Optimal Backpressure Routing for Wireless Networks with Multi-Receiver Diversity

Size: px
Start display at page:

Download "Optimal Backpressure Routing for Wireless Networks with Multi-Receiver Diversity"

Transcription

1 CONFERENCE ON INFORMATION SCIENCES AND SYSTEMS (CISS), INVITED PAPER ON OPTIMIZATION OF COMM. NETWORKS, MARCH Optimal Backpressure Routig for Wireless Networks with Multi-Receiver Diversity Michael J. Neely Uiversity of Souther Califoria mjeely Abstract We cosider the problem of optimal schedulig ad routig i a ad-hoc wireless etwork with multiple traffic streams ad time varyig chael reliability. Each packet trasmissio ca be overheard by a subset of receiver odes, with a trasmissio success probability that may vary from receiver to receiver ad may also vary with time. We develop a simple backpressure routig algorithm that maximizes etwork throughput ad expeds a average power that ca be pushed arbitrarily close to the miimum average power required for etwork stability, with a correspodig tradeoff i etwork delay. The algorithm ca be implemeted i a distributed maer usig oly local li error probability iformatio, ad supports a blid trasmissio mode (where error probabilities are ot required) i special cases whe the power metric is eglected ad whe there is oly a sigle destiatio for all traffic streams. Idex Terms Broadcast advatage, distributed algorithms, dyamic cotrol, mobility, queueig aalysis, schedulig I. INTRODUCTION I this paper, we cosider a multi-ode, multi-hop wireless etwork with ureliable chaels. Each trasmissio li has a associated error probability that may vary with time due to exteral factors such as eviromet chages or user mobility. May previous studies assume that accurate chael iformatio is available so that error probabilities are relatively small ad ca be eglected. However, i this work we cosider the opposite case where precise chael iformatio is difficult or impossible to obtai, but where simple estimates of chael quality ca be made based o limited chael feedback. A motivatig example is a uderwater sesor etwork that uses acoustic chaels with large propagatio delays. This is a particularly challegig eviromet due to time varyig wave ripple, complex sigal reflectios betwee surface ad groud, ad large delay spreads [1] [2]. While it may ot be practical to assume that a accurate chael quality ca be determied at the time of packet trasmissio, it is reasoable to estimate the error probability based o past sigal stregth values ad/or ACK/NACK history from previous trasmissios. The problem of ureliable chaels is also importat i other cotexts, such as mobile etworks where kowledge of which receivers are withi trasmissio rage may be ucertai, or i dese ad-hoc etworks where upredictable trasmissios of other odes ca act as radom iter-chael iterferece. It is imperative to develop flexible mathematical models of such etworks, ad to develop robust etworkig This material is based o work supported by the Natioal Sciece Foudatio grat OCE error X broadcastig Fig. 1. A multi-hop etwork with chael errors ad multi-receiver diversity. I this example there is a sigle destiatio idicated by the star ode. Note that a closest-to-the-destiatio heuristic might result i data beig routed from ode 1 to 2 to 3, resultig i a deadlock. strategies that exploit all system resources to operate efficietly i these extreme eviromets. I this paper, we desig robust algorithms by exploitig the broadcast advatage of wireless etworks. Specifically, our etwork model icludes the fact that a sigle packet trasmissio might be overheard by a subset of receiver odes withi rage of the trasmitter. This creates a multi-receiver diversity gai, where the probability of successful receptio by at least oe ode withi a subset of receivers ca be much larger tha the correspodig success probability of just oe receiver aloe. Hece, it is desirable to desig flexible routig algorithms that do ot require a sigle ext hop receiver to be specified i advace. Such algorithms ca dyamically adjust routig ad schedulig decisios i respose to the radom outcome of each trasmissio. The wireless broadcast advatage has bee used i various cotexts, for example, i [3] for the desig of wireless multicast algorithms, ad i [4] for the desig of miimum eergy disjoit paths. Our model ad problem formulatio is closest to the work by Zorzi ad Rao i [5], ad more recetly by Biswas ad Morris i [6], where efficiet methods of usig multi-receiver diversity for packet forwardig are explored. We ote that such formulatios ievitably ivolve situatios where the same packet is redudatly distributed over differet etwork odes. A fudametal decisio is whether to allow the differet versios of the packet to simultaeously propagate throughout the etwork, or to desigate oly a sigle copy that is allowed to proceed. The work i [5] cosiders the

2 CONFERENCE ON INFORMATION SCIENCES AND SYSTEMS (CISS), INVITED PAPER ON OPTIMIZATION OF COMM. NETWORKS, MARCH simple heuristic that shifts packet forwardig resposibilities to the receiver that is closest to the destiatio. While this scheme has may desirable properties, especially for large adhoc etworks, it is clear that for a give etwork of fixed size, the closest-to-destiatio heuristic either maximizes throughput or miimizes average power expediture. Further, this scheme ca lead to a udesirable deadlock mode if data is cosistetly forwarded to a particular ode for which there are o other ext-hop receivers that are closer to the destiatio (see Fig. 1). Thus, it is ofte better to route packets alog paths that temporarily take them further from the destiatio, especially if these paths evetually lead to lis that are more reliable ad/or that are ot as heavily utilized by other traffic streams. The work i [6] cosiders a routig heuristic based o a estimated delivery cost, computed by a estimate of the expected umber of hops required to reach the destiatio alog a traditioal shortest path. However, this method is ot ecessarily optimal i terms of eergy or throughput. There are several difficulties associated with developig a throughput optimal algorithm i this cotext. First, idividual odes might oly kow the error probabilities o their ow outgoig lis, ad may ot kow the error rates or traffic loads o other portios of the etwork. Secod, eve if cetralized etwork kowledge were fully available, a optimal algorithm would eed to specify a cotigecy pla for each possible radom trasmissio outcome. For example, suppose a give ode trasmits a packet for which there are k potetial receivers. There are 2 k possible outcomes of this sigle trasmissio (oe for each possible subset of successful receivers). A optimal algorithm would require a decisio for each possible outcome, perhaps also allowig for redudat packet forwardig. Hece, the desig of a optimal algorithm must overcome these geometric complexity issues. This is further complicated if there are multiple simultaeous packet trasmissios ad multiple traffic streams sharig the same etwork, ad if the etwork topology ad li error probabilities are chagig with time. I this paper, we overcome these challeges with a simple solutio that uses the cocept of backpressure routig ad Lyapuov drift. We first show that it is possible to restrict attetio to algorithms that do ot allow redudat forwardig, without loss of optimality. We the show that the optimal packet commodity to trasmit at each etwork ode ca be determied by a backpressure idex that compares the curret queue backlog of each commodity to the backlog i the potetial receivers. Oce a packet from this optimal commodity is trasmitted, the resposibility of forwardig the packet to its destiatio is shifted to the receiver ode that maximizes the differetial backlog. Resposibility is retaied by the origial trasmitter if o suitable receivers are foud o a give trasmissio attempt. Backpressure techiques of this type were first applied to multi-hop wireless etworks by Tassiulas ad Ephremides i [7], where throughput optimal algorithms were developed usig Lyapuov drift theory. Lyapuov theory has sice bee a powerful mathematical tool for the developmet of stable schedulig strategies for wireless etworks ad switchig systems [7]-[18], icludig our ow work i [15]-[18] that applies backpressure cocepts to solve problems of optimal power allocatio, routig, ad fair flow cotrol i wireless etworks with mobility. Related work o eergy efficiet wireless schedulig is developed i [19]-[22]. The work i [7]-[22] does ot cosider the broadcast advatage of wireless etworks, ad assumes that all trasmissios are fully reliable. Lyapuov schedulig for wireless MIMO dowlis with multiple trasmit ad receive ateas is cosidered i [23], ad related MIMO results are developed for chaels with errors i [24] [25]. Recet work i [26] cosiders backpressure techiques i combiatio with etwork codig, ad work i [27] cosiders backpressure strategies for cooperative trasmissio (where multiple odes ca trasmit redudat iformatio simultaeously for a power ehacemet at the receiver). Complexity issues of cooperative commuicatio uder the wireless broadcast advatage are discussed i [28]. We do ot cosider etwork codig or cooperative trasmissio i this paper, ad restrict attetio to the multi-user diversity problem for etworks with errors, as described above. It is likely that our formulatio ca be exteded to cosider more sophisticated cotrol actios by augmetig the set of decisio optios available to the etwork cotroller, i which case redudat packet forwardig may be required for optimality. I the ext sectio, we develop a simple etwork model i terms of (potetially time varyig) li error probabilities, ad specify the cotrol decisio optios for this model. I Sectio III, we specify the etwork capacity ad the miimum average power for stability associated with this model. I Sectio IV we develop the dyamic cotrol algorithm, ad i Sectio VI we exted the formulatio to iclude dyamic resource allocatio with variable rate ad power optios, where li error probabilities ca deped o trasmissio decisios. II. THE BASIC NETWORK MODEL We cosider a timeslotted system with slots ormalized to itegral uits t {0, 1, 2,.... There are N etwork odes ad L potetial trasmissio lis (possibly a sigle li for each ode pair (a, b)). All data arrives radomly to the etwork i (t) represet the umber of packets that exogeously arrive to etwork ode durig slot t that are iteded for delivery to etwork ode c. All packets destied for a particular ode c are defied as commodity c packets. Arrivals are assumed to be i.i.d. over timeslots, ad we packetized uits, ad we let A let λ = E{A (t) represet the arrival rate of commodity c data ito source ode (i uits of packets/slot). Iteral etwork queues store packets accordig to their commodities. Each packet is assumed to have a appropriate header field with commodity ad packet umber idetifiers. We assume that at most oe packet ca be trasmitted from ay give ode durig a sigle timeslot, ad let µ (t) represet the umber of packets trasmitted by ode durig slot t (where µ (t) {0, 1). Trasmissio opportuities are determied by a uderlyig radom access or time divisio multiple access (TDMA) structure, ad we let χ (t) represet a 0/1 process which is 1 if ad oly if ode is allowed to trasmit durig slot t. Each packet trasmissio is assumed to exped a costat amout of power P tra, ad is successfully

3 CONFERENCE ON INFORMATION SCIENCES AND SYSTEMS (CISS), INVITED PAPER ON OPTIMIZATION OF COMM. NETWORKS, MARCH received by the other odes of the etwork accordig to receptio probabilities q (t) (for, k {1,..., N). For coveiece, we defie the etwork topology state process S(t) as the collective process of all ode trasmissio capabilities ad li coditios at time t, so that trasmissio opportuities ad li probabilities ca be determied as fuctioals of S(t) as follows: seder ode receiver ode k Cotrol Iformatio Cotrol Iformatio Packet Trasmissio ACK/NACK Fial Istructios t t+1 χ (t) = ˆχ (S(t)) q (t) = ˆq (S(t)) Let K (t) represet the set cosistig of all potetial receivers for ode durig slot t (which ca potetially chage from slot to slot if the etwork is mobile). The set K (t) ca geerally cotai all N 1 other etwork odes, although it typically has a much smaller size ad cosists oly of those odes withi realistic trasmissio rage of ode. Error evets for a sigle packet trasmissio ca be correlated over various lis, ad hece a more complete characterizatio of each trasmitter is give by probabilities q,ω (t), where Ω is a subset of odes withi the receiver set K (t), ad q,ω (t) represets the probability that the set of all odes that successfully receive the packet trasmitted by ode is exactly give by the subset Ω. This probability is also determied as a fuctioal of the topology state process: q,ω (t) = ˆq,Ω (S(t)) The error evets of differet packet trasmissios from differet odes may also be correlated, ad these correlatios i priciple are also determied by the topology state process S(t). However, we shall fid that these additioal correlatios are irrelevat to etwork capacity ad optimal cotrol. For aalytical purposes, the etwork topology state S(t) is assumed to take values i a fiite (but arbitrarily large) state space S. We ote that the success probabilities of a give li or set of lis are completely determied by the etwork topology state process S(t). That is, give S(t), these probabilities are ot affected by the trasmissio decisios µ (t) for {1,..., N. This assumptio is reasoable if all trasmittig odes use orthogoal sigals, or if iterchael iterferece ca be approximated as radomly ad idepedetly ifluecig the chael probabilities. A more geeral model where chael probabilities ca deped o trasmissio decisios is cosidered i Sectio VI. A. A Timig Diagram for Oe Timeslot The timig diagram of Fig. 2 illustrates our model of iformatio exchage betwee odes. The evets that take place betwee a trasmittig ode ad a potetial receiver ode k durig a sigle timeslot are outlied i the diagram. At the begiig of the timeslot, chael probability iformatio ad ay ecessary cotrol iformatio is passed betwee the two odes. This ca possibly take place over a dedicated cotrol chael, or might be implemeted by appedig header iformatio to packets trasmitted o previous slots. Next, the trasmitter ode observes the trasmissio opportuity process ˆχ (S(t)). If ˆχ (S(t)) = 0 the ode does ot trasmit, while if ˆχ (S(t)) = 1 the ode ca decide whether Fig. 2. A timig diagram illustratig the evets withi a sigle timeslot. or ot it desires to trasmit a packet. If it decides to trasmit, it chooses a particular packet ad trasmits it with power P tra, for a fixed amout of time as idicated i the timig diagram. Every potetial receiver ode the provides immediate ACK/NACK feedback to the trasmitter, iformig the trasmitter if the packet was successfully received. The absece of a ACK sigal is cosidered to be equivalet to a NACK (this treats the case whe the receiver ode did ot detect ay trasmissio). The trasmitter ode accumulates all of the ACK resposes, ad the trasmits a fial message that iforms the successful receivers of all other successful receivers. This fial trasmissio possibly also provides istructios for future packet forwardig. The 3-part hadshake of the timig diagram (trasmissio, ACK/NACK, ad fial message) is desiged to clealy describe a system where trasmissio outcomes are kow to all relevat odes at the ed of a sigle timeslot. This facilitates mathematical aalysis. However, i practice the last two steps of the hadshake may take place by appedig this iformatio to the packet header of future packet trasmissios. This creates a system with delayed feedback iformatio, which i priciple does ot affect throughput optimality (provided some regularity assumptios hold cocerig the timeliess of the feedback) but may affect ed-to-ed etwork delay, as discussed i more detail i Sectio VII. Throughout this paper, we make the idealistic assumptio of perfect cotrol iformatio, so that the cotrol sigals themselves are ot subject to errors. I particular, for the timig diagram of Fig. 2, it is assumed that if a packet trasmitted at ode was successfully received at ode k, the the chael from k to ad from to k is good eough for the remaiig parts of the hadshake to be successful. This is a reasoable assumptio if forward ad backward chaels are relatively similar for the duratio of a timeslot, or if the dedicated cotrol chael is reliable. The possibility of cotrol chael errors ca create aother situatio of delayed feedback iformatio, ad this is also briefly discussed i more detail i Sectio VII. B. Network Objective ad Cotrol Decisio Variables The goal is to desig a cotrol algorithm that stabilizes the etwork wheever possible. Further, the average power cost should be as small as possible. Specifically, for a power vector P = (P 1,..., P N ), we defie the separable cost fuctio h(p ) = h 1 (P 1 ) h N (P N ), where each compoet h (P ) is o-egative, cotiuous, ad has the property that h (0) = 0. The power expeded o each timeslot t is give by the vector P (t) =P tra (µ 1 (t),..., µ N (t)), ad the time

4 CONFERENCE ON INFORMATION SCIENCES AND SYSTEMS (CISS), INVITED PAPER ON OPTIMIZATION OF COMM. NETWORKS, MARCH average power cost h is defied: h = lim t 1 t t 1 τ=0 E {h(p (τ)) Note that choosig h(p ) = N =1 P coicides with the objective of miimizig the time average expected power expediture. Uder our simple etwork model, we have P (t) {0, P tra for all t, so that h (P (t)) {0, h (P tra ). I this case, the h ( ) fuctio plays oly a limited role i geeralizig the miimum average power objective, although it shall be more meaigful i the exteded formulatio of Sectio VI that cosiders a cotiuum of power optios. I Sectio III we show that throughput ad eergy optimality ca be achieved without usig redudat packet forwardig. This allows the followig more detailed etwork queueig variables ad cotrol decisio variables to be defied. At each timeslot t, every etwork ode makes a trasmissio decisio µ (t) subject to µ (t) {0, 1 ad µ (t) = 0 wheever χ (t) = 0. It the chooses a packet commodity to trasmit by selectig cotrol variables µ (t) subject to: µ (t) {0, 1, N c=1 µ (t) µ (t), µ c (t) = 0 (1) That is, µ (t) represets a opportuity for commodity c packet trasmissio by ode durig slot t. This ca be either 0 or 1, but ca be 1 for at most oe commodity c. We set (t) = 0 as it does ot make sese to retrasmit a packet that has already reached its destiatio. We say that µ (t) is a trasmissio opportuity because it is useful to imagie the possibility of choosig these decisio variables idepedet of queue backlog. I cases whe a trasmissio opportuity arises but there is o commodity c packet available, the o packet is actually trasmitted. We let H (t) represet the radom variable that is 1 if a packet trasmitted from ode was successfully received by receiver k, ad zero otherwise. After receivig ACK/NACK feedback, ode selects a ew ode to take resposibility for the packet (possibly choosig itself), ad iforms its receivers of the choice. This is doe accordig to cotrol decisio µ c variables β (t), represetig the umber of commodity c packets whose resposibility is shifted from ode to ode k durig slot t, where: β (t) {0, 1, β (t) µ (t)h (t) β (t) = 0, N k=1 β (t) 1 (2) That is, the β (t) variables are either 0 or 1, ca be 1 oly if a commodity c trasmissio opportuity occurs o slot t ad H (t) = 1, ad ca be 1 for at most oe receiver ode k (where such a ode k is ecessarily i the set of potetial receivers K (t)). If β (t) = 0 for all k K (t), the ode retais resposibility for the packet. It shall be coveiet to also allow these decisio variables to be idepedet of queue backlog, ad so both β (t) ad µ (t) ca potetially equal 1, regardless of whether or ot ode was holdig a commodity c packet that it actually trasmitted. I this case, the H (t) value is viewed as a radom variable that is distributed the same as if a packet had actually bee trasmitted. The actual cotrol decisios β (t) i the case of o packet trasmissio are irrelevat as they do ot affect the system. However, it is useful to formally allow choosig o-zero β (t) values i this case. Specifically, we fid it useful for mathematical proofs to imagie the existece of a statioary radomized cotrol policy that chooses decisio variables idepedetly of queue backlog, but where o packets are actually trasferred if these decisios attempt trasmissio from a empty queue. Packets are stored at every ode accordig to their commodity, ad we defie U (t) as the curret umber of commodity c packets i ode at the begiig of slot t. The U (t) process takes values i the set of o-egative itegers, ad evolves accordig to the followig queueig dyamics: [ U (t + 1) max U (t) ] N k=1 β (t), 0 + N a=1 β a (t) + A (t) (3) The expressio above is a iequality rather tha a equality because the actual edogeous arrivals to ode may be less tha N a=1 β a (t) if there are little or o actual commodity c packets trasmitted from the other odes a. We formally defie U () (t) to be zero for all ad all t. III. NETWORK CAPACITY AND MINIMUM POWER Here we defie the optimal throughput ad average power cost operatig poits. The etwork layer capacity regio Λ is defied as the closure of all iput rate matrices (λ ) that ca be stabilized by the etwork accordig to some cotrol algorithm, perhaps a algorithm that uses redudat packet forwardig. We ote that this otio of capacity assumes that etwork cotrol actios are withi the scope of the system model described i Sectio II, ad i particular this model does ot iclude the possibility of cooperative trasmissio or etwork codig, which ca potetially improve performace. Suppose that the etwork topology state process S(t) takes values o a fiite state space S, ad has well defied time average probabilities π s for each s S. For each ode, let H deote the set of all subsets Ω of {1,..., N {. For each subset Ω, recall that ˆq,Ω (s) is the probability that Ω is exactly the set of all successful receivers of a packet trasmitted by ode, give such a packet is trasmitted whe the topology state is give by S(t) = s. Theorem 1: (Network Capacity ad Miimum Cost) The etwork capacity regio Λ cosists of all rate matrices (λ ) for which there exist multi-commodity flow variables {f together with probabilities α (s), θ (Ω ) for all, k, c, all topology states s S, ad all subsets Ω H, such that: c a f c f ab 0, f cb f a + λ b s S π s α (s) = 0, f aa = 0 (4) f b for all c (5) [ ˆq,Ω (s)θ (Ω ) Ω H where (4) holds for all a, b, c {1,..., N, (6) holds for all lis (, k), ad where the probabilities θ (Ω ) satisfy for all Ω H : θ (Ω N ) = 0 if k / {Ω {, k=1 θ (Ω ) = 1 ] (6)

5 CONFERENCE ON INFORMATION SCIENCES AND SYSTEMS (CISS), INVITED PAPER ON OPTIMIZATION OF COMM. NETWORKS, MARCH ad for all s S the α (s) probabilities satisfy: N c=1 α (s) 1, α (s) = 0 if ˆχ (s) = 0 Furthermore, the miimum average power cost required for etwork stability is give by the value h that miimizes the followig metric: h = [ s S π N ] N s c=1 α (s)h (P tra ) over all {f, α =1 (s), θ (Ω ) variables that satisfy (4)-(6). The above theorem is similar to the capacity theorem of [16] [15], where the costraits (4) represet o-egativity ad flow efficiecy costraits for the flow variables {f ab, the costraits (5) represet flow coservatio costraits, ad the costraits (6) represet li costraits for each li (, k). Each α (s) value ca be iterpreted as the coditioal probability that ode trasmits a commodity c packet give that S(t) = s. Each θ (Ω ) value ca be iterpreted as the coditioal probability that ode shifts packet forwardig resposibilities to ode k, give that ode trasmits a commodity c packet that is heard exactly by the subset Ω of receivers. With this iterpretatio, the theorem ca be simplified accordig to the followig corollary. For each iput rate matrix λ = (λ ) Λ, we defie Φ(λ) as the miimum power cost h required to stabilize the system. Suppose that the iput rate matrix is iterior to the capacity regio, so that there exists a positive value ɛ such that (λ + ɛ1 ) Λ, where 1 is a idicator fuctio equal to 1 if ad oly if c, ad zero else. Corollary 1: If the topology state S(t) is i.i.d. over timeslots, the a rate matrix (λ + ɛ1 ) is i the capacity regio Λ if ad oly if there exists a statioary radomized algorithm that chooses cotrol decisio variables µ (t) ad β (t) (accordig to the costraits specified i Sectio II-B) based oly o the curret topology state S(t) (ad hece idepedet of curret queue backlog), to yield: E a { β a (t) + λ + ɛ { E β b (t) c b (7) E {h(p (t)) = Φ(λ + ɛ) (8) where ɛ = (ɛ1 ) ad P (t) = P tra (µ 1 (t),..., µ N (t)). The expectatios i (7) ad (8) are take with respect to the radom topology state S(t) ad the radom cotrol decisios based o this topology state, ad do ot deped o queue backlog. The above theorem ad its corollary demostrate that for ay rate matrix (λ ) Λ, there exists a statioary radomized algorithm (with probabilities precisely matched to the etwork traffic rates ad topology state probabilities) that ca achieve a multi-commodity flow that supports the iput rate matrix by routig all data to its proper destiatio, ad that icurs a average power cost exactly give by h. However, eve if all topology state probabilities π s were fully kow, the geometric complexity of the optimizatio problem i Theorem 1 demostrates the extreme difficulty of directly solvig for the parameters required to implemet such a policy. Theorem 1 is prove by first showig that the costraits (4)-(6) are ecessary for etwork stability. The sufficiecy part of the theorem is prove by costructig a stabilizig algorithm for ay rate matrix (λ ) that is iterior to the capacity regio (so that (λ + ɛ1 ) Λ, for some positive value ɛ). Such stabilizig policies ca be costructed with resultig average power costs that are arbitrarily close to h (by choosig ɛ arbitrarily small), with a correspodig tradeoff i ed-to-ed etwork delay. The proof of ecessity uses the fiite state space assumptio for the topology state variable S(t), ad is related to similar proofs of capacity ad miimum eergy i [16] [17] [15] (proof omitted for brevity). Sufficiecy does ot require the fiite state space property, ad is prove i the ext sectio, where a simple dyamic cotrol algorithm is costructed that ca be implemeted i real time. IV. THE DYNAMIC CONTROL ALGORITHM We have the followig dyamic cotrol algorithm, defied i terms of a o-egative cotrol parameter V that determies the degree to which we emphasize power cost miimizatio. Diversity Backpressure Routig (DIVBAR): Every timeslot t, each etwork ode observes the queue backlogs i each of its potetial receiver odes k K (t), ad observes the curret li chael probabilities associated with its receivers. Each ode determies if χ (t) = 1 (i.e., it determies if a trasmissio opportuity is available o the curret slot). If so, it performs the followig operatios: 1) For each commodity c ad each receiver k K (t), the differetial backlog weights W (t) are computed as follows: W (t) = max[u (t) U k (t), 0] (9) That is, the weight W (t) is equal to the differece betwee the commodity c backlog i ode ad the commodity c backlog i ode k (maxed with zero). 2) The receivers k K (t) are priority raed accordig to the W (t) weights, so that receivers with larger weights are ordered with higher priority (breakig ties arbitrarily). We defie k(, c, t, b) as the ode k K (t) with the bth largest weight W (t) for commodity c. Thus, by defiitio we have: W,k(,c,t,1) (t) W,k(,c,t,2) (t) W,k(,c,t,3) (t)... 3) Defie φ (t) as the probability that a packet trasmissio from ode is correctly recieved by ode k, but is ot received by ay other odes k K (t) that are raed with higher priority tha ode k accordig to the commodity c ra orderig of the previous step. 4) Defie the optimal commodity c (t) as the commodity c {1,..., N that maximizes (breakig ties arbitrarily): K (t) b=1 W,k(,c,t,b) (t)φ,k(,c,t,b) (t)

6 CONFERENCE ON INFORMATION SCIENCES AND SYSTEMS (CISS), INVITED PAPER ON OPTIMIZATION OF COMM. NETWORKS, MARCH where K (t) deotes the umber of odes i the set K (t). Defie W (t) as the resultig maximum value: W (t) = K (t) b=1 W (c ),k(,c,t,b)(t)φ(c ),k(,c,t,b)(t) 5) If W(t) V h (P tra ) > 0, ode trasmits a packet of commodity c (t). Else, ode remais idle for slot t. 6) After receivig ACK/NACK feedback about the successful recipiets of the trasmissio, ode shifts resposibility of packet forwardig to the successful receiver k with the largest positive differetial backlog W (c (t)) (t). If o successful receivers have positive differetial backlog, ode retais resposibility of the packet. The above algorithm is fully distributed, i that each ode oly requires queue backlog ad li probability values for each of its eighborig odes (i.e., each ode withi K (t)). The queue backlogs ca be passed durig the cotrol iformatio phase of the timeslot, or ca be based o backlog updates received i the headers of previous packets. We ote that, as i the Dyamic Routig ad Power Cotrol (DRPC) policy of [15] [16], the algorithm ca be implemeted without loss of throughput optimality by usig out of date backlog iformatio, provided that some regularity coditios hold (see Chapter of [15]). The li error probabilities ca be obtaied based o cotrol iformatio exchage at the begiig of the timeslot (such as a pilot sigal ad a correspodig SINR measuremet, as i [16]), or ca be estimated based o previous ACK/NACK history. The above algorithm cosiders the geeral case where li error evets ca be correlated. However, computatio of the φ (t) probabilities ca be greatly simplified uder the assumptio that error evets are idepedet over each li. I this case, φ (t) is obtaied from a simple multiplicatio of the appropriate success or error probabilities of the correspodig lis. A. Algorithm Performace To facilitate mathematical aalysis, we assume the etwork topology state S(t) is i.i.d. over timeslots. 1 Note that this also icludes the case whe the topology state does ot chage over time. Defie the followig costat µ i max to be the largest umber of edogeous packet arrivals that ay sigle ode ca receive durig a timeslot. Further, defie A 2 max as a upper boud o the secod momet of the total exogeous arrivals to ay ode durig a timeslot, so that: { (N max E c=1 A (t) ) 2 A 2 max We assume the iput rate matrix is iterior to the capacity regio Λ (so that stability is possible), ad defie ɛ max as the largest scalar such that (λ + ɛ max 1 ) Λ. Theorem 2: (Algorithm Performace) If topology state variatios S(t) are i.i.d. over timeslots, ad if the iput rate matrix is strictly iterior to the capacity regio Λ, the the 1 The same algorithm ca be show to be throughput optimal for o-i.i.d. topology state variatios usig a similar T -slot Lyapuov drift argumet, see [16][15] for such a aalysis for a related algorithm. DIVBAR algorithm stabilizes all queues of the system (ad hece provides maximum throughput). Furthermore, average etwork cogestio ad average power cost satisfies: lim sup t lim sup t 1 t 1 t t 1 τ=0,c { E U (τ) NB + V h max ɛ max t 1 E {h (µ (τ)p tra ) h + NB/V τ=0 where h max = h (P tra ), ad where B is defied: B= (µi max + A max ) (10) 2 Note that choosig the cotrol parameter V to be zero leads to the best cogestio boud but does ot lead to ay power efficiecy guaratees. The parameter V ca be icreased to drive average power cost arbitrarily close to the miimum cost h required for etwork stability, with a correspodig liear icrease i average etwork cogestio (ad hece, by Little s Theorem, average delay). B. Chael Blid Packet Trasmissio I the special case whe power optimizatio is eglected (so that V = 0) ad there is a sigle destiatio for all packets, the DIVBAR algorithm ca be sigificatly simplified to allow for blid packet trasmissios. Specifically, because there is just a sigle commodity, the steps (1)-(5) ca be avoided ad the algorithm reduces to havig ode trasmit a packet wheever possible (i.e., wheever χ (t) = 1). It the receives ACK/NACK feedback from the various receivers, ad chooses the receiver k with the largest positive differetial backlog U (t) U k (t), breakig ties arbitrarily ad retaiig the packet if o receiver has a positive differetial backlog. Note that the backlog of each receiver ca simply be icluded i the ACK/NACK sigal. The algorithm thus achieves throughput optimality without requirig chael probability iformatio. This is a remarkable property, ad eables perfect throughput optimality to be achieved eve whe chael probabilities are rapidly chagig due to dramatic ode mobility. No effort is eeded to estimate error rates, or to track them if they vary with time. V. PERFORMANCE ANALYSIS Here we prove Theorem 2. The proof uses the followig result from [15] [17] [18] cocerig performace optimal Lyapuov schedulig, which is a simple but importat extesio of classical Lyapuov stability results of [7]-[16]. Let U(t) = (U (t)) represet the matrix of queue backlog values, ad assume these backlogs evolve accordig to a give probability law ad are affected by a cotrol process P (t) = (P 1 (t),..., P N (t)). Let h(p ) be ay o-egative fuctio of P, ad let h represet a target value for the time,c (U average of h(p (t)). Let L(U) = 1 2 ) 2 represet a quadratic Lyapuov fuctio, ad defie the oe step Lyapuov drift (U(t)) as follows: (U(t)) =E {L(U(t + 1)) L(U(t)) U(t)

7 CONFERENCE ON INFORMATION SCIENCES AND SYSTEMS (CISS), INVITED PAPER ON OPTIMIZATION OF COMM. NETWORKS, MARCH Theorem 3: (Lyapuov Optimizatio [15] [17][18]) If there exist positive costats B, V, ɛ such that for all timeslots t ad for all queue backlogs U(t), the Lyapuov drift satisfies: (U(t)) + V E {h(p (t)) U(t) B ɛ,c U (t) + V h the all queues are stable, ad time average cogestio ad etwork cost satisfies:,c U = 1 lim sup t t t 1 τ=0,c { E U (τ) B + V h ɛ h= 1 t 1 lim sup E {h(p (τ)) h + B/V t t τ=0 The above theorem suggests the strategy of miimizig the metric (U(t)) + V E {h(p (t)) U(t) every timeslot t, which is the motivatio behid DIVBAR. A. Proof of the DIVBAR Performace Theorem (Theorem 2) The coditioal Lyapuov drift ca be computed from the queue backlog expressio (3) accordig to stadard drift techiques (see [7][15][16]), ad is give by: (U(t)) NB {,c U (t)e k β (t) a β a (t) λ U(t) where B is defied i (10). Addig the cost metric to both sides (where P (t) = P tra (µ 1 (t),..., µ N (t))), we have: (U(t)) + V E {h(p (t)) U(t) NB + V E {h(p (t)) U(t) { U (t)e β (t),c a k β a (t) λ U(t) (11) The DIVBAR algorithm is desiged to choose cotrol actios that greedily miimize the right had side of the above iequality over all possible choices of the cotrol variables µ (t), µ (t), ad β (t) that satisfy the costraits (1) ad (2). This ca be see by switchig the sums to ote that: { U (t)e β (t) β a (t) U(t) =,c k a { [ ] E β ab (t) U(t) U a (t) U b (t) ab c which reveals the differetial backlog metric (details omitted for brevity). It follows that the right had side of (11) uder the DIVBAR algorithm is less tha or equal to the correspodig expressio whe the cotrol variables are replaced with ay others, ad i particular those of the statioary radomized algorithm from (7) ad (8) of Corollary 1, ad so: (U(t)) + V E {h(p (t)) U(t) NB + V Φ(λ + ɛ),c U (t)ɛ The above iequality is i the exact form for applicatio of the Lyapuov Optimizatio Theorem (Theorem 3), ad we thus have (otig that Φ(λ + ɛ) h max ): U (NB + V h max )/ɛ (12),c h Φ(λ + ɛ) + NB/V (13) The above performace bouds hold for ay value ɛ > 0 such that (λ +ɛ1 ) Λ, ad hece the bouds ca be optimized separately over all such ɛ. Lettig ɛ ɛ max i (12) yields the cogestio boud of Theorem 2, ad lettig ɛ 0 i (13) yields the power cost boud of Theorem 2. VI. VARIABLE RATE AND POWER CONTROL Cosider ow a system with variable rate ad power cotrol optios, so that every timeslot the trasmissio rates µ(t) = (µ 1 (t),..., µ N (t)) ca be chose such that µ (t) {0, 1,..., µ out max for all t (for some pre-specified iteger µ out max), ad trasmissio power to support these rates is chose accordig a power vector P (t) = (P 1 (t),..., P N (t)), where 0 P (t) P peak for all t ad all (for some peak trasmissio power P peak ). Note that the µ (t) variable is still iteger valued, but there is o loger ay multiple access process χ (t) that places further restrictios o µ (t). Defie I(t) =(µ(t); P (t)) as the collective trasmissio cotrol decisios of all etwork odes durig slot t, ad defie I as the set of all possible optios for I(t). We assume that error probabilities are fuctios of I(t) ad the curret topology state S(t), so that: q,ω (t) = ˆq,Ω (I(t), S(t)) If m packets are trasmitted by ode, the each of them is assumed to have the same q,ω (t) probability. Correlatios i the error evets of differet packets withi the batch of m are arbitrary ad do ot affect capacity or optimal cotrol decisios. The cotrol objective of stabilizig the etwork ad miimizig h is the same as before. Usig similar reasoig, it ca agai be show that it is possible to restrict to algorithms that do ot allow redudat forwardig, without loss of optimality. A similar Lyapuov argumet the leads to the followig optimal policy: 1) Compute W (t) = max[u (t) U k (t), 0] as before. For each ode ad each commodity c, we agai ra order the receivers k K (t) with priority give by the largest values of W (t), ad defie k(, c, t, b) as ˆφ before. We defie (I(t), S(t)) as the probability that a packet trasmissio from ode durig slot t is correctly received by ode k, but ot received by ay other odes k K (t) that are raed with higher priority tha ode k accordig to the commodity c orderig. 2) Defie: G,c,t,b (I(t), S(t)) = W,k(,c,t,b) (t) ˆφ,k(,c,t,b) (I(t), S(t)) Choose a etwork-collaborative cotrol actio I (t) = (µ (t), P (t)) I ad a collectio of optimal commodities c (t) {1,..., N (for all odes ) that joitly

8 CONFERENCE ON INFORMATION SCIENCES AND SYSTEMS (CISS), INVITED PAPER ON OPTIMIZATION OF COMM. NETWORKS, MARCH maximizes the metric: K (t) G,c (t),t,b(i (t), S(t)) V h (P(t)) b=1 3) If K (t) [ b=1 G,c (t),t,b (I (t), S(t)) ] > V h (P(t)), ode trasmits µ (t) commodity c (t) packets (usig idle fill if there are ot eough such packets). 4) After receivig ACK/NACK feedback from each receiver about each of the µ (t) trasmitted packets, ode shifts resposibility of each packet to the successful receiver with the largest positive differetial backlog W (c (t)) (t). If o receivers of a give packet have positive differetial backlog, ode retais resposibility of the packet. Choosig the appropriate cotrol actio I(t) = (µ(t); P (t)) effectively optimizes over all multiple access decisios, but yields a optimizatio problem i step 2 that ca be quite difficult to solve ad may require full cetralized coordiatio. However, distributed implemetatio is possible if all odes trasmit with orthogoal sigals, ad costat factor throughput optimality results ca be achieved i a decetralized maer for some etworks models (such as the ode exclusive spectrum sharig model of [29]), if power optimizatio is eglected ad simple radom access methods are employed. We further ote that if simple radom access methods are used as i [16][15], ad if trasmissios are idepedet of queue backlog, the radom trasmissios themselves ca be viewed as part of the chael state, ad the algorithm thus achieves efficiet performace with respect to this (suboptimal) radom access. VII. DISCUSSION OF EXTENSIONS We ote that delay ca be improved by modifyig the differetial backlog W (t) by addig a estimated hop cout differetial to the destiatio, as i the EDRPC policy of [16] [15]. Optimizatio of geeral utility ad fairess metrics ca also be achieved i cases whe the iput rate matrix is either iside or outside of the capacity regio Λ by usig the simple ad optimal flow cotrol techiques of [15] [18] together with DIVBAR. Fially, we ote that Lyapuov drift theorems hold if drift expressios are off by a additive costat, so that queue backlog estimates ca be used i replacemet of actual queue backlogs [15]. This does ot effect throughput optimality, but may icrease delay by a costat proportioal to the differece betwee the true ad estimated backlog values. This eables DIVBAR to be implemeted with delayed feedback, provided the feedback delay is bouded. A queueig implemetatio of this would provide separate storage for data that has bee correctly received but has ot yet bee ackowledged by a fial istructio message from the trasmitter. REFERENCES [1] M. Stojaovic, J. A. Catipovic, ad J. G. Proakis. Phase-coheret digital commuicatios for uderwater acoustic chaels. IEEE Joural of Oceaic Egieerig, vol. 19, o. 1, Ja [2] D. B. Kilfoyle, J. C. Preisig, ad A. B. Baggeroer. Spatial modulatio experimets i the uderwater acoustic chael. IEEE Joural of Oceaic Egieerig, vol. 30, o. 2, April [3] J. E. Wieselthier, G. D. Nguye, ad A. Ephremides. Algorithms for eergy-efficiet multicastig i ad hoc wireless etworks. Proc. IEEE Military Commuicatios Coferece, pp , [4] A. Sriivas ad E. Modiao. Miimum eergy disjoit path routig i wireless ad-hoc etworks. IEEE Proc. of Mobicom, September [5] M. Zorzi ad R. Rao. Geographic radom forwardig (geraf) for ad hoc ad sesor etworks: Multihop performace. IEEE Tras. o Mobile Computig, Vol. 2, o. 4, Oct.-Dec [6] S. Biswas ad R. Morris. Exor: Opportuistic multi-hop routig for wireless etworks. Proc. of Sigcomm, [7] L. Tassiulas ad A. Ephremides. Stability properties of costraied queueig systems ad schedulig policies for maximum throughput i multihop radio etworks. IEEE Trasacatios o Automatic Cotrol, Vol. 37, o. 12, Dec [8] L. Tassiulas ad A. Ephremides. Dyamic server allocatio to parallel queues with radomly varyig coectivity. IEEE Tras. o Iform. Theory, vol. 39, pp , March [9] P.R. Kumar ad S.P. Mey. Stability of queueig etworks ad schedulig policies. IEEE Tras. o Automatic Cotrol, Feb [10] N. McKeow, V. Aatharam, ad J. Walrad. Achievig 100% throughput i a iput-queued switch. Proc. INFOCOM, [11] N. Kahale ad P. E. Wright. Dyamic global packet routig i wireless etworks. IEEE Proceedigs of INFOCOM, [12] M. Adrews, K. Kumara, K. Ramaa, A. Stolyar, ad P. Whitig. Providig quality of service over a shared wireless li. IEEE Commuicatios Magazie, [13] E. Leoardi, M. Melia, F. Neri, ad M. Ajmoe Marso. Bouds o average delays ad queue size averages ad variaces i iput-queued cell-based switches. Proc. INFOCOM, [14] M. J. Neely, E. Modiao, ad C. E. Rohrs. Power allocatio ad routig i multi-beam satellites with time varyig chaels. IEEE Trasactios o Networkig, Feb [15] M. J. Neely. Dyamic Power Allocatio ad Routig for Satellite ad Wireless Networks with Time Varyig Chaels. PhD thesis, Massachusetts Istitute of Techology, LIDS, [16] M. J. Neely, E. Modiao, ad C. E Rohrs. Dyamic power allocatio ad routig for time varyig wireless etworks. IEEE Joural o Selected Areas i Commuicatios, Jauary [17] M. J. Neely. Eergy optimal cotrol for time varyig wireless etworks. Proceedigs of IEEE INFOCOM, March [18] M. J. Neely, E. Modiao, ad C. Li. Fairess ad optimal stochastic cotrol for heterogeeous etworks. Proceedigs of IEEE INFOCOM, March [19] R. Cruz ad A. Sathaam. Optimal routig, li schedulig, ad power cotrol i multi-hop wireless etworks. IEEE Proceedigs of INFOCOM, April [20] X. Liu, E. K. P. Chog, ad N. B. Shroff. A framework for opportuistic schedulig i wireless etworks. Computer Networks, vol. 41, o. 4, pp , March [21] A. Stolyar. Maximizig queueig etwork utility subject to stability: Greedy primal-dual algorithm. Queueig Systems, [22] J. W. Lee, R. R. Mazumdar, ad N. B. Shroff. Opportuistic power schedulig for dyamic multiserver wireless systems. to appear i IEEE Tras. o Wireless Systems, [23] C. Swaack, E. Uysal-Biyikoglu, ad G. Worell. Low complexity multiuser schedulig for maximizig throughput i the mimo broadcast chael. Proc. of 42d Allerto Coferece o Commuicatio, Cotrol, ad Computig, September [24] M. Kobayashi. O the Use of Multiple Ateas for the Dowli of Wireless Systems. PhD thesis, Ecole Natioale Superieure des Telecommuicatios, Paris, [25] M. Kobayashi, G. Caire, ad D. Gesbert. Impact of multiple trasmit ateas i a queued SDMA/TDMA dowli. I Proceedigs of 6th IEEE Workshop o Sigal Processig Advaces i Wireless Commuicatios (SPAWC), Jue [26] T. Ho ad H. Viswaatha. Dyamic algorithms for multicast with itrasessio etwork codig. I Proceedigs of 43rd Allerto Coferece o Comuicatio, Cotrol ad Computig, September [27] E. Yeh ad R. Berry. Throughput optimal cotrol of cooperative relay etworks. Proc. of Iteratioal Symposium o Iformatio Theory, Adelaide, Australia, pp , September [28] A. Khadai, J. Abouadi, E. Modiao, ad L. Zheg. Cooperative routig i wireless etworks. Allerto Coferece o Commuicatios, Cotrol, ad Computig, pp , Oct [29] X. Li ad N. B. Shroff. The impact of imperfect schedulig o crosslayer rate cotrol i wireless etworks. IEEE Proc. of INFOCOM, 2005.

The Fundamental Capacity-Delay Tradeoff in Large Mobile Ad Hoc Networks

The Fundamental Capacity-Delay Tradeoff in Large Mobile Ad Hoc Networks The Fudametal Capacity-Delay Tradeoff i Large Mobile Ad Hoc Networks Xiaoju Li ad Ness B. Shroff School of Electrical ad Computer Egieerig, Purdue Uiversity West Lafayette, IN 47907, U.S.A. {lix, shroff}@ec.purdue.edu

More information

Taking DCOP to the Real World: Efficient Complete Solutions for Distributed Multi-Event Scheduling

Taking DCOP to the Real World: Efficient Complete Solutions for Distributed Multi-Event Scheduling Taig DCOP to the Real World: Efficiet Complete Solutios for Distributed Multi-Evet Schedulig Rajiv T. Maheswara, Milid Tambe, Emma Bowrig, Joatha P. Pearce, ad Pradeep araatham Uiversity of Souther Califoria

More information

Modified Line Search Method for Global Optimization

Modified Line Search Method for Global Optimization Modified Lie Search Method for Global Optimizatio Cria Grosa ad Ajith Abraham Ceter of Excellece for Quatifiable Quality of Service Norwegia Uiversity of Sciece ad Techology Trodheim, Norway {cria, ajith}@q2s.tu.o

More information

On the Capacity of Hybrid Wireless Networks

On the Capacity of Hybrid Wireless Networks O the Capacity of Hybrid ireless Networks Beyua Liu,ZheLiu +,DoTowsley Departmet of Computer Sciece Uiversity of Massachusetts Amherst, MA 0002 + IBM T.J. atso Research Ceter P.O. Box 704 Yorktow Heights,

More information

Stock Market Trading via Stochastic Network Optimization

Stock Market Trading via Stochastic Network Optimization PROC. IEEE CONFERENCE ON DECISION AND CONTROL (CDC), ATLANTA, GA, DEC. 2010 1 Stock Market Tradig via Stochastic Network Optimizatio Michael J. Neely Uiversity of Souther Califoria http://www-rcf.usc.edu/

More information

Chapter 7 Methods of Finding Estimators

Chapter 7 Methods of Finding Estimators Chapter 7 for BST 695: Special Topics i Statistical Theory. Kui Zhag, 011 Chapter 7 Methods of Fidig Estimators Sectio 7.1 Itroductio Defiitio 7.1.1 A poit estimator is ay fuctio W( X) W( X1, X,, X ) of

More information

Capacity of Wireless Networks with Heterogeneous Traffic

Capacity of Wireless Networks with Heterogeneous Traffic Capacity of Wireless Networks with Heterogeeous Traffic Migyue Ji, Zheg Wag, Hamid R. Sadjadpour, J.J. Garcia-Lua-Aceves Departmet of Electrical Egieerig ad Computer Egieerig Uiversity of Califoria, Sata

More information

Asymptotic Growth of Functions

Asymptotic Growth of Functions CMPS Itroductio to Aalysis of Algorithms Fall 3 Asymptotic Growth of Fuctios We itroduce several types of asymptotic otatio which are used to compare the performace ad efficiecy of algorithms As we ll

More information

Properties of MLE: consistency, asymptotic normality. Fisher information.

Properties of MLE: consistency, asymptotic normality. Fisher information. Lecture 3 Properties of MLE: cosistecy, asymptotic ormality. Fisher iformatio. I this sectio we will try to uderstad why MLEs are good. Let us recall two facts from probability that we be used ofte throughout

More information

Vladimir N. Burkov, Dmitri A. Novikov MODELS AND METHODS OF MULTIPROJECTS MANAGEMENT

Vladimir N. Burkov, Dmitri A. Novikov MODELS AND METHODS OF MULTIPROJECTS MANAGEMENT Keywords: project maagemet, resource allocatio, etwork plaig Vladimir N Burkov, Dmitri A Novikov MODELS AND METHODS OF MULTIPROJECTS MANAGEMENT The paper deals with the problems of resource allocatio betwee

More information

*The most important feature of MRP as compared with ordinary inventory control analysis is its time phasing feature.

*The most important feature of MRP as compared with ordinary inventory control analysis is its time phasing feature. Itegrated Productio ad Ivetory Cotrol System MRP ad MRP II Framework of Maufacturig System Ivetory cotrol, productio schedulig, capacity plaig ad fiacial ad busiess decisios i a productio system are iterrelated.

More information

Domain 1 - Describe Cisco VoIP Implementations

Domain 1 - Describe Cisco VoIP Implementations Maual ONT (642-8) 1-800-418-6789 Domai 1 - Describe Cisco VoIP Implemetatios Advatages of VoIP Over Traditioal Switches Voice over IP etworks have may advatages over traditioal circuit switched voice etworks.

More information

Analyzing Longitudinal Data from Complex Surveys Using SUDAAN

Analyzing Longitudinal Data from Complex Surveys Using SUDAAN Aalyzig Logitudial Data from Complex Surveys Usig SUDAAN Darryl Creel Statistics ad Epidemiology, RTI Iteratioal, 312 Trotter Farm Drive, Rockville, MD, 20850 Abstract SUDAAN: Software for the Statistical

More information

Discrete Mathematics and Probability Theory Spring 2014 Anant Sahai Note 13

Discrete Mathematics and Probability Theory Spring 2014 Anant Sahai Note 13 EECS 70 Discrete Mathematics ad Probability Theory Sprig 2014 Aat Sahai Note 13 Itroductio At this poit, we have see eough examples that it is worth just takig stock of our model of probability ad may

More information

Non-life insurance mathematics. Nils F. Haavardsson, University of Oslo and DNB Skadeforsikring

Non-life insurance mathematics. Nils F. Haavardsson, University of Oslo and DNB Skadeforsikring No-life isurace mathematics Nils F. Haavardsso, Uiversity of Oslo ad DNB Skadeforsikrig Mai issues so far Why does isurace work? How is risk premium defied ad why is it importat? How ca claim frequecy

More information

I. Chi-squared Distributions

I. Chi-squared Distributions 1 M 358K Supplemet to Chapter 23: CHI-SQUARED DISTRIBUTIONS, T-DISTRIBUTIONS, AND DEGREES OF FREEDOM To uderstad t-distributios, we first eed to look at aother family of distributios, the chi-squared distributios.

More information

Your organization has a Class B IP address of 166.144.0.0 Before you implement subnetting, the Network ID and Host ID are divided as follows:

Your organization has a Class B IP address of 166.144.0.0 Before you implement subnetting, the Network ID and Host ID are divided as follows: Subettig Subettig is used to subdivide a sigle class of etwork i to multiple smaller etworks. Example: Your orgaizatio has a Class B IP address of 166.144.0.0 Before you implemet subettig, the Network

More information

INVESTMENT PERFORMANCE COUNCIL (IPC)

INVESTMENT PERFORMANCE COUNCIL (IPC) INVESTMENT PEFOMANCE COUNCIL (IPC) INVITATION TO COMMENT: Global Ivestmet Performace Stadards (GIPS ) Guidace Statemet o Calculatio Methodology The Associatio for Ivestmet Maagemet ad esearch (AIM) seeks

More information

Output Analysis (2, Chapters 10 &11 Law)

Output Analysis (2, Chapters 10 &11 Law) B. Maddah ENMG 6 Simulatio 05/0/07 Output Aalysis (, Chapters 10 &11 Law) Comparig alterative system cofiguratio Sice the output of a simulatio is radom, the comparig differet systems via simulatio should

More information

Interference Alignment and the Generalized Degrees of Freedom of the X Channel

Interference Alignment and the Generalized Degrees of Freedom of the X Channel Iterferece Aligmet ad the Geeralized Degrees of Freedom of the X Chael Chiachi Huag, Viveck R. Cadambe, Syed A. Jafar Electrical Egieerig ad Computer Sciece Uiversity of Califoria Irvie Irvie, Califoria,

More information

Research Article Sign Data Derivative Recovery

Research Article Sign Data Derivative Recovery Iteratioal Scholarly Research Network ISRN Applied Mathematics Volume 0, Article ID 63070, 7 pages doi:0.540/0/63070 Research Article Sig Data Derivative Recovery L. M. Housto, G. A. Glass, ad A. D. Dymikov

More information

1 Computing the Standard Deviation of Sample Means

1 Computing the Standard Deviation of Sample Means Computig the Stadard Deviatio of Sample Meas Quality cotrol charts are based o sample meas ot o idividual values withi a sample. A sample is a group of items, which are cosidered all together for our aalysis.

More information

Domain 1: Designing a SQL Server Instance and a Database Solution

Domain 1: Designing a SQL Server Instance and a Database Solution Maual SQL Server 2008 Desig, Optimize ad Maitai (70-450) 1-800-418-6789 Domai 1: Desigig a SQL Server Istace ad a Database Solutio Desigig for CPU, Memory ad Storage Capacity Requiremets Whe desigig a

More information

Systems Design Project: Indoor Location of Wireless Devices

Systems Design Project: Indoor Location of Wireless Devices Systems Desig Project: Idoor Locatio of Wireless Devices Prepared By: Bria Murphy Seior Systems Sciece ad Egieerig Washigto Uiversity i St. Louis Phoe: (805) 698-5295 Email: bcm1@cec.wustl.edu Supervised

More information

Chapter 6: Variance, the law of large numbers and the Monte-Carlo method

Chapter 6: Variance, the law of large numbers and the Monte-Carlo method Chapter 6: Variace, the law of large umbers ad the Mote-Carlo method Expected value, variace, ad Chebyshev iequality. If X is a radom variable recall that the expected value of X, E[X] is the average value

More information

On the Use of Adaptive OFDM to Preserve Energy in Ad Hoc Wireless Networks

On the Use of Adaptive OFDM to Preserve Energy in Ad Hoc Wireless Networks O the Use of Adaptive OFDM to Preserve Eergy i Ad Hoc Wireless etworks Kamol Kaemarugsi ad Prashat Krishamurthy Telecommuicatios Program, School of Iformatio Sciece, Uiversity of Pittsburgh 135 orth Bellefield

More information

In nite Sequences. Dr. Philippe B. Laval Kennesaw State University. October 9, 2008

In nite Sequences. Dr. Philippe B. Laval Kennesaw State University. October 9, 2008 I ite Sequeces Dr. Philippe B. Laval Keesaw State Uiversity October 9, 2008 Abstract This had out is a itroductio to i ite sequeces. mai de itios ad presets some elemetary results. It gives the I ite Sequeces

More information

Lecture 13. Lecturer: Jonathan Kelner Scribe: Jonathan Pines (2009)

Lecture 13. Lecturer: Jonathan Kelner Scribe: Jonathan Pines (2009) 18.409 A Algorithmist s Toolkit October 27, 2009 Lecture 13 Lecturer: Joatha Keler Scribe: Joatha Pies (2009) 1 Outlie Last time, we proved the Bru-Mikowski iequality for boxes. Today we ll go over the

More information

How to read A Mutual Fund shareholder report

How to read A Mutual Fund shareholder report Ivestor BulletI How to read A Mutual Fud shareholder report The SEC s Office of Ivestor Educatio ad Advocacy is issuig this Ivestor Bulleti to educate idividual ivestors about mutual fud shareholder reports.

More information

Department of Computer Science, University of Otago

Department of Computer Science, University of Otago Departmet of Computer Sciece, Uiversity of Otago Techical Report OUCS-2006-09 Permutatios Cotaiig May Patters Authors: M.H. Albert Departmet of Computer Sciece, Uiversity of Otago Micah Colema, Rya Fly

More information

Subject CT5 Contingencies Core Technical Syllabus

Subject CT5 Contingencies Core Technical Syllabus Subject CT5 Cotigecies Core Techical Syllabus for the 2015 exams 1 Jue 2014 Aim The aim of the Cotigecies subject is to provide a groudig i the mathematical techiques which ca be used to model ad value

More information

5 Boolean Decision Trees (February 11)

5 Boolean Decision Trees (February 11) 5 Boolea Decisio Trees (February 11) 5.1 Graph Coectivity Suppose we are give a udirected graph G, represeted as a boolea adjacecy matrix = (a ij ), where a ij = 1 if ad oly if vertices i ad j are coected

More information

Convexity, Inequalities, and Norms

Convexity, Inequalities, and Norms Covexity, Iequalities, ad Norms Covex Fuctios You are probably familiar with the otio of cocavity of fuctios. Give a twicedifferetiable fuctio ϕ: R R, We say that ϕ is covex (or cocave up) if ϕ (x) 0 for

More information

Engineering Data Management

Engineering Data Management BaaERP 5.0c Maufacturig Egieerig Data Maagemet Module Procedure UP128A US Documetiformatio Documet Documet code : UP128A US Documet group : User Documetatio Documet title : Egieerig Data Maagemet Applicatio/Package

More information

Week 3 Conditional probabilities, Bayes formula, WEEK 3 page 1 Expected value of a random variable

Week 3 Conditional probabilities, Bayes formula, WEEK 3 page 1 Expected value of a random variable Week 3 Coditioal probabilities, Bayes formula, WEEK 3 page 1 Expected value of a radom variable We recall our discussio of 5 card poker hads. Example 13 : a) What is the probability of evet A that a 5

More information

Institute of Actuaries of India Subject CT1 Financial Mathematics

Institute of Actuaries of India Subject CT1 Financial Mathematics Istitute of Actuaries of Idia Subject CT1 Fiacial Mathematics For 2014 Examiatios Subject CT1 Fiacial Mathematics Core Techical Aim The aim of the Fiacial Mathematics subject is to provide a groudig i

More information

CHAPTER 3 THE TIME VALUE OF MONEY

CHAPTER 3 THE TIME VALUE OF MONEY CHAPTER 3 THE TIME VALUE OF MONEY OVERVIEW A dollar i the had today is worth more tha a dollar to be received i the future because, if you had it ow, you could ivest that dollar ad ear iterest. Of all

More information

Incremental calculation of weighted mean and variance

Incremental calculation of weighted mean and variance Icremetal calculatio of weighted mea ad variace Toy Fich faf@cam.ac.uk dot@dotat.at Uiversity of Cambridge Computig Service February 009 Abstract I these otes I eplai how to derive formulae for umerically

More information

COMPARISON OF THE EFFICIENCY OF S-CONTROL CHART AND EWMA-S 2 CONTROL CHART FOR THE CHANGES IN A PROCESS

COMPARISON OF THE EFFICIENCY OF S-CONTROL CHART AND EWMA-S 2 CONTROL CHART FOR THE CHANGES IN A PROCESS COMPARISON OF THE EFFICIENCY OF S-CONTROL CHART AND EWMA-S CONTROL CHART FOR THE CHANGES IN A PROCESS Supraee Lisawadi Departmet of Mathematics ad Statistics, Faculty of Sciece ad Techoology, Thammasat

More information

Recovery time guaranteed heuristic routing for improving computation complexity in survivable WDM networks

Recovery time guaranteed heuristic routing for improving computation complexity in survivable WDM networks Computer Commuicatios 30 (2007) 1331 1336 wwwelseviercom/locate/comcom Recovery time guarateed heuristic routig for improvig computatio complexity i survivable WDM etworks Lei Guo * College of Iformatio

More information

Lecture 3. denote the orthogonal complement of S k. Then. 1 x S k. n. 2 x T Ax = ( ) λ x. with x = 1, we have. i = λ k x 2 = λ k.

Lecture 3. denote the orthogonal complement of S k. Then. 1 x S k. n. 2 x T Ax = ( ) λ x. with x = 1, we have. i = λ k x 2 = λ k. 18.409 A Algorithmist s Toolkit September 17, 009 Lecture 3 Lecturer: Joatha Keler Scribe: Adre Wibisoo 1 Outlie Today s lecture covers three mai parts: Courat-Fischer formula ad Rayleigh quotiets The

More information

Induced Cooperative Multi-user Diversity Relaying for Multi-hop Cellular Networks

Induced Cooperative Multi-user Diversity Relaying for Multi-hop Cellular Networks Iduced Cooperative Multi-user Diversity Relayig for Multi-hop Cellular Networks Keiva Navaie ad Halim Yaikomeroglu Broadbad Commuicatios ad Wireless Systems (BCWS) Cetre Departmet of Systems ad Computer

More information

The following example will help us understand The Sampling Distribution of the Mean. C1 C2 C3 C4 C5 50 miles 84 miles 38 miles 120 miles 48 miles

The following example will help us understand The Sampling Distribution of the Mean. C1 C2 C3 C4 C5 50 miles 84 miles 38 miles 120 miles 48 miles The followig eample will help us uderstad The Samplig Distributio of the Mea Review: The populatio is the etire collectio of all idividuals or objects of iterest The sample is the portio of the populatio

More information

0.7 0.6 0.2 0 0 96 96.5 97 97.5 98 98.5 99 99.5 100 100.5 96.5 97 97.5 98 98.5 99 99.5 100 100.5

0.7 0.6 0.2 0 0 96 96.5 97 97.5 98 98.5 99 99.5 100 100.5 96.5 97 97.5 98 98.5 99 99.5 100 100.5 Sectio 13 Kolmogorov-Smirov test. Suppose that we have a i.i.d. sample X 1,..., X with some ukow distributio P ad we would like to test the hypothesis that P is equal to a particular distributio P 0, i.e.

More information

The analysis of the Cournot oligopoly model considering the subjective motive in the strategy selection

The analysis of the Cournot oligopoly model considering the subjective motive in the strategy selection The aalysis of the Courot oligopoly model cosiderig the subjective motive i the strategy selectio Shigehito Furuyama Teruhisa Nakai Departmet of Systems Maagemet Egieerig Faculty of Egieerig Kasai Uiversity

More information

Chatpun Khamyat Department of Industrial Engineering, Kasetsart University, Bangkok, Thailand ocpky@hotmail.com

Chatpun Khamyat Department of Industrial Engineering, Kasetsart University, Bangkok, Thailand ocpky@hotmail.com SOLVING THE OIL DELIVERY TRUCKS ROUTING PROBLEM WITH MODIFY MULTI-TRAVELING SALESMAN PROBLEM APPROACH CASE STUDY: THE SME'S OIL LOGISTIC COMPANY IN BANGKOK THAILAND Chatpu Khamyat Departmet of Idustrial

More information

THE ABRACADABRA PROBLEM

THE ABRACADABRA PROBLEM THE ABRACADABRA PROBLEM FRANCESCO CARAVENNA Abstract. We preset a detailed solutio of Exercise E0.6 i [Wil9]: i a radom sequece of letters, draw idepedetly ad uiformly from the Eglish alphabet, the expected

More information

MARTINGALES AND A BASIC APPLICATION

MARTINGALES AND A BASIC APPLICATION MARTINGALES AND A BASIC APPLICATION TURNER SMITH Abstract. This paper will develop the measure-theoretic approach to probability i order to preset the defiitio of martigales. From there we will apply this

More information

Notes on exponential generating functions and structures.

Notes on exponential generating functions and structures. Notes o expoetial geeratig fuctios ad structures. 1. The cocept of a structure. Cosider the followig coutig problems: (1) to fid for each the umber of partitios of a -elemet set, (2) to fid for each the

More information

SECTION 1.5 : SUMMATION NOTATION + WORK WITH SEQUENCES

SECTION 1.5 : SUMMATION NOTATION + WORK WITH SEQUENCES SECTION 1.5 : SUMMATION NOTATION + WORK WITH SEQUENCES Read Sectio 1.5 (pages 5 9) Overview I Sectio 1.5 we lear to work with summatio otatio ad formulas. We will also itroduce a brief overview of sequeces,

More information

CHAPTER 3 DIGITAL CODING OF SIGNALS

CHAPTER 3 DIGITAL CODING OF SIGNALS CHAPTER 3 DIGITAL CODING OF SIGNALS Computers are ofte used to automate the recordig of measuremets. The trasducers ad sigal coditioig circuits produce a voltage sigal that is proportioal to a quatity

More information

Project Deliverables. CS 361, Lecture 28. Outline. Project Deliverables. Administrative. Project Comments

Project Deliverables. CS 361, Lecture 28. Outline. Project Deliverables. Administrative. Project Comments Project Deliverables CS 361, Lecture 28 Jared Saia Uiversity of New Mexico Each Group should tur i oe group project cosistig of: About 6-12 pages of text (ca be loger with appedix) 6-12 figures (please

More information

INVESTMENT PERFORMANCE COUNCIL (IPC) Guidance Statement on Calculation Methodology

INVESTMENT PERFORMANCE COUNCIL (IPC) Guidance Statement on Calculation Methodology Adoptio Date: 4 March 2004 Effective Date: 1 Jue 2004 Retroactive Applicatio: No Public Commet Period: Aug Nov 2002 INVESTMENT PERFORMANCE COUNCIL (IPC) Preface Guidace Statemet o Calculatio Methodology

More information

Chapter 5 O A Cojecture Of Erdíos Proceedigs NCUR VIII è1994è, Vol II, pp 794í798 Jeærey F Gold Departmet of Mathematics, Departmet of Physics Uiversity of Utah Do H Tucker Departmet of Mathematics Uiversity

More information

MTO-MTS Production Systems in Supply Chains

MTO-MTS Production Systems in Supply Chains NSF GRANT #0092854 NSF PROGRAM NAME: MES/OR MTO-MTS Productio Systems i Supply Chais Philip M. Kamisky Uiversity of Califoria, Berkeley Our Kaya Uiversity of Califoria, Berkeley Abstract: Icreasig cost

More information

Estimating Probability Distributions by Observing Betting Practices

Estimating Probability Distributions by Observing Betting Practices 5th Iteratioal Symposium o Imprecise Probability: Theories ad Applicatios, Prague, Czech Republic, 007 Estimatig Probability Distributios by Observig Bettig Practices Dr C Lych Natioal Uiversity of Irelad,

More information

IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 27, NO. 2, FEBRUARY 2009 117

IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 27, NO. 2, FEBRUARY 2009 117 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 27, NO. 2, FEBRUARY 2009 7 Capacity ad Delay of Hybrid ireless Broadbad Access Networks Pa Li, Chi Zhag, Studet Member, IEEE, ad Yuguag Fag, Fellow,

More information

The Forgotten Middle. research readiness results. Executive Summary

The Forgotten Middle. research readiness results. Executive Summary The Forgotte Middle Esurig that All Studets Are o Target for College ad Career Readiess before High School Executive Summary Today, college readiess also meas career readiess. While ot every high school

More information

Matrix Model of Trust Management in P2P Networks

Matrix Model of Trust Management in P2P Networks Matrix Model of Trust Maagemet i P2P Networks Miroslav Novotý, Filip Zavoral Faculty of Mathematics ad Physics Charles Uiversity Prague, Czech Republic miroslav.ovoty@mff.cui.cz Abstract The trust maagemet

More information

Running Time ( 3.1) Analysis of Algorithms. Experimental Studies ( 3.1.1) Limitations of Experiments. Pseudocode ( 3.1.2) Theoretical Analysis

Running Time ( 3.1) Analysis of Algorithms. Experimental Studies ( 3.1.1) Limitations of Experiments. Pseudocode ( 3.1.2) Theoretical Analysis Ruig Time ( 3.) Aalysis of Algorithms Iput Algorithm Output A algorithm is a step-by-step procedure for solvig a problem i a fiite amout of time. Most algorithms trasform iput objects ito output objects.

More information

Tradigms of Astundithi and Toyota

Tradigms of Astundithi and Toyota Tradig the radomess - Desigig a optimal tradig strategy uder a drifted radom walk price model Yuao Wu Math 20 Project Paper Professor Zachary Hamaker Abstract: I this paper the author iteds to explore

More information

C.Yaashuwanth Department of Electrical and Electronics Engineering, Anna University Chennai, Chennai 600 025, India..

C.Yaashuwanth Department of Electrical and Electronics Engineering, Anna University Chennai, Chennai 600 025, India.. (IJCSIS) Iteratioal Joural of Computer Sciece ad Iformatio Security, A New Schedulig Algorithms for Real Time Tasks C.Yaashuwath Departmet of Electrical ad Electroics Egieerig, Aa Uiversity Cheai, Cheai

More information

Universal coding for classes of sources

Universal coding for classes of sources Coexios module: m46228 Uiversal codig for classes of sources Dever Greee This work is produced by The Coexios Project ad licesed uder the Creative Commos Attributio Licese We have discussed several parametric

More information

3 Basic Definitions of Probability Theory

3 Basic Definitions of Probability Theory 3 Basic Defiitios of Probability Theory 3defprob.tex: Feb 10, 2003 Classical probability Frequecy probability axiomatic probability Historical developemet: Classical Frequecy Axiomatic The Axiomatic defiitio

More information

iprox sensors iprox inductive sensors iprox programming tools ProxView programming software iprox the world s most versatile proximity sensor

iprox sensors iprox inductive sensors iprox programming tools ProxView programming software iprox the world s most versatile proximity sensor iprox sesors iprox iductive sesors iprox programmig tools ProxView programmig software iprox the world s most versatile proximity sesor The world s most versatile proximity sesor Eato s iproxe is syoymous

More information

The Stable Marriage Problem

The Stable Marriage Problem The Stable Marriage Problem William Hut Lae Departmet of Computer Sciece ad Electrical Egieerig, West Virgiia Uiversity, Morgatow, WV William.Hut@mail.wvu.edu 1 Itroductio Imagie you are a matchmaker,

More information

(VCP-310) 1-800-418-6789

(VCP-310) 1-800-418-6789 Maual VMware Lesso 1: Uderstadig the VMware Product Lie I this lesso, you will first lear what virtualizatio is. Next, you ll explore the products offered by VMware that provide virtualizatio services.

More information

Confidence Intervals for One Mean

Confidence Intervals for One Mean Chapter 420 Cofidece Itervals for Oe Mea Itroductio This routie calculates the sample size ecessary to achieve a specified distace from the mea to the cofidece limit(s) at a stated cofidece level for a

More information

Section 11.3: The Integral Test

Section 11.3: The Integral Test Sectio.3: The Itegral Test Most of the series we have looked at have either diverged or have coverged ad we have bee able to fid what they coverge to. I geeral however, the problem is much more difficult

More information

Case Study. Normal and t Distributions. Density Plot. Normal Distributions

Case Study. Normal and t Distributions. Density Plot. Normal Distributions Case Study Normal ad t Distributios Bret Halo ad Bret Larget Departmet of Statistics Uiversity of Wiscosi Madiso October 11 13, 2011 Case Study Body temperature varies withi idividuals over time (it ca

More information

Multi-server Optimal Bandwidth Monitoring for QoS based Multimedia Delivery Anup Basu, Irene Cheng and Yinzhe Yu

Multi-server Optimal Bandwidth Monitoring for QoS based Multimedia Delivery Anup Basu, Irene Cheng and Yinzhe Yu Multi-server Optimal Badwidth Moitorig for QoS based Multimedia Delivery Aup Basu, Iree Cheg ad Yizhe Yu Departmet of Computig Sciece U. of Alberta Architecture Applicatio Layer Request receptio -coectio

More information

Stochastic Online Scheduling with Precedence Constraints

Stochastic Online Scheduling with Precedence Constraints Stochastic Olie Schedulig with Precedece Costraits Nicole Megow Tark Vredeveld July 15, 2008 Abstract We cosider the preemptive ad o-preemptive problems of schedulig obs with precedece costraits o parallel

More information

where: T = number of years of cash flow in investment's life n = the year in which the cash flow X n i = IRR = the internal rate of return

where: T = number of years of cash flow in investment's life n = the year in which the cash flow X n i = IRR = the internal rate of return EVALUATING ALTERNATIVE CAPITAL INVESTMENT PROGRAMS By Ke D. Duft, Extesio Ecoomist I the March 98 issue of this publicatio we reviewed the procedure by which a capital ivestmet project was assessed. The

More information

Entropy of bi-capacities

Entropy of bi-capacities Etropy of bi-capacities Iva Kojadiovic LINA CNRS FRE 2729 Site école polytechique de l uiv. de Nates Rue Christia Pauc 44306 Nates, Frace iva.kojadiovic@uiv-ates.fr Jea-Luc Marichal Applied Mathematics

More information

Convention Paper 6764

Convention Paper 6764 Audio Egieerig Society Covetio Paper 6764 Preseted at the 10th Covetio 006 May 0 3 Paris, Frace This covetio paper has bee reproduced from the author's advace mauscript, without editig, correctios, or

More information

Rainbow options. A rainbow is an option on a basket that pays in its most common form, a nonequally

Rainbow options. A rainbow is an option on a basket that pays in its most common form, a nonequally Raibow optios INRODUCION A raibow is a optio o a basket that pays i its most commo form, a oequally weighted average of the assets of the basket accordig to their performace. he umber of assets is called

More information

Class Meeting # 16: The Fourier Transform on R n

Class Meeting # 16: The Fourier Transform on R n MATH 18.152 COUSE NOTES - CLASS MEETING # 16 18.152 Itroductio to PDEs, Fall 2011 Professor: Jared Speck Class Meetig # 16: The Fourier Trasform o 1. Itroductio to the Fourier Trasform Earlier i the course,

More information

Here are a couple of warnings to my students who may be here to get a copy of what happened on a day that you missed.

Here are a couple of warnings to my students who may be here to get a copy of what happened on a day that you missed. This documet was writte ad copyrighted by Paul Dawkis. Use of this documet ad its olie versio is govered by the Terms ad Coditios of Use located at http://tutorial.math.lamar.edu/terms.asp. The olie versio

More information

1 Correlation and Regression Analysis

1 Correlation and Regression Analysis 1 Correlatio ad Regressio Aalysis I this sectio we will be ivestigatig the relatioship betwee two cotiuous variable, such as height ad weight, the cocetratio of a ijected drug ad heart rate, or the cosumptio

More information

TIGHT BOUNDS ON EXPECTED ORDER STATISTICS

TIGHT BOUNDS ON EXPECTED ORDER STATISTICS Probability i the Egieerig ad Iformatioal Scieces, 20, 2006, 667 686+ Prited i the U+S+A+ TIGHT BOUNDS ON EXPECTED ORDER STATISTICS DIMITRIS BERTSIMAS Sloa School of Maagemet ad Operatios Research Ceter

More information

THE REGRESSION MODEL IN MATRIX FORM. For simple linear regression, meaning one predictor, the model is. for i = 1, 2, 3,, n

THE REGRESSION MODEL IN MATRIX FORM. For simple linear regression, meaning one predictor, the model is. for i = 1, 2, 3,, n We will cosider the liear regressio model i matrix form. For simple liear regressio, meaig oe predictor, the model is i = + x i + ε i for i =,,,, This model icludes the assumptio that the ε i s are a sample

More information

The Power of Free Branching in a General Model of Backtracking and Dynamic Programming Algorithms

The Power of Free Branching in a General Model of Backtracking and Dynamic Programming Algorithms The Power of Free Brachig i a Geeral Model of Backtrackig ad Dyamic Programmig Algorithms SASHKA DAVIS IDA/Ceter for Computig Scieces Bowie, MD sashka.davis@gmail.com RUSSELL IMPAGLIAZZO Dept. of Computer

More information

Hypergeometric Distributions

Hypergeometric Distributions 7.4 Hypergeometric Distributios Whe choosig the startig lie-up for a game, a coach obviously has to choose a differet player for each positio. Similarly, whe a uio elects delegates for a covetio or you

More information

A probabilistic proof of a binomial identity

A probabilistic proof of a binomial identity A probabilistic proof of a biomial idetity Joatho Peterso Abstract We give a elemetary probabilistic proof of a biomial idetity. The proof is obtaied by computig the probability of a certai evet i two

More information

1 The Gaussian channel

1 The Gaussian channel ECE 77 Lecture 0 The Gaussia chael Objective: I this lecture we will lear about commuicatio over a chael of practical iterest, i which the trasmitted sigal is subjected to additive white Gaussia oise.

More information

Lecture 4: Cauchy sequences, Bolzano-Weierstrass, and the Squeeze theorem

Lecture 4: Cauchy sequences, Bolzano-Weierstrass, and the Squeeze theorem Lecture 4: Cauchy sequeces, Bolzao-Weierstrass, ad the Squeeze theorem The purpose of this lecture is more modest tha the previous oes. It is to state certai coditios uder which we are guarateed that limits

More information

Sequences and Series

Sequences and Series CHAPTER 9 Sequeces ad Series 9.. Covergece: Defiitio ad Examples Sequeces The purpose of this chapter is to itroduce a particular way of geeratig algorithms for fidig the values of fuctios defied by their

More information

International Journal on Emerging Technologies 1(2): 48-56(2010) ISSN : 0975-8364

International Journal on Emerging Technologies 1(2): 48-56(2010) ISSN : 0975-8364 e t Iteratioal Joural o Emergig Techologies (): 48-56(00) ISSN : 0975-864 Dyamic load balacig i distributed ad high performace parallel eterprise computig by embeddig MPI ad ope MP Sadip S. Chauha, Sadip

More information

Evaluating Model for B2C E- commerce Enterprise Development Based on DEA

Evaluating Model for B2C E- commerce Enterprise Development Based on DEA , pp.180-184 http://dx.doi.org/10.14257/astl.2014.53.39 Evaluatig Model for B2C E- commerce Eterprise Developmet Based o DEA Weli Geg, Jig Ta Computer ad iformatio egieerig Istitute, Harbi Uiversity of

More information

Cantilever Beam Experiment

Cantilever Beam Experiment Mechaical Egieerig Departmet Uiversity of Massachusetts Lowell Catilever Beam Experimet Backgroud A disk drive maufacturer is redesigig several disk drive armature mechaisms. This is the result of evaluatio

More information

Determining the sample size

Determining the sample size Determiig the sample size Oe of the most commo questios ay statisticia gets asked is How large a sample size do I eed? Researchers are ofte surprised to fid out that the aswer depeds o a umber of factors

More information

Chair for Network Architectures and Services Institute of Informatics TU München Prof. Carle. Network Security. Chapter 2 Basics

Chair for Network Architectures and Services Institute of Informatics TU München Prof. Carle. Network Security. Chapter 2 Basics Chair for Network Architectures ad Services Istitute of Iformatics TU Müche Prof. Carle Network Security Chapter 2 Basics 2.4 Radom Number Geeratio for Cryptographic Protocols Motivatio It is crucial to

More information

Normal Distribution.

Normal Distribution. Normal Distributio www.icrf.l Normal distributio I probability theory, the ormal or Gaussia distributio, is a cotiuous probability distributio that is ofte used as a first approimatio to describe realvalued

More information

Advanced Methods for Security Constrained Financial Transmission Rights (FTR)

Advanced Methods for Security Constrained Financial Transmission Rights (FTR) dvaced Methods for Security ostraied Fiacial Trasmissio Rights (FTR) Stephe Elbert, Steve.Elbert@PNN.gov Kara Kalsi, Kurt Glaesema, Mark Rice, Maria Vlachopoulou, Nig Zhou Pacific Northwest Natioal aboratory,

More information

Perfect Packing Theorems and the Average-Case Behavior of Optimal and Online Bin Packing

Perfect Packing Theorems and the Average-Case Behavior of Optimal and Online Bin Packing SIAM REVIEW Vol. 44, No. 1, pp. 95 108 c 2002 Society for Idustrial ad Applied Mathematics Perfect Packig Theorems ad the Average-Case Behavior of Optimal ad Olie Bi Packig E. G. Coffma, Jr. C. Courcoubetis

More information

CME 302: NUMERICAL LINEAR ALGEBRA FALL 2005/06 LECTURE 8

CME 302: NUMERICAL LINEAR ALGEBRA FALL 2005/06 LECTURE 8 CME 30: NUMERICAL LINEAR ALGEBRA FALL 005/06 LECTURE 8 GENE H GOLUB 1 Positive Defiite Matrices A matrix A is positive defiite if x Ax > 0 for all ozero x A positive defiite matrix has real ad positive

More information

Page 1. Real Options for Engineering Systems. What are we up to? Today s agenda. J1: Real Options for Engineering Systems. Richard de Neufville

Page 1. Real Options for Engineering Systems. What are we up to? Today s agenda. J1: Real Options for Engineering Systems. Richard de Neufville Real Optios for Egieerig Systems J: Real Optios for Egieerig Systems By (MIT) Stefa Scholtes (CU) Course website: http://msl.mit.edu/cmi/ardet_2002 Stefa Scholtes Judge Istitute of Maagemet, CU Slide What

More information

Automatic Tuning for FOREX Trading System Using Fuzzy Time Series

Automatic Tuning for FOREX Trading System Using Fuzzy Time Series utomatic Tuig for FOREX Tradig System Usig Fuzzy Time Series Kraimo Maeesilp ad Pitihate Soorasa bstract Efficiecy of the automatic currecy tradig system is time depedet due to usig fixed parameters which

More information

Overview on S-Box Design Principles

Overview on S-Box Design Principles Overview o S-Box Desig Priciples Debdeep Mukhopadhyay Assistat Professor Departmet of Computer Sciece ad Egieerig Idia Istitute of Techology Kharagpur INDIA -721302 What is a S-Box? S-Boxes are Boolea

More information

FOUNDATIONS OF MATHEMATICS AND PRE-CALCULUS GRADE 10

FOUNDATIONS OF MATHEMATICS AND PRE-CALCULUS GRADE 10 FOUNDATIONS OF MATHEMATICS AND PRE-CALCULUS GRADE 10 [C] Commuicatio Measuremet A1. Solve problems that ivolve liear measuremet, usig: SI ad imperial uits of measure estimatio strategies measuremet strategies.

More information