UPSSS! JavaScript nie działa sprawdź ustawienia przeglądarki

Logowanie Rejestracja

Informacje o produkcie

Distributed network protocols for anonymous stations in cooperative and noncooperative settings


Distributed network protocols for anonymous stations in cooperative and noncooperative settings
  Cena:

Ilość

przechowalnia

28,00 zł

Dostępność: jest w magazynie sklepu
Dostępna ilość: 1
Najniższy koszt wysyłki to tylko 13,00 zł

Najedź aby zobaczyć wszystkie koszty wysyłki

Specyfikacja książki
Ilość stron
179
Okładka
miękka
Format
B5
Rok wydania
2006 - wyd. I
Język
angielski
  Cena:

Ilość

przechowalnia

28,00 zł

This work studies the effects of station anonymity and/or noncooperative (selfish) station behavior upon the design and performance of selected distributed network protocols. A set of stations interconnected by a computer network are assumed to be engaged in a collective activity for which some inter-station consistency is required. The considered protocols relate to group communication (where message delivery in agreed order is required), random multiple access over a single channel (where collision avoidance is required), and packet forwarding (where maintenance of end-to-end paths is required). Design of distributed protocols for anonymous stations permits to dispense with costly authentication mechanisms, and provides a "security lower bound" for systems in which station identities are relied upon. Modeling of noncooperative settings (where stations are allowed to behave selfishly) is justified in view of the increased autonomy and user configurability of today's network components; at the same time, as a departure from the traditional engineering paradigm, it opens up interesting areas of networking research.
In Chapter 2, two agreed order multicast protocols are proposed for anonymous stations in a cooperative setting. Both are LAN-oriented i.e., designed to yield high group throughput by exploiting the natural synchronism of single-broadcast transmission media. A faulty communication environment is assumed with occasional reception misses - failures by one or more stations to receive a message. The Distributed Precedence Graph (DPG) protocol is network-independent in that multicast and reception primitives clearly define a protocol-to-network interface, and missed message recovery is separated from message ordering. In contrast, the Jamming Protocol (JP) is network-dependent and exploits in a novel way the "ideal synchronism" of LAN broadcast to couple missed message recovery and message ordering. Both proposed protocols are insensitive to membership control mechanisms and do without the knowledge of station identities. Even so, they compare favorably with existing agreed order multicast protocols like LANSIS or TOTEM, as demonstrated by the analysis of logicaltime properties and simulation of stochastic performance.
Chapters 3 through 5 are devoted to game-theoretic analysis of random access MAC-layer protocols in a noncooperative setting. Chapter 3 deals with CSMA/CA, the multiple access mechanism of the IEEE 802.11 MAC protocol. It is argued that in an ad hoc wireless LAN, CSMA/CA is vulnerable to a selfish backoff attack consisting in systematic selection of short backoff times. Network performance under a backoff attack is evaluated via extension of Bianchi's Markovian approximation, as well as using more accurate models. A sufficient condition of solvability is given and a game-theoretic analysis is carried out with stations' bandwidth shares regarded as payoffs. Launching a backoff attack is established as a strictly dominating action, the payoff structure of the resulting one-shot CSMA/CA game being similar to a multiplayer Prisoners' Dilemma (with a unique and Pareto non-optmal Nash equilibrium). The notion of a Nash capacity is introduced to show that selfish station behavior reduces the achievable bandwidth utilization roughly by half compared with a cooperative setting.
Incentive-based discouragement of backoff attacks is proposed in Chapter 4 using the concept of a repeated CSMA/CA game, where a station is allowed to reconfigure its CSMA/CA mechanism based upon past payoffs. Assuming that most stations are interested in enforcing cooperative behavior of the other stations, a satisfactory equilibrium play is shown possible. In particular, a selfish station can be expected to incline to honest play for fear of the other stations inclining to selfish play. This idea is elaborated upon and gives rise to two proposed strategies of playing the repeated CSMA/CA game, called Cooperation via Randomized Inclination to Selfish Play (CRISP) and Selfish Play to Elicit Live-and-Let-live (SPELL). By exploiting another property of the one-shot CSMA/CA game, called coarse profile observability, CRISP and SPELL are shown to asymptotically achieve Pareto optimality and the related Nash equilibrium becomes subgame perfect. It is shown how to extend the proposed strategies if some stations disengage the CSMA/CA backoff scheme. Chapter 4 concludeswith a discussion of QoS-sensitive station behavior, in which case the CSMA/CA game may change from a Prisoners' Dilemma into a form of a queuing game.
Selfish behavior of CSMA/CA stations studied in Chapters 3 and 4 only affects specific protocol parameters, leaving the very principle of contention unchanged. In Chapter 5 we look at a slightly more general model of single-channel contention. In the proposed framework, the MAC protocol proceeds in cycles, each of which accommodates the stations' requests to transmit. This subsumes various existing random token-like protocols e.g., CSMA/CA, Random Token, slotted ALOHA, and HIPERLAN/1. A selection configuration determines how the requests are placed, whereas a winner policy decides the winner of a contention. It is argued that while the winner policy must be agreed upon by all the stations, the selection configuration is entirely within a station's discretion. For a wide class of winner policies, called Random Token with Extraneous Collision Detection (RT/ECD), the notions of verifiability and viability are introduced to factor out brute-force selfish behavior, and shown to lead to two interesting winner policies, RT/ECD-0 and RT/ECD--. The resulting games are found to be a multiplayer Prisoners' Dilemma and an anti-coordination game, respectively. Related repeated games are again found to potentially lead to cooperative behavior.
In Chapter 6, selfishness considerations are extended to the network layer of a mobile ad hoc network (MANET). Each MANET station generates source packets and also is supposed to forward transit packets for pairs of stations currently out of each other's reception range. It is argued that forwarding transit packets becomes a dual liability: it shortens the station's battery life "in exchange for" sacrificing a portion of the channel bandwidth that could be used for source packets. Under the introduced packet anonymity model (no packet reveals its source station except at destination), undetectable selfish manipulation of the local congestion controls is possible. The issue is addressed in a noncooperative game-theoretic framework: each MANET station selects its own strategy (congestion control settings) at will, so as to maximize some defined payoff (a combination of source packet throughput and a reputation measure). For a class of Drop-and-Throttle (D&T) congestion control mechanisms, a generic model is formulated, whereby a station adjusts its D&T threshold in response to the other stations' adjusted thresholds; this model is subsequently translated into a simple extensive-form game. Types of reachable Nash equilibria are discussed and the notion of robustness of a Nash equilibrium is introduced. Finally, a novel packet forwarding protocol called Fair Forwarding with Forced Transmissions (F3T) is presented, which, if properly configured, leads to a desirable type of Nash equilibrium.

Spis treści:

CONTENTS 
LIST OF SYMBOLS AND ABBREVIATIONS

1. INTRODUCTION 
1.1. Motivation, objective, and scope 
1.2. Outline of the contents

2. MEMBERSHIP INSENSITIVE AGREED ORDER MULTICAST 
2.1. Introduction and related work 
2.2. Message ordering constraints 
2.3. Jamming Protocol 
2.3.1. LAN model 
2.3.2. JP mechanisms and data structures 
2.3.3. Jamming and lag detection 
2.3.3.1. Ethernet 
2.3.3.2. Token Ring 
2.3.4. "Catch-up" mechanism 
2.4. Evaluation of JP 
2.4.1. Logical-time properties of JP 
2.4.2. Stochastic performance of JP 
2.5. Distributed Precedence Graph protocol 
2.5.1. Network service model 
2.5.1.1. Wired or wireless single-channel LAN 
2.5.1.2. Tag seąuencer 
2.5.1.3. Bounded-deiay WAN with sparse group messages 
2.5.2. DPG data structures and PDU types 
2.5.3. Maintenance of strong ADO 
2.5.4. Missed message recovery and flow control 
2.6. Evaluation of DPG 
2.6.1. Logical-time properties of DPG 
2.6.2. Stochastic performance of DPG 
2.7. Conclusion

3. BACKOFF ATTACK AND ONE-SHOT CSMA/CA GAME 
3.1. Introduction and related work 
3.2. Network operation 
3.2.1. CSMA/CA contention under IEEE 802.11 DCF protocol 
3.2.2. Network model 
3.3. Solvability of EBM 
3.3.1. Uniform CSMA/CA configuration profile 
3.3.2. Non-uniform CSMA/CA configuration profile 
3.4. Station performance 
3.5. Backoff attack incentives 
3.5.1. Selfish configurations 
3.5.2. CSMA/CA game and Nash capacity 
3.5.3. Selfish and greedy configurations 
3.5.3.1. Incentives in a binary set of actions 
3.5.3.2. Incentives in a temary set of actions 
3.5.4. Mixed and Monte Carlo models 
3.6. Conclusion

4. BACKOFF ATTACK DISCOURAGEMENT and REPEATED CSMA/CA GAME 
4.1. Introduction and related work 
4.2. Morę properties of one-shot CSMA/CA game 
4.3. Repeated CSMA/CA game 
4.3.1. Introduction to repeated games 
4.3.2. Application to CSMA/CA game 
4.4. CRISP strategy 
4.4.1. Strategy description 
4.4.2. Uncorrelated Pareto optimality and subgame perfection 
4.4.3. Performance 
4.4.4. Enforceability 
4.4.5. Ternary CRISP 
4.5. SPELL strategy 
4.5.1. Strategy Description 
4.5.2. Uncorrelated Pareto optimality and subgame perfection 
4.5.3. Ternary SPELL 
4.6. QoS game 
4.6.1. One-shot QoS game 
4.6.2. Dynamie QoS game scenarios 
4.6.3. Stochastic QoS game 
4.7. Conclusion and comments

5. NONCOOPERATIVE CHANNEL CONTENTION IN AD HOC LANS WITH ANONYMOUS STATIONS 
5.1. Introduction 
5.2. WLAN model 
5.3. Framework for a winner policy 
5.3.1. Reąuests to transmit 
5.3.2. Verifiability 
5.3.3. Performance objectives 
5.3.4. Viability 
5.4. Binary RT/ECD game 
5.4.1. One-shot RT/ECD game 
5.4.2. Stage-by-stage repeated RT/ECD game 
5.4.2.1. Nash strategy 
5.4.2.2. Idea Bag strategy 
5.5. Cycle-by-cycle repeated RT/ECD game 
5.5.1. Reinforcement learning strategies 
5.5.2. Evolutionary stability 
5.6. Conclusion

6. FAIR PACKET FORWARDING IN MOBILE AD HOC NETWORKS 
6.1. Introduction 
6.2. Related work 
6.2.1. Reputation systems 
6.2.2. Micropayment schemes 
6.2.3. Game-theoretic approaches 
6.3. Forwarding model and related selfish behavior 
6.3.1. Forwarding model 
6.3.2. Discussion of the model 
6.3.3. Undetectable selfish behavior 
6.4. Game-theoretic model 
6.4.1. D&Tgame 
6.4.2. Reachability of Nash eąuilibria 
6.5. F3T protocol 
6.5.1. Protocol description 
6.5.2. D&T game payoffs under F3T 
6.6. F3T protocol configuration and performance 
6.7. Conclusion and future research

7. SUMMARY OF MAIN RESULTS

BIBLIOGRAPHY 
Summary in English 
Summary in Polish

  Cena:

Ilość

przechowalnia

28,00 zł

Inni klienci kupujący ten produkt zakupili również
Zapytaj o szczegóły
Imię i nazwisko:
E-mail:
Twoje pytanie:
Przepisz kod:
weryfikator
Księgarnia Techniczna zamieszcza w ofercie głównie podręczniki akademickie oraz książki techniczne przede wszystkim z dziedzin takich jak mechanika techniczna, podstawy konstrukcji, technologia gastronomiczna. Główne wydawnictwa w ofercie to Politechnika Warszawska, Politechnika Wrocławska, Politechnika Świętokrzyska oraz POLSL
Księgarnia Techniczna - XML Sitemap
©Sprzedaż Książek. Wszelkie Prawa Zastrzeżone. All Rights Reserved

Wykonanie: inż. Agnieszka Kamińska