Publications (89)32.7 Total impact
 [Show abstract] [Hide abstract]
ABSTRACT: There has been considerable research on the performance analysis of ondemand caching replacement policies like LeastRecentlyUsed (LRU), FirstInFirstOut (FIFO) or Random (RND). Much progress has been made on the analysis of a single cache running these algorithms. However it has been almost impossible to extend the results to networks of caches. In this paper, we introduce a TimeToLive (TTL) based caching model, that assigns a timer to each content stored in the cache and redraws it every time the content is requested (at each hit/miss). We derive the performance metrics (hit/miss ratio and rate, occupancy) of a TTLbased cache in isolation fed by stationary and ergodic request processes with general TTL distributions. Moreover we propose an iterative procedure to analyze TTLbased cache networks under the assumptions that requests are described by renewal processes (that generalize Poisson processes or the standard IRM assumption). We validate our theoretical findings through eventdriven and MonteCarlo simulations based on the Fourier Amplitude Sensitivity Test to explore the space of the input parameters. We observe that our analytic model predicts remarkably well all metrics of interest with relative errors smaller than 1%1%.Computer Networks. 01/2014;  [Show abstract] [Hide abstract]
ABSTRACT: This paper studies the performance of PeertoPeer storage and backup systems (P2PSS). These systems are based on three pillars: data fragmentation and dissemination among the peers, redundancy mechanisms to cope with peers churn and repair mechanisms to recover lost or temporarily unavailable data. Usually, redundancy is achieved either by using replication or by using erasure codes. A new class of network coding (regenerating codes) has been proposed recently. Therefore, we will adapt our work to these three redundancy schemes. We introduce two mechanisms for recovering lost data and evaluate their performance by modeling them through absorbing Markov chains. Specifically, we evaluate the quality of service provided to users in terms of durability and availability of stored data for each recovery mechanism and deduce the impact of its parameters on the system performance. The first mechanism is centralized and based on the use of a single server that can recover multiple losses at once. The second mechanism is distributed: reconstruction of lost fragments is iterated sequentially on many peers until that the required level of redundancy is attained. The key assumptions made in this work, in particular, the assumptions made on the recovery process and peer ontimes distribution, are in agreement with the analysis in [1] and in [2] respectively. The models are thereby general enough to be applicable to many distributed environments as shown through numerical computations. We find that, in stable environments such as local area or research institute networks where machines are usually highly available, the distributedrepair scheme in erasurecoded systems offers a reliable, scalable and cheap storage/backup solution. For the case of highly dynamic environments, in general, the distributedrepair scheme is inefficient, in particular to maintain high data availability, unless the data redundancy is high. Using regenerating codes overcomes this limitation of the distributedrepair scheme. P2PSS with centralizedrepair scheme are efficient in any environment but have the disadvantage of relying on a centralized authority. However, the analysis of the overhead cost (e.g. computation, bandwidth and complexity cost) resulting from the different redundancy schemes with respect to their advantages (e.g. simplicity), is left for future work.Computer Networks. 01/2014; 
Conference Paper: Estimating endtoend delays under changing conditions
[Show abstract] [Hide abstract]
ABSTRACT: We consider the problem of estimating the endtoend latency of intermittently connected paths in disruption/delay tolerant networks. This is useful when performing source routing, in which a complete path is chosen for a packet to travel from source to destination (when intermediate nodes are really low complexity devices that can only forward packets but cannot perform route computations), or in linear network topologies. While computing the time to traverse such a path may be straightforward in fixed, static networks, doing so becomes much more challenging in dynamic networks, in which the state of an edge in one timeslot (i.e., its presence or absence) is random, and may depend on its state in the previous timeslot. The traversal time is due to both time spent waiting for edges to appear and time spent crossing them once they become available. We compute the expected traversal time (ETT) for a dynamic path in a number of special cases of stochastic edge dynamics models, and for three different edge failure models, culminating in a surprisingly nontrivial yet realistic ``hybrid network" setting in which the initial configuration of edge states for the entire path is known. We show that the ETT for this "initial configuration" setting can be computed in quadratic time (as a function of path length), by an algorithm based on probability generating functions. We also give several lineartime upper and lower bounds on the ETT, which we evaluate, along with our ETT algorithm, using numerical simulations.Proceedings of the 8th ACM MobiCom workshop on Challenged networks; 09/2013  [Show abstract] [Hide abstract]
ABSTRACT: In source routing, a complete path is chosen for a packet to travel from source to destination. While computing the time to traverse such a path may be straightforward in a fixed, static graph, doing so becomes much more challenging in dynamic graphs, in which the state of an edge in one time slot (i.e., its presence or absence) is random, and may depend on its state in the previous time step. The traversal time is due to both time spent waiting for edges to appear and time spent crossing them once they become available. We compute the expected traversal time (ETT) for a dynamic path in a number of special cases of stochastic edge dynamics models, and for three edge failure models, culminating in a surprisingly challenging yet realistic setting in which the initial configuration of edge states for the entire path is known. We show that the ETT for this "initial configuration" setting can be computed in quadratic time, by an algorithm based on probability generating functions. We also give several lineartime upper and lower bounds on the ETT.03/2013; 
Article: Predicting the Impact of Measures Against P2P Networks: Transient Behavior and Phase Transition
[Show abstract] [Hide abstract]
ABSTRACT: The paper has two objectives. The first is to study rigorously the transient behavior of some peertopeer (P2P) networks whenever information is replicated and disseminated according to epidemiclike dynamics. The second is to use the insight gained from the previous analysis in order to predict how efficient are measures taken against P2P networks. We first introduce a stochastic model that extends a classical epidemic model and characterize the P2P swarm behavior in presence of freeriding peers. We then study a second model in which a peer initiates a contact with another peer chosen randomly. In both cases, the network is shown to exhibit phase transitions: A small change in the parameters causes a large change in the behavior of the network. We show, in particular, how phase transitions affect measures of content providers against P2P networks that distribute nonauthorized music, books, or articles and what is the efficiency of countermeasures. In addition, our analytical framework can be generalized to characterize the heterogeneity of cooperative peers.IEEE/ACM Transactions on Networking 01/2013; 21(3):935949. · 2.01 Impact Factor  [Show abstract] [Hide abstract]
ABSTRACT: Two independent Poisson streams of jobs flow into a singleserver service system having a limited common buffer that can hold at most one job. If a typei job (i=1,2) finds the server busy, it is blocked and routed to a separate typei retrial (orbit) queue that attempts to redispatch its jobs at its specific Poisson rate. This creates a system with three dependent queues. Such a queueing system serves as a model for two competing job streams in a carrier sensing multiple access system. We study the queueing system using multidimensional probability generating functions, and derive its necessary and sufficient stability conditions while solving a boundary value problem. Various performance measures are calculated and numerical results are presented.Queueing Systems 06/2012; · 0.44 Impact Factor 
Conference Paper: Analysis of TTLbased cache networks
[Show abstract] [Hide abstract]
ABSTRACT: Many researchers have been working on the performance analysis of caching in InformationCentric Networks (ICNs) under various replacement policies like Least Recently Used (LRU), FIFO or Random (RND). However, no exact results are provided, and many approximate models do not scale even for the simple network of two caches connected in tandem. In this paper, we introduce a TimeToLive based policy (TTL), that assigns a timer to each content stored in the cache and redraws the timer each time the content is requested (at each hit/miss). We show that our TTL policy is more general than LRU, FIFO or RND, since it is able to mimic their behavior under an appropriate choice of its parameters. Moreover, the analysis of networks of TTLbased caches appears simpler not only under the Independent Reference Model (IRM, on which many existing results rely) but also with the Renewal Model for requests. In particular, we determine exact formulas for the performance metrics of interest for a linear network and a tree network with one root cache and N leaf caches. For more general networks, we propose an approximate solution with the relative errors smaller than 103 and 102 for exponentially distributed and constant TTLs respectively.Performance Evaluation Methodologies and Tools (VALUETOOLS), 2012 6th International Conference on; 01/2012  [Show abstract] [Hide abstract]
ABSTRACT: In this paper we study the behavior of a continuous time random walk (CTRW) on a stationary and ergodic time varying dynamic graph. We establish conditions under which the CTRW is a stationary and ergodic process. In general, the stationary distribution of the walker depends on the walker rate and is difficult to characterize. However, we characterize the stationary distribution in the following cases: i) the walker rate is significantly larger or smaller than the rate in which the graph changes (timescale separation), ii) the walker rate is proportional to the degree of the node that it resides on (coupled dynamics), and iii) the degrees of node belonging to the same connected component are identical (structural constraints). We provide examples that illustrate our theoretical findings.Sigmetrics Performance Evaluation Review  SIGMETRICS. 12/2011;  [Show abstract] [Hide abstract]
ABSTRACT: The paper has two objectives. The first is to study rigorously the transient behavior of some peertopeer (P2P) networks whenever information is replicated and disseminated according to epidemiclike dynamics. The second is to use the insight gained from the previous analysis in order to predict how efficient are measures taken against P2P networks. We first introduce a stochastic model which extends a classical epidemic model, and characterize the P2P swarm behavior in presence of free riding peers. We then study a second model in which a peer initiates a contact with another peer chosen randomly. In both cases the network is shown to exhibit phase transitions: a small change in the parameters causes a large change in the behavior of the network. We show, in particular, how phase transitions affect measures of content providers against P2P networks that distribute nonauthorized music or books, and what is the efficiency of countermeasures.INFOCOM, 2011 Proceedings IEEE; 05/2011 
Article: Optimal threshold control by the robots of web search engines with obsolescence of documents
[Show abstract] [Hide abstract]
ABSTRACT: A typical web search engine consists of three principal parts: crawling engine, indexing engine, and searching engine. The present work aims to optimize the performance of the crawling engine. The crawling engine finds new web pages and updates web pages existing in the database of the web search engine. The crawling engine has several robots collecting information from the Internet. We first calculate various performance measures of the system (e.g., probability of arbitrary page loss due to the buffer overflow, probability of starvation of the system, the average time waiting in the buffer). Intuitively, we would like to avoid system starvation and at the same time to minimize the information loss. We formulate the problem as a multicriteria optimization problem and attributing a weight to each criterion. We solve it in the class of threshold policies. We consider a very general web page arrival process modeled by Batch Marked Markov Arrival Process and a very general service time modeled by Phasetype distribution. The model has been applied to the performance evaluation and optimization of the crawler designed by INRIA Maestro team in the framework of the RIAM INRIACanon research project.Computer Networks. 01/2011;  [Show abstract] [Hide abstract]
ABSTRACT: In this paper we study the dynamic aspects of the coverage of a mobile sensor network resulting from continuous movement of sensors. As sensors move around, initially uncovered locations are likely to be covered at a later time. A larger area is covered as time continues, and intruders that might never be detected in a stationary sensor network can now be detected by moving sensors. However, this improvement in coverage is achieved at the cost that a location is covered only part of the time, alternating between covered and not covered. We characterize area coverage at specific time instants and during time intervals, as well as the time durations that a location is covered and uncovered. We further characterize the time it takes to detect a randomly located intruder. For mobile intruders, we take a game theoretic approach and derive optimal mobility strategies for both sensors and intruders. Our results show that sensor mobility brings about unique dynamic coverage properties not present in a stationary sensor network, and that mobility can be exploited to compensate for the lack of sensors to improve coverage.IEEE Transactions on Parallel and Distributed Systems 01/2011; abs/1101.0376. · 1.80 Impact Factor  [Show abstract] [Hide abstract]
ABSTRACT: With the introduction of the new generation high speed routers, it becomes possible to improve the Quality of Service, the Quality of Experience for users and the network efficiency for ISPs with the help of "flowaware" traffic management. An example of the "flowaware" traffic management is the AlcatelLucent framework "Semantic Networking," where shortlived and longlived TCP flows are treated differently. Shortlived flows are processed with high priority and longlived flows are controlled in a "flowaware" fashion. To control efficiently the longlived flows, one needs to know an estimation of the Round Trip Time (RTT). In the present work, we provide an online RTT estimation algorithm which is passive and can deal with a oneway traffic. The oneway traffic requirement is essential for the application of the algorithm for "flowaware" traffic management inside the network. To the best of our knowledge, there was no online oneway traffic RTT estimators. Tests on the Internet demonstrate high accuracy of the proposed estimator. The results show that, 75% (resp. 99%) of the time, the RTT estimation is within 10% (resp. 20%) of the RTT at the source.01/2010;  [Show abstract] [Hide abstract]
ABSTRACT: This work investigates distributed transmission scheduling in wireless networks. Due to interference constraints, "neighboring links" cannot be simultaneously activated, otherwise transmissions will fail. Here, we consider any binary model of interference. We follow the model described by Bui, Sanghavi, and Srikant in SBS07,SBS09. We suppose that time is slotted and during each slot we have two phases: one control phase which determines what links will be activated and send data during the second phase. We assume random arrivals on each link during each slot, therefore a queue is associated to each link. Since nodes do not have a global knowledge of the network, our aim (like in SBS07,SBS09) is to design for the control phase, a distributed algorithm which determines a set of non interfering links. To be efficient the control phase should be as short as possible; this is done by exchanging control messages during a constant number of minislots (constant overhead). In this article we design the first fully distributed local algorithm with the following properties: it works for any arbitrary binary interference model; it has a constant overhead (independent of the size of the network and the values of the queues); and it needs no knowledge. Indeed contrary to other existing algorithms, we do not need to know the values of the queues of the "neighboring links", which are difficult to obtain in a wireless network with interference. We prove that this algorithm gives a maximal set of active links (in each interference set, there is at least one active edge). We also give sufficient conditions for stability under Markovian assumptions. Finally the performance of our algorithm (throughput, stability) is investigated and compared via simulations to that of previously proposed schemes.Sigmetrics 2010. 01/2010;  [Show abstract] [Hide abstract]
ABSTRACT: The peertopeer (P2P) paradigm have emerged as a cheap, scalable, selfrepairing and faulttolerant storage solution. This solution relies on erasure codes to generate additional redundant fragments of each "block of data" in order to increase the reliability and availability and overcome the churn. When the amount of unreachable fragments attains a predefined threshold, due to permanent departures or long disconnections of peers, a recovery process is initiated to compensate the missing fragments, requiring multiple fragments of data of a given "block" to be downloaded in parallel for an enhanced service. Recent modeling efforts that address the availability and the durability of data have assumed the recovery process to follow an exponential distribution, an assumption made mainly in the absence of studies characterizing the "real" distribution of the recovery process. This work aims at filling this gap and better understanding the behavior of these systems through simulation while taking into consideration the heterogeneity of peers, the underlying network topologies, the propagation delays and the transport protocol. To that end, the distributed storage protocol is implemented in the NS2 network simulator. This paper describes a realistic simulation model that captures the behavior of P2P storage systems. We provide some experiments results that show how modeling the availability and durability can be impacted by the recovery times distribution which is impacted in turn by the characteristics of the the network and the context.10/2009;  [Show abstract] [Hide abstract]
ABSTRACT: Peertopeer storage systems rely on data fragmentation and distributed storage. Unreachable fragments are continuously recovered, requiring multiple fragments of data (constituting a ldquoblockrdquo) to be downloaded in parallel. Recent modeling efforts have assumed the recovery process to follow an exponential distribution, an assumption made mainly in the absence of studies characterizing the ldquorealrdquo distribution of the recovery process. This work aims at filling this gap through a simulation study. To that end, we implement the distributed storage protocol in the NS2 network simulator and run a total of seven experiments covering a large variety of scenarios. We show that the fragment download time follows approximately an exponential distribution. We also show that the block download time and the recovery time essentially follow a hypoexponential distribution with many distinct phases (maximum of as many exponentials). We use expectation maximization and least square estimation algorithms to fit the empirical distributions. We also provide a good approximation of the number of phases of the hypoexponential distribution that applies in all scenarios considered. Last, we test the goodness of our fits using statistical (KolmogorovSmirnov test) and graphical methods.Teletraffic Congress, 2009. ITC 21 2009. 21st International; 10/2009 
Conference Paper: Analysis of relay protocols for throwboxequipped DTNs
[Show abstract] [Hide abstract]
ABSTRACT: This paper addresses the design and performance evaluation of relay strategies for opportunistic Delay Tolerant Networks (DTNs) augmented with throwboxes. By opportunistic we mean that a node does not have any knowledge regarding its past and future contact opportunities with the other nodes. We consider a network model composed of both mobile relay nodes and throwboxes, where throwboxes are stationary wireless devices acting simply as fixed relays. We propose and evaluate various relay strategies, where the goal is to take advantage of the presence of throwboxes to minimize resources consumption at mobile nodes. Under Markovian assumptions we introduce a mathematical framework which allows us to calculate the main performance metrics (average delivery delay, overhead, etc.) of each proposed relay scheme. The obtained results highlight the various tradeoffs that are left to network designers when adding throwboxes to a DTN, and draw insights on the effectiveness of these strategies.Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks, 2009. WiOPT 2009. 7th International Symposium on; 07/2009 
Conference Paper: Distributed Storage Management of Evolving Files in Delay Tolerant Ad Hoc Networks
[Show abstract] [Hide abstract]
ABSTRACT: This work focuses on a class of distributed storage systems whose content may evolve over time. Each component or node of the storage system is mobile and the set of all nodes forms a delay tolerant (ad hoc) network (DTN). The goal of the paper is to study efficient ways for distributing evolving files within DTNs and for managing dynamically their content. We specify to dynamic files where not only the latest version is useful but also previous ones; we restrict however to files where a file has no use if another more recent version is available. The DTN is composed of fixed number of nodes including a single source. At some points in time the source makes available a new version of a single file F. We consider both the cases when (a) nodes do not cooperate and (b) nodes cooperate. In case (a) only the source may transmit a copy of F to a node that it meets, while in case (b) any node may transmit a copy of F to a node that it meets. Scenario (a) is studied under the assumption that the source updates F at discrete times t = 0,1,.. .. Within each slot [t,t + 1) there is a fixed probability that a node meets the source. A file management policy is a set of rules specifying when the source transmits a copy of F to a node (say node i) that it meets; this decision only depends on the age of the version of F (if any) that node i is carrying, where the age is k if this version was created k1 slots ago. We And the optimal static (resp. dynamic) policy which maximizes a general utility function under a constraint on the number of transmissions within a slot. In particular, we show the existence of a threshold dynamic policy. In scenario (b) F is updated at random points in time. Similar to scenerio (a) we assume that each node knows the age of the file it carries (the case where nodes only know the date of creation of a file is studied in (E. Altman et al., 2008)). Under Markovian assumptions regarding nodes mobility and update frequency of F, we study the stability of th e system (aging of the nodes) and derive an (approximate) optimal static policy. We then revisit scenario (a) when the source does not know the number of nodes and the probability that the source meets a node in a slot, and we derive a stochastic approximation algorithm which we show to converge to the optimal static policy found in the complete information setting. Numerical results illustrate the respective performance of optimal static and dynamic policies as well as the benefit of node cooperation.INFOCOM 2009, IEEE; 05/2009 
Article: Performance Analysis of Centralized versus Distributed Recovery Schemes in P2P Storage Systems
[Show abstract] [Hide abstract]
ABSTRACT: This report studies the performance of PeertoPeer Storage Systems (P2PSS) in terms of data lifetime and availability. Two schemes for recovering lost data are modeled through absorbing Markov chains and their performance are evaluated and compared. The first scheme relies on a centralized controller that can recover multiple losses at once, whereas the second scheme is distributed and recovers one loss at a time. The impact of each system parameter on the performance is evaluated, and guidelines are derived on how to engineer the system and tune its key parameters in order to provide desired lifetime and/or availability of data. We find that, in stable environments such as local area or research laboratory networks where machines are usually highly available, the distributedrepair scheme offers a reliable, scalable and cheap storage/backup solution. This is in contrast with the case of highly dynamic environments, where the distributedrepair scheme is inefficient as long as the storage overhead is kept reasonable. P2PSS with centralizedrepair scheme are efficient in any environment but have the disadvantage of relying on a centralized authority. Our analysis also suggests that the use of large size fragments reduces the efficiency of the recovery mechanism.01/2009;  [Show abstract] [Hide abstract]
ABSTRACT: This work focuses on a class of distributed storage systems whose content may evolve over time. Each component or node of the storage system is mobile and the set of all nodes forms a delay tolerant (ad hoc) network (DTN). The goal of the paper is to study efficient ways for distributing evolving files within DTNs and for managing dynamically their content. We specify to dynamic files where not only the latest version is useful but also previous ones; we restrict however to files where a file has no use if another more recent version is available. There are N+1 mobile nodes including a single} source which at some points in time makes available a new version of a single} file F. We consider both the cases when (a) nodes do not cooperate and (b) nodes cooperate. In case (a) only the source may transmit a copy of the latest version of F to a node that it meets, while in case (b) any node may transmit a copy of F to a node that it meets. A file management policy is a set of rules specifying when a node may send a copy of F to a node that it meets. The objective is to find file management policies which maximize some system utility functions under a constraint on the resource consumption. Both myopic (static) and statedependent (dynamic) policies are considered, where the state of a node is the age of the copy of F it carries. Scenario (a) is studied under the assumption that the source updates F at discrete times t=0,1,\ldots. During a slot [t,t+1) the source meets any node with a fixed probability q. We find the optimal static (resp. dynamic) policy which maximizes a general utility function under a constraint on the number of transmissions within a slot. In particular, we show the existence of a threshold dynamic policy. In scenario (b) F is updated at random points in time, with the consequence that between two meetings with the source a node does not know the age evolution of the version of F it holds. Under Markovian assumptions regarding nodes mobility and update frequency of F, we study the stability of the system (aging of the nodes) and derive an (approximate) optimal static policy. We then revisit scenario (a) when the source does not know parameter N (node population) and q (node meeting probability) and derive a stochastic approximation algorithm which we show to converge to the optimal static policy found in the complete information setting. Numerical results illustrate the respective performance of optimal static and dynamic policies as well as the benefit of node cooperation.01/2009;  Perform. Eval. 01/2008; 65:463483.
Publication Stats
2k  Citations  
32.70  Total Impact Points  
Top Journals
Institutions

1992–2013

University of NiceSophia Antipolis
Nice, ProvenceAlpesCôte d'Azur, France 
Tilburg University
Tilburg, North Brabant, Netherlands


2007–2008

Universiteit Twente
Enschede, Overijssel, Netherlands 
The University of Arizona
 Department of Electrical and Computer Engineering
Tucson, Arizona, United States


1998

University of Wisconsin, Madison
 Department of Electrical and Computer Engineering
Madison, MS, United States
