Self-Sustainability of Energy Harvesting Systems: Concept, Analysis, and Design

Save this PDF as:
 WORD  PNG  TXT  JPG

Size: px
Start display at page:

Download "Self-Sustainability of Energy Harvesting Systems: Concept, Analysis, and Design"

Transcription

1 1 Self-Sustainability of Energy arvesting Systems: Concet, Analysis, and Design Sudarshan Guruacharya and Ekram ossain Deartment of Electrical and Comuter Engineering, University of Manitoba, Canada s: arxiv: v2 [cs.it] 2 May 217 Abstract Ambient energy harvesting is touted as a low cost solution to rolong the life of low-owered devices, reduce the carbon footrint, and make the system self-sustainable. Most research to date have focused either on the hysical asects of energy conversion rocess or on otimal consumtion olicy of the harvested energy at the system level. A concet closely associated with the energy harvesting system is that of self-sustainability. In the context of energy harvesting systems, self-sustainability refers to the ability of the system to rovide the necessary energy for the consumer, without the need for external grid ower. owever, although intuitively understood, to the best of our knowledge, the idea of self-sustainability is yet to be made recise and studied as a erformance metric. In this aer, we rovide a mathematical definition of the concet of self-sustainability of an energy harvesting system, based on the comlementary idea of eventual energy outage. In articular, we analyse the harveststore-consume system with infinite battery caacity, stochastic energy arrivals, and fixed energy consumtion rate. Using the random walk theory, two basic insights are obtained: 1) The system is self-sustainable when the rate of energy consumtion is strictly less the rate of energy harvest, and 2) assuming the first condition is satisfied, the robability of eventual energy outage is exonentially uer bounded. This uer bound guarantees that the eventual energy outage robability can be made arbitrarily small simly by increasing the initial battery energy. These gives us a valuable design insight. General formulas are given for the self-sustainability robability in the form of integral equations. For the secial case when the energy arrival follows a Poisson rocess, we are able to find the exact formulas for the eventual outage robability. We also show that the harvest-store-consume system is mathematically equivalent to a GI/G/1 queueing system, which allows us to easily find the outage robability, in case the necessary condition for self-sustainability is violated. Monte-Carlo simulations verify our analysis. Index Terms Energy harvesting, self-sustainability, eventual energy outage, random walks, renewal theory, martingale I. INTRODUCTION In recent years, ambient energy harvesting and its alications have become a toic to great interest. Basically, a device is assumed to be able to harvest energy from random energy source in their surrounding environment for their future use. Such methods are useful when devices are low owered and energy constrained. They can hel in extending the life time of a device and lower its maintenance cost. Furthermore, they can hel lower carbon dioxide emissions and fight climate change. The readers are referred to [1] [5] for general surveys of this field. The work was suorted by a CRD grant from the Natural Sciences and Engineering Research Council of Canada (NSERC). Aart from traditional sources of ambient energy such as solar, wind, and wave, in the ast decade research has extended to energy scavenging techniques from diverse energy sources [6], [7] such as thermal [8], [9], ressure and vibrations [8], [1], ambient radio-frequency (RF) radiation [11], [12], bodily motions [13], magnetic field [14], ambient sound [15], and ambient light [16], [17]. Such energy harvesting techniques have been alied in the context of various technologies such as wireless sensor networks [18] [21], telecommunications [22] [24], cellular networks [25], cognitive radio network [26], [27], vehicular network [28], health care [29], [3], IoT (Internet of things) [31], [32], IoE (Internet of energy) [33], and smart grid [34] [36] technology. A. Nature of Energy Sources and General Architectures The ambient energy sources can be broadly classified as steady energy source or stochastic energy source. These diverse sources of energy are converted into electricity, which is then either directly consumed, or stored in a rechargeable battery for future use. As with harvesting, consumtion of energy can also be either steady or stochastic. Three general architectures of energy harvesting network are harvest-consume, harvest-store-consume, and harvest-store/consume [5]. In the harvest-consume model, the harvested energy is immediately consumed by the consumer. This model is aroriate when a steady suly of harvested energy can be guaranteed and a battery-free circuit is desired. The major roblem with this aroach is that, due to ossible random nature of the energy source, when the harvested energy is less than the minimum oerational energy required by consuming device, the device is disabled. We say that the consumer has exerienced an energy outage. In harvest-store-consume model, the harvested energy is first transferred to an energy storage facility or a rechargeable battery. The consumer then accesses the harvested energy from the battery. By this method the consumer can ensure a steady suly of energy, even though the harvested energy is randomly varying. This method is aroriate when the harvested energy is very small and is highly fluctuating, as in ambient RF energy harvesting. We can also have a hybrid model given by harveststore/consume. In this model the harvest energy is either stored for some future use or directly consumed. If the harvested energy is above minimum requirement of the consumer, then the excess energy can also be stored. This model is aroriate when the harvested energy does not fluctuate very raidly.

2 2 B. Motivation: Energy Outage, Eventual Energy Outage, and Self-Sustainability A lot of research attention has been aid either on the hysical asects of energy harvesting mechanism [6] [17] or on the consumtion olicy of the harvested energy [18] [36]. Work dealing with former issue tend to focus on hysical modeling and otimization of energy harvesting devices, with the goal of imroving the efficiency of energy conversion rocess. Work dealing with the latter issue tend to focus on otimizing some erformance metric, under energy constrains of the energy harvesting device. Clearly, the work of the latter category is affected by the general architecture assumed. Also, the exact nature of the metric deends on what the harvested energy is being used for. For instance, in the field of energy harvesting communication, when the harvest-consume model is used, a ossible natural erformance metric is the joint energy and information outage robability [37], [38]. If instead the harvest-store-consume model is used, a ossible erformance metric is the data rate of the communication system [23]. While there have been lenty of work that have examined systems that have energy harvesting caability, very few work have focused on examining the concet of energy outage of the consumer. This is unfortunate because energy outage should be one of the key erformance metric of any energy harvesting system. The consumer is said to exerience energy outage if there is no energy available for its consumtion. For the harvest-consume model, the energy outage is determined by the randomness of the harvester-to-consumer energy transmission channel or the inherent randomness of the energy source. In [37], [38] the energy arrival rocess is assumed to be a stationary, ergodic rocess. As such, knowing the distribution of energy arrival allows us to calculate the energy outage robability. For the harvest-store-consume model, the harvested energy is modeled as discrete ackets of ossibly variable size [23]. The energy arrival rocess at the battery is generally modeled as a Poisson rocess, allowing the battery state to be modeled as a Markov chain [39]. owever, research on energy outage is sorely lacking for harvest-store-consume architecture. Another concet closely associated with energy harvesting systems is the concet of self-sustainability. Self-sustainability is commonly understood as the ability to suly one s own needs without external assistance. In the context of energy harvesting systems, it would mean the ability of the system to rovide the necessary energy for the consumer, without the need for external grid ower. To the best of our knowledge, this concet is yet to be mathematically defined. In this aer, we define this concet as follows: The self-sustainability of an energy harvesting system is the robability that the consumer will not eventually exerience an energy outage. ere we need to make a distinction between energy outage and eventual energy outage. We say that a consumer exeriences an eventual energy outage if the consumer undergoes an energy outage within finite time. If the consumer has to wait for infinite amount of time to exerience an energy outage, then we say that the energy harvesting system is self-sustainable. The eventual energy outage and the self-sustainability are comlementary in that the sum of their robabilities is unity. Symbolically, we will denote the self-sustainability robability by φ and eventual energy outage robability by ψ such that φ + ψ = 1. We will say that a system is self-sustainable if φ > (or equivalently ψ < 1) and unsustainable if φ = (or equivalently ψ = 1). Given these basic definitions, we can quickly oint out that for harvest-consume system, if both the energy harvest and consumtion rates are steady, then the system is selfsustainable if the rate of consumtion is less than the rate of harvest. Likewise, if the energy harvest is stochastic and the consumtion is steady (or vice versa), then the consumer will almost surely exerience an eventual outage. ence the harvest-consume system is not a self-sustaining architecture for stochastic energy harvest or consumtion. In general, we want the answers to the following questions: 1) Under what condition is self-sustainability ossible? 2) Can we come u with a formula or a bound for selfsustainability robability? 3) ow should we design a system, rovided a constraint on the eventual energy outage robability? C. Contributions The main contribution of this work is the establishment of the concet of self-sustainability for energy harvesting system. We secifically study the harvest-store-consume architecture, for the case where energy arrives imulsively, storage caacity is infinite and the rate of consumtion is constant. We analyze the system based on random walks, renewal theory, and martingales. In articular, we are able to adat many ideas from the ruin theory of actuarial science, making our work cross-discilinary. This leads to the following findings: 1) The necessary condition for harvest-store-consume system to be self-sustaining is simly that the rate of consumtion be strictly less than the rate of harvest. That is, λ X >, where λ is the arrival rate, X is the average energy of an imulse, and is the fixed consumed ower. In this aer, we refer this condition as the selfsustainability condition. 2) We rovide general formulas for the self-sustainability robability of harvest-store-consume system. In articular, we demonstrate the relationshi between the selfsustainability robability and the maximum of the underlying random walk using three different formulas. Two of these formulas are renewal tye and Weiner-of tye integral equations. 3) Since finding the analytical solution to the integral equations is difficult, using the concet of martingales, we rovide an exonential uer bound to eventual energy outage robability for harvest-store-consume system, rovided that the self-sustainability condition is satisfied. That is, ψ(u ) e r u, where u is the initial battery energy and r is some constant. We show that the eventual energy outage can be made arbitrarily small by simly increasing the initial battery energy. This leads

3 3 to simle design guidelines, given the eventual energy outage constraint. 4) Using the renewal tye integral equation for the selfsustainability robability, we rovide an asymtotic formula for the eventual energy outage robability based on key renewal theorem for defective distributions. That is, ψ(u ) Ce r u, for some constant C. 5) For the secial case when the arrival of energy ackets is modeled as a Poisson rocess, we give exact formulas for the eventual energy outage robability. That is, ψ(u ) = (1 r u λ )e r. 6) We rove that harvest-store-consume system is mathematically equivalent to a GI/G/1 queuing system and draw arallels between the two systems. This allows us to imort the results from queueing theory when the self-sustainability condition is not satisfied. While this is certainly not a new observation [23], we rovide a systematic roof of the equivalence and connect it to the idea of energy outage. From here on, without any ambiguity, we will simly refer to the energy outage as the outage and the eventual energy outage as the eventual outage. D. Organization The rest of the aer is organized as follows: Section II discusses the system model and assumtions, and also defines the robability of self-sustainability. Section III gives the random walk analysis of the energy surlus rocess and the self-sustainability robability. Section IV gives an exonential uer bound on eventual outage robability (which is the comlement of self-sustainability robability). Section V gives an asymtotic aroximation of the eventual outage robability and discusses the comutation of the adjustment coefficient, while Section VI studies the secial case when the energy arrivals is a Poisson rocess. Section VII examines the battery energy rocess, while Section VIII gives a numerical verification of the obtained formulas. Lastly, Section IX concludes the aer. II. SYSTEM MODEL, ASSUMPTIONS, AND TE CONCEPT OF SELF-SUSTAINABILITY A. Definitions and Assumtions We will consider a harvest-store-consume (SC) system. That is, all the harvested energy is first collected, before being consumed. Thus, the consumer obtains the harvested energy indirectly from the storage facility (or the rechargeable battery). The harvested energy can arrive into the storage in continuous fashion (as in solar or wind or grid ower source) or in imulsive fashion. We will restrict our analysis to the case of imulsive energy arrivals. Thus, our main hysical assumtion is as follows: Assumtion 1. arvested energy arrives as imulses into the storage system. In other words, the harvested energy arrival is a countable rocess. The harvested energy arrives in the form of ackets into the storage system, and the size of each energy ackets may vary randomly. As such, we can model the energy surlus rocess of the system at any time t as U(t) = u t N(t) (u, t)dt + X i h(t t i ). (1) ere u > is the initial battery energy and (u, t) > is the ower consumtion from the battery, i.e. du dt = (u, t). The X i R + is the amount of energy in an i-th energy acket that arrives at time t i, while N(t) = max{i : t i t} is the total number of energy ackets that have arrived at the storage by time t. Lastly, h(t) is the transient resonse of battery charging rocess, such that h(t) = for t < and h( ) = 1. Clearly, if U(t) >, then the system is roducing more energy than it is consuming. Likewise, if U(t) <, then the consumer takes in energy from the grid to comensate for energy deficit. We will now define a few concets that will be used in the aer: i=1 Definition 1 (Defective and roer distributions). A random variable with distribution F is said to be defective if lim x F (x) < 1, the amount of defect being 1 F ( ). The random variable is said to be roer if lim x F (x) = 1. Definition 2 (Renewal rocess). A sequence {S n } is a renewal rocess if S n = X 1 + +X n and S =, where {X i } are mutually indeendent, non-negative random variables with common distribution F X such that F X () =. When F X is a roer distribution, the renewal rocess is said to be a ersistent renewal rocess. If F X is defective then the renewal rocess is said to be a transient or terminating renewal rocess. The variables X i in Definition 2 is often interreted as interarrival time. owever, X i need not always be time, as we will see later in the aer. We will now make further mathematical assumtions required to simlify our analysis. We will be working with the resulting simlified model for the rest of the aer. Assumtion 2. The storage caacity of the battery is infinite. Assumtion 3. The rate of energy consumtion is constant, i.e. (u, t) =. Assumtion 4. The transient resonse of battery charging rocess, h(t), is a unit ste function. Assumtion 5. The energy ackets {E i } arrive at corresonding times {t i }, for i =, 1, 2..., such that = t < t 1 < t 2 <, where the energy acket E arrives at time. The arrival times {t i } is (i) a renewal rocess. That is, the inter-arrival times {A i }, where A i = t i+1 t i, are such that A, A 1, A 2,... are indeendent and identically distributed. (ii) To avoid more than one energy arrivals at a time, we assume F A () =. (iii) Lastly, the exected inter-arrival time is finite, E[A i ] <. Assumtion 6. The amount of energy {X i } corresonding to the energy ackets {E i } for i =, 1, 2,..., is (i) a nonnegative, continuous random variable; (ii) {X i } are indeendent and identically distributed; (iii) to avoid energy ackets of

4 ρ = 1.2 Thus, the same results for our basic model can be shown to be valid for the dual model as well by a trivial variation in the arguments. For the rest of the work, we will focus on the first model given in (1). Energy Surlus Outage Time Fig. 1. Energy surlus, U(t), versus time, t. zero size, we assume F X () = ; and (iv) lastly, the exected energy in a acket is finitie, E[X i ] <. Assumtion 7. The inter-arrival times {A i } and the energy acket sizes {X i } are indeendent of each other. ere Assumtions 2 4 essentially simlify the hysical setu of our roblem, whereas Assumtions 5 7 are essentially technical in nature to facilitate the mathematical analysis. Note that in Assumtion 5, while not necessary, we have secifically assume that there is an arrival of an energy acket at time zero. Of all these Assumtions 2 7, Assumtion 2 regarding the infiniteness of the battery caacity is erhas be the most objectionable. owever, for the time being, we shall kee it for reliminary analysis of the system. Given these assumtions, our initial mathematical model of the energy surlus (1) simlifies to N(t) U(t) = u t + X i. (2) i= This is an instance of a random walk on real line through continuous time with downward drift. As shown in Fig. 1, the grah of U(t) versus t will look like a sawtooth wave with descending rams and with random jum discontinuities. The system exeriences an outage just before t = 12. B. Dual Model It is also ossible to have a dual model of the model given in (1). If instead of Assumtion 1, the harvested energy arrives steadily and the energy consumtion is imulsive, then we have the dual model given by N(t) U(t) = u X i h(t t i ) + i=1 t (u, t)dt. (3) Notice that in this dual model there is only a change in the signs and an interchange in the interretation of the symbols. C. Concet of Self-Sustainability Let W (t) be the battery energy at time t. We say that the system undergoes outage when W (t) =. In other words, the case W (t) = reresents the situation when the battery is emty, and as such, the consumer needs to fetch its required energy from the grid to sustain its consumtion. We define the outage robability of the system as P out = P (W (t) = ), (4) that is, robability of finding the battery emty at any time t. Now consider the first time that the battery is emty, τ = inf{t > : W (t) =, W () = u } = inf{t > : U(t), U() = u }. We will refer to τ as the first time to outage. If τ =, then the system become self-sustaining and the system will not face an eventual outage. The occurrence of eventual outage is equivalent to the event {τ < }. owever, since τ itself is a random variable, we can only describe it robabilistically. Thus, we are interested in knowing the robability φ(u,, T ) = P su t T t N(t) i= X i u, that the energy surlus U(t) will not fall below zero through time t [, T ] where T <, and N(t) φ(u,, ) = P su t X i u, t i= of avoiding an eventual outage. Note that this is equivalent to the robability φ(u,, T ) = 1 P (τ < T u, ) = 1 ψ(u,, T ), where ψ(u,, T ) is the robability of an outage occurring within a finite time T. Definition 3. We define the self-sustainability robability of an energy harvesting system as the robability that the first time to outage τ is at infinity. Likewise, the eventual outage robability is the robability that the first time to outage τ is finite. That is, φ(u,, ) = 1 P (τ < u, ) = 1 ψ(u,, ), (5) where φ(u,, ) denotes the self-sustainability robability and ψ(u,, ) denotes the eventual outage robability. Since is held constant and T =, from here on, we will dro these two arameters in the argument, and simly refer to the self-sustainability robability and eventual outage robability as a function of u as given by φ(u ) and ψ(u ).

5 5 III. ENERGY SURPLUS PROCESS AND EVALUATION OF SELF-SUSTAINABILITY PROBABILITY Let the exected inter-arrival time be E[A i ] and the arrival rate of energy ackets be defined as λ = 1 E[A i]. Also, let the average energy acket size be E[X i ] = X. Intuitively, if we want our system to be self-sustainable, then we would want the exected surlus energy to be ositive E[U(t)] >. For this to be true, it is sufficient that the consumtion rate,, be less than the energy arrival rate, λ X. That is, λ X >. We can state this condition rigorously as follows: Proosition 1. If λ X >, then lim t E[U(t)] > almost surely. Proof: The exected value of U(t) for any t is N(t) E[U(t)] = u t + E (a) i= X i = u t + E[N(t)]E[X i ] ( ) E[N(t)] = u + X t. t ere, the equality (a) is due to Wald s identity. From elementary renewal theorem, E[N(t)] t λ for large t almost surely [4, Th 5.8.4]. Thus, we have for large t, E[U(t)] = u + (λ X )t, almost surely. Given the assumtion λ X >, lim t E[U(t)] +, roving the statement. It should be noted that when the condition λ X > is satisfied, the roosition does not tell us that an outage will never occur. Rather, it tells us that there is a chance of such non-occurrence of outage. A. Random Walk Analysis In general, it is difficult to determine the outage event. We therefore have to condition on the arrival rocess. We can tell that an outage has occurred if a new arrival finds the battery emty. The mathematical trick here is to reduce the continuous time rocess into a discrete time rocess by counting over the arrivals. In the analysis of random walks and ascending ladder rocess, we will basically follow the aroach laid out by Feller in [42]. A modern treatment of the subject can be found in [41]. Proosition 2. Let the sequences {Z i ; i } and {S i ; i 1} be defined as Z i = A i X i and S n = n 1 i= Z i where S =. The surlus energy {U n }, observed immediately before the arrival of n-th energy acket, forms a discrete time random walk over real line, such that U n = u S n. (6) Proof: Let us denote t n = t n ɛ where ɛ >, as time immediately before the arrival of n-th energy acket at t n. Since t n t n as ɛ, we can decomose the arrival time t n as t n = n 1 i= A i. If we follow the value of U(t) immediately before each arrival at t n, we have N(t n ) U n = u t n + i= X i n 1 = u (A i X i ). i= Let Z i = A i X i for n =, 1, 2,.... ere, Z i is no longer a non-negative random variable; rather it can take any real value, Z i R. Since both {X i } and {A i } are indeendent and identical to each other, {Z i } are indeendent and identical to each other too. Our exression now becomes n 1 U n = u Z i, which is a discrete time random walk over real line. Defining {S n ; n 1} such that S n = n 1 i= Z i with initialization S =, gives us U n = u S n, which is our desired result. Remark: ere, the {U i } records the troughs of the sawtooth wave U(t) and can be interreted as the energy surlus that the n-th arrival finds the system in. The existence of self-sustainable system can be directly roved with the hel of the following theorem from the theory of random walk on real line: i= Theorem 1. [41, Ch 8, Th 2.4] For any random walk with F Z not degenerate at, one of the following ossibilities occur: 1) (Oscillating Case) If E[Z i ] =, then P (lim su n S n = + ) = 1, P (lim inf n S n = ) = 1; 2) (Drift to + ) If E[Z i ] >, then P (lim n S n = + ) = 1; 3) (Drift to ) If E[Z i ] <, then P (lim n S n = ) = 1. The following roositions easily follow from the above theorem. Proosition 3. The SC system is self-sustaining only if it satisfies the self-sustainability condition given by λ X >. Proof: Consider the contra-ositive of the statement: If λ X, then the SC system will exerience eventual outage. Since E[Z i ] = E[A i X i ] = E[A] X, we have that λ X is equivalent to E[Z i ]. If λ X <, then this is equivalent to the condition E[Z i ] >. ence, from Case 2 of Theorem 1, S n + almost surely. Since U n = u S n, by the definition of limit, P ( lim n S n = + ) = P ( c, n : n > n, S n > c) = P ( c, n : n > n, u U n > c) = P ( c, n : n > n, U n < u c) Taking c = u, we have P ( n : n > n, U n < ) = 1. That is, U n < almost surely. Thus, the SC system will exerience eventual outage. Similarly, if λ X =, this is equivalent to the condition E[Z i ] =. Thus, according to the Case 1 of Theorem 1

6 6 lim su n S n = + and lim inf n S n = almost surely. It suffices to consider the latter case. From the definition of limit suremum, we have P (lim su S n = + ) n = P ( c, n : n > n, su S m > c) m n = P ( c, n : n > n, su (u U m ) > c) m n = P ( c, n : n > n, u inf U m > c) m n = P ( c, n : n > n, inf U m < u c). m n Taking c = u, we have that P ( n : n > n, inf m n U m < ) = 1. ere the inf m n U m < imlies the existence of a subsequence U mk which is strictly less than zero. Thus, combining the cases for λ X < and λ X =, we conclude that the SC system will exerience eventual outage. Remark: Since λ X > is a necessary, but not sufficient, condition for self-sustainability, its satisfaction does not guarantee that outage will not occur. owever, it rovides an easily-to-check condition under which self-sustainability is ossible. We will refer to this condition λ X > (or equivalently E[Z i ] < ) as the self-sustainability condition. Condition 1 (Self-sustainability). An SC system is said to be self-sustainable when λ X >, or equivalently, E[Z i ] <. In the following sections, we will investigate the two mathematical cases that arises when this condition is or is not satisfied. Below we give an immediate corollary of the Proosition 3. Corollary 1. The SC system will exerience an eventual outage almost surely if the self-sustainability condition is not satisfied. If the self-sustainability condition is satisfied, then the robability of eventual outage will be less than unity. That is, { = 1, if λ X ψ(u ) < 1, if λ X >. B. Ascending Ladder Process Now that we have succeeded in converting a continuous time rocess into discrete time rocess, consider the random walk {S n ; n 1} as S n = Z + + Z n 1, or recursively as S n+1 = S n + Z n, with initial value S =. Thus, we have U n as U n = u S n. Consider the ascending ladder rocess defined by M n = max i n S i, which is the artial maximum of artial sums. Since M = S =, the {M i ; i 1} is a ositive non-decreasing sequence, hence the name ascending ladder rocess. We can also relate the values of M n by the recursion M n = max(m n 1, S n ). Also, let M = su i S i be the maximum value attained by S i through the entire duration of its run. We have the key observation that the eventual outage is equivalent to {τ < } {M > u }. ence, φ(u ) = F M (u ) and ψ(u ) = 1 F M (u ). (7) Given the sequence {S i ; i 1}, the first strict ascending ladder oint (σ 1, 1 ) is the first term in this sequence for which S i >. That is, σ 1 = inf{n 1 : S n > } and 1 = S σ1. In other words, the eoch of the first entry into the strictly ositive half-axis is defined by {σ 1 = n} = {S 1,..., S n 1, S n > }. (8) The σ 1 is called the first ladder eoch while 1 is called the first ladder height. Let the joint distribution of (σ 1, 1 ) be denoted by P (σ 1 = n, 1 x) = F,n (x). (9) The marginal distributions are given by P (σ 1 = n) = F,n ( ) (1) P ( 1 x) = F,n (x) = F (x). (11) n=1 The two variables have the same defect 1 F ( ). We can iteratively define the ladder eochs {σ n } and ladder heights { n } as σ n+1 = inf{k 1 : S k+σn > S σn }, (12) n+1 = S σn+1 S σn. (13) The airs (σ i, i ) are mutually indeendent and have the same common distribution given in (9). These ladder heights are related to record maximum at time n by M n = π n i=1 i, (14) where π n is the number of ladder oints u until time n, i.e. π n = min{k : σ σ k n}. An imortant observation related to the ladder oints is that the sums of {σ i } and { i }, σ σ n and n, form (ossibly terminating) renewal rocesses with interrenewal interval σ i and i. Clearly, we can have M < if and only if the ascending ladder rocess is terminating. As er Definition 2, this renewal rocess is terminating if the underlying distribution F is defective. Accordingly, let be a defective random variable with F () = and F ( ) = θ < 1, then the amount of defect given by 1 θ reresents the robability of termination. In other words, θ reresent the robability of another renewal, while 1 θ reresents the robability that the inter-renewal interval is infinite. Thus the termination eoch is a Bernoulli random variable with failure being interreted as termination with robability 1 θ. The sum n has a defective distribution given by the n-fold convolution F (n), whose total mass equals F (n) ( ) = F n ( ) = θ n. (15) This is easily seen by re-interreting the n ladder eochs as n successes of a Bernoulli random variable. The defect 1 θ n is thus the robability of termination before the n-th ladder eoch.

7 7 Let us define the renewal function for the ladder heights by the sum on n-fold convolutions of F as ζ(x) = F (n) (x), (16) n= where F (n) is the n-fold convolution of F defined recursively as F (i) (x) = F (i 1) (x t)df (t) where i = 1, 2,..., n; and F () (x) is a unit ste function at the origin. Remark: ere the Lalace transform of F (n) can be () obtained recursively as: F (r) = 1/r, being a unit ste function. Similarly, F (1) () (r) = L[F f () (x)] = F (r) f (r) = (2) (1) f (r)/r. And, F (r) = L[F f (1) (x)] = F (r) f (r) = [ f (r)] 2 (n) /r. And so on, until F (r) = [ f (r)] n /r. Thus, the Lalace transform of the renewal function is ζ(r) = 1 r n= [ f (r)] n 1 = r(1 f, since for the geometric sum (r)) f (r) < f () = 1. The renewal function ζ(x) is equivalent to the exected number of ladder oints in the stri [, x], where the origin counts as a renewal eoch. Thus, ζ() = 1. For terminating renewal rocess, the exected number of eochs ever occurring is finite, as from (15), we have ζ( ) = n= F (n) ( ) = n= θ n = 1 1 θ. If the ascending ladder rocess terminates after n-th eoch, then n = M, the all time maximum attained by the random walk S n. The robability that the n-th ladder eoch is the last and that {M x} is given by P (M x, terminate after n) = (1 θ)f (n) (x). (17) Using (15), the marginalization of (17) over M shows us that the robability of the ladder rocess terminating after the n-th eoch follows a geometric distribution, P (terminate after n) = (1 θ)f (n) ( ) = (1 θ)θn. Similarly, marginalizing (17) over n, we have P (M x) = (1 θ) F (n) (x) = (1 θ)ζ(x). (18) n= We now need a criteria to determine whether the ascending ladder rocess terminates or not, as well as a method to find the value of θ. The following roosition also immediately follows from Theorem 1 and our discussion about ascending ladder rocess: Proosition 4. Given the self-sustainability condition, λ X >, the ascending ladder height rocess { i } of an SC system is terminating almost surely. The robability of selfsustainability given in terms of the renewal function ζ is φ(u ) = (1 θ)ζ(u ). (19) Proof: From Case 3 of Theorem 1, since the random walk S n drifts to when E[Z i ] <, the maximum M < almost surely, and thus { i } terminates. Equation (19) is obtained from (7) and (18). We now relate the self-sustainability robability with two convolution formulas. Proosition 5. Given the self-sustainability condition, the self-sustainability robability satisfies the following equivalent integral equations: φ(u ) = (1 θ) + φ(u ) = u φ(u x)f (x)dx, (2) φ(x)f Z (u x)dx. (21) Proof: We begin with the fact that φ(u ) = P (M u ). The roofs of the two equations follow from the standard renewal tye argument: (1) The event {M u } occurs if the ascending ladder rocess terminates with M or else if 1 assumes some ositive value x u and the residual rocess attains the age u x. So, φ(u ) = P (M = ) + = (1 θ) + u u P ( 1 u 1 = x)f (x)dx P (age u x)f (x)dx. Since the ascending ladder rocess renews at 1, the robability of the age of the residual rocess is P (age u x) = φ(u x). Thus φ(u ) = P (M u ) satisfies the renewal equation φ(u ) = (1 θ) + u φ(u x)f (x)dx. (2) The event {M u } occurs if and only if max(z, Z + Z 1, Z + Z 1 + Z 2,...) u. Conditioning on Z = y, this is equivalent to Z = y u and max(, Z 1, Z 1 + Z 2,...) u y. ere, P (max(, Z 1, Z 1 +Z 2,...) u y) = φ(u y). Deconditioning over all ossible y, we get φ(u ) = u φ(u y)f Z (y)dy, which by change of variable x = u y becomes φ(u ) = φ(x)f Z (u x)dx. Equations (2) and (21) relates φ with (a defective random variable) and Z (a roer random variable) resectively. Equation (2) can be recognized as a renewal equation while (21) can be recognized as a Wiener-of integral. Corollary 2. Given the self-sustainability condition, φ(u ) is a roer distribution with (i) φ(u ) = for u <, (ii) φ() = 1 θ, and (iii) φ( ) = 1. Proof: (i) Follows from the fact that u only takes non-negative values. (ii) By utting u = in (2). (iii) Putting u = in (2), we have φ( ) = (1 θ) + φ( ) f (x)dx. Recalling that is defective, with F ( ) = θ, we have φ( ) = 1. Remark: Corollary 2 can also be roved as a consequence of Proosition 4, since ζ() = 1 and ζ( ) = 1/(1 θ). In the above corollary, it is remarkable that the system can be self-sustaining even when there is no initial battery energy. In other words, this is robability that the random walk U n starting from the origin will always be ositive. This also

8 8 allows us to interret θ as the eventual outage robability when there is no initial battery energy, i.e. ψ() = θ. The corollary also guarantees that as u becomes large, the eventual outage becomes zero. Thus, a ossible strategy in reducing the eventual outage is to simly increase the initial battery energy. We will later show that the rate at which the eventual outage decreases with u is exonential. Remark: Equations (19) and (2) can also be exressed in terms of Lalace transform. Taking the Lalace transform of (2), we have φ(r) = 1 θ + r φ(r) f (r), φ(r) 1 θ = r(1 f (r)). (22) The exact relationshi between and Z is given by the well known Weiner-of factorization identity [42, Ch XII.3] [41, Ch VIII.3] in terms of their moment generating functions (MGFs) as: 1 M Z = (1 M )(1 M ), (23) where is the descending ladder height, defined in a manner similar to the ascending ladder height rocess. The is defined over (, ]. The above identity is also written in terms of convolution as F Z = F + F F F. (24) Likewise, the distributions of Z,, and are related to the renewal function ζ by F (x) = ζ F Z (x), x >, (25a) F (x) = ζ F Z (x), x, (25b) where ζ is the renewal function defined by F in a manner similar to ζ given in (16). When E[Z] <, while F is defective, F is roer. Since is defined over (, ], this gives us the condition that F () = 1. Lastly, since is a roer distribution, the descending ladder rocess is a roer renewal rocess. The factorization (23) is in itself difficult to erform exlicitly. As such, we will focus on obtaining bounds and asymtotic aroximations of φ (or equivalently, ψ). IV. BOUND ON EVENTUAL OUTAGE PROBABILITY Thus far we have described the energy surlus rocess and related the various associated concets to the eventual outage / self-sustainability robability. These robabilities can be evaluated by solving the formulas given in Proositions 4 and 5. owever, doing so is not trivial. As such, we wish for a simle bound to estimate the eventual outage robability. ere we will establish a tight exonential bound for the eventual outage robability using the concet of martingales. Definition 4 (Adjustment Coefficient). The value r is said to the adjustment coefficient of X if E[ex(r X)] = e r x df X = 1 Definition 5 (Martingale). A rocess {X i } is said to be a martingale if E[X n+1 X n,..., X ] = X n. Proosition 6. Let {Z i }, {S i } be as before. Suose there exists an adjustment coefficient r > such that E[ex(r Z i )] = 1, then ex(r S n ) for n =, 1, 2,... is a martingale. Proof: We have E[ex(rS n+1 ) S n,..., S 1 ] = E[ex(r(S n + Z n+1 )) S n,..., S 1 ] = E[ex(rZ n+1 )] E[ex(rS n ) S n,..., S 1 ] = E[ex(rZ n+1 )] ex(rs n ). Since there exists a constant r > such that E[ex(r Z n+1 )] = 1, then E[ex(r S n+1 ) S n,..., S 1 ] = ex(r S n ), satisfying the definition of a martingale. Remark: Note that since Z is a roer random variable, it is trivially true that E[ex(r Z i )] = 1 if r =. In the following roosition, we show that there exists a non-trivial value of r > for which this roerty holds true. We will now give the conditions under which the adjustment coefficient will exist. Proosition 7. Suose that E[Z i ] <, which is the selfsustainability condition. Also, assume that there is r 1 > such that the moment generating function (MGF) M Z (r) = E[e rzi ] < for all r 1 < r < r 1, and that lim r r1 M Z (r) =. Then, there is a unique adjustment coefficient r >. Proof: Since the MGF exists in a neighborhood of zero, derivative of every order exists in ( r 1, r 1 ). By definition, M Z () = 1. Since E[Z i ] < by assumtion, we have M Z () = E[Z i] <, which means that M Z (r) is decreasing in the neighborhood of. Also, since M Z (h) = E[Z2 i erzi ] > (due to the fact that the exectation of ositive random variable is ositive), it follows that M Z is convex on ( r 1, r 1 ). Again, by assumtion, we have lim r r1 M Z (r) =. It now follows that there exists a unique s (, r 1 ) such that M Z (s) < 1 and M Z (s) = ; and that on the interval (s, r 1) the function M Z ( ) is strictly increasing to +. As such, there exists a unique r (, r 1 ) such that M Z (r ) = 1. Since the MGF does not exist on [r 1, ), it follows that r is unique on (, ). Proosition 8. Assume that the self-sustainability condition E[Z i ] < holds, and that the adjustment coefficient r > exists. Then the eventual outage robability is bounded by ψ(u ) = P (τ(u ) < ) ex( r u ), u >. (26) Proof: Put τ(u ) = inf(n : S n > u ). The τ(u ) is the the first assage time that the random walk {S n } exceeds u > in the ositive direction. The event that {τ(u ) k} is equivalent to the union of events k i= {S i > u }. Similarly, for a >, let σ(a) = inf{n : S n < a}. ere σ(a) denotes the first assage time that the random walk {S n } exceeds a < in the negative direction. Since E[Z i ] <, by Case 3 of Theorem 1, P (σ(a) < ) = 1. ence, (τ(u ) σ(a)) min(τ(u ), σ(a)) is a stoing time with P ((τ(u ) σ(a)) < ) = 1. Since ex(r S n )

9 9 for n =, 1, 2,... is a martingale, using otional samling theorem, we now get 1 = E[e r S ] = E[ex(r S (τ(u) σ(a)))] = E[ex(r S τ(u)) τ(u ) < σ(a)] + E[ex(r S σ(a) ) σ(a) < τ(u )] E[ex(r S τ(u)) τ(u ) < σ(a)] e r u P (τ(u ) < σ(a)) as S τ(u) > u. Since P (lim a σ(a) = ) = 1, thus letting a we get P (τ(u ) < ) = lim a P (τ(u ) < σ(a)) e r u, as required. Remark: This remarkable roosition guarantees that the eventual outage robability can be made arbitrarily small simly by increasing the initial battery energy. Thus, during the design of a system, where we are willing to tolerate an arbitrarily small eventual outage robability, our task is to determine the initial battery energy. We can use the above bound to roughly calculate the required initial battery energy. The following corollaries immediately follows. Corollary 3. Assuming that the self-sustainability condition holds and r > exists, then lim u φ(u ) =. Corollary 4. Assuming that the self-sustainability condition holds and the adjustment coefficient r > exists, for a given tolerance ɛ >, if the eventual outage robability is constrained at ψ(u ) = ɛ, then the maximum initial battery energy required is u = 1 r log( 1 ɛ ). V. ASYMPTOTIC APPROXIMATION OF EVENTUAL OUTAGE PROBABILITY While the exonential bound given in the revious section is simle to use, we can sharen our estimates using the key renewal theorem for defective distribution. The basic idea behind this aroach is to transform a defective distribution into a roer distribution using the adjustment coefficient, and then aly the key renewal theorem for roer distribution. A. Asymtotic Aroximation First we define the renewal equation and then give the related theorem. Definition 6. The renewal equation is the convolution equation for the form Z = z + F Z, where Z is an unknown function on [, ), z is a known funciton on [, ) and F is a known non-negative measure on [, ). Often F is assumed to be a robability distribution. If F ( ) = 1, then the renewal equation is roer. If F ( ) < 1, then the renewal equation is defective. Theorem 2. [41, Ch. V, Pro 7.6,. 164] [42, Ch. IX.6, Theo. 2,. 376] Suose that for defective distribution F, there exists an adjustment coefficient r > such that µ = te r t df < exists. If in the defective renewal equation Z = z + F Z, z( ) = lim t z(t) exists and e r t (z(t) z( )) is directly Riemann integrable, then the solution of the renewal equation satisfies µe r t [Z( ) Z(t)] z( ) r + e r s [z( ) z(s)]ds. (27) Proosition 9. Given the self-sustainability condition, the eventual outage robability is asymtotically given by ψ(u ) 1 θ r e r u, (28) µ where r is the adjustment coefficient of and µ = xe r x f (x)dx <. Proof: Recall that (2) is in the form of a renewal equation with the defective distribution of. As such, in Theorem 2, we have z(t) 1 θ and Z φ(u ). Thus, the integral in (27) vanishes and µ e r u [φ( ) φ(u )] 1 θ r. From Corollary 2, we know that φ is roer. Thus φ( ) φ(u ) = 1 φ(u ) = ψ(u ). ence, we have the desired result. Note that the adjustment coefficient r for is the same as that for Z. This is easily demonstrated, since M Z (r ) = E[e r z ] = 1 where r >, we have from the Weiner-of factorization (23), (1 M (r ))(1 M (r )) = 1 M Z (r ) =. ere since is defined over (, ], we have M (r ) = (a) < (b) = 1, e r x f ( x)dx, f ( x)dx, since in (a) r > and f is ositive, while in (b) is a roer distribution. Thus, 1 M (r ) >, and this imlies that M (r ) = 1, making r the adjustment factor of as well. The above result is asymtotic in the sense that higher values of u gives more accurate results. owever, without further modeling assumtions, this is as far as we can roceed, since exlicit evaluation of F is difficult. In Section VI, we will exlore the case when the energy acket arrival is modeled as Poisson rocess. B. Comuting the Adjustment Coefficient The comutation of the adjustment coefficient r is not trivial. As given by Definition 4, the adjustment coefficient must satisfy the condition E[e r Z ] = M Z (r ) = 1. For the comutational urose, it is more convenient to use the cumulant generating function (CGF) of Z rather than its MGF. The CGF of Z is defined as K Z (r) = log M Z (r). In terms of CGF, the adjustment coefficient satisfies the condition K Z (r ) = log M Z (r ) = log 1 =. ence, we see that the

10 1 adjustment coefficient r is a real ositive root of CGF K Z (r), and can be found by solving the equation K Z (r ) =. (29) Since Z i = A i X i, due to the linearity of CGFs, K Z (r) = K A (r) + K X ( r). For the imortant case when the energy acket arrival is a Poisson rocess, the inter-arrival time A i Ex(λ). As such, M A (r) = λ/(λ r), and K A (r) = log(1 r λ ). Thus for) the Poisson arrivals, the equation (29) becomes log (1 r λ + K X ( r ) =, which after exonentiation can be exressed in fixed oint form as r = λ (1 M X( r )). (3) Given the Poisson arrival, some ossible distributions for the energy ackets sizes and their corresonding solutions are as follows: 1) If we further assume that size of the energy ackets is exonentially distributed, X i Ex(1/ X), then we have M X (r) = 1/(1 r X). So, the solution to (3) is r = 1 X ( ) λ X 1. (31) 2) If instead we assume that the size of the energy ackets is deterministic, X i = c, then we have M X (r) = e cr. So the solution to (3) is obtained by solving the equation e cr + r λ 1 =. (32) 3) If instead we assume that the size of the energy ackets is uniformly distributed, X i U(, 2 X), then we have M X (r) = e2 Xr 1 2 Xr. So the solution to (3) is obtained by solving the equation e 2 Xr 2 X λ r2 + 2 Xr + 1 =. (33) A simle aroximation of r can be obtained by making a formal ower exansion of M X ( r) in terms of the moments X2 of X i u to second order term as M X ( r) 1 Xr+ 2 r2. Using this exression in (3) and solving for r >, we obtain r 2 ( ) λ X λx 2 1. (34) More generally, since K Z (r) can be exanded in terms of the mean and variance of Z as K Z (r) µ Z r+ σ2 Z 2 r 2, we have the aroximate solution r for (29) as r 2µ Z σz 2. (35) Since µ Z = E[Z] <, the above aroximation will correctly give r >. This value can be used as an initial oint for a root finding algorithm. Better aroximations can be found by including higher order terms in the exansion and reverting the series using Lagrange inversion. VI. SPECIAL CASE: EVALUATION OF EVENTUAL OUTAGE PROBABILITY FOR POISSON ARRIVALS In general, the density of Z i is given in terms of the densities of A i and X i as f Z (z) = 1 ( ) z + x f A f X (x)dx. (36) max(, z) When the energy acket arrival is assumed to be a Poisson rocess, the inter-arrival time A i is exonentially distributed, A i Ex(λ). As such, the density of Z i = A i X i is f Z (z) = λ e λz max(, z) When z, the density of Z has the form f Z (z) = λ e λz e λx fx (x)dx. (37) e λx fx (x)dx, z. Since the above integral is indeendent of z, the right tail of the density is exonential. That is, f Z (z) = Ce λz for z, where C is some constant given by C = λ e λx fx (x)dx. Now, from (25a), we have f (x) = ζ f Z (x) = C e λ(x s) ζ (s)ds. Again we see that regardless of the exression for ζ, f takes an exonential form given by f (x) = Ce λx e s ζ (s)ds, where the integral on the right hand is some constant. Since we know that is defective, we can re-write f in a manner similar to roer exonential distribution as f (x) = ( λ r ) e λx, such that θ = F ( ) = 1 r λ. ence, the amount of defect is 1 θ = r λ. When f (x) is multilied by e r x we have the roer distribution e r x f (x) = ( ) λ r e ( λ r )x, and the mean of this roer exonential distribution is µ = ( λ ) 1. r Thus, we have from (28), ( ψ(u ) 1 r λ ) e r u. (38) When X is also exonentially distributed, we have the exact value of r from (31). ence, we have ψ(u ) u e r. (39) λ X In fact, the (38) and (39) are not just asymtotic aroximations, but also exact formulas (see [42, Ch XII.5, Ex 5(b)]). From these arguments, we have the following roosition: Proosition 1. Assume that the self-sustainability condition holds and the adjustment coefficient r > exists. If the energy

11 11 ackets arrive into an SC system as a Poisson rocess, then the eventual outage robability is given by ( ) ψ(u ) = 1 r e r u. (4) λ Furthermore, if the energy acket size is also exonentially distributed, then ψ(u ) = { ex 1 X ( ) } λ X λ X 1 u. (41) VII. BATTERY ENERGY PROCESS So far we have directed our attention to the energy surlus U(t) and the case when the self-sustainability condition is satisfied. For the sake of comleteness, let us now consider the battery energy W (t) at time t and the case when the selfsustainability condition is not satisfied. A. Equivalence with Queueing Systems We will first rove that the battery energy rocess is a Lindley rocess. Making this identification will then allow us to comare the SC system to a GI/G/1 queue, which in turn will allow us to exloit the results from queueing theory, for which the Lindley rocess was first studied. When the self-sustainability condition is not satisfied, the battery energy rocess W (t) is stationary and ergodic. Thus, it makes better sense to talk about the outage robability P (W (t) = ) of the system rather than the eventual outage robability which is always unity, i.e. ψ(u ) = 1. Definition 7. [41, Ch 3.6] A discrete-time stochastic rocess {Y i } is a Lindley rocess if and only if {Y i } satisfies the recurrence relation Y n+1 = max(, Y n + X n ), n =, 1,... (42) where Y = y and {X i } are indeendent and identically distributed. This recursive equation is called Lindley recursion. 1 Proosition 11. The battery energy rocess W (t) observed just before the energy acket arrivals is a Lindley rocess and satisfies the recursion W n+1 = max(, W n + Z n), for n =, 1,..., where W = u and Z n = Z n. Proof: As with the energy surlus in the revious section, let W n be the amount of battery energy immediately before the arrival of n-th energy acket. The initial battery energy immediately before the arrival of the first energy acket E is W = u. The amount of battery energy just before the arrival of next (n+1)-th energy acket, W n+1, is then the sum of W n and X n, the amount of energy contributed by the n-th acket into the battery, minus the amount consumed during the interarrival eriod of the (n + 1)-th acket, A n. Thus we have W n+1 = W n + X n A n if W n + X n A n. Likewise, the battery will be emty, W n+1 =, if W n +X n A n. Putting both of them together, we have { Wn + Z W n+1 = n, if W n + Z n, if W n + Z n (43). 1 The Lindley rocess is referred to as queueing rocess in [42] TABLE I QUEUEING ANALOGUE Parameter Energy arvesting Queueing - Consumer Server - Battery Buffer E n n-th energy acket n-th customer λ Packet arrival rate Customer arrival rate W n Battery energy Virtual waiting time X n Packet size Service time A n Energy consumed Inter-arrival time (scaled) P (W (t) = ) System outage System idle - Self-sustainability System always busy ρ = λ X Utilization factor Utilization factor (traffic intensity) where Z n = X n A n. Since {X n } and {A n } are IID, the {Z n} is clearly IID. We can write (43) in comact form as W n+1 = max(, W n + Z n), which is a Lindley recursion as given in (42). Since {W i } satisfies the Lindley recursion, {W i } is a Lindley rocess. The fact that {W i } is a Lindley rocess gives rise to a number of imortant consequences, as stated in the following roositions. Proosition 12. The SC system is equivalent to a GI/G/1 queueing system. Proof: The roosition follows from the fact that the virtual waiting time (i.e. the amount of time the server will have to work until the system is emty, rovided that no new customers arrive, or equivalently the waiting time of a customer in a first-in-first-out queueing disciline) of an n-th customer arriving into a GI/G/1 queueing system is a Lindley rocess [41, Ch 3.6, Ex 6.1]. Since SC Consume system is also a Lindley rocess by Proosition 11, the equivalence is established. Proosition 13. Suose < E[ Z i ] <. The SC system is ergodic and stationary if and only if E[Z i ] >, i.e. λ X <, when the self-sustainability condition is not satisfied. Under this condition, there will exist a unique stationary distribution for W n, indeendent of the initial condition W, which is given by the Lindley s integral equation: F W (w) = F Z (w x)df W (x), w (44) Proof: This is a standard result from queueing theory. See [41, Coro 6.6]. Remark: We see that when the self-sustainability condition is not satisfied, the battery energy rocess of the system is ergodic and stationary. This allows us to easily comute the outage robability, P (W (t) = ), for such case by invoking the ergodicity and stationarity of W (t). The equation (44) is again a Weiner-of integral; and excet for some secial cases, its general solution is difficult to obtain. B. Discussion 1) By comaring the Lindley recursion for the energy harvesting system with that of a queueing system, we can translate the terms and concets of one system into those

Self-Sustainability of Energy Harvesting Systems: Concept, Analysis, and Design

Self-Sustainability of Energy Harvesting Systems: Concept, Analysis, and Design 1 Self-Sustainability of Energy arvesting Systems: Concept, Analysis, and Design Sudarshan Guruacharya and Ekram ossain Department of Electrical and Computer Engineering, University of Manitoba, Canada

More information

1 Gambler s Ruin Problem

1 Gambler s Ruin Problem Coyright c 2017 by Karl Sigman 1 Gambler s Ruin Problem Let N 2 be an integer and let 1 i N 1. Consider a gambler who starts with an initial fortune of $i and then on each successive gamble either wins

More information

Analysis of some entrance probabilities for killed birth-death processes

Analysis of some entrance probabilities for killed birth-death processes Analysis of some entrance robabilities for killed birth-death rocesses Master s Thesis O.J.G. van der Velde Suervisor: Dr. F.M. Sieksma July 5, 207 Mathematical Institute, Leiden University Contents Introduction

More information

Lecture 6. 2 Recurrence/transience, harmonic functions and martingales

Lecture 6. 2 Recurrence/transience, harmonic functions and martingales Lecture 6 Classification of states We have shown that all states of an irreducible countable state Markov chain must of the same tye. This gives rise to the following classification. Definition. [Classification

More information

Elementary Analysis in Q p

Elementary Analysis in Q p Elementary Analysis in Q Hannah Hutter, May Szedlák, Phili Wirth November 17, 2011 This reort follows very closely the book of Svetlana Katok 1. 1 Sequences and Series In this section we will see some

More information

Improved Capacity Bounds for the Binary Energy Harvesting Channel

Improved Capacity Bounds for the Binary Energy Harvesting Channel Imroved Caacity Bounds for the Binary Energy Harvesting Channel Kaya Tutuncuoglu 1, Omur Ozel 2, Aylin Yener 1, and Sennur Ulukus 2 1 Deartment of Electrical Engineering, The Pennsylvania State University,

More information

MATH 2710: NOTES FOR ANALYSIS

MATH 2710: NOTES FOR ANALYSIS MATH 270: NOTES FOR ANALYSIS The main ideas we will learn from analysis center around the idea of a limit. Limits occurs in several settings. We will start with finite limits of sequences, then cover infinite

More information

MATHEMATICAL MODELLING OF THE WIRELESS COMMUNICATION NETWORK

MATHEMATICAL MODELLING OF THE WIRELESS COMMUNICATION NETWORK Comuter Modelling and ew Technologies, 5, Vol.9, o., 3-39 Transort and Telecommunication Institute, Lomonosov, LV-9, Riga, Latvia MATHEMATICAL MODELLIG OF THE WIRELESS COMMUICATIO ETWORK M. KOPEETSK Deartment

More information

4. Score normalization technical details We now discuss the technical details of the score normalization method.

4. Score normalization technical details We now discuss the technical details of the score normalization method. SMT SCORING SYSTEM This document describes the scoring system for the Stanford Math Tournament We begin by giving an overview of the changes to scoring and a non-technical descrition of the scoring rules

More information

Statics and dynamics: some elementary concepts

Statics and dynamics: some elementary concepts 1 Statics and dynamics: some elementary concets Dynamics is the study of the movement through time of variables such as heartbeat, temerature, secies oulation, voltage, roduction, emloyment, rices and

More information

On Wald-Type Optimal Stopping for Brownian Motion

On Wald-Type Optimal Stopping for Brownian Motion J Al Probab Vol 34, No 1, 1997, (66-73) Prerint Ser No 1, 1994, Math Inst Aarhus On Wald-Tye Otimal Stoing for Brownian Motion S RAVRSN and PSKIR The solution is resented to all otimal stoing roblems of

More information

Estimation of the large covariance matrix with two-step monotone missing data

Estimation of the large covariance matrix with two-step monotone missing data Estimation of the large covariance matrix with two-ste monotone missing data Masashi Hyodo, Nobumichi Shutoh 2, Takashi Seo, and Tatjana Pavlenko 3 Deartment of Mathematical Information Science, Tokyo

More information

ANALYTIC NUMBER THEORY AND DIRICHLET S THEOREM

ANALYTIC NUMBER THEORY AND DIRICHLET S THEOREM ANALYTIC NUMBER THEORY AND DIRICHLET S THEOREM JOHN BINDER Abstract. In this aer, we rove Dirichlet s theorem that, given any air h, k with h, k) =, there are infinitely many rime numbers congruent to

More information

LECTURE 7 NOTES. x n. d x if. E [g(x n )] E [g(x)]

LECTURE 7 NOTES. x n. d x if. E [g(x n )] E [g(x)] LECTURE 7 NOTES 1. Convergence of random variables. Before delving into the large samle roerties of the MLE, we review some concets from large samle theory. 1. Convergence in robability: x n x if, for

More information

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK Towards understanding the Lorenz curve using the Uniform distribution Chris J. Stehens Newcastle City Council, Newcastle uon Tyne, UK (For the Gini-Lorenz Conference, University of Siena, Italy, May 2005)

More information

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Technical Sciences and Alied Mathematics MODELING THE RELIABILITY OF CISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Cezar VASILESCU Regional Deartment of Defense Resources Management

More information

Elements of Asymptotic Theory. James L. Powell Department of Economics University of California, Berkeley

Elements of Asymptotic Theory. James L. Powell Department of Economics University of California, Berkeley Elements of Asymtotic Theory James L. Powell Deartment of Economics University of California, Berkeley Objectives of Asymtotic Theory While exact results are available for, say, the distribution of the

More information

On Code Design for Simultaneous Energy and Information Transfer

On Code Design for Simultaneous Energy and Information Transfer On Code Design for Simultaneous Energy and Information Transfer Anshoo Tandon Electrical and Comuter Engineering National University of Singaore Email: anshoo@nus.edu.sg Mehul Motani Electrical and Comuter

More information

p-adic Measures and Bernoulli Numbers

p-adic Measures and Bernoulli Numbers -Adic Measures and Bernoulli Numbers Adam Bowers Introduction The constants B k in the Taylor series exansion t e t = t k B k k! k=0 are known as the Bernoulli numbers. The first few are,, 6, 0, 30, 0,

More information

8 STOCHASTIC PROCESSES

8 STOCHASTIC PROCESSES 8 STOCHASTIC PROCESSES The word stochastic is derived from the Greek στoχαστικoς, meaning to aim at a target. Stochastic rocesses involve state which changes in a random way. A Markov rocess is a articular

More information

Improved Bounds on Bell Numbers and on Moments of Sums of Random Variables

Improved Bounds on Bell Numbers and on Moments of Sums of Random Variables Imroved Bounds on Bell Numbers and on Moments of Sums of Random Variables Daniel Berend Tamir Tassa Abstract We rovide bounds for moments of sums of sequences of indeendent random variables. Concentrating

More information

On Doob s Maximal Inequality for Brownian Motion

On Doob s Maximal Inequality for Brownian Motion Stochastic Process. Al. Vol. 69, No., 997, (-5) Research Reort No. 337, 995, Det. Theoret. Statist. Aarhus On Doob s Maximal Inequality for Brownian Motion S. E. GRAVERSEN and G. PESKIR If B = (B t ) t

More information

Online Appendix to Accompany AComparisonof Traditional and Open-Access Appointment Scheduling Policies

Online Appendix to Accompany AComparisonof Traditional and Open-Access Appointment Scheduling Policies Online Aendix to Accomany AComarisonof Traditional and Oen-Access Aointment Scheduling Policies Lawrence W. Robinson Johnson Graduate School of Management Cornell University Ithaca, NY 14853-6201 lwr2@cornell.edu

More information

On a Markov Game with Incomplete Information

On a Markov Game with Incomplete Information On a Markov Game with Incomlete Information Johannes Hörner, Dinah Rosenberg y, Eilon Solan z and Nicolas Vieille x{ January 24, 26 Abstract We consider an examle of a Markov game with lack of information

More information

Convex Optimization methods for Computing Channel Capacity

Convex Optimization methods for Computing Channel Capacity Convex Otimization methods for Comuting Channel Caacity Abhishek Sinha Laboratory for Information and Decision Systems (LIDS), MIT sinhaa@mit.edu May 15, 2014 We consider a classical comutational roblem

More information

The non-stochastic multi-armed bandit problem

The non-stochastic multi-armed bandit problem Submitted for journal ublication. The non-stochastic multi-armed bandit roblem Peter Auer Institute for Theoretical Comuter Science Graz University of Technology A-8010 Graz (Austria) auer@igi.tu-graz.ac.at

More information

Optimal Random Access and Random Spectrum Sensing for an Energy Harvesting Cognitive Radio with and without Primary Feedback Leveraging

Optimal Random Access and Random Spectrum Sensing for an Energy Harvesting Cognitive Radio with and without Primary Feedback Leveraging 1 Otimal Random Access and Random Sectrum Sensing for an Energy Harvesting Cognitive Radio with and without Primary Feedback Leveraging Ahmed El Shafie, Member, IEEE, arxiv:1401.0340v3 [cs.it] 27 Ar 2014

More information

General Linear Model Introduction, Classes of Linear models and Estimation

General Linear Model Introduction, Classes of Linear models and Estimation Stat 740 General Linear Model Introduction, Classes of Linear models and Estimation An aim of scientific enquiry: To describe or to discover relationshis among events (variables) in the controlled (laboratory)

More information

Optimism, Delay and (In)Efficiency in a Stochastic Model of Bargaining

Optimism, Delay and (In)Efficiency in a Stochastic Model of Bargaining Otimism, Delay and In)Efficiency in a Stochastic Model of Bargaining Juan Ortner Boston University Setember 10, 2012 Abstract I study a bilateral bargaining game in which the size of the surlus follows

More information

1 Probability Spaces and Random Variables

1 Probability Spaces and Random Variables 1 Probability Saces and Random Variables 1.1 Probability saces Ω: samle sace consisting of elementary events (or samle oints). F : the set of events P: robability 1.2 Kolmogorov s axioms Definition 1.2.1

More information

John Weatherwax. Analysis of Parallel Depth First Search Algorithms

John Weatherwax. Analysis of Parallel Depth First Search Algorithms Sulementary Discussions and Solutions to Selected Problems in: Introduction to Parallel Comuting by Viin Kumar, Ananth Grama, Anshul Guta, & George Karyis John Weatherwax Chater 8 Analysis of Parallel

More information

Sharp gradient estimate and spectral rigidity for p-laplacian

Sharp gradient estimate and spectral rigidity for p-laplacian Shar gradient estimate and sectral rigidity for -Lalacian Chiung-Jue Anna Sung and Jiaing Wang To aear in ath. Research Letters. Abstract We derive a shar gradient estimate for ositive eigenfunctions of

More information

Robustness of classifiers to uniform l p and Gaussian noise Supplementary material

Robustness of classifiers to uniform l p and Gaussian noise Supplementary material Robustness of classifiers to uniform l and Gaussian noise Sulementary material Jean-Yves Franceschi Ecole Normale Suérieure de Lyon LIP UMR 5668 Omar Fawzi Ecole Normale Suérieure de Lyon LIP UMR 5668

More information

Model checking, verification of CTL. One must verify or expel... doubts, and convert them into the certainty of YES [Thomas Carlyle]

Model checking, verification of CTL. One must verify or expel... doubts, and convert them into the certainty of YES [Thomas Carlyle] Chater 5 Model checking, verification of CTL One must verify or exel... doubts, and convert them into the certainty of YES or NO. [Thomas Carlyle] 5. The verification setting Page 66 We introduce linear

More information

arxiv:cond-mat/ v2 25 Sep 2002

arxiv:cond-mat/ v2 25 Sep 2002 Energy fluctuations at the multicritical oint in two-dimensional sin glasses arxiv:cond-mat/0207694 v2 25 Se 2002 1. Introduction Hidetoshi Nishimori, Cyril Falvo and Yukiyasu Ozeki Deartment of Physics,

More information

An Analysis of TCP over Random Access Satellite Links

An Analysis of TCP over Random Access Satellite Links An Analysis of over Random Access Satellite Links Chunmei Liu and Eytan Modiano Massachusetts Institute of Technology Cambridge, MA 0239 Email: mayliu, modiano@mit.edu Abstract This aer analyzes the erformance

More information

On the Role of Finite Queues in Cooperative Cognitive Radio Networks with Energy Harvesting

On the Role of Finite Queues in Cooperative Cognitive Radio Networks with Energy Harvesting On the Role of Finite Queues in Cooerative Cognitive Radio Networks with Energy Harvesting Mohamed A. Abd-Elmagid, Tamer Elatt, and Karim G. Seddik Wireless Intelligent Networks Center (WINC), Nile University,

More information

PHYS 301 HOMEWORK #9-- SOLUTIONS

PHYS 301 HOMEWORK #9-- SOLUTIONS PHYS 0 HOMEWORK #9-- SOLUTIONS. We are asked to use Dirichlet' s theorem to determine the value of f (x) as defined below at x = 0, ± /, ± f(x) = 0, - < x

More information

Asymptotically Optimal Simulation Allocation under Dependent Sampling

Asymptotically Optimal Simulation Allocation under Dependent Sampling Asymtotically Otimal Simulation Allocation under Deendent Samling Xiaoing Xiong The Robert H. Smith School of Business, University of Maryland, College Park, MD 20742-1815, USA, xiaoingx@yahoo.com Sandee

More information

Department of Mathematics

Department of Mathematics Deartment of Mathematics Ma 3/03 KC Border Introduction to Probability and Statistics Winter 209 Sulement : Series fun, or some sums Comuting the mean and variance of discrete distributions often involves

More information

Combining Logistic Regression with Kriging for Mapping the Risk of Occurrence of Unexploded Ordnance (UXO)

Combining Logistic Regression with Kriging for Mapping the Risk of Occurrence of Unexploded Ordnance (UXO) Combining Logistic Regression with Kriging for Maing the Risk of Occurrence of Unexloded Ordnance (UXO) H. Saito (), P. Goovaerts (), S. A. McKenna (2) Environmental and Water Resources Engineering, Deartment

More information

On the capacity of the general trapdoor channel with feedback

On the capacity of the general trapdoor channel with feedback On the caacity of the general tradoor channel with feedback Jui Wu and Achilleas Anastasooulos Electrical Engineering and Comuter Science Deartment University of Michigan Ann Arbor, MI, 48109-1 email:

More information

CONVOLVED SUBSAMPLING ESTIMATION WITH APPLICATIONS TO BLOCK BOOTSTRAP

CONVOLVED SUBSAMPLING ESTIMATION WITH APPLICATIONS TO BLOCK BOOTSTRAP Submitted to the Annals of Statistics arxiv: arxiv:1706.07237 CONVOLVED SUBSAMPLING ESTIMATION WITH APPLICATIONS TO BLOCK BOOTSTRAP By Johannes Tewes, Dimitris N. Politis and Daniel J. Nordman Ruhr-Universität

More information

Topic 7: Using identity types

Topic 7: Using identity types Toic 7: Using identity tyes June 10, 2014 Now we would like to learn how to use identity tyes and how to do some actual mathematics with them. By now we have essentially introduced all inference rules

More information

Convex Analysis and Economic Theory Winter 2018

Convex Analysis and Economic Theory Winter 2018 Division of the Humanities and Social Sciences Ec 181 KC Border Conve Analysis and Economic Theory Winter 2018 Toic 16: Fenchel conjugates 16.1 Conjugate functions Recall from Proosition 14.1.1 that is

More information

Brownian Motion and Random Prime Factorization

Brownian Motion and Random Prime Factorization Brownian Motion and Random Prime Factorization Kendrick Tang June 4, 202 Contents Introduction 2 2 Brownian Motion 2 2. Develoing Brownian Motion.................... 2 2.. Measure Saces and Borel Sigma-Algebras.........

More information

Elements of Asymptotic Theory. James L. Powell Department of Economics University of California, Berkeley

Elements of Asymptotic Theory. James L. Powell Department of Economics University of California, Berkeley Elements of Asymtotic Theory James L. Powell Deartment of Economics University of California, Berkeley Objectives of Asymtotic Theory While exact results are available for, say, the distribution of the

More information

Approximate Dynamic Programming for Dynamic Capacity Allocation with Multiple Priority Levels

Approximate Dynamic Programming for Dynamic Capacity Allocation with Multiple Priority Levels Aroximate Dynamic Programming for Dynamic Caacity Allocation with Multile Priority Levels Alexander Erdelyi School of Oerations Research and Information Engineering, Cornell University, Ithaca, NY 14853,

More information

2 Asymptotic density and Dirichlet density

2 Asymptotic density and Dirichlet density 8.785: Analytic Number Theory, MIT, sring 2007 (K.S. Kedlaya) Primes in arithmetic rogressions In this unit, we first rove Dirichlet s theorem on rimes in arithmetic rogressions. We then rove the rime

More information

Availability and Maintainability. Piero Baraldi

Availability and Maintainability. Piero Baraldi Availability and Maintainability 1 Introduction: reliability and availability System tyes Non maintained systems: they cannot be reaired after a failure (a telecommunication satellite, a F1 engine, a vessel

More information

arxiv: v1 [physics.data-an] 26 Oct 2012

arxiv: v1 [physics.data-an] 26 Oct 2012 Constraints on Yield Parameters in Extended Maximum Likelihood Fits Till Moritz Karbach a, Maximilian Schlu b a TU Dortmund, Germany, moritz.karbach@cern.ch b TU Dortmund, Germany, maximilian.schlu@cern.ch

More information

Lecture 4: Law of Large Number and Central Limit Theorem

Lecture 4: Law of Large Number and Central Limit Theorem ECE 645: Estimation Theory Sring 2015 Instructor: Prof. Stanley H. Chan Lecture 4: Law of Large Number and Central Limit Theorem (LaTeX reared by Jing Li) March 31, 2015 This lecture note is based on ECE

More information

Positive decomposition of transfer functions with multiple poles

Positive decomposition of transfer functions with multiple poles Positive decomosition of transfer functions with multile oles Béla Nagy 1, Máté Matolcsi 2, and Márta Szilvási 1 Deartment of Analysis, Technical University of Budaest (BME), H-1111, Budaest, Egry J. u.

More information

RANDOM WALKS AND PERCOLATION: AN ANALYSIS OF CURRENT RESEARCH ON MODELING NATURAL PROCESSES

RANDOM WALKS AND PERCOLATION: AN ANALYSIS OF CURRENT RESEARCH ON MODELING NATURAL PROCESSES RANDOM WALKS AND PERCOLATION: AN ANALYSIS OF CURRENT RESEARCH ON MODELING NATURAL PROCESSES AARON ZWIEBACH Abstract. In this aer we will analyze research that has been recently done in the field of discrete

More information

Stochastic integration II: the Itô integral

Stochastic integration II: the Itô integral 13 Stochastic integration II: the Itô integral We have seen in Lecture 6 how to integrate functions Φ : (, ) L (H, E) with resect to an H-cylindrical Brownian motion W H. In this lecture we address the

More information

MATH 361: NUMBER THEORY EIGHTH LECTURE

MATH 361: NUMBER THEORY EIGHTH LECTURE MATH 361: NUMBER THEORY EIGHTH LECTURE 1. Quadratic Recirocity: Introduction Quadratic recirocity is the first result of modern number theory. Lagrange conjectured it in the late 1700 s, but it was first

More information

Robust Predictive Control of Input Constraints and Interference Suppression for Semi-Trailer System

Robust Predictive Control of Input Constraints and Interference Suppression for Semi-Trailer System Vol.7, No.7 (4),.37-38 htt://dx.doi.org/.457/ica.4.7.7.3 Robust Predictive Control of Inut Constraints and Interference Suression for Semi-Trailer System Zhao, Yang Electronic and Information Technology

More information

On split sample and randomized confidence intervals for binomial proportions

On split sample and randomized confidence intervals for binomial proportions On slit samle and randomized confidence intervals for binomial roortions Måns Thulin Deartment of Mathematics, Usala University arxiv:1402.6536v1 [stat.me] 26 Feb 2014 Abstract Slit samle methods have

More information

Anytime communication over the Gilbert-Eliot channel with noiseless feedback

Anytime communication over the Gilbert-Eliot channel with noiseless feedback Anytime communication over the Gilbert-Eliot channel with noiseless feedback Anant Sahai, Salman Avestimehr, Paolo Minero Deartment of Electrical Engineering and Comuter Sciences University of California

More information

Age of Information: Whittle Index for Scheduling Stochastic Arrivals

Age of Information: Whittle Index for Scheduling Stochastic Arrivals Age of Information: Whittle Index for Scheduling Stochastic Arrivals Yu-Pin Hsu Deartment of Communication Engineering National Taiei University yuinhsu@mail.ntu.edu.tw arxiv:80.03422v2 [math.oc] 7 Ar

More information

Analysis of M/M/n/K Queue with Multiple Priorities

Analysis of M/M/n/K Queue with Multiple Priorities Analysis of M/M/n/K Queue with Multile Priorities Coyright, Sanjay K. Bose For a P-riority system, class P of highest riority Indeendent, Poisson arrival rocesses for each class with i as average arrival

More information

Power Aware Wireless File Downloading: A Constrained Restless Bandit Approach

Power Aware Wireless File Downloading: A Constrained Restless Bandit Approach PROC. WIOP 204 Power Aware Wireless File Downloading: A Constrained Restless Bandit Aroach Xiaohan Wei and Michael J. Neely, Senior Member, IEEE Abstract his aer treats ower-aware throughut maximization

More information

arxiv: v1 [physics.data-an] 26 Oct 2012

arxiv: v1 [physics.data-an] 26 Oct 2012 Constraints on Yield Parameters in Extended Maximum Likelihood Fits Till Moritz Karbach a, Maximilian Schlu b a TU Dortmund, Germany, moritz.karbach@cern.ch b TU Dortmund, Germany, maximilian.schlu@cern.ch

More information

PETER J. GRABNER AND ARNOLD KNOPFMACHER

PETER J. GRABNER AND ARNOLD KNOPFMACHER ARITHMETIC AND METRIC PROPERTIES OF -ADIC ENGEL SERIES EXPANSIONS PETER J. GRABNER AND ARNOLD KNOPFMACHER Abstract. We derive a characterization of rational numbers in terms of their unique -adic Engel

More information

HAUSDORFF MEASURE OF p-cantor SETS

HAUSDORFF MEASURE OF p-cantor SETS Real Analysis Exchange Vol. 302), 2004/2005,. 20 C. Cabrelli, U. Molter, Deartamento de Matemática, Facultad de Cs. Exactas y Naturales, Universidad de Buenos Aires and CONICET, Pabellón I - Ciudad Universitaria,

More information

2 Asymptotic density and Dirichlet density

2 Asymptotic density and Dirichlet density 8.785: Analytic Number Theory, MIT, sring 2007 (K.S. Kedlaya) Primes in arithmetic rogressions In this unit, we first rove Dirichlet s theorem on rimes in arithmetic rogressions. We then rove the rime

More information

Information collection on a graph

Information collection on a graph Information collection on a grah Ilya O. Ryzhov Warren Powell October 25, 2009 Abstract We derive a knowledge gradient olicy for an otimal learning roblem on a grah, in which we use sequential measurements

More information

Information collection on a graph

Information collection on a graph Information collection on a grah Ilya O. Ryzhov Warren Powell February 10, 2010 Abstract We derive a knowledge gradient olicy for an otimal learning roblem on a grah, in which we use sequential measurements

More information

HENSEL S LEMMA KEITH CONRAD

HENSEL S LEMMA KEITH CONRAD HENSEL S LEMMA KEITH CONRAD 1. Introduction In the -adic integers, congruences are aroximations: for a and b in Z, a b mod n is the same as a b 1/ n. Turning information modulo one ower of into similar

More information

Chapter 1 Fundamentals

Chapter 1 Fundamentals Chater Fundamentals. Overview of Thermodynamics Industrial Revolution brought in large scale automation of many tedious tasks which were earlier being erformed through manual or animal labour. Inventors

More information

Understanding and Using Availability

Understanding and Using Availability Understanding and Using Availability Jorge Luis Romeu, Ph.D. ASQ CQE/CRE, & Senior Member Email: romeu@cortland.edu htt://myrofile.cos.com/romeu ASQ/RD Webinar Series Noviembre 5, J. L. Romeu - Consultant

More information

Homework Solution 4 for APPM4/5560 Markov Processes

Homework Solution 4 for APPM4/5560 Markov Processes Homework Solution 4 for APPM4/556 Markov Processes 9.Reflecting random walk on the line. Consider the oints,,, 4 to be marked on a straight line. Let X n be a Markov chain that moves to the right with

More information

1 Extremum Estimators

1 Extremum Estimators FINC 9311-21 Financial Econometrics Handout Jialin Yu 1 Extremum Estimators Let θ 0 be a vector of k 1 unknown arameters. Extremum estimators: estimators obtained by maximizing or minimizing some objective

More information

Pretest (Optional) Use as an additional pacing tool to guide instruction. August 21

Pretest (Optional) Use as an additional pacing tool to guide instruction. August 21 Trimester 1 Pretest (Otional) Use as an additional acing tool to guide instruction. August 21 Beyond the Basic Facts In Trimester 1, Grade 8 focus on multilication. Daily Unit 1: Rational vs. Irrational

More information

Feedback-error control

Feedback-error control Chater 4 Feedback-error control 4.1 Introduction This chater exlains the feedback-error (FBE) control scheme originally described by Kawato [, 87, 8]. FBE is a widely used neural network based controller

More information

Modeling Residual-Geometric Flow Sampling

Modeling Residual-Geometric Flow Sampling 1 Modeling Residual-Geometric Flow Samling Xiaoming Wang, Xiaoyong Li, and Dmitri Loguinov Abstract Traffic monitoring and estimation of flow arameters in high seed routers have recently become challenging

More information

Recursive Estimation of the Preisach Density function for a Smart Actuator

Recursive Estimation of the Preisach Density function for a Smart Actuator Recursive Estimation of the Preisach Density function for a Smart Actuator Ram V. Iyer Deartment of Mathematics and Statistics, Texas Tech University, Lubbock, TX 7949-142. ABSTRACT The Preisach oerator

More information

COMMUNICATION BETWEEN SHAREHOLDERS 1

COMMUNICATION BETWEEN SHAREHOLDERS 1 COMMUNICATION BTWN SHARHOLDRS 1 A B. O A : A D Lemma B.1. U to µ Z r 2 σ2 Z + σ2 X 2r ω 2 an additive constant that does not deend on a or θ, the agents ayoffs can be written as: 2r rθa ω2 + θ µ Y rcov

More information

The Poisson Regression Model

The Poisson Regression Model The Poisson Regression Model The Poisson regression model aims at modeling a counting variable Y, counting the number of times that a certain event occurs during a given time eriod. We observe a samle

More information

t 0 Xt sup X t p c p inf t 0

t 0 Xt sup X t p c p inf t 0 SHARP MAXIMAL L -ESTIMATES FOR MARTINGALES RODRIGO BAÑUELOS AND ADAM OSȨKOWSKI ABSTRACT. Let X be a suermartingale starting from 0 which has only nonnegative jums. For each 0 < < we determine the best

More information

Weakly Short Memory Stochastic Processes: Signal Processing Perspectives

Weakly Short Memory Stochastic Processes: Signal Processing Perspectives Weakly Short emory Stochastic Processes: Signal Processing Persectives by Garimella Ramamurthy Reort No: IIIT/TR/9/85 Centre for Security, Theory and Algorithms International Institute of Information Technology

More information

Plotting the Wilson distribution

Plotting the Wilson distribution , Survey of English Usage, University College London Setember 018 1 1. Introduction We have discussed the Wilson score interval at length elsewhere (Wallis 013a, b). Given an observed Binomial roortion

More information

6 Stationary Distributions

6 Stationary Distributions 6 Stationary Distributions 6. Definition and Examles Definition 6.. Let {X n } be a Markov chain on S with transition robability matrix P. A distribution π on S is called stationary (or invariant) if π

More information

Generalized Coiflets: A New Family of Orthonormal Wavelets

Generalized Coiflets: A New Family of Orthonormal Wavelets Generalized Coiflets A New Family of Orthonormal Wavelets Dong Wei, Alan C Bovik, and Brian L Evans Laboratory for Image and Video Engineering Deartment of Electrical and Comuter Engineering The University

More information

Sums of independent random variables

Sums of independent random variables 3 Sums of indeendent random variables This lecture collects a number of estimates for sums of indeendent random variables with values in a Banach sace E. We concentrate on sums of the form N γ nx n, where

More information

A Social Welfare Optimal Sequential Allocation Procedure

A Social Welfare Optimal Sequential Allocation Procedure A Social Welfare Otimal Sequential Allocation Procedure Thomas Kalinowsi Universität Rostoc, Germany Nina Narodytsa and Toby Walsh NICTA and UNSW, Australia May 2, 201 Abstract We consider a simle sequential

More information

Chapter 7: Special Distributions

Chapter 7: Special Distributions This chater first resents some imortant distributions, and then develos the largesamle distribution theory which is crucial in estimation and statistical inference Discrete distributions The Bernoulli

More information

Journal of Mathematical Analysis and Applications

Journal of Mathematical Analysis and Applications J. Math. Anal. Al. 44 (3) 3 38 Contents lists available at SciVerse ScienceDirect Journal of Mathematical Analysis and Alications journal homeage: www.elsevier.com/locate/jmaa Maximal surface area of a

More information

LORENZO BRANDOLESE AND MARIA E. SCHONBEK

LORENZO BRANDOLESE AND MARIA E. SCHONBEK LARGE TIME DECAY AND GROWTH FOR SOLUTIONS OF A VISCOUS BOUSSINESQ SYSTEM LORENZO BRANDOLESE AND MARIA E. SCHONBEK Abstract. In this aer we analyze the decay and the growth for large time of weak and strong

More information

s v 0 q 0 v 1 q 1 v 2 (q 2) v 3 q 3 v 4

s v 0 q 0 v 1 q 1 v 2 (q 2) v 3 q 3 v 4 Discrete Adative Transmission for Fading Channels Lang Lin Λ, Roy D. Yates, Predrag Sasojevic WINLAB, Rutgers University 7 Brett Rd., NJ- fllin, ryates, sasojevg@winlab.rutgers.edu Abstract In this work

More information

System Reliability Estimation and Confidence Regions from Subsystem and Full System Tests

System Reliability Estimation and Confidence Regions from Subsystem and Full System Tests 009 American Control Conference Hyatt Regency Riverfront, St. Louis, MO, USA June 0-, 009 FrB4. System Reliability Estimation and Confidence Regions from Subsystem and Full System Tests James C. Sall Abstract

More information

Some Results on the Generalized Gaussian Distribution

Some Results on the Generalized Gaussian Distribution Some Results on the Generalized Gaussian Distribution Alex Dytso, Ronit Bustin, H. Vincent Poor, Daniela Tuninetti 3, Natasha Devroye 3, and Shlomo Shamai (Shitz) Abstract The aer considers information-theoretic

More information

B8.1 Martingales Through Measure Theory. Concept of independence

B8.1 Martingales Through Measure Theory. Concept of independence B8.1 Martingales Through Measure Theory Concet of indeendence Motivated by the notion of indeendent events in relims robability, we have generalized the concet of indeendence to families of σ-algebras.

More information

ECE 534 Information Theory - Midterm 2

ECE 534 Information Theory - Midterm 2 ECE 534 Information Theory - Midterm Nov.4, 009. 3:30-4:45 in LH03. You will be given the full class time: 75 minutes. Use it wisely! Many of the roblems have short answers; try to find shortcuts. You

More information

On the Relationship Between Packet Size and Router Performance for Heavy-Tailed Traffic 1

On the Relationship Between Packet Size and Router Performance for Heavy-Tailed Traffic 1 On the Relationshi Between Packet Size and Router Performance for Heavy-Tailed Traffic 1 Imad Antonios antoniosi1@southernct.edu CS Deartment MO117 Southern Connecticut State University 501 Crescent St.

More information

Session 5: Review of Classical Astrodynamics

Session 5: Review of Classical Astrodynamics Session 5: Review of Classical Astrodynamics In revious lectures we described in detail the rocess to find the otimal secific imulse for a articular situation. Among the mission requirements that serve

More information

Various Proofs for the Decrease Monotonicity of the Schatten s Power Norm, Various Families of R n Norms and Some Open Problems

Various Proofs for the Decrease Monotonicity of the Schatten s Power Norm, Various Families of R n Norms and Some Open Problems Int. J. Oen Problems Comt. Math., Vol. 3, No. 2, June 2010 ISSN 1998-6262; Coyright c ICSRS Publication, 2010 www.i-csrs.org Various Proofs for the Decrease Monotonicity of the Schatten s Power Norm, Various

More information

BEST POSSIBLE DENSITIES OF DICKSON m-tuples, AS A CONSEQUENCE OF ZHANG-MAYNARD-TAO

BEST POSSIBLE DENSITIES OF DICKSON m-tuples, AS A CONSEQUENCE OF ZHANG-MAYNARD-TAO BEST POSSIBLE DENSITIES OF DICKSON m-tuples, AS A CONSEQUENCE OF ZHANG-MAYNARD-TAO ANDREW GRANVILLE, DANIEL M. KANE, DIMITRIS KOUKOULOPOULOS, AND ROBERT J. LEMKE OLIVER Abstract. We determine for what

More information

arxiv: v1 [cs.lg] 31 Jul 2014

arxiv: v1 [cs.lg] 31 Jul 2014 Learning Nash Equilibria in Congestion Games Walid Krichene Benjamin Drighès Alexandre M. Bayen arxiv:408.007v [cs.lg] 3 Jul 204 Abstract We study the reeated congestion game, in which multile oulations

More information

1 1 c (a) 1 (b) 1 Figure 1: (a) First ath followed by salesman in the stris method. (b) Alternative ath. 4. D = distance travelled closing the loo. Th

1 1 c (a) 1 (b) 1 Figure 1: (a) First ath followed by salesman in the stris method. (b) Alternative ath. 4. D = distance travelled closing the loo. Th 18.415/6.854 Advanced Algorithms ovember 7, 1996 Euclidean TSP (art I) Lecturer: Michel X. Goemans MIT These notes are based on scribe notes by Marios Paaefthymiou and Mike Klugerman. 1 Euclidean TSP Consider

More information