Books by Rafit Izhak Ratzin
Papers by Rafit Izhak Ratzin
International Journal of Distributed Sensor Networks, 2011
We study game-theoretic mechanisms for routing in wireless ad hoc networks. Our major results inc... more We study game-theoretic mechanisms for routing in wireless ad hoc networks. Our major results include a combination of theoretical bounds and extensive simulations, showing that VCG-based routing in wireless ad-hoc networks exhibits small frugality ratio with high probability. Game-theoretic mechanisms capture the noncooperative and selfish behavior of nodes in a resourceconstrained environment. There have been some recent proposals to use these mechanisms (in particular VCG) for routing in wireless ad-hoc networks, and some frugality bounds are known when the connectivity graph is essentially complete. We are the first to show frugality bounds for random geometric graphs, a well-known model for ad-hoc wireless connectivity. In addition, we generalize the model of agent behavior by allowing sets of nodes to form communities to maximize total profit. We are the first to analyze the performance of VCG under such a community model. While some recent truthful protocols for the traditional (individual) agent model have improved upon the frugality of VCG by selecting paths to minimize not only the cost but the overpayment, we show that extending such protocols to the community model requires solving NP-complete problems which are provably hard to approximate.
Lecture Notes in Computer Science
We study game-theoretic mechanisms for routing in ad-hoc networks. Game-theoretic mechanisms capt... more We study game-theoretic mechanisms for routing in ad-hoc networks. Game-theoretic mechanisms capture the non-cooperative and selfish behavior of nodes in a resource-constrained environment. There have been some recent proposals to use incentive-based mechanisms (in particular, VCG) for routing in wireless ad-hoc networks, and some frugality bounds are known when the connectivity graph is essentially complete. We show frugality bounds for random geometric graphs, a wellknown model for ad-hoc wireless connectivity. Our main result demonstrates that VCG-based routing in ad-hoc networks exhibits small frugality ratio (i.e., overpayment) with high probability. In addition, we study a more realistic generalization where sets of agents can form communities to maximize total profit. We also analyze the performance of VCG under such a community model and show similar bounds. While some recent truthful protocols for the traditional (individual) agent model have improved upon the frugality of VCG by selecting paths to minimize not only the cost but the overpayment, we show that extending such protocols to the community model requires solving NP-complete problems which are provably hard to approximate.
Lecture Notes in Computer Science, 2009
Recent research efforts have shown that the popular Bit-Torrent protocol does not strictly enforc... more Recent research efforts have shown that the popular Bit-Torrent protocol does not strictly enforce fairness and allows free-riding, mainly via optimistic unchokes. This paper proposes a BitTorrent-like protocol, that encourages peers of similar upload bandwidth to be buddiespeers collaborating for mutual benefit. Buddy peers mostly satisfy their download needs through their buddies and perform optimistic unchokes only when absolutely necessary. As a result, the buddy protocol improves fairness via explicit cooperation between buddies, and limits bandwidth spent on random optimistic unchokes, leading to a system more robust against free-riders. We implemented the buddy protocol on top of an existing BitTorrent implementation and ran experiments on a controlled PlanetLab testbed to evaluate its impact. Our results show that the buddy protocol promotes fairness, discourages free-riding, and improves the robustness of the system as compared to regular BitTorrent. It also provides incentives to be adopted by all the peers in the system.
IEEE Transactions on Parallel and Distributed Systems, 2012
We propose a BitTorrent-like protocol based on an online learning (reinforcement learning) mechan... more We propose a BitTorrent-like protocol based on an online learning (reinforcement learning) mechanism, which can replace the peer selection mechanisms in the regular BitTorrent protocol. We model the peers' interactions in the BitTorrent-like network as a repeated stochastic game, where the strategic behaviors of the peers are explicitly considered. A peer that applies the reinforcement learning (RL)-based mechanism uses the observations on the associated peers' statistical reciprocal behaviors to determine its best responses and estimate the corresponding impact on its expected utility. The policy determines the peer's resource reciprocations such that the peer can maximize its long-term performance. We have implemented the proposed mechanism and incorporated it into an existing BitTorrent client. Our experiments performed on a controlled Planetlab testbed confirm that the proposed protocol 1) promotes fairness and provides incentives to contributed resources, i.e., high capacity peers improve their download completion time by up to 33 percent, 2) improves the system stability and robustness, i.e., reduces the peer selection fluctuations by 57 percent, and (3) discourages free-riding, i.e., peers reduce their uploads to free-riders by 64 percent as compared to the regular BitTorrent protocol.
Abstract—Although the popular BitTorrent protocol strives to limit free-riding via its tit-for-ta... more Abstract—Although the popular BitTorrent protocol strives to limit free-riding via its tit-for-tat incentives, recent research efforts have shown that it does not strictly enforce fairness. Freeriding opportunities indeed exist, mainly via optimistic unchokes, a BitTorrent mechanism that facilitates the continuous discovery of better peers to interact with. Our results in this work also show that increasing numbers of free-riders can considerably hurt the performance of compliant peers. In an effort to address this problem, this paper proposes a BitTorrent-like protocol that dynamically organizes peers of similar upload bandwidth in teams — groups of peers collaborating for mutual benefit. Team members mostly satisfy their data download needs inside their team and only perform optimistic unchokes when absolutely necessary. We show that, as a result, the team protocol improves peer performance via explicit cooperation within teams. At the same time, it limits bandwidth spent on optim...
A number of previous measurement studies [10, 12, 17] have shown the existence of path exploratio... more A number of previous measurement studies [10, 12, 17] have shown the existence of path exploration and slow convergence in the global Internet routing system, and a number of protocol enhancements have been proposed to remedy the problem [21, 15, 4, 20, 5]. However all the previous measurements were conducted over a small number of testing prefixes. There has been no systematic study to quantify the pervasiveness of BGP slow convergence in the operational Internet, nor there is any known effort to deploy any of the proposed solutions. In this paper we present our measurement results from identifying BGP slow convergence events across the entire global routing table. Our data shows that the severity of path exploration and slow convergence varies depending on where prefixes are originated and where the observations are made in the Internet routing hierarchy. In general, routers in tier-1 ISPs observe less path exploration, hence shorter convergence delays than routers in edge ASes, a...
In this paper, we conduct a systematic study on the pervasiveness and persistency of one specific... more In this paper, we conduct a systematic study on the pervasiveness and persistency of one specific phenomenon in the global routing system: a small set of highly active prefixes account for a large number of routing updates. Our data analysis shows that this phenomenon is commonly observed across all the monitored routers that belong to different ISPs, and is persistent over our 3-year study period. Our data further shows that the majority of these high active prefixes are transient while a small number of them are persistent over time. The cause of these highly active prefixes include topological failures, BGP slow convergence, protocol defects, and failure of turning on protocol protection mechanisms.
Peer-to-peer (P2P) content sharing protocols dominate the traffic on the Internet [IPO09], and th... more Peer-to-peer (P2P) content sharing protocols dominate the traffic on the Internet [IPO09], and thus have become an important piece in building scalable Internet application. Peers in P2P network are typically independent entities that together form a self-organizing, self-maintaining network with no central authority. As a result, P2P network performance is highly dependent on the amount of voluntary resources individual peers contribute to the system. However, peers in P2P systems have self-interest to control their degree of collaboration and contribution [SP03, FS02, Pap01], as cooperation may incur significant communication and computation costs. Thus, rational peers may refuse to contribute their fair share of resources [AH00, SPD02, HCW05]. In some scenarios this may lead to the “tragedy of commons” [G68], when maximizing peers' own utilities may effectively decrease the overall utility of the system. Hence, mechanisms that incentivize peers to actively cooperate and contr...
Recent research efforts have shown that the popular BitTorrent protocol does not provide fair res... more Recent research efforts have shown that the popular BitTorrent protocol does not provide fair resource reciprocation and may allow free-riding. In this paper, we propose a BitTorrentlike protocol that replaces the peer selection mechanisms in the regular BitTorrent protocol with a novel reinforcement learning based mechanism. Due to the inherent opration of P2P systems, which involves repeated interactions among peers over a long period of time, the peers can efficiently identify free-riders as well as desirable collaborators by learning the behavior of their associated peers. Thus, it can help peers improve their download rates and discourage free-riding, while improving fairness in the system. We model the peers' interactions in the BitTorrent-like network as a repeated interaction game, where we explicitly consider the strategic behavior of the peers. A peer, which applies the reinforcement learning based mechanism, uses a partial history of the observations on associated peers' statistical reciprocal behaviors to determine its best responses and estimate the corresponding impact on its expected utility. The policy determines the peer's resource reciprocations with other peers, which would maximize the peer's long-term performance, thereby making foresighted decisions. We have implemented the proposed reinforcement-learning based mechanism and incorporated it into an existing BitTorrent client. We have performed extensive experiments on a controlled Planetlab test bed. Our results confirm that our proposed protocol (1) promotes fairness in terms of incentives to each peer's contribution e.g. high capacity peers improve their download completion time by up to 33%, (2) improves the system stability and robustness e.g. reducing the peer selection fluctuations by 57%, and (3) discourages free-riding e.g. peers reduce by 64% their upload to free-rider, in comparison to the regular BitTorrent protocol.
Recent research efforts have shown that the popular Bit-Torrent protocol does not provide fair re... more Recent research efforts have shown that the popular Bit-Torrent protocol does not provide fair resource reciprocation and allows free-riding. In this paper, we propose a novelforesighted resource reciprocation mechanism that replaces the peer selection mechanism with a reinforcement learning mechani sm that adopts a foresighted resource reciprocation policy. We model the peer interactions in the BitTorrent-like network as a stochastic-game, where we explicitly consider the strategic behavior peers. The peers can observe partial historic info rmation of associated peers’ statistical reciprocal behavior s, through which the peers can estimate the impact on their expected uti lity and then adopt their best response. The policy determines th e peer’s optimal resource reciprocations, and enables the pe er to maximize the long-term performance. The mechanism improve s fairness as it relies on long-term history. Moreover, it hurts the free-riders directly since foresighted peers are discou...
Abstract—In this paper, we propose a BitTorrent-like protocol
that replaces the peer selection me... more Abstract—In this paper, we propose a BitTorrent-like protocol
that replaces the peer selection mechanisms in the regular
BitTorrent protocol with a novel reinforcement learning based
mechanism. The inherent operation of P2P systems, which
involves repeated interactions among peers over a long time
period, allows peers to efficiently identify free-riders as well
as desirable collaborators by learning the behavior of their
associated peers. Thus, it can help peers improve their download
rates and discourage free-riding (FR), while improving fairness.
We model the peers’ interactions in the BitTorrent-like network
as a repeated interaction game, where we explicitly consider the
strategic behavior of the peers. A peer that applies the reinforcement
learning based mechanism uses a partial history of the
observations on associated peers’ statistical reciprocal behaviors
to determine its best responses and estimate the corresponding
impact on its expected utility. The policy determines the peer’s
resource reciprocations with other peers, which would maximize
the peer’s long-term performance.
In this chapter, we discuss P2P systems that have been deployed in file sharing and real-time med... more In this chapter, we discuss P2P systems that have been deployed in file sharing and real-time media streaming. We discuss the limitations of the implementations for existing P2P-based file sharing and media streaming applications in detail. More advanced resource reciprocation strategies, where peers make foresighted decisions on their resource distribution in a way that maximizes their cumulative utilities are discussed.
Recent research efforts have shown that the popular Bit-Torrent protocol does not strictly enforc... more Recent research efforts have shown that the popular Bit-Torrent protocol does not strictly enforce fairness and allows free-riding, mainly via optimistic unchokes. This paper proposes a BitTorrent-like protocol, that encourages peers of similar upload bandwidth to be buddies-peers collaborating for mutual benefit. Buddy peers mostly satisfy their download needs through their buddies and perform optimistic unchokes only when absolutely necessary. As a result, the buddy protocol improves fairness via explicit cooperation between buddies, and limits bandwidth spent on random optimistic unchokes, leading to a system more robust against free-riders. We implemented the buddy protocol on top of an existing BitTorrent implementation and ran experiments on a controlled PlanetLab testbed to evaluate its impact. Our results show that the buddy protocol promotes fairness, discourages free-riding, and improves the robustness of the system as compared to regular BitTorrent. It also provides incentives to be adopted by all the peers in the system.
We study game-theoretic mechanisms for routing in ad-hoc networks. Game-theoretic mechanisms capt... more We study game-theoretic mechanisms for routing in ad-hoc networks. Game-theoretic mechanisms capture the non-cooperative and selfish behavior of nodes in a resource-constrained environment. There have been some recent proposals to use incentive-based mechanisms (in particular, VCG) for routing in wireless ad-hoc networks, and some frugality bounds are known when the connectivity graph is essentially complete. We show frugality bounds for random geometric graphs, a wellknown model for ad-hoc wireless connectivity. Our main result demonstrates that VCG-based routing in ad-hoc networks exhibits small frugality ratio (i.e., overpayment) with high probability. In addition, we study a more realistic generalization where sets of agents can form communities to maximize total profit. We also analyze the performance of VCG under such a community model and show similar bounds. While some recent truthful protocols for the traditional (individual) agent model have improved upon the frugality of VCG by selecting paths to minimize not only the cost but the overpayment, we show that extending such protocols to the community model requires solving NP-complete problems which are provably hard to approximate.
A number of previous measurement studies have shown the existence of path exploration and slow co... more A number of previous measurement studies have shown the existence of path exploration and slow convergence in the global Internet routing system, and a number of protocol enhancements have been proposed to remedy the problem . However all the previous measurements were conducted over a small number of testing prefixes. There has been no systematic study to quantify the pervasiveness of BGP slow convergence in the operational Internet, nor there is any known effort to deploy any of the proposed solutions.
Uploads
Books by Rafit Izhak Ratzin
Papers by Rafit Izhak Ratzin
that replaces the peer selection mechanisms in the regular
BitTorrent protocol with a novel reinforcement learning based
mechanism. The inherent operation of P2P systems, which
involves repeated interactions among peers over a long time
period, allows peers to efficiently identify free-riders as well
as desirable collaborators by learning the behavior of their
associated peers. Thus, it can help peers improve their download
rates and discourage free-riding (FR), while improving fairness.
We model the peers’ interactions in the BitTorrent-like network
as a repeated interaction game, where we explicitly consider the
strategic behavior of the peers. A peer that applies the reinforcement
learning based mechanism uses a partial history of the
observations on associated peers’ statistical reciprocal behaviors
to determine its best responses and estimate the corresponding
impact on its expected utility. The policy determines the peer’s
resource reciprocations with other peers, which would maximize
the peer’s long-term performance.
that replaces the peer selection mechanisms in the regular
BitTorrent protocol with a novel reinforcement learning based
mechanism. The inherent operation of P2P systems, which
involves repeated interactions among peers over a long time
period, allows peers to efficiently identify free-riders as well
as desirable collaborators by learning the behavior of their
associated peers. Thus, it can help peers improve their download
rates and discourage free-riding (FR), while improving fairness.
We model the peers’ interactions in the BitTorrent-like network
as a repeated interaction game, where we explicitly consider the
strategic behavior of the peers. A peer that applies the reinforcement
learning based mechanism uses a partial history of the
observations on associated peers’ statistical reciprocal behaviors
to determine its best responses and estimate the corresponding
impact on its expected utility. The policy determines the peer’s
resource reciprocations with other peers, which would maximize
the peer’s long-term performance.
that replaces the peer selection mechanisms in the regular
BitTorrent protocol with a novel reinforcement learning based
mechanism. The inherent operation of P2P systems, which
involves repeated interactions among peers over a long time
period, allows peers to efficiently identify free-riders as well
as desirable collaborators by learning the behavior of their
associated peers. Thus, it can help peers improve their download rates and discourage free-riding (FR), while improving fairness.
We model the peers’ interactions in the BitTorrent-like network
as a repeated interaction game, where we explicitly consider the strategic behavior of the peers. A peer that applies the reinforcement learning based mechanism uses a partial history of the observations on associated peers’ statistical reciprocal behaviors to determine its best responses and estimate the corresponding impact on its expected utility. The policy determines the peer’s resource reciprocations with other peers, which would maximize the peer’s long-term performance.