Donald F. Towsley

University of Massachusetts Amherst, Amherst Center, Massachusetts, United States

Are you Donald F. Towsley?

Claim your profile

Publications (439)205.73 Total impact

  • J. Zhao · J.C.S. Lui · D. Towsley · P. Wang · X. Guan ·
    [Show abstract] [Hide abstract]
    ABSTRACT: Random walk-based graph sampling methods have become increasingly popular and important for characterizing large-scale complex networks. While powerful, they are known to exhibit problems when the graph is loosely connected, which slows down the convergence of a random walk and can result in poor estimation accuracy. In this work, we observe that many graphs under study, called target graphs, usually do not exist in isolation. In many situations, a target graph is often related to an auxiliary graph and an affiliation graph, and the target graph becomes better connected when viewed from these three graphs as a whole, or what we called a hybrid social-affiliation network. This viewpoint brings extra benefits to the graph sampling framework, e.g., when directly sampling a target graph is difficult or inefficient, we can efficiently sample it with the assistance of auxiliary and affiliation graphs. We propose three sampling methods on such a hybrid social-affiliation network to estimate target graph characteristics, and conduct extensive experiments on both synthetic and real datasets, to demonstrate the effectiveness of these new sampling methods.
  • R. Soltani · B. Bash · D. Goeckel · S. Guha · D. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: Covert communication, also known as low probability of detection (LPD) communication, prevents the adversary from knowing that a communication is taking place. Recent work has demonstrated that, in a three-party scenario with a transmitter (Alice), intended recipient (Bob), and adversary (Warden Willie), the maximum number of bits that can be transmitted reliably from Alice to Bob without detection by Willie, when additive white Gaussian noise (AWGN) channels exist between all parties, is on the order of the square root of the number of channel uses. In this paper, we begin consideration of network scenarios by studying the case where there are additional 'friendly' nodes present in the environment that can produce artificial noise to aid in hiding the communication. We establish achievability results by considering constructions where the system node closest to the warden produces artificial noise and demonstrate a significant improvement in the throughput achieved covertly, without requiring close coordination between Alice and the noise-generating node. Conversely, under mild restrictions on the communication strategy, we demonstrate no higher covert throughput is possible. Extensions to the consideration of the achievable covert throughput when multiple wardens randomly located in the environment collaborate to attempt detection of the transmitter are also considered.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We study the emergence of long-range connectivity in multilayer networks (also termed multiplex, composite and overlay networks) obtained by merging the connectivity subgraphs of multiple percolating instances of an underlying backbone network. Multilayer networks have applications ranging from studying long-range connectivity in a communication or social network formed with hybrid technologies, a transportation network connecting the same cities via rail, road and air, in studying outbreaks of flu epidemics involving multiple viral strains, studying temporal flow of information in dynamic networks, and potentially in studying conductivity properties of graphene-like stacked lattices. For a homogenous multilayer network---formed via merging $M$ random site-percolating instances of the same graph $G$ with single-layer site-occupation probability $q$---we argue that when $q$ exceeds a threshold $q_c(M) = \Theta(1/\sqrt{M})$, a spanning cluster appears in the multilayer network. Using a configuration model approach, we find $q_c(M)$ exactly for random graphs with arbitrary degree distributions, which have many applications in mathematical sociology. For multilayer percolation in a general graph $G$, we show that $q_c/\sqrt{M} < q_c(M) < \sqrt{-\ln(1-p_c)}/{\sqrt{M}}, \forall M \in {\mathbb Z}^+$, where $q_c$ and $p_c$ are the site and bond percolation thresholds of $G$, respectively. We show a close connection between multilayer percolation and mixed (site-bond) percolation, since both provide a smooth bridge between pure-site and pure-bond percolations. We find excellent approximations and bounds on layered percolation thresholds for regular lattices using the aforesaid connection, and provide several exact results (via numerical simulations), and a specialized bound for the multilayer kagome lattice using a site-to-bond transformation technique.
  • Ting He · D. Goeckel · R. Raghavendra · D. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: We consider the problem of endhost-based shortest path routing in a network with unknown, time-varying link qualities. Endhost-based routing is needed when internal nodes of the network do not have the scope or capability to provide globally optimal paths to given source-destination pairs, as can be the case in networks consisting of autonomous subnetworks or those with endhost-based routing restrictions. Assuming the source can probe links along selected paths, we formulate the problem as an online learning problem, where an existing solution achieves a performance loss (called regret) that is logarithmic in time with respect to (wrt) an offline algorithm that knows the link qualities. Current solutions assume coupled probing and routing; in contrast, we give a simple algorithm based on decoupled probing and routing, whose regret is only constant in time. We then extend our solution to support multi-path probing and cooperative learning between multiple sources, where we show an inversely proportional decay in regret wrt the probing rate. We also show that without the decoupling, the regret grows at least logarithmically in time, thus establishing decoupling as critical for obtaining constant regret. Although our analysis assumes certain conditions (i.i.d.) on link qualities, our solution applies with straightforward amendments to much broader scenarios where these conditions are relaxed. The efficacy of the proposed solution is verified by trace-driven simulations.
    INFOCOM, 2013 Proceedings IEEE; 01/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: Characterizing user pair relationships is important for applications such as friend recommendation and interest targeting in online social networks (OSNs). Due to the large scale nature of such networks, it is infeasible to enumerate all user pairs and so sampling is used. In this paper, we show that it is a great challenge even for OSN service providers to characterize user pair relationships even when they possess the complete graph topology. The reason is that when sampling techniques (i.e., uniform vertex sampling (UVS) and random walk (RW)) are naively applied, they can introduce large biases, in particular, for estimating similarity distribution of user pairs with constraints such as existence of mutual neighbors, which is important for applications such as identifying network homophily. Estimating statistics of user pairs is more challenging in the absence of the complete topology information, since an unbiased sampling technique such as UVS is usually not allowed, and exploring the OSN graph topology is expensive. To address these challenges, we present asymptotically unbiased sampling methods to characterize user pair properties based on UVS and RW techniques respectively. We carry out an evaluation of our methods to show their accuracy and efficiency. Finally, we apply our methods to two Chinese OSNs, Doudan and Xiami, and discover significant homophily is present in these two networks.
    Data Engineering (ICDE), 2013 IEEE 29th International Conference on; 01/2013
  • A.I. Bejan · R.J. Gibbens · Yeon-sup Lim · D. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: Mobile communication platforms of individual agents and ground vehicles, ships and aircrafts of both civil and military services often operate in highly dynamic conditions with constantly changing infrastructure and access to communication resources. Efficient techniques for rapid and yet stable communication of such fleets with their control centres and between cooperating vehicles within the fleet is a challenging but important area of study with the potential to facilitate the analysis and design of efficient and robust communication systems. Multipath extensions of data transmission protocols aim to take advantage of path diversity to achieve efficient and robust bandwidth allocation while maintaining stability. Such multipath resource pooling extensions of routing and congestion control intrinsically implement decentralisation with implicit resource sharing. In this paper, we build on the recent theoretical work on fluid model approximations of multipath TCP and study their application to the scenarios in which a convoy with two communication nodes (representing the convoy's head and tail) establishes channels with a set of radio/WiFi towers and a satellite relaying information to a remote destination; these channels have time-varying capacities which depend on the position and dynamics of the convoy. The paper studies the performance of a multipath TCP controller and demonstrates how path diversity can be implicitly utilised to spread flows across available paths. Furthermore, we study the patterns of sub-flows governed by dynamic control according to the motion of the convoy and investigate the trade-offs between resource utilisation and the speed of response by the sub-flows.
    Teletraffic Congress (ITC), 2013 25th International; 01/2013
  • Upendra Sharma · Prashant Shenoy · Donald F. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper we present a simple and effective approach for resource provisioning to achieve a percentile bound on the end to end response time of a multi-tier application. We, at first, model the multi-tier application as an open tandem network of M/G/1-PS queues and develop a method that produces a near optimal application configuration, i.e, number of servers at each tier, to meet the percentile bound in a homogeneous server environment -- using a single type of server. We then extend our solution to a K-server case and our technique demonstrates a good accuracy, independent of the variability of service-times. Our approach demonstrates a provisioning error of no more than 3% compared to a 140% worst case provisioning error obtained by techniques based on an M/M/1-FCFS queue model. In addition, we extend our approach to handle a heterogenous server environment, i.e., with multiple types of servers. We find that fewer high-capacity servers are preferable for high percentile provisioning. Finally, we extend our approach to account for the rental cost of each server-type and compute a cost efficient application configuration with savings of over 80%. We demonstrate the applicability of our approach in a real world system by employing it to provision the two tiers of the java implementation of TPC-W -- a multi-tier transactional web benchmark that represents an e-commerce web application, i.e. an online bookstore.
    Proceedings of the 9th international conference on Autonomic computing; 09/2012
  • Boulat A. Bash · Dennis Goeckel · Don Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: We present a square root limit on low probability of detection (LPD) communication over additive white Gaussian noise (AWGN) channels. Specifically, if a warden has an AWGN channel to the transmitter with non-zero noise power, we prove that o(√n) bits can be sent from the transmitter to the receiver in n AWGN channel uses with probability of detection by the warden less than e for any ϵ >; 0, and, if a lower bound on the noise power on the warden's channel is known, then O(√n) bits can be covertly sent in n channel uses. Conversely, trying to transmit more than O(√n) bits either results in detection by the warden with probability one or a non-zero probability of decoding error as n → ∞. Further, we show that LPD communication on the AWGN channel allows one to send a nonzero symbol on every channel use, in contrast to what might be expected from the square root law found recently in image-based steganography.
    Information Theory Proceedings (ISIT), 2012 IEEE International Symposium on; 07/2012
  • Ping Yi · Ting Zhu · Bo Jiang · Bing Wang · D. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: Vehicle-to-grid provides a viable approach that feeds the battery energy stored in electric vehicles (EVs) back to the power grid. Meanwhile, since EVs are mobile, the energy in EVs can be easily transported from one place to another. Based on these two observations, we introduce a novel concept called EV energy network for energy transmission and distribution using EVs. We present a concrete example to illustrate the usage of an EV energy network, and then study the optimization problem of how to deploy energy routers in an EV energy network. We prove that the problem is NP-hard and develop a greedy heuristic solution. Simulations using real-world data shows that our method is efficient.
    Communications (ICC), 2012 IEEE International Conference on; 01/2012
  • B. Ribeiro · D. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: Estimating characteristics of large graphs via sampling is vital in the study of complex networks. In this work, we study the Mean Squared Error (MSE) associated with different sampling methods for the degree distribution. These sampling methods include independent random vertex (RV) and random edge (RE) sampling, and crawling methods such as random walks (RWs) and the widely used Metropolis-Hastings algorithm for uniformly sampling vertices (MHRWu). We see that the RW MSE is upper bounded by a quantity that is proportional to the RE MSE and inversely proportional to the spectral gap of the RW transition probability matrix. We also determine conditions under which RW is preferable to RV. Finally, we present an approximation of the MHRWu MSE. We evaluate the accuracy of our approximations and bounds through simulations on large real world graphs.
    Decision and Control (CDC), 2012 IEEE 51st Annual Conference on; 01/2012
  • Bo Jiang · R. Brockett · Weibo Gong · D. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper we present simple stochastic differential equations that lead to lower-tail and/or upper tail power law behaviors. We also present a model with bi-directional Poisson counters that exhibits power law behavior near a critical point, which might be of interest to statistical physics.
    Decision and Control (CDC), 2012 IEEE 51st Annual Conference on; 01/2012
  • Source
    Yeon-sup Lim · D.S. Menasche · B. Ribeiro · D. Towsley · P. Basu ·
    [Show abstract] [Hide abstract]
    ABSTRACT: A well known way to find the most central nodes in a network consists of coupling random walk sampling (or one of its variants) with a method to identify the most central nodes in the subgraph induced by the samples. Although it is commonly assumed that degree information is collected during the sampling step, in previous works this information has not been used at the identification step [10], [18]. In this paper, we showed that using degree information at the identification step in a very naive way, namely setting the degree as an alias to other centrality metrics, yields promising results.
    Network Science Workshop (NSW), 2011 IEEE; 07/2011
  • Source
    Boulat A. Bash · Dennis Goeckel · Don Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: Low power ad hoc wireless networks operate in conditions where channels are subject to fading. Cooperative diversity mitigates fading in these networks by establishing virtual antenna arrays through clustering the nodes. A cluster in a cooperative diversity network is a collection of nodes that cooperatively transmits a single packet. There are two types of clustering schemes: static and dynamic. In static clustering all nodes start and stop transmission simultaneously, and nodes do not join or leave the cluster while the packet is being transmitted. Dynamic clustering allows a node to join an ongoing cooperative transmission of a packet as soon as the packet is received. In this paper we take a broad view of the cooperative network by examining packet flows, while still faithfully implementing the physical layer at the bit level. We evaluate both clustering schemes using simulations on large multi-flow networks. We demonstrate that dynamically-clustered cooperative networks substantially outperform both statically-clustered cooperative networks and classical point-to-point networks.
    INFOCOM, 2011 Proceedings IEEE; 05/2011
  • Source
    Ting Zhu · D. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: Wireless communication consumes a significant amount of energy in wireless networks. Most multi-hop wireless routing protocols are adopted from wired network routing protocols and use pre-selected single or multiple minimum cost paths to forward data packets. However unlike wired networks, a wireless network has unique characteristics such as unreliable wireless links and multiple receivers associated with a single packet transmission. These particular characteristics of wireless network make it energy inefficient by using preselected paths to forward data packets. In this paper we propose a scalable, opportunistic, and energy efficient routing protocol (E<sup>2</sup>R) that uses an opportunistic forwarding scheme to deliver control messages and data packets in a multi-hop wireless network. To further reduce the overhead of control message, we introduce a novel greedy forwarding algorithm and an efficient self-suppression scheme. Extensive analysis and simulations show that E<sup>2</sup>R can provide high packet delivery ratio and effectively reduce redundant packet transmissions which results in less energy consumption.
    Computer Communications Workshops (INFOCOM WKSHPS), 2011 IEEE Conference on; 05/2011
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Follower networks such as Twitter and Digg are becoming popular form of social information networks. This paper seeks to gain insights into how they evolve and the relationship between their structure and their ability to spread information. By studying the Douban follower network, which is a popular online social network in China, we provide some evidences showing its suitability for information spreading. For example, it exhibits an unbalanced bow-tie structure with a large out-component, which indicates that the majority of users can spread information widely; the effective diameter of the strongly connected component is shrinking as the user base grows, which facilitates spreading; and the transitivity property shows that people in a follower network tend to shorten the path of information flow, i.e., it takes fewer hops to spread information. Also, we observe the following users' behaviors, a user's following activity decays exponentially during her lifetime and the following behaviors differ according to the age of the account. These findings provide a deep understanding on the evolution of follower networks, and can provide guidelines on how to build an efficient information diffusion system.
    Computer Communications Workshops (INFOCOM WKSHPS), 2011 IEEE Conference on; 05/2011
  • C. Capar · D. Goeckel · Benyuan Liu · D. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: Consider a one-dimensional wireless network with n nodes uniformly and independently distributed at random in the interval. In addition, m eavesdropper nodes are uniformly and independently distributed in. For a randomly selected source-destination pair, we consider the problem of securely delivering a message from the source to the destination and we present achievable results on the number of eavesdropper nodes that can be tolerated by the network. Our constructions make use of cooperative jamming, in which nodes located close to the eavesdroppers generate artificial noise. For the one-dimensional network case, our results provide an improvement to the connectivity properties of the recently-introduced secrecy graph which is disconnected for any positive number of eavesdroppers without cooperative jamming. We consider cases of both known and unknown eavesdropper locations. For known eavesdropper locations, we show that a message can be securely delivered from the source to the destination with probability one as the number of nodes n goes to infinity, for any number of independent eavesdroppers m(n) satisfying m(n) = o(√n / log n). For unknown eavesdropper locations, we present a construction which can tolerate m(n) = o(n/log n) under the assumption of independent eavesdroppers, but which is fragile in the face of collaborating eavesdroppers.
    Information Sciences and Systems (CISS), 2011 45th Annual Conference on; 04/2011
  • Source
    Cagatay Capar · Dennis Goeckel · Donald F. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: The capability of nodes to broadcast their message to the entire wireless network when nodes employ cooperation is considered. We employ an asymptotic analysis using an extended random network setting and show that the broadcast performance strongly depends on the path loss exponent of the medium. In particular, as the size of the random network grows, the probability of broadcast in a one-dimensional network goes to zero for path loss exponents larger than one, and goes to a nonzero value for path loss exponents less than one. In two-dimensional networks, the same behavior is observed for path loss exponents above and below two, respectively.
    IEEE Transactions on Information Theory 04/2011; abs/1104.3209(9). DOI:10.1109/TIT.2013.2252418 · 2.33 Impact Factor
  • Source
    Yan Cai · Patrick P. C. Lee · Weibo Gong · Donald F. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: Traffic burstiness is known to be undesirable for a router as it increases the router’s queue length and hence the queueing delays of data flows. This poses a security problem in which an attacker intentionally introduces traffic burstiness into routers. We consider a correlation attack, whose fundamental characteristic is to correlate multiple attack flows to generate synchronized small attack bursts, in an attempt to aggregate the bursts into a large burst at a target router.In this paper, we develop an analytical, fluid-based framework that models how the correlation attack disrupts router queues and how it can be mitigated. Using Poisson Counter Stochastic Differential Equations (PCSDEs), our framework captures the dynamics of a router queue for special cases and gives the closed-form average router queue length as a function of the inter-flow correlation. To mitigate the correlation attack, we apply our analytical framework to model different pacing schemes including Markov ON–OFF pacing and rate limiting, which are respectively designed to break down the inter-flow correlation and suppress the peak rates of bursts. We verify that our fluid models conform to packet-level ns2 simulation results.
    Computer Networks 02/2011; 55(3):734-747. DOI:10.1016/j.comnet.2010.10.016 · 1.26 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The workloads in modern Chip-multiprocessors (CMP) are becoming increasingly diversified, creating different resource demands on hardware substrate. It is necessary to allocate hardware resources based on the needs of the workloads in order to improve ...
    SIGMETRICS 2011, Proceedings of the 2011 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems, San Jose, CA, USA, 07-11 June 2011 (Co-located with FCRC 2011); 01/2011
  • Source
    Peter B. Key · Laurent Massoulié · Donald F. Towsley ·
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we investigate the benefits that accrue from the use of multiple paths by a session coupled with rate control over those paths. In particular, we study data transfers under two classes of multipath control, coordinated control where the rates over the paths are determined as a function of all paths, and uncoordinated control where the rates are determined independently over each path. We show that coordinated control exhibits desirable load balancing properties; for a homogeneous static random paths scenario, we show that the worst-case throughput performance of uncoordinated control behaves as if each user has but a single path (scaling like log(log(N) )/ log(N) where N is the system size, measured in number of resources), whereas coordinated control yields a worstcase throughput allocation bounded away from zero. We then allow users to change their set of paths and introduce the notion of a Nash equilibrium. We show that both coordinated and uncoordinated control lead to Nash equilibria corresponding to desirable welfare maximizing states, provided in the latter case, the rate controllers over each path do not exhibit any round-trip time (RTT) bias (unlike TCP Reno). Finally, we show in the case of coordinated control that more paths are better, leading to greater welfare states and throughput capacity, and that simple path reselection polices that shift to paths with higher net benefit can achieve these states.
    Communications of the ACM 01/2011; 54(1):109-116. DOI:10.1145/1866739.1866762 · 3.62 Impact Factor

Publication Stats

25k Citations
205.73 Total Impact Points


  • 1970-2013
    • University of Massachusetts Amherst
      • • School of Computer Science
      • • Department of Electrical and Computer Engineering
      Amherst Center, Massachusetts, United States
  • 2010
    • Pacific Northwest National Laboratory
      Richland, Washington, United States
  • 2002-2010
    • Microbiology Department at UMass Amherst
      Amherst Center, Massachusetts, United States
    • Northrop Grumman
      Falls Church, Virginia, United States
    • Columbia University
      • Department of Electrical Engineering
      New York City, New York, United States
  • 2008
    • University of Massachusetts Lowell
      • Department of Computer Science
      Lowell, Massachusetts, United States
  • 2006
    • Università degli Studi di Palermo
      Palermo, Sicily, Italy
  • 2003
    • University of Florida
      Gainesville, Florida, United States
  • 1999-2002
    • AT&T Labs
      • Research
      Austin, Texas, United States
  • 1994
    • National Chung Cheng University
      • Department of Computer Science and Information Engineering
      Chia-i-hsien, Taiwan, Taiwan
  • 1992
    • Tilburg University
      Tilburg, North Brabant, Netherlands
  • 1990
    • The University of Western Ontario
      London, Ontario, Canada
  • 1989
    • Yale University
      • Department of Computer Science
      New Haven, CT, United States