Distributed computing meets game theory

Lecture Notes in Computer Science, 2007

In this work, we investigate the problem of resolving conflicts in a distributed environment using only local knowledge. The contribution of this paper is twofold. First, we present a self-stabilizing algorithm to deal with this problem. Self-stabilizing algorithms protect against transient failures. The second result gives a particular implementation and analysis based on probabilistic procedures. Thus, the stabilization time is computed in terms of computation steps, then approximated according to the needed synchronizations.

2010

2007

Replicating data objects onto servers across a system can alleviate access delays. The selection of data objects and servers requires solving a constraint optimization problem, which is NP-complete in general. A majority of conventional replica placement techniques falter on issues of scalability or solution quality. To counteract such issues, we propose a game theoretical replica placement technique, in which computational agents compete for the allocation or reallocation of replicas onto their servers in order to reduce the user perceived access delays. The technique is based upon six well-defined axioms, each guaranteeing certain basic game theoretical properties. This eccentric method of designing game theoretical techniques using axioms is unique in the literature and takes away from the designers the cumbersome mathematical details of game theory. The distinctive feature of these axioms is that when amassed together, their individual properties constrict into one system-wide performance enhancement property, which in our case is the reduction of access time. The control of the proposed technique is "semi-distributed" in nature, wherein all the heavy processing is done on the servers of the distributed system and the central body is only required to take a binary decision: (0) not to replicate or (1) to replicate. This semi-distributed approach makes the technique scalable and helps solutions to converge in a fast turn-around time without loosing much of the solution quality. Experimental comparisons are made against: 1) branch and bound, 2) greedy, 3) genetic, 4) Dutch auction, and 5) English auction. As attested by the results, the proposed technique maintains superior solution quality in terms of lower communication cost and reduced execution time.

2009

We use ideas from distributed computing to study dynamic environments in which computational nodes, or decision makers, follow adaptive heuristics [16], i.e., simple and unsophisticated rules of behavior, e.g., repeatedly "best replying" to others' actions, and minimizing "regret", that have been extensively studied in game theory and economics. We explore when convergence of such simple dynamics to an equilibrium is guaranteed in asynchronous computational environments, where nodes can act at any time. Our research agenda, distributed computing with adaptive heuristics, lies on the borderline of computer science (including distributed computing and learning) and game theory (including game dynamics and adaptive heuristics). We exhibit a general non-termination result for a broad class of heuristics with bounded recall-that is, simple rules of behavior that depend only on recent history of interaction between nodes. We consider implications of our result across a wide variety of interesting and timely applications: game theory, circuit design, social networks, routing and congestion control. We also study the computational and communication complexity of asynchronous dynamics and present some basic observations regarding the effects of asynchrony on no-regret dynamics. We believe that our work opens a new avenue for research in both distributed computing and game theory.

Sajjad Haider , Naveed Riaz Ansari , Muhammad Akbar , Mohammad Raza Perwez , Khawaja MoyeezUllah Ghori 1 National University of Modern Languages, Islamabad, 2 Shaheed Zulfiqar Ali Bhutto Institute of Science & Technology, Islamabad Abstract. Distributed systems are responsible for providing the main execution platform for High Performance Computing (HPC). As distributed systems can be homogeneous (cluster) as well as heterogeneous (grid and cloud etc), they are prone to different kinds of problems. The issues in distributed systems can be Security, Quality of Service, Resource Selection and Fault Tolerance etc. Fault tolerance is responsible for handling the reliability and availability of distributed systems. It is not feasible to ignore job failures in distributed environments where long and persistent commitments of resources are required. In this paper we have presented a comprehensive classification of errors, failures and faults that can be encountered in a Distributed environ...

ACM Computing Surveys, 1993

The consensus problem is concerned with the agreement on a system status by the fault-free segment of a processor population in spite of the possible inadvertent or even malicious spread of disinformation by the faulty segment of that population. The resulting protocols are useful throughout fault-tolerant parallel and distributed systems and will impact the design of decision systems to come. This paper surveys research on the consensus problem, compares approaches, outlines applications, and suggests directions for future work.

Proceedings of the first …, 2002

2009

We use ideas from distributed computing to study dynamic environments in which computational nodes, or decision makers, follow adaptive heuristics (Hart 2005), i.e., simple and unsophisticated rules of behavior, e.g., repeatedly "best replying" to others' actions, and minimizing "regret", that have been extensively studied in game theory and economics. We explore when convergence of such simple dynamics to an equilibrium is guaranteed in asynchronous computational environments, where nodes can act at any time. Our research agenda, distributed computing with adaptive heuristics, lies on the borderline of computer science (including distributed computing and learning) and game theory (including game dynamics and adaptive heuristics). We exhibit a general non-termination result for a broad class of heuristics with bounded recall---that is, simple rules of behavior that depend only on recent history of interaction between nodes. We consider implications of our result across a wide variety of interesting and timely applications: game theory, circuit design, social networks, routing and congestion control. We also study the computational and communication complexity of asynchronous dynamics and present some basic observations regarding the effects of asynchrony on no-regret dynamics. We believe that our work opens a new avenue for research in both distributed computing and game theory.

2015

Consider a distributed information network with harmful procedures called attackers (e.g., viruses); each attacker uses a probability distribution to choose a node of the network to damage. Opponent to the attackers is the system protector scanning and cleaning from attackers some part of the network (e.g., an edge or a simple path), which it chooses independently using an-other probability distribution. Each attacker wishes to maximize the proba-bility of escaping its cleaning by the system protector; towards a conflicting objective, the system protector aims at maximizing the expected number of cleaned attackers. In [8, 9], we model this network scenario as a non-cooperative strategic game on graphs. We focus on two basic cases for the protector; where it may choose a single edge or a simple path of the network. The two games obtained are called as the Path and the Edge model, respec-tively. For these games, we are interested in the associated Nash equilibria, where no network ent...

… of the 2nd international conference on …, 1976

A technique is described which permits distributed resources to be shared (services to be offered) in a resilient manner. The essence of the technique is to a priori declare one of the server hosts primary and the others backups. Any of the servers can perform the primary duties. ...

The theory of mechanism design in economics/game theory deals with a center who wishes to maximize an objective function which depends on a vector of information variables. The value of each variable is known only to a selfish agent, which is not controlled by the center. In order to obtain its objective the center constructs a game, in which every agent participates and reveals its information, because these actions maximize its utility. However, several crucial new issues arise when one tries to transform existing economic mechanisms into protocols to be used in computational environments. In this paper we deal with two such issues: 1. The communication structure, and 2. the representation (syntax) of the agents' information. The existing literature on mechanism design implicitly assumes that these two features are not relevant. In particular, it assumes a communication structure in which every agent is directly connected to the center. We present new protocols that can be implemented in a large variety of communication structures, and discuss the sensitivity of these protocols to the way in which information is presented.

ACM Transactions on Autonomous and Adaptive Systems, 2016

Self-stabilizing systems tolerate transient faults by always returning to a legitimate system state within a finite time. This goal is challenged by several system features such as arbitrary system states after faults, various process execution models, and constrained process communication means. This work designs self-stabilizing distributed algorithms from the perspective of game theory, achieving an intended system goal through private goals of processes. We propose a generic game design for identifying a maximal independent set (MIS) or a maximal weighted independent set (MWIS) among all processes in a distributed system. From the generic game several specific games can be defined which differ in whether and how neighboring players influence each other. Turning the game designs into self-stabilizing algorithms, we obtain the first algorithms for the MWIS problem and also the first self-stabilizing MIS algorithm that considers node degree (including an analysis of its performance...

2006

Game theory has an elegant way of modeling some structural aspects of social games. The predicted outcome of the social games holds as long as ?the rules of the game? are kept. Therefore, a game authority (which enforces the ?rules?) is implied. We present the first design for that game authority, and the first suiting middleware for executing an algorithmic mechanism in distributed systems. The middleware restricts the agents to ?play by the rules?, and excludes non-selfish agents since we consider them as Byzantine. We base our design on a self-stabilizing Byzantine agreement that allows processors to audit each other?s actions. We show that when the agents are restricted to act selfishly the resource allocation is asymptotically optimal (according to our novel performance ratio; multi-round anarchy cost). Our design also includes services that allow owners to share a collaborative effort for coalition optimization using group-preplay negotiation. Since there are no guarantees for...

Lecture Notes in Computer Science, 2003

We propose a notion of distributed games as a framework to formalize and solve distributed synthesis problems. In general the problem of solving distributed games is undecidable. We give two theorems allowing to simplify, and possibly to solve, some distributed games. We show how several approaches to distributed synthesis found in the literature can be formalized and solved in the proposed setting.

2007

Distributed algorithm designers often assume that system processes execute the same predefined software. Alternatively, when they do not assume that, designers turn to noncooperative games and seek an outcome that corresponds to a rough consensus when no coordination is allowed. We argue that both assumptions are inapplicable in many real distributed systems, e.g., the Internet, and propose designing self-stabilizing and Byzantine fault-tolerant distributed game authorities. Once established, the game authority can secure the execution of any complete information game. As a result, we reduce costs that are due to the processes' freedom of choice. Namely, we reduce the price of malice.