ2 2,2 = 0.4, m 1,2 = 8, m 2,2 = 3, ϕ 2,2 = 0.3. ϕ 1,2 vares wthn [0.8; 0.98]. r 2 (π1 M, π 2) s the expected long run average SINR for user 2 when user 1 adopts strategy π1 M. processes of order 1. The decson process s modeled as a restless bandt, or equvalently as a Markov Decson Process. The myopc channel selecton strategy s nearly optmal when the channels are smlarly correlated. Instead we suggest to adopt a randomzed strategy when one channel shows hgher autocorrelaton. When two users are present, they nterfere wth each other, and we model the compettve learnng process as a stochastc game. We fnally propose two ways to approxmate a best response selecton strategy for the transmtters. Acknowledgments: Ths research was supported by Agence Natonale de la Recherche, wth reference ANR-09- VERS-001, and Orange France Telecom Grant on Content- Centrc Networkng. We would lke to thank Alexey Punovsky for very helpful dscusson. REFERENCES [1] J. C. Gttns, R. Weber, and K. D. Glazebrook, Mult-armed bandt allocaton ndces. Wley Onlne Lbrary, 1989, vol. 25. [2] C. H. Papadmtrou and J. N. Tstskls, The complexty of optmal queueng network control, Mathematcs of Operatons Research, vol. 24, [3] P. Whttle, Restless bandts: Actvty allocaton n a changng world, Journal of appled probablty, pp , [4] R. Weber and G. Wess, On an ndex polcy for restless bandts, Journal of Appled Probablty, pp , [5] R. Aguero, M. Garca, and L. Mufoz, BEAR: A bursty error autoregressve model for ndoor wreless envronments, n Personal, Indoor and Moble Rado Communcatons, PIMRC IEEE 18th Internatonal Symposum on. IEEE, 2007, pp [6] N. Bäuerle and U. Reder, Markov Decson Processes wth applcatons to fnance. Sprnger Verlag, [7] W. Thompson, On the lkelhood that one unknown probablty exceeds another n vew of the evdence of two samples, Bometrka, vol. 25, no. 3/4, pp , [8] J. Flar and K. Vreze, Compettve Markov decson processes. Sprnger Verlag, [9] E. Altman, K. Avrachenkov, N. Bonneau, M. Debbah, R. El-Azouz, D. Sadoc Menasche, Constraned cost-coupled stochastc games wth ndependent state processes, Operatons Research Letters, vol. 36, no. 2, pp , 2008.

Luby s Alg. for Maximal Independent Sets using Pairwise Independence

Luby s Alg. for Maximal Independent Sets using Pairwise Independence Lecture Notes for Randomzed Algorthms Luby s Alg. for Maxmal Independent Sets usng Parwse Independence Last Updated by Erc Vgoda on February, 006 8. Maxmal Independent Sets For a graph G = (V, E), an ndependent

More information

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ). REVIEW OF RISK MANAGEMENT CONCEPTS LOSS DISTRIBUTIONS AND INSURANCE Loss and nsurance: When someone s subject to the rsk of ncurrng a fnancal loss, the loss s generally modeled usng a random varable or

More information

INVESTIGATION OF VEHICULAR USERS FAIRNESS IN CDMA-HDR NETWORKS

INVESTIGATION OF VEHICULAR USERS FAIRNESS IN CDMA-HDR NETWORKS 21 22 September 2007, BULGARIA 119 Proceedngs of the Internatonal Conference on Informaton Technologes (InfoTech-2007) 21 st 22 nd September 2007, Bulgara vol. 2 INVESTIGATION OF VEHICULAR USERS FAIRNESS

More information

Stochastic Games on a Multiple Access Channel

Stochastic Games on a Multiple Access Channel Stochastc Games on a Multple Access Channel Prashant N and Vnod Sharma Department of Electrcal Communcaton Engneerng Indan Insttute of Scence, Bangalore 560012, Inda Emal: prashant2406@gmal.com, vnod@ece.sc.ernet.n

More information

A Lyapunov Optimization Approach to Repeated Stochastic Games

A Lyapunov Optimization Approach to Repeated Stochastic Games PROC. ALLERTON CONFERENCE ON COMMUNICATION, CONTROL, AND COMPUTING, OCT. 2013 1 A Lyapunov Optmzaton Approach to Repeated Stochastc Games Mchael J. Neely Unversty of Southern Calforna http://www-bcf.usc.edu/

More information

Forecasting the Direction and Strength of Stock Market Movement

Forecasting the Direction and Strength of Stock Market Movement Forecastng the Drecton and Strength of Stock Market Movement Jngwe Chen Mng Chen Nan Ye cjngwe@stanford.edu mchen5@stanford.edu nanye@stanford.edu Abstract - Stock market s one of the most complcated systems

More information

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module LOSSLESS IMAGE COMPRESSION SYSTEMS Lesson 3 Lossless Compresson: Huffman Codng Instructonal Objectves At the end of ths lesson, the students should be able to:. Defne and measure source entropy..

More information

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek HE DISRIBUION OF LOAN PORFOLIO VALUE * Oldrch Alfons Vascek he amount of captal necessary to support a portfolo of debt securtes depends on the probablty dstrbuton of the portfolo loss. Consder a portfolo

More information

Learning the Best K-th Channel for QoS Provisioning in Cognitive Networks

Learning the Best K-th Channel for QoS Provisioning in Cognitive Networks 000 001 002 003 004 005 006 007 008 009 010 011 012 013 014 015 016 017 018 019 020 021 022 023 024 025 026 027 028 029 030 031 032 033 034 035 036 037 038 039 040 041 042 043 044 045 046 047 048 049 050

More information

What is Candidate Sampling

What is Candidate Sampling What s Canddate Samplng Say we have a multclass or mult label problem where each tranng example ( x, T ) conssts of a context x a small (mult)set of target classes T out of a large unverse L of possble

More information

An Alternative Way to Measure Private Equity Performance

An Alternative Way to Measure Private Equity Performance An Alternatve Way to Measure Prvate Equty Performance Peter Todd Parlux Investment Technology LLC Summary Internal Rate of Return (IRR) s probably the most common way to measure the performance of prvate

More information

Recurrence. 1 Definitions and main statements

Recurrence. 1 Definitions and main statements Recurrence 1 Defntons and man statements Let X n, n = 0, 1, 2,... be a MC wth the state space S = (1, 2,...), transton probabltes p j = P {X n+1 = j X n = }, and the transton matrx P = (p j ),j S def.

More information

How To Solve An Onlne Control Polcy On A Vrtualzed Data Center

How To Solve An Onlne Control Polcy On A Vrtualzed Data Center Dynamc Resource Allocaton and Power Management n Vrtualzed Data Centers Rahul Urgaonkar, Ulas C. Kozat, Ken Igarash, Mchael J. Neely urgaonka@usc.edu, {kozat, garash}@docomolabs-usa.com, mjneely@usc.edu

More information

On the Optimal Control of a Cascade of Hydro-Electric Power Stations

On the Optimal Control of a Cascade of Hydro-Electric Power Stations On the Optmal Control of a Cascade of Hydro-Electrc Power Statons M.C.M. Guedes a, A.F. Rbero a, G.V. Smrnov b and S. Vlela c a Department of Mathematcs, School of Scences, Unversty of Porto, Portugal;

More information

CALL ADMISSION CONTROL IN WIRELESS MULTIMEDIA NETWORKS

CALL ADMISSION CONTROL IN WIRELESS MULTIMEDIA NETWORKS CALL ADMISSION CONTROL IN WIRELESS MULTIMEDIA NETWORKS Novella Bartoln 1, Imrch Chlamtac 2 1 Dpartmento d Informatca, Unverstà d Roma La Sapenza, Roma, Italy novella@ds.unroma1.t 2 Center for Advanced

More information

Credit Limit Optimization (CLO) for Credit Cards

Credit Limit Optimization (CLO) for Credit Cards Credt Lmt Optmzaton (CLO) for Credt Cards Vay S. Desa CSCC IX, Ednburgh September 8, 2005 Copyrght 2003, SAS Insttute Inc. All rghts reserved. SAS Propretary Agenda Background Tradtonal approaches to credt

More information

Analysis of Energy-Conserving Access Protocols for Wireless Identification Networks

Analysis of Energy-Conserving Access Protocols for Wireless Identification Networks From the Proceedngs of Internatonal Conference on Telecommuncaton Systems (ITC-97), March 2-23, 1997. 1 Analyss of Energy-Conservng Access Protocols for Wreless Identfcaton etworks Imrch Chlamtac a, Chara

More information

Dynamic Pricing for Smart Grid with Reinforcement Learning

Dynamic Pricing for Smart Grid with Reinforcement Learning Dynamc Prcng for Smart Grd wth Renforcement Learnng Byung-Gook Km, Yu Zhang, Mhaela van der Schaar, and Jang-Won Lee Samsung Electroncs, Suwon, Korea Department of Electrcal Engneerng, UCLA, Los Angeles,

More information

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic Lagrange Multplers as Quanttatve Indcators n Economcs Ivan Mezník Insttute of Informatcs, Faculty of Busness and Management, Brno Unversty of TechnologCzech Republc Abstract The quanttatve role of Lagrange

More information

An Analysis of Central Processor Scheduling in Multiprogrammed Computer Systems

An Analysis of Central Processor Scheduling in Multiprogrammed Computer Systems STAN-CS-73-355 I SU-SE-73-013 An Analyss of Central Processor Schedulng n Multprogrammed Computer Systems (Dgest Edton) by Thomas G. Prce October 1972 Techncal Report No. 57 Reproducton n whole or n part

More information

Frequency Selective IQ Phase and IQ Amplitude Imbalance Adjustments for OFDM Direct Conversion Transmitters

Frequency Selective IQ Phase and IQ Amplitude Imbalance Adjustments for OFDM Direct Conversion Transmitters Frequency Selectve IQ Phase and IQ Ampltude Imbalance Adjustments for OFDM Drect Converson ransmtters Edmund Coersmeer, Ernst Zelnsk Noka, Meesmannstrasse 103, 44807 Bochum, Germany edmund.coersmeer@noka.com,

More information

Properties of Indoor Received Signal Strength for WLAN Location Fingerprinting

Properties of Indoor Received Signal Strength for WLAN Location Fingerprinting Propertes of Indoor Receved Sgnal Strength for WLAN Locaton Fngerprntng Kamol Kaemarungs and Prashant Krshnamurthy Telecommuncatons Program, School of Informaton Scences, Unversty of Pttsburgh E-mal: kakst2,prashk@ptt.edu

More information

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence 1 st Internatonal Symposum on Imprecse Probabltes and Ther Applcatons, Ghent, Belgum, 29 June 2 July 1999 How Sets of Coherent Probabltes May Serve as Models for Degrees of Incoherence Mar J. Schervsh

More information

denote the location of a node, and suppose node X . This transmission causes a successful reception by node X for any other node

denote the location of a node, and suppose node X . This transmission causes a successful reception by node X for any other node Fnal Report of EE359 Class Proect Throughput and Delay n Wreless Ad Hoc Networs Changhua He changhua@stanford.edu Abstract: Networ throughput and pacet delay are the two most mportant parameters to evaluate

More information

Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy

Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy 4.02 Quz Solutons Fall 2004 Multple-Choce Questons (30/00 ponts) Please, crcle the correct answer for each of the followng 0 multple-choce questons. For each queston, only one of the answers s correct.

More information

Can Auto Liability Insurance Purchases Signal Risk Attitude?

Can Auto Liability Insurance Purchases Signal Risk Attitude? Internatonal Journal of Busness and Economcs, 2011, Vol. 10, No. 2, 159-164 Can Auto Lablty Insurance Purchases Sgnal Rsk Atttude? Chu-Shu L Department of Internatonal Busness, Asa Unversty, Tawan Sheng-Chang

More information

RequIn, a tool for fast web traffic inference

RequIn, a tool for fast web traffic inference RequIn, a tool for fast web traffc nference Olver aul, Jean Etenne Kba GET/INT, LOR Department 9 rue Charles Fourer 90 Evry, France Olver.aul@nt-evry.fr, Jean-Etenne.Kba@nt-evry.fr Abstract As networked

More information

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College Feature selecton for ntruson detecton Slobodan Petrovć NISlab, Gjøvk Unversty College Contents The feature selecton problem Intruson detecton Traffc features relevant for IDS The CFS measure The mrmr measure

More information

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis The Development of Web Log Mnng Based on Improve-K-Means Clusterng Analyss TngZhong Wang * College of Informaton Technology, Luoyang Normal Unversty, Luoyang, 471022, Chna wangtngzhong2@sna.cn Abstract.

More information

A Probabilistic Theory of Coherence

A Probabilistic Theory of Coherence A Probablstc Theory of Coherence BRANDEN FITELSON. The Coherence Measure C Let E be a set of n propostons E,..., E n. We seek a probablstc measure C(E) of the degree of coherence of E. Intutvely, we want

More information

Efficient On-Demand Data Service Delivery to High-Speed Trains in Cellular/Infostation Integrated Networks

Efficient On-Demand Data Service Delivery to High-Speed Trains in Cellular/Infostation Integrated Networks IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. XX, NO. XX, MONTH 2XX 1 Effcent On-Demand Data Servce Delvery to Hgh-Speed Trans n Cellular/Infostaton Integrated Networks Hao Lang, Student Member,

More information

NON-CONSTANT SUM RED-AND-BLACK GAMES WITH BET-DEPENDENT WIN PROBABILITY FUNCTION LAURA PONTIGGIA, University of the Sciences in Philadelphia

NON-CONSTANT SUM RED-AND-BLACK GAMES WITH BET-DEPENDENT WIN PROBABILITY FUNCTION LAURA PONTIGGIA, University of the Sciences in Philadelphia To appear n Journal o Appled Probablty June 2007 O-COSTAT SUM RED-AD-BLACK GAMES WITH BET-DEPEDET WI PROBABILITY FUCTIO LAURA POTIGGIA, Unversty o the Scences n Phladelpha Abstract In ths paper we nvestgate

More information

CS 2750 Machine Learning. Lecture 3. Density estimation. CS 2750 Machine Learning. Announcements

CS 2750 Machine Learning. Lecture 3. Density estimation. CS 2750 Machine Learning. Announcements Lecture 3 Densty estmaton Mlos Hauskrecht mlos@cs.ptt.edu 5329 Sennott Square Next lecture: Matlab tutoral Announcements Rules for attendng the class: Regstered for credt Regstered for audt (only f there

More information

Cooperative Load Balancing in IEEE 802.11 Networks with Cell Breathing

Cooperative Load Balancing in IEEE 802.11 Networks with Cell Breathing Cooperatve Load Balancng n IEEE 82.11 Networks wth Cell Breathng Eduard Garca Rafael Vdal Josep Paradells Wreless Networks Group - Techncal Unversty of Catalona (UPC) {eduardg, rvdal, teljpa}@entel.upc.edu;

More information

Optimization of Resource Allocation in Wireless Systems Based on Game Theory

Optimization of Resource Allocation in Wireless Systems Based on Game Theory Internatonal Journal of Computer Scences and Engneerng Open Access Research Paper Volume-4, Issue-1 E-ISSN: 347-693 Optmzaton of Resource Allocaton n Wreless Systems Based on Game Theory Sara Rah 1*, Al

More information

DEFINING %COMPLETE IN MICROSOFT PROJECT

DEFINING %COMPLETE IN MICROSOFT PROJECT CelersSystems DEFINING %COMPLETE IN MICROSOFT PROJECT PREPARED BY James E Aksel, PMP, PMI-SP, MVP For Addtonal Informaton about Earned Value Management Systems and reportng, please contact: CelersSystems,

More information

Stochastic Bandits with Side Observations on Networks

Stochastic Bandits with Side Observations on Networks Stochastc Bandts wth Sde Observatons on Networks Swapna Buccapatnam, Atlla Erylmaz Department of ECE The Oho State Unversty Columbus, OH - 430 buccapat@eceosuedu, erylmaz@osuedu Ness B Shroff Departments

More information

Data Broadcast on a Multi-System Heterogeneous Overlayed Wireless Network *

Data Broadcast on a Multi-System Heterogeneous Overlayed Wireless Network * JOURNAL OF INFORMATION SCIENCE AND ENGINEERING 24, 819-840 (2008) Data Broadcast on a Mult-System Heterogeneous Overlayed Wreless Network * Department of Computer Scence Natonal Chao Tung Unversty Hsnchu,

More information

The OC Curve of Attribute Acceptance Plans

The OC Curve of Attribute Acceptance Plans The OC Curve of Attrbute Acceptance Plans The Operatng Characterstc (OC) curve descrbes the probablty of acceptng a lot as a functon of the lot s qualty. Fgure 1 shows a typcal OC Curve. 10 8 6 4 1 3 4

More information

Efficient Reinforcement Learning in Factored MDPs

Efficient Reinforcement Learning in Factored MDPs Effcent Renforcement Learnng n Factored MDPs Mchael Kearns AT&T Labs mkearns@research.att.com Daphne Koller Stanford Unversty koller@cs.stanford.edu Abstract We present a provably effcent and near-optmal

More information

Project Networks With Mixed-Time Constraints

Project Networks With Mixed-Time Constraints Project Networs Wth Mxed-Tme Constrants L Caccetta and B Wattananon Western Australan Centre of Excellence n Industral Optmsaton (WACEIO) Curtn Unversty of Technology GPO Box U1987 Perth Western Australa

More information

Availability-Based Path Selection and Network Vulnerability Assessment

Availability-Based Path Selection and Network Vulnerability Assessment Avalablty-Based Path Selecton and Network Vulnerablty Assessment Song Yang, Stojan Trajanovsk and Fernando A. Kupers Delft Unversty of Technology, The Netherlands {S.Yang, S.Trajanovsk, F.A.Kupers}@tudelft.nl

More information

Calculation of Sampling Weights

Calculation of Sampling Weights Perre Foy Statstcs Canada 4 Calculaton of Samplng Weghts 4.1 OVERVIEW The basc sample desgn used n TIMSS Populatons 1 and 2 was a two-stage stratfed cluster desgn. 1 The frst stage conssted of a sample

More information

Efficient Project Portfolio as a tool for Enterprise Risk Management

Efficient Project Portfolio as a tool for Enterprise Risk Management Effcent Proect Portfolo as a tool for Enterprse Rsk Management Valentn O. Nkonov Ural State Techncal Unversty Growth Traectory Consultng Company January 5, 27 Effcent Proect Portfolo as a tool for Enterprse

More information

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12 14 The Ch-squared dstrbuton PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 1 If a normal varable X, havng mean µ and varance σ, s standardsed, the new varable Z has a mean 0 and varance 1. When ths standardsed

More information

Cross-Selling in a Call Center with a Heterogeneous Customer Population

Cross-Selling in a Call Center with a Heterogeneous Customer Population OPERATIONS RESEARCH Vol. 57, No. 2, March Aprl 29, pp. 299 313 ssn 3-364X essn 1526-5463 9 572 299 nforms do 1.1287/opre.18.568 29 INFORMS Cross-Sellng n a Call Center wth a Heterogeneous Customer Populaton

More information

1. Measuring association using correlation and regression

1. Measuring association using correlation and regression How to measure assocaton I: Correlaton. 1. Measurng assocaton usng correlaton and regresson We often would lke to know how one varable, such as a mother's weght, s related to another varable, such as a

More information

Coordinated Denial-of-Service Attacks in IEEE 802.22 Networks

Coordinated Denial-of-Service Attacks in IEEE 802.22 Networks Coordnated Denal-of-Servce Attacks n IEEE 82.22 Networks Y Tan Department of ECE Stevens Insttute of Technology Hoboken, NJ Emal: ytan@stevens.edu Shamk Sengupta Department of Math. & Comp. Sc. John Jay

More information

Efficient Bandwidth Management in Broadband Wireless Access Systems Using CAC-based Dynamic Pricing

Efficient Bandwidth Management in Broadband Wireless Access Systems Using CAC-based Dynamic Pricing Effcent Bandwdth Management n Broadband Wreless Access Systems Usng CAC-based Dynamc Prcng Bader Al-Manthar, Ndal Nasser 2, Najah Abu Al 3, Hossam Hassanen Telecommuncatons Research Laboratory School of

More information

Blind Estimation of Transmit Power in Wireless Networks

Blind Estimation of Transmit Power in Wireless Networks Bln Estmaton of Transmt Power n Wreless Networks Murtaza Zafer (IBM Research), Bongjun Ko (IBM Research), Chatschk Bskan (IBM Research) an Ivan Ho (Imperal College, UK) Transmt-power Estmaton: Problem

More information

1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP)

1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP) 6.3 / -- Communcaton Networks II (Görg) SS20 -- www.comnets.un-bremen.de Communcaton Networks II Contents. Fundamentals of probablty theory 2. Emergence of communcaton traffc 3. Stochastc & Markovan Processes

More information

1 Example 1: Axis-aligned rectangles

1 Example 1: Axis-aligned rectangles COS 511: Theoretcal Machne Learnng Lecturer: Rob Schapre Lecture # 6 Scrbe: Aaron Schld February 21, 2013 Last class, we dscussed an analogue for Occam s Razor for nfnte hypothess spaces that, n conjuncton

More information

How To Understand The Results Of The German Meris Cloud And Water Vapour Product

How To Understand The Results Of The German Meris Cloud And Water Vapour Product Ttel: Project: Doc. No.: MERIS level 3 cloud and water vapour products MAPP MAPP-ATBD-ClWVL3 Issue: 1 Revson: 0 Date: 9.12.1998 Functon Name Organsaton Sgnature Date Author: Bennartz FUB Preusker FUB Schüller

More information

Real-Time Process Scheduling

Real-Time Process Scheduling Real-Tme Process Schedulng ktw@cse.ntu.edu.tw (Real-Tme and Embedded Systems Laboratory) Independent Process Schedulng Processes share nothng but CPU Papers for dscussons: C.L. Lu and James. W. Layland,

More information

APPLICATION OF PROBE DATA COLLECTED VIA INFRARED BEACONS TO TRAFFIC MANEGEMENT

APPLICATION OF PROBE DATA COLLECTED VIA INFRARED BEACONS TO TRAFFIC MANEGEMENT APPLICATION OF PROBE DATA COLLECTED VIA INFRARED BEACONS TO TRAFFIC MANEGEMENT Toshhko Oda (1), Kochro Iwaoka (2) (1), (2) Infrastructure Systems Busness Unt, Panasonc System Networks Co., Ltd. Saedo-cho

More information

Joint Dynamic Radio Resource Allocation and Mobility Load Balancing in 3GPP LTE Multi-Cell Network

Joint Dynamic Radio Resource Allocation and Mobility Load Balancing in 3GPP LTE Multi-Cell Network 288 FENG LI, LINA GENG, SHIHUA ZHU, JOINT DYNAMIC RADIO RESOURCE ALLOCATION AND MOBILITY LOAD BALANCING Jont Dynamc Rado Resource Allocaton and Moblty Load Balancng n 3GPP LTE Mult-Cell Networ Feng LI,

More information

Survey on Virtual Machine Placement Techniques in Cloud Computing Environment

Survey on Virtual Machine Placement Techniques in Cloud Computing Environment Survey on Vrtual Machne Placement Technques n Cloud Computng Envronment Rajeev Kumar Gupta and R. K. Paterya Department of Computer Scence & Engneerng, MANIT, Bhopal, Inda ABSTRACT In tradtonal data center

More information

An Intelligent Policy System for Channel Allocation of Information Appliance

An Intelligent Policy System for Channel Allocation of Information Appliance Tamkang Journal of Scence and Engneerng, Vol. 5, No., pp. 63-68 (2002) 63 An Intellgent Polcy System for Channel Allocaton of Informaton Applance Cheng-Yuan Ku, Chang-Jnn Tsao 2 and Davd Yen 3 Department

More information

Downlink Power Allocation for Multi-class. Wireless Systems

Downlink Power Allocation for Multi-class. Wireless Systems Downlnk Power Allocaton for Mult-class 1 Wreless Systems Jang-Won Lee, Rav R. Mazumdar, and Ness B. Shroff School of Electrcal and Computer Engneerng Purdue Unversty West Lafayette, IN 47907, USA {lee46,

More information

AN APPROACH TO WIRELESS SCHEDULING CONSIDERING REVENUE AND USERS SATISFACTION

AN APPROACH TO WIRELESS SCHEDULING CONSIDERING REVENUE AND USERS SATISFACTION The Medterranean Journal of Computers and Networks, Vol. 2, No. 1, 2006 57 AN APPROACH TO WIRELESS SCHEDULING CONSIDERING REVENUE AND USERS SATISFACTION L. Bada 1,*, M. Zorz 2 1 Department of Engneerng,

More information

Testing and Debugging Resource Allocation for Fault Detection and Removal Process

Testing and Debugging Resource Allocation for Fault Detection and Removal Process Internatonal Journal of New Computer Archtectures and ther Applcatons (IJNCAA) 4(4): 93-00 The Socety of Dgtal Informaton and Wreless Communcatons, 04 (ISSN: 0-9085) Testng and Debuggng Resource Allocaton

More information

Dynamic optimization of the LNG value chain

Dynamic optimization of the LNG value chain Proceedngs of the 1 st Annual Gas Processng Symposum H. Alfadala, G.V. Rex Reklats and M.M. El-Halwag (Edtors) 2009 Elsever B.V. All rghts reserved. 1 Dynamc optmzaton of the LNG value chan Bjarne A. Foss

More information

One-Shot Games for Spectrum Sharing among Co-Located Radio Access Networks

One-Shot Games for Spectrum Sharing among Co-Located Radio Access Networks One-Shot Games for Spectrum Sharng among Co-Located Rado Access etwors Sofonas Halu, Alexs A. Dowhuszo, Olav Tronen and Lu We Department of Communcatons and etworng, Aalto Unversty, P.O. Box 3000, FI-00076

More information

Simulation and optimization of supply chains: alternative or complementary approaches?

Simulation and optimization of supply chains: alternative or complementary approaches? Smulaton and optmzaton of supply chans: alternatve or complementary approaches? Chrstan Almeder Margaretha Preusser Rchard F. Hartl Orgnally publshed n: OR Spectrum (2009) 31:95 119 DOI 10.1007/s00291-007-0118-z

More information

QoS-Aware Spectrum Sharing in Cognitive Wireless Networks

QoS-Aware Spectrum Sharing in Cognitive Wireless Networks QoS-Aware Spectrum Sharng n Cogntve reless Networks Long Le and Ekram Hossan Abstract e consder QoS-aware spectrum sharng n cogntve wreless networks where secondary users are allowed to access the spectrum

More information

Customer Lifetime Value Modeling and Its Use for Customer Retention Planning

Customer Lifetime Value Modeling and Its Use for Customer Retention Planning Customer Lfetme Value Modelng and Its Use for Customer Retenton Plannng Saharon Rosset Enat Neumann Ur Eck Nurt Vatnk Yzhak Idan Amdocs Ltd. 8 Hapnna St. Ra anana 43, Israel {saharonr, enatn, ureck, nurtv,

More information

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting Causal, Explanatory Forecastng Assumes cause-and-effect relatonshp between system nputs and ts output Forecastng wth Regresson Analyss Rchard S. Barr Inputs System Cause + Effect Relatonshp The job of

More information

Adaptive Sampling for Energy Conservation in Wireless Sensor Networks for Snow Monitoring Applications

Adaptive Sampling for Energy Conservation in Wireless Sensor Networks for Snow Monitoring Applications Adaptve Samplng for Energy Conservaton n Wreless Sensor Networks for Snow Montorng Applcatons Cesare Alpp *, Guseppe Anastas, Crstan Galpert *, Francesca Mancn, Manuel Rover * * Dp. d Elettronca e Informazone

More information

Stochastic Protocol Modeling for Anomaly Based Network Intrusion Detection

Stochastic Protocol Modeling for Anomaly Based Network Intrusion Detection Stochastc Protocol Modelng for Anomaly Based Network Intruson Detecton Juan M. Estevez-Tapador, Pedro Garca-Teodoro, and Jesus E. Daz-Verdejo Department of Electroncs and Computer Technology Unversty of

More information

MAC Layer Service Time Distribution of a Fixed Priority Real Time Scheduler over 802.11

MAC Layer Service Time Distribution of a Fixed Priority Real Time Scheduler over 802.11 Internatonal Journal of Software Engneerng and Its Applcatons Vol., No., Aprl, 008 MAC Layer Servce Tme Dstrbuton of a Fxed Prorty Real Tme Scheduler over 80. Inès El Korb Ecole Natonale des Scences de

More information

Joint Scheduling of Processing and Shuffle Phases in MapReduce Systems

Joint Scheduling of Processing and Shuffle Phases in MapReduce Systems Jont Schedulng of Processng and Shuffle Phases n MapReduce Systems Fangfe Chen, Mural Kodalam, T. V. Lakshman Department of Computer Scence and Engneerng, The Penn State Unversty Bell Laboratores, Alcatel-Lucent

More information

Schedulability Bound of Weighted Round Robin Schedulers for Hard Real-Time Systems

Schedulability Bound of Weighted Round Robin Schedulers for Hard Real-Time Systems Schedulablty Bound of Weghted Round Robn Schedulers for Hard Real-Tme Systems Janja Wu, Jyh-Charn Lu, and We Zhao Department of Computer Scence, Texas A&M Unversty {janjaw, lu, zhao}@cs.tamu.edu Abstract

More information

2008/8. An integrated model for warehouse and inventory planning. Géraldine Strack and Yves Pochet

2008/8. An integrated model for warehouse and inventory planning. Géraldine Strack and Yves Pochet 2008/8 An ntegrated model for warehouse and nventory plannng Géraldne Strack and Yves Pochet CORE Voe du Roman Pays 34 B-1348 Louvan-la-Neuve, Belgum. Tel (32 10) 47 43 04 Fax (32 10) 47 43 01 E-mal: corestat-lbrary@uclouvan.be

More information

RELIABILITY, RISK AND AVAILABILITY ANLYSIS OF A CONTAINER GANTRY CRANE ABSTRACT

RELIABILITY, RISK AND AVAILABILITY ANLYSIS OF A CONTAINER GANTRY CRANE ABSTRACT Kolowrock Krzysztof Joanna oszynska MODELLING ENVIRONMENT AND INFRATRUCTURE INFLUENCE ON RELIABILITY AND OPERATION RT&A # () (Vol.) March RELIABILITY RIK AND AVAILABILITY ANLYI OF A CONTAINER GANTRY CRANE

More information

A GENERIC HANDOVER DECISION MANAGEMENT FRAMEWORK FOR NEXT GENERATION NETWORKS

A GENERIC HANDOVER DECISION MANAGEMENT FRAMEWORK FOR NEXT GENERATION NETWORKS A GENERIC HANDOVER DECISION MANAGEMENT FRAMEWORK FOR NEXT GENERATION NETWORKS Shanthy Menezes 1 and S. Venkatesan 2 1 Department of Computer Scence, Unversty of Texas at Dallas, Rchardson, TX, USA 1 shanthy.menezes@student.utdallas.edu

More information

Enabling P2P One-view Multi-party Video Conferencing

Enabling P2P One-view Multi-party Video Conferencing Enablng P2P One-vew Mult-party Vdeo Conferencng Yongxang Zhao, Yong Lu, Changja Chen, and JanYn Zhang Abstract Mult-Party Vdeo Conferencng (MPVC) facltates realtme group nteracton between users. Whle P2P

More information

In some supply chains, materials are ordered periodically according to local information. This paper investigates

In some supply chains, materials are ordered periodically according to local information. This paper investigates MANUFACTURING & SRVIC OPRATIONS MANAGMNT Vol. 12, No. 3, Summer 2010, pp. 430 448 ssn 1523-4614 essn 1526-5498 10 1203 0430 nforms do 10.1287/msom.1090.0277 2010 INFORMS Improvng Supply Chan Performance:

More information

Heuristic Static Load-Balancing Algorithm Applied to CESM

Heuristic Static Load-Balancing Algorithm Applied to CESM Heurstc Statc Load-Balancng Algorthm Appled to CESM 1 Yur Alexeev, 1 Sher Mckelson, 1 Sven Leyffer, 1 Robert Jacob, 2 Anthony Crag 1 Argonne Natonal Laboratory, 9700 S. Cass Avenue, Argonne, IL 60439,

More information

Optimal Bidding Strategies for Generation Companies in a Day-Ahead Electricity Market with Risk Management Taken into Account

Optimal Bidding Strategies for Generation Companies in a Day-Ahead Electricity Market with Risk Management Taken into Account Amercan J. of Engneerng and Appled Scences (): 8-6, 009 ISSN 94-700 009 Scence Publcatons Optmal Bddng Strateges for Generaton Companes n a Day-Ahead Electrcty Market wth Rsk Management Taken nto Account

More information

Robust Design of Public Storage Warehouses. Yeming (Yale) Gong EMLYON Business School

Robust Design of Public Storage Warehouses. Yeming (Yale) Gong EMLYON Business School Robust Desgn of Publc Storage Warehouses Yemng (Yale) Gong EMLYON Busness School Rene de Koster Rotterdam school of management, Erasmus Unversty Abstract We apply robust optmzaton and revenue management

More information

Cross-Selling in a Call Center with a Heterogeneous Customer Population

Cross-Selling in a Call Center with a Heterogeneous Customer Population OPERATIONS RESEARCH Vol. 57, No. 2, March Aprl 2009, pp. 299 313 ssn 0030-364X essn 1526-5463 09 5702 0299 nforms do 10.1287/opre.1080.0568 2009 INFORMS Cross-Sellng n a Call Center wth a Heterogeneous

More information

A Design Method of High-availability and Low-optical-loss Optical Aggregation Network Architecture

A Design Method of High-availability and Low-optical-loss Optical Aggregation Network Architecture A Desgn Method of Hgh-avalablty and Low-optcal-loss Optcal Aggregaton Network Archtecture Takehro Sato, Kuntaka Ashzawa, Kazumasa Tokuhash, Dasuke Ish, Satoru Okamoto and Naoak Yamanaka Dept. of Informaton

More information

Marginal Revenue-Based Capacity Management Models and Benchmark 1

Marginal Revenue-Based Capacity Management Models and Benchmark 1 Margnal Revenue-Based Capacty Management Models and Benchmark 1 Qwen Wang 2 Guanghua School of Management, Pekng Unversty Sherry Xaoyun Sun 3 Ctgroup ABSTRACT To effcently meet customer requrements, a

More information

Optimal Scheduling in the Hybrid-Cloud

Optimal Scheduling in the Hybrid-Cloud Optmal Schedulng n the Hybrd-Cloud Mark Shfrn Faculty of Electrcal Engneerng Technon, Israel Emal: shfrn@tx.technon.ac.l Ram Atar Faculty of Electrcal Engneerng Technon, Israel Emal: atar@ee.technon.ac.l

More information

Power-of-Two Policies for Single- Warehouse Multi-Retailer Inventory Systems with Order Frequency Discounts

Power-of-Two Policies for Single- Warehouse Multi-Retailer Inventory Systems with Order Frequency Discounts Power-of-wo Polces for Sngle- Warehouse Mult-Retaler Inventory Systems wth Order Frequency Dscounts José A. Ventura Pennsylvana State Unversty (USA) Yale. Herer echnon Israel Insttute of echnology (Israel)

More information

An Interest-Oriented Network Evolution Mechanism for Online Communities

An Interest-Oriented Network Evolution Mechanism for Online Communities An Interest-Orented Network Evoluton Mechansm for Onlne Communtes Cahong Sun and Xaopng Yang School of Informaton, Renmn Unversty of Chna, Bejng 100872, P.R. Chna {chsun,yang}@ruc.edu.cn Abstract. Onlne

More information

A Dynamic Load Balancing Algorithm in Heterogeneous Network

A Dynamic Load Balancing Algorithm in Heterogeneous Network 06 7th Internatonal Conference on Intellgent Systems Modellng and Smulaton A Dynamc Load Balancng Algorthm n Heterogeneous etwork Zhxong Dng Xngjun Wang and Wenmng Yang Shenzhen Key Lab of Informaton Securty

More information

A User-Centric Approach for Dynamic Resource Allocation in CDMA systems based on Hopfield Neural Networks

A User-Centric Approach for Dynamic Resource Allocation in CDMA systems based on Hopfield Neural Networks A User-Centrc Approach for Dynamc esource Allocaton n CDA systems based on Hopfeld eural etworks. García. Agustí J. érez-omero Unverstat ompeu Fabra (UF) Unverstat oltècnca de Catalunya (UC) Barcelona

More information

The Greedy Method. Introduction. 0/1 Knapsack Problem

The Greedy Method. Introduction. 0/1 Knapsack Problem The Greedy Method Introducton We have completed data structures. We now are gong to look at algorthm desgn methods. Often we are lookng at optmzaton problems whose performance s exponental. For an optmzaton

More information

The Stochastic Guaranteed Service Model with Recourse for Multi-Echelon Warehouse Management

The Stochastic Guaranteed Service Model with Recourse for Multi-Echelon Warehouse Management The Stochastc Guaranteed Servce Model wth Recourse for Mult-Echelon Warehouse Management Jörg Rambau, Konrad Schade 1 Lehrstuhl für Wrtschaftsmathematk Unverstät Bayreuth Bayreuth, Germany Abstract The

More information

Underwriting Risk. Glenn Meyers. Insurance Services Office, Inc.

Underwriting Risk. Glenn Meyers. Insurance Services Office, Inc. Underwrtng Rsk By Glenn Meyers Insurance Servces Offce, Inc. Abstract In a compettve nsurance market, nsurers have lmted nfluence on the premum charged for an nsurance contract. hey must decde whether

More information

On the Interaction between Load Balancing and Speed Scaling

On the Interaction between Load Balancing and Speed Scaling On the Interacton between Load Balancng and Speed Scalng Ljun Chen and Na L Abstract Speed scalng has been wdely adopted n computer and communcaton systems, n partcular, to reduce energy consumpton. An

More information

Sketching Sampled Data Streams

Sketching Sampled Data Streams Sketchng Sampled Data Streams Florn Rusu, Aln Dobra CISE Department Unversty of Florda Ganesvlle, FL, USA frusu@cse.ufl.edu adobra@cse.ufl.edu Abstract Samplng s used as a unversal method to reduce the

More information

On the Interaction between Load Balancing and Speed Scaling

On the Interaction between Load Balancing and Speed Scaling On the Interacton between Load Balancng and Speed Scalng Ljun Chen, Na L and Steven H. Low Engneerng & Appled Scence Dvson, Calforna Insttute of Technology, USA Abstract Speed scalng has been wdely adopted

More information

Extending Probabilistic Dynamic Epistemic Logic

Extending Probabilistic Dynamic Epistemic Logic Extendng Probablstc Dynamc Epstemc Logc Joshua Sack May 29, 2008 Probablty Space Defnton A probablty space s a tuple (S, A, µ), where 1 S s a set called the sample space. 2 A P(S) s a σ-algebra: a set

More information

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by 6 CHAPTER 8 COMPLEX VECTOR SPACES 5. Fnd the kernel of the lnear transformaton gven n Exercse 5. In Exercses 55 and 56, fnd the mage of v, for the ndcated composton, where and are gven by the followng

More information

Transition Matrix Models of Consumer Credit Ratings

Transition Matrix Models of Consumer Credit Ratings Transton Matrx Models of Consumer Credt Ratngs Abstract Although the corporate credt rsk lterature has many studes modellng the change n the credt rsk of corporate bonds over tme, there s far less analyss

More information

Using Series to Analyze Financial Situations: Present Value

Using Series to Analyze Financial Situations: Present Value 2.8 Usng Seres to Analyze Fnancal Stuatons: Present Value In the prevous secton, you learned how to calculate the amount, or future value, of an ordnary smple annuty. The amount s the sum of the accumulated

More information

Statistical Methods to Develop Rating Models

Statistical Methods to Develop Rating Models Statstcal Methods to Develop Ratng Models [Evelyn Hayden and Danel Porath, Österrechsche Natonalbank and Unversty of Appled Scences at Manz] Source: The Basel II Rsk Parameters Estmaton, Valdaton, and

More information

Portfolio Loss Distribution

Portfolio Loss Distribution Portfolo Loss Dstrbuton Rsky assets n loan ortfolo hghly llqud assets hold-to-maturty n the bank s balance sheet Outstandngs The orton of the bank asset that has already been extended to borrowers. Commtment

More information