Results 1  10
of
13
Mdpop: Faithful distributed implementation of efficient social choice problems
 In AAMAS’06  Autonomous Agents and Multiagent Systems
, 2006
"... In the efficient social choice problem, the goal is to assign values, subject to side constraints, to a set of variables to maximize the total utility across a population of agents, where each agent has private information about its utility function. In this paper we model the social choice problem ..."
Abstract

Cited by 42 (15 self)
 Add to MetaCart
In the efficient social choice problem, the goal is to assign values, subject to side constraints, to a set of variables to maximize the total utility across a population of agents, where each agent has private information about its utility function. In this paper we model the social choice problem as a distributed constraint optimization problem (DCOP), in which each agent can communicate with other agents that share an interest in one or more variables. Whereas existing DCOP algorithms can be easily manipulated by an agent, either by misreporting private information or deviating from the algorithm, we introduce MDPOP, the first DCOP algorithm that provides a faithful distributed implementation for efficient social choice. This provides a concrete example of how the methods of mechanism design can be unified with those of distributed optimization. Faithfulness ensures that no agent can benefit by unilaterally deviating from any aspect of the protocol, neither informationrevelation, computation, nor communication, and whatever the private information of other agents. We allow for payments by agents to a central bank, which is the only central authority that we require. To achieve faithfulness, we carefully integrate the VickreyClarkeGroves (VCG) mechanism with the DPOP algorithm, such that each agent is only asked to perform computation, report
Evaluating the Performance of DCOP Algorithms in a Real World, Dynamic Problem
, 2008
"... Complete algorithms have been proposed to solve problems modelled as distributed constraint optimization (DCOP). However, there are only few attempts to address real world scenarios using this formalism, mainly because of the complexity associated with those algorithms. In the present work we compar ..."
Abstract

Cited by 22 (1 self)
 Add to MetaCart
Complete algorithms have been proposed to solve problems modelled as distributed constraint optimization (DCOP). However, there are only few attempts to address real world scenarios using this formalism, mainly because of the complexity associated with those algorithms. In the present work we compare three complete algorithms for DCOP, aiming at studying how they perform in complex and dynamic scenarios of increasing sizes. In order to assess their performance we measure not only standard quantities such as number of cycles to arrive to a solution, size and quantity of exchanged messages, but also computing time and quality of the solution which is related to the particular domain we use. This study can shed light in the issues of how the algorithms perform when applied to problems other than those reported in the literature (graph coloring, meeting scheduling, and distributed sensor network).
Evaluation of CBR on Live Networks
"... Abstract. A large class of problems in multiagent systems can be solved by distributed constraint optimization (DCOP). Several algorithms have been created to solve these problems, however, no extensive evaluation of current DCOP algorithms on live networks exists in the literature. This paper uses ..."
Abstract

Cited by 6 (4 self)
 Add to MetaCart
Abstract. A large class of problems in multiagent systems can be solved by distributed constraint optimization (DCOP). Several algorithms have been created to solve these problems, however, no extensive evaluation of current DCOP algorithms on live networks exists in the literature. This paper uses DCOPolis—a framework for comparing and deploying DCOP software in heterogeneous environments—to contribute an analysis of two stateoftheart DCOP algorithms run in various network environments solving a number of different problem types. Then, we use this empirical validation to evaluate the use of both cyclebased runtime and concurrent constraint checks. 1
Caching schemes for DCOP search algorithms
 In Proceedings of AAMAS
, 2009
"... Distributed Constraint Optimization (DCOP) is useful for solving agentcoordination problems. Anyspace DCOP search algorithms require only a small amount of memory but can be sped up by caching information. However, their current caching schemes do not exploit the cached information when deciding w ..."
Abstract

Cited by 4 (3 self)
 Add to MetaCart
Distributed Constraint Optimization (DCOP) is useful for solving agentcoordination problems. Anyspace DCOP search algorithms require only a small amount of memory but can be sped up by caching information. However, their current caching schemes do not exploit the cached information when deciding which information to preempt from the cache when a new piece of information needs to be cached. Our contributions are threefold: (1) We frame the problem as an optimization problem. (2) We introduce three new caching schemes (MaxPriority, MaxEffort and MaxUtility) that exploit the cached information in a DCOPspecific way. (3) We evaluate how the resulting speed up depends on the search strategy of the DCOP search algorithm. Our experimental results show that, on all tested DCOP problem classes, our MaxEffort and MaxUtility schemes speed up ADOPT (which uses bestfirst search) more than the other tested caching schemes, while our MaxPriority scheme speeds up BnBADOPT (which uses depthfirst branchandbound search) at least as much as the other tested caching schemes.
Improving dpop with function filtering
 In AAMAS
, 2010
"... DPOP is an algorithm for distributed constraint optimization which has, as main drawback, the exponential size of some of its messages. Recently, some algorithms for distributed cluster tree elimination have been proposed. They also suffer from exponential size messages. However, using the strategy ..."
Abstract

Cited by 3 (1 self)
 Add to MetaCart
DPOP is an algorithm for distributed constraint optimization which has, as main drawback, the exponential size of some of its messages. Recently, some algorithms for distributed cluster tree elimination have been proposed. They also suffer from exponential size messages. However, using the strategy of cost function filtering, in practice these algorithms obtain important reductions in maximum message size and total communication cost. In this paper, we explain the relation between DPOP and these algorithms, and show how cost function filtering can be combined with DPOP. We present experimental evidence of the benefits of this new approach.
A Distributed Constraint Optimization Approach to Wireless Network Optimization
"... We present a new algorithm called Variable Message Size (VMS) ADOPT for solving Distributed Constraint Optimization Problems (DCOP) which trades off message size and memory usage for running time. The algorithm is applied to a wireless network optimization problem, in which small robots act as wirel ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
We present a new algorithm called Variable Message Size (VMS) ADOPT for solving Distributed Constraint Optimization Problems (DCOP) which trades off message size and memory usage for running time. The algorithm is applied to a wireless network optimization problem, in which small robots act as wireless routers with the objective of maximizing signal strength in the network by repositioning themselves. Memory and bandwidth are limited resources in this application; our algorithm incorporates features of ADOPT and DPOP and introduces a parameter which controls the memory usage and message size at each agent. Boundederror approximation can also be used to trade off solution
Recent advances in dynamic, distributed constraint optimization
 Swiss Federal Institute of Technology (EPFL), Lausanne (Switzerland
, 2006
"... Many new technologies like sensor networks, RFID and the internet provide companies with a wealth of interconnected, realtime data sources and have significant potential for automating certain parts of their business processes. The potential of these technologies goes well beyond the limited role o ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
Many new technologies like sensor networks, RFID and the internet provide companies with a wealth of interconnected, realtime data sources and have significant potential for automating certain parts of their business processes. The potential of these technologies goes well beyond the limited role of simply collecting, storing and sending data to a central location for processing. We believe that these ”smart items ” will certainly evolve to more complex embedded systems that can also communicate between themselves, aggregate data and cooperate to take optimal decisions locally, with little or no influence from a central system. We believe that many key enabling technologies will come from the field of Distributed Artificial Intelligence. Specifically, within that field, Distributed Constraint Optimization is an excellent formalism for problem solving in multiagent systems. Many reallife problems like planning, scheduling and resource allocation can be modeled in this framework, and effectively solved in a distributed fashion. This paper is a brief introduction to this area. We present some important issues that arise in this domain, and some of our recent results.
Stochastic dominance in stochastic DCOPs for risksensitive applications
 In Proceedings of AAMAS
, 2012
"... Distributed constraint optimization problems (DCOPs) are wellsuited for modeling multiagent coordination problems where the primary interactions are between local subsets of agents. However, one limitation of DCOPs is the assumption that the constraint rewards are without uncertainty. Researchers ..."
Abstract

Cited by 1 (1 self)
 Add to MetaCart
Distributed constraint optimization problems (DCOPs) are wellsuited for modeling multiagent coordination problems where the primary interactions are between local subsets of agents. However, one limitation of DCOPs is the assumption that the constraint rewards are without uncertainty. Researchers have thus extended DCOPs to Stochastic DCOPs (SDCOPs), where rewards are sampled from known probability distribution reward functions, and introduced algorithms to find solutions with the largest expected reward. Unfortunately, such a solution might be very risky, that is, very likely to result in a poor reward. Thus, in this paper, we make three contributions: (1) we propose a stricter objective for SDCOPs, namely to find a solution with the most stochastically dominating probability distribution reward function; (2) we introduce an algorithm to find such solutions; and (3) we show that stochastically dominating solutions can indeed be less risky than expected reward maximizing solutions.
Analyzing the Performance of Distributed Algorithms
"... Abstract — A large class of problems in multiagent systems can be solved by distributed constraint optimization (DCOP). Several algorithms have been created to solve these problems, however, no extensive evaluation of current DCOP algorithms on live networks exists in the literature. This paper uses ..."
Abstract
 Add to MetaCart
Abstract — A large class of problems in multiagent systems can be solved by distributed constraint optimization (DCOP). Several algorithms have been created to solve these problems, however, no extensive evaluation of current DCOP algorithms on live networks exists in the literature. This paper uses DCOPolis—a framework for comparing and deploying DCOP software in heterogeneous environments—to contribute an analysis of two stateoftheart DCOP algorithms solving a number of different problem types. Then, we use this empirical validation to evaluate the use of both cyclebased runtime and concurrent constraint checks. I.
Message Passing for Distributed QoSSecurity Tradeoffs ∗
"... Information Assurance (IA) is of growing concern to the field of distributed systems. However, IA cannot be considered in isolation, as it interacts with Quality of Service (QoS); in the presence of limited resources, the security mechanisms employed for IA (e.g., firewalls, antivirus, encryption) u ..."
Abstract
 Add to MetaCart
Information Assurance (IA) is of growing concern to the field of distributed systems. However, IA cannot be considered in isolation, as it interacts with Quality of Service (QoS); in the presence of limited resources, the security mechanisms employed for IA (e.g., firewalls, antivirus, encryption) usually adversely affect QoS levels delivered by a system. The system therefore needs to make a tradeoff between IA and QoS. This tradeoff is complicated by the fact that users ’ relative preferences over QoS/IA aspects change based on the situation, the interests of different users conflict, and tradeoff decisions made at one node in the distributed system typically affect other nodes as well. We address the problem of distributed computation of tradeoff among various aspects of QoS and IA in a way that maximizes the satisfaction of all stakeholders. Specifically, we want the nodes in the system to make coordinated decisions as to what local actions to take to optimize QoS/IA levels delivered by the system. Our first contribution is formulating this problem as a Distributed Constraint Optimization Problem (DCOP). This entails quantifying various aspects of the system in order to be able to compare options in the course of optimization, as well as encoding the effects of various decisions on the quantities we want to optimize. The DCOPs we obtain have cost functions with many local configurations that result in the same cost. In addition, the corresponding factor graphs contain many cycles. To deal with these issues, our second contribution is a value propagation phase that helps nodes reach a consistent set of decisions even in cyclic factor graphs with nonunique local minimizers. We present experimental results comparing the performance of the maxsum algorithm with and without value propagation against other algorithms implemented in the Frodo [6] framework applied to two different kinds of problems.