T. Roughgarden. Designing networks for selfish users is hard. In Proc. 42nd IEEE FOCS, pp. 472--481, 2001.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....further motivates the need for equilibrium aware protocols. B. Related Work Game theory [18] is a very mature topic. The current challenges of game theory applied to computer networks are summarized by Papadimitrou [21] Several papers (for example [6] 22] 15] 16] 3] 27] 1] 25] [26], 17] 12] 5] 17] have applied tools from microeconomics and game theory to computer networks over the last fifteen years. A thorough literature survey is beyond the scope of the paper. We now consider some of the most directly relevant related work, and compare them with our approach. In ....

T. Roughgarden, "Designing networks for selfish users is hard," in IEEE Symposium on Foundations of Computer Science, 2001, pp. 472--481.

....further motivates the need for equilibrium aware protocols. B. Related Work Game theory [18] is a very mature topic. The current challenges of game theory applied to computer networks are summarized by Papadimitrou [21] Several papers (for example [6] 22] 16] 15] 3] 27] 1] 26] [25], 17] 12] 5] 17] have applied tools from microeconomics and game theory to computer networks over the last fifteen years. A thorough literature survey is beyond the scope of the paper. We now consider some of the most directly relevant related work, and compare them with our approach. In ....

T. Roughgarden, "Designing networks for selfish users is hard," in IEEE Symposium on Foundations of Computer Science, 2001, pp. 472--481.

....further motivates the need for equilibrium aware protocols. B. Related Work Game theory [18] is a very mature topic. The current challenges of game theory applied to computer networks are summarized by Papadimitrou [21] Several papers (for example [6] 22] 15] 16] 3] 27] 1] 25] [26], 17] 12] 5] 17] have applied tools from microeconomics and game theory to computer networks over the last fifteen years. A thorough literature survey is beyond the scope of the paper. We now consider some of the most directly relevant related work, and compare them with our approach. In ....

T. Roughgarden, "Designing networks for selfish users is hard," in IEEE Symposium on Foundations of Computer Science, 2001, pp. 472--481.

.... there are open questions regarding the complexity of executing various mechanism optimally (e.g. 4, 16, 42, 43] or approximately (e.g. 25, 33] the complexity of manipulating various mechanisms (e.g. 2, 3, 9, 44] the complexity of designing mechanisms (that lead to desirable outcomes) [10, 40], and the complexity of deciding what information to elicit from the players in various mechanisms [11] Another avenue involves studying more sophisticated equilibrium notions which take into account that players have limited memory (e.g. 1, 14, 32, 39, 41] or limited capability to solve ....

Tim Roughgarden. Designing networks for selfish users is hard. In FOCS, 2001.

.... of earlier work is our paper which studied the complexity of multiagent mechanism design without side payments and only for two specific notions of nonmanipulability [3] Roughgarden has studied the complexity of designing a good network topology for agents that selfishly choose the links they use [14]. This is related to mechanism design, but differs significantly in that the designer only has restricted control over the rules of the game because there is no party that can impose the outcome (or side payments) Also, there is no explicit reporting of preferences. 9 Conclusions and future ....

Tim Roughgarden. Designing networks for selfish users is hard. In FOCS, 2001.

....further motivates the need for equilibrium aware protocols. B. Related Work Game theory [8] is a very mature topic. The current challenges of game theory applied to computer networks are summarized by Papadimitrou [9] Several papers (for example [10] 11] 12] 13] 14] 15] 16] 17] [18], 19] 20] 21] 19] have looked at applying tools from microeconomic theory and game theory to computer networks over the last fifteen years. A thorough literature survey is beyond the scope of the paper. We now consider some of the most directly relevant related work and compare them with ....

Tim Roughgarden, "Designing networks for selfish users is hard," in IEEE Symposium on Foundations of Computer Science, 2001, pp. 472--481.

....to a server promising minimum cost, taking into account the behavior of other units of flow. Assuming infinitesimal small units of flow, we come to the following fractional variant of the integral assignment model. This fractional model has been frequently considered in the literature, see, e.g. [8, 24, 25, 26, 27]) The fractional model does not distinguish between mixed and pure strategies. There are several equivalent ways to define a Nash equilibrium in this model. We use the characterization of Wardrop [29] see also [24] A fractional assignment is in Nash equilibrium if x j i 0 implies C j # Cq ....

T. Roughgarden. Designing networks for selfish users is hard. In Proc. 42nd IEEE FOCS, pp. 472--481, 2001.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In FOCS '01, pages 472--481.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001. Full version available from http://www.cs.cornell.edu/timr.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001.

....independently of each other, although in Chapter 5 we assume some of the results (but not the proof techniques) of Chapter 3. Finally, Chapter 4 is meant to be read following Chapter 3. 1. 6 Bibliographic Notes Most of the work reported in this thesis has appeared previously in research papers [159, 160, 161, 162, 163, 164, 165]. Chapter 4 and portions of Chapter 3 and Appendix A are joint work with Eva Tardos and appeared in [164, 165] the rest of Chapter 3 is drawn from [160, 163] The results of Chapter 5 appeared in [159] the results of Chapter 6 in [161] and Theorem A.3.1 of Section A.3 in [162] Preliminaries ....

.... in this thesis has appeared previously in research papers [159, 160, 161, 162, 163, 164, 165] Chapter 4 and portions of Chapter 3 and Appendix A are joint work with Eva Tardos and appeared in [164, 165] the rest of Chapter 3 is drawn from [160, 163] The results of Chapter 5 appeared in [159], the results of Chapter 6 in [161] and Theorem A.3.1 of Section A.3 in [162] Preliminaries In this chapter we present the basic definitions and preliminary technical results needed in the rest of this work. In Section 2.1 we formally define the tra#c model discussed in Chapter 1. In Section ....

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001.

....Nash equilibrium with minimum possible user disutility (accounting for disutility due to both latency and taxes) for the case of homogeneous tra#c. In [9] we showed, among other results, that this optimization problem essentially reduces to the network design problem of avoiding Braess s Paradox [25] and is hard to approximate to within any reasonable factor. Organization Section 2 formally introduces our model and states some preliminaries. In Section 3 we show that edge taxes always su#ce to induce a minimum latency routing of tra#c, even when network users are heterogeneous. In Section 4 ....

....undersaturated edges can be scattered throughout the network in an arbitrary way; for this reason, making this intuitive argument precise takes some work. The proof of Lemma 3. 3 hinges on defining an appropriate ordering on the vertices of G; this idea was also used, for different purposes, in [25]. To describe the properties we desire of this ordering, let f be a Nash flow for the well behaved instance (G , #, #) and, for a vertex v, let d(v) denote the length of a shortest s v path, using # as edge lengths. The proof approach (see below) motivates an ordering satisfying: P1) if f ....

[Article contains additional citation context not shown here]

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001.

.... here are provably worse in general graphs than in networks of parallel links [23, 24] that Braess s Paradox the counterintuitive phenomenon that removing arcs from a network may decrease the cost of selfish routing grows increasingly severe as the underlying network becomes more complex [22], and that worst case examples for selfish routing with respect to the maximum (rather than the total) latency grow in severity with the network size [21] Organization After reviewing some technical preliminaries in Section 2, in Section 3 we give an upper bound on the price of anarchy with ....

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001.

....famous Braess s Paradox shows this intuition to be incorrect. We consider strategies for pricing network edges to reduce the cost of a Nash equilibrium. Since levying a sufficiently large tax on an edge effectively removes it from the network, our study generalizes previous work on network design [22]. In this paper, we prove the following results. In a large class of networks including all networks with linear latency functions marginal cost taxes do not improve the cost of the Nash equilibrium. # Supported in part by NSF grant CCR0105678. Supported in part by an NSF CAREER ....

....that of edge removal. In every network with linear latency functions, taxes cannot improve over removing edges. There are networks with nonlinear latency functions, however, in which taxes are radically more powerful than edge removal. Strong inapproximability results known for network design [22] carry over to the problem of computing optimal taxes, proving this problem computationally intractable. 1. INTRODUCTION Selfish Routing and Marginal Cost Pricing networks. We focus on a simple model of selfish routing, defined by Wardrop [28] and first studied in the theoretical computer ....

[Article contains additional citation context not shown here]

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001. Full version available from http://www.cs.cornell.edu/timr.

....taxes that induce a Nash equilibrium with minimum possible user disutility (accounting for disutility due to both latency and taxes) for the case of homogeneous traffic. In [8] we showed that this optimization problem essentially reduces to the network design problem of avoiding Braess s Paradox [23] and is hard to approximate to within any reasonable factor. Organization. Section 2 formally introduces our model and states some preliminaries. In Section 3 we show that edge taxes always suffice to induce a minimum latency routing of traffic, even when network users are heterogeneous. In ....

....undersaturated edges can be scattered throughout the network in an arbitrary way; for this reason, making this intuitive argument precise takes some work. The proof of Lemma 3. 3 hinges on defining an appropriate ordering on the vertices of G; this idea was also used, for different purposes, in [23]. To describe the properties we desire of this ordering, let f be a Nash flow for the well behaved instance (G , #, #) and, for a vertex v, let d(v) denote the length of a shortest s v path, using # as edge lengths. The proof approach of Lemma 3.3 (see Appendix A) motivates an ordering ....

[Article contains additional citation context not shown here]

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001.

....famous Braess s Paradox shows this intuition to be incorrect. We consider strategies for pricing network edges to reduce the cost of a Nash equilibrium. Since levying a su#ciently large tax on an edge e#ectively removes it from the network, our study generalizes previous work on network design [23]. In this paper, we prove the following results. # Department of Computer Science, New York University, 251 Mercer Street, New York, NY 10012. Supported by NSF grant CCR0105678. Email: cole cs.nyu.edu. Department of Computer Science, New York University, 251 Mercer Street, New York, NY 10012. ....

....that of edge removal. In every network with linear latency functions, taxes cannot improve over removing edges. There are networks with nonlinear latency functions, however, in which taxes are radically more powerful than edge removal. Strong inapproximability results known for network design [23] carry over to the problem of computing optimal taxes, proving this problem computationally intractable. Categories and Subject Descriptors F.2.0 [Analysis of Algorithms and Problem Complexity ] General General Terms Algorithms, Theory Keywords Selfish routing, network pricing, Nash ....

[Article contains additional citation context not shown here]

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001. Full version available from http://www.cs.cornell.edu/timr.

.... capacity to system resources, Cocchi et al. 7] investigate the control of selfish users through various pricing policies, Shenker [34] demonstrates that an appropriate (centralized) protocol at a network switch induces selfish users to exhibit good flow control behavior, and Roughgarden [30] studies the problem of designing networks that exhibit good performance when used selfishly. An area of research that typically does not study Nash equilibria but is nevertheless concerned with controlling selfish behavior is that of algorithmic mechanism design [1, 14, 23, 24, 29] In this ....

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd Annual Symposium on Foundations of Computer Science, pages 472--481, 2001.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proc. 42nd IEEE FOCS, pp. 472--481, 2001.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd IEEE Symposium on Foundations of Computer Science (FOCS), pages 472--481, Las Vegas, NV, October 14--17, 2001. IEEE Computer Society Press, Los Alamitos, CA.

No context found.

Tim Roughgarden. Designing networks for selfish users is hard. In FOCS, 2001.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the Annual Symposium on Foundations of Computer Science (FOCS), 2001.

No context found.

Roughgarden, T. (2001). Designing networks for selfish users is hard. In: Proceedings of the 42nd Annual IEEE Symposium on Foundation of Computer Science. pp. 472--481. (to appear in a special issue of Journal of Computer and System Sciences).

No context found.

Tim Roughgarden. Designing networks for selfish users is hard. In Proc. 42nd Ann. Symp. on Foundations of Computer Science, pages 472--481, 2001.

No context found.

T. Roughgarden, "Designing networks for selfish users is Hard," in IEEE Symposium on Foundations of Computer Science, 2001, pp. 472--481.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proc. of the 42nd IEEE Symposium on Foundations of Computer Science (FOCS), pages 472--481, 2001.

No context found.

T. Roughgarden, "Designing networks for selfish users is Hard," in IEEE Symposium on Foundations of Computer Science, 2001, pp. 472--481.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proc. 42nd IEEE Symp. on Found. of Comp. Science, pages 472--481, 2001.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the Annual Symposium on Foundations of Computer Science (FOCS), 2001.

No context found.

T. Roughgarden. Designing networks for selfish users is hard. In Proceedings of the 42nd IEEE Symposium on Foundations of Computer Science, pages 472--481, 2001.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC