Results 1  10
of
85
Asynchronous backtracking without adding links: a new member in the ABT family
, 2005
"... ..."
(Show Context)
BnBADOPT: An asynchronous branchandbound DCOP algorithm
 In Proceedings of AAMAS
, 2008
"... Abstract. Distributed constraint optimization problems (DCOPs) are a popular way of formulating and solving agentcoordination problems. It is often desirable to solve DCOPs optimally with memorybounded and asynchronous algorithms. We thus introduce BranchandBound ADOPT (BnBADOPT), a memoryboun ..."
Abstract

Cited by 64 (21 self)
 Add to MetaCart
(Show Context)
Abstract. Distributed constraint optimization problems (DCOPs) are a popular way of formulating and solving agentcoordination problems. It is often desirable to solve DCOPs optimally with memorybounded and asynchronous algorithms. We thus introduce BranchandBound ADOPT (BnBADOPT), a memorybounded asynchronous DCOP algorithm that uses the message passing and communication framework of ADOPT, a well known memorybounded asynchronous DCOP algorithm, but changes the search strategy of ADOPT from bestfirst search to depthfirst branchandbound search. Our experimental results show that BnBADOPT is up to one order of magnitude faster than ADOPT on a variety of large DCOPs and faster than NCBB, a memorybounded synchronous DCOP algorithm, on most of these DCOPs. 1
Asynchronous forwardbounding for distributed constraints optimization
 In: Proc. 1st Intern. Workshop on Distributed and Speculative Constraint Processing. (2005
, 2006
"... A new search algorithm for solving distributed constraint optimization problems (DisCOPs) is presented. Agents assign variables sequentially and propagate their assignments asynchronously. The asynchronous forwardbounding algorithm (AFB) is a distributed optimization search algorithm that keeps one ..."
Abstract

Cited by 47 (8 self)
 Add to MetaCart
(Show Context)
A new search algorithm for solving distributed constraint optimization problems (DisCOPs) is presented. Agents assign variables sequentially and propagate their assignments asynchronously. The asynchronous forwardbounding algorithm (AFB) is a distributed optimization search algorithm that keeps one consistent partial assignment at all times. Forward bounding propagates the bounds on the cost of solutions by sending copies of the partial assignment to all unassigned agents concurrently. The algorithm is described in detail and its correctness proven. Experimental evaluation of AFB on random MaxDisCSPs reveals a phase transition as the tightness of the problem increases. This effect is analogous to the phase transition of MaxCSP when local consistency maintenance is applied [3]. AFB outperforms Synchronous Branch & Bound (SBB) as well as the asynchronous stateoftheart ADOPT algorithm, for the harder problem instances. Both asynchronous algorithms outperform SBB by a large factor. 1
Impact of problem centralization in distributed constraint optimization algorithms
 In AAMAS
, 2005
"... Recent progress in Distributed Constraint Optimization Problems (DCOP) has led to a range of algorithms now available which differ in their amount of problem centralization. Problem centralization can have a significant impact on the amount of computation required by an agent but unfortunately the d ..."
Abstract

Cited by 47 (4 self)
 Add to MetaCart
Recent progress in Distributed Constraint Optimization Problems (DCOP) has led to a range of algorithms now available which differ in their amount of problem centralization. Problem centralization can have a significant impact on the amount of computation required by an agent but unfortunately the dominant evaluation metric of “number of cycles ” fails to account for this cost. We analyze the relative performance of two recent algorithms for DCOP: OptAPO, which performs partial centralization, and Adopt, which maintains distribution of the DCOP. Previous comparison of Adopt and OptAPO has found that OptAPO requires fewer cycles than Adopt. We extend the cycles metric to define “CycleBased Runtime (CBR) ” to account for both the amount of computation required in each cycle and the communication latency between cycles. Using the CBR metric, we show that Adopt outperforms OptAPO under a range of communication latencies. We also ask: What level of centralization is most suitable for a given communication latency? We use CBR to create performance curves for three algorithms that vary in degree of centralization, namely Adopt, OptAPO, and centralized Branch and Bound search.
Message delay and DisCSP search algorithms
 ANN MATH ARTIF INTELL (2006 ) 46 : 415–439
, 2006
"... Distributed constraint satisfaction problems (DisCSPs) are composed of agents, each holding its own variables, that are connected by constraints to variables of other agents. Due to the distributed nature of the problem, message delay can have unexpected effects on the behavior of distributed searc ..."
Abstract

Cited by 32 (18 self)
 Add to MetaCart
(Show Context)
Distributed constraint satisfaction problems (DisCSPs) are composed of agents, each holding its own variables, that are connected by constraints to variables of other agents. Due to the distributed nature of the problem, message delay can have unexpected effects on the behavior of distributed search algorithms on DisCSPs. This has been recently shown in experimental studies of asynchronous backtracking algorithms (Bejar et al., Artif. Intell., 161:117–148, 2005; Silaghi and Faltings, Artif. Intell., 161:25–54, 2005). To evaluate the impact of message delay on the run of DisCSP search algorithms, a model for distributed performance measures is presented. The model counts the number of non concurrent constraints checks, to arrive at a solution, as a non concurrent measure of distributed computation. A simpler version measures distributed computation cost by the nonconcurrent number of steps of computation. An algorithm for computing these distributed measures of computational effort is described. The realization of the model for measuring performance of distributed search algorithms is a simulator which includes the cost of message delays. Two families of distributed search algorithms on DisCSPs are investigated. Algorithms that run a single search process, and multiple search processes algorithms. The two families of algorithms are described and associated with existing algorithms. The performance of three representative algorithms of these two families is measured on randomly generated instances of DisCSPs with delayed messages. The delay of messages is found to have a strong negative effect on single search process algorithms, whether synchronous or asynchronous. Multi
A general, fully distributed multiagent planning algorithm
 In Proceedings of AAMAS’10
, 2010
"... We present a fully distributed multiagent planning algorithm. Our methodology uses distributed constraint satisfaction to coordinate between agents, and local planning to ensure the consistency of these coordination points. To solve the distributed CSP efficiently, we must modify existing methods t ..."
Abstract

Cited by 31 (6 self)
 Add to MetaCart
(Show Context)
We present a fully distributed multiagent planning algorithm. Our methodology uses distributed constraint satisfaction to coordinate between agents, and local planning to ensure the consistency of these coordination points. To solve the distributed CSP efficiently, we must modify existing methods to take advantage of the structure of the underlying planning problem. In multiagent planning domains with limited agent interaction, our algorithm empirically shows scalability beyond state of the art centralized solvers. Our work also provides a novel, realworld setting for testing and evaluating distributed constraint satisfaction algorithms in structured domains and illustrates how existing techniques can be altered to address such structure. Categories and Subject Descriptors
Dynamic Ordering for Asynchronous Backtracking on DisCSPs
, 2006
"... An algorithm that performs asynchronous backtracking on distributed CSPs, with dynamic ordering of agents is proposed, ABT DO. Agents propose reorderings of lower priority agents and send these proposals whenever they send assignment messages. Changes of ordering triggers a different computation of ..."
Abstract

Cited by 30 (8 self)
 Add to MetaCart
(Show Context)
An algorithm that performs asynchronous backtracking on distributed CSPs, with dynamic ordering of agents is proposed, ABT DO. Agents propose reorderings of lower priority agents and send these proposals whenever they send assignment messages. Changes of ordering triggers a different computation of Nogoods. The dynamic ordered asynchronous backtracking algorithm uses polynomial space, similarly to standard ABT. TheABT DO algorithm with three different ordering heuristics is compared to standard ABT on randomly generated DisCSPs. A Nogoodtriggered heuristic, inspired by dynamic backtracking, is found to outperform static order ABT by a large factor in runtime and improve the network load.
Synchronous vs asynchronous search on DisCSPs
 In Proc. EUMAS’03
, 2003
"... Abstract. Distributed constraint satisfaction problems (DisCSP s) are composed of agents, each holding its variables, that are connected by constraints to variables of other agents. There are two known measures of performance for distributed search the computational effort which represents the tota ..."
Abstract

Cited by 30 (12 self)
 Add to MetaCart
Abstract. Distributed constraint satisfaction problems (DisCSP s) are composed of agents, each holding its variables, that are connected by constraints to variables of other agents. There are two known measures of performance for distributed search the computational effort which represents the total search time and the number of messages sent which represents the network load. Due to the distributed nature of the problem, the behavior of the experimental environment is extremely important. However, most experimental studies have used a perfect simulator with instantaneous message delivery. The present paper investigates two families of distributed search algorithms on DisCSPs, Synchronous and Asynchronous search. Improved versions of the two families of algorithms are presented and investigated. The performance of the algorithms of these two extended families is measured on randomly generated instances of DisCSPs. The results of the investigation are twofold. First, the delay of messages is found to deteriorate the performance of asynchronous search by a large margin. This shows that a correct (and realistic) experimental scenario is important. Second, when messages are delayed, synchronous search performs better than asynchronous search in terms of computational effort as well as in network load. It turns out that asynchronous search fails to use its multiple computing power to an advantage. 1
Using additional information in DisCSPs search
 In DCR
, 2004
"... Abstract. A method of volunteering information during asynchronous search on DisCSPs is presented. The meeting scheduling problem (MSP) is formulated as a distributed search problem. In order to implement asynchronous backtracking (ABT) for the MSP, a multivariable version of ABT is described. Agen ..."
Abstract

Cited by 23 (3 self)
 Add to MetaCart
Abstract. A method of volunteering information during asynchronous search on DisCSPs is presented. The meeting scheduling problem (MSP) is formulated as a distributed search problem. In order to implement asynchronous backtracking (ABT) for the MSP, a multivariable version of ABT is described. Agents participate in multiple meetings, where each meeting is represented by a variable that needs to be assigned a timeslot. Assignments are constrained by arrivaltime constraints, since meetings take place in different locations. All constraints are local to their agents. Additional information is in the form of Nogoods. During search for a consistent schedule for all meetings, agents can generate and send additional Nogoods to those sent by the ABT algorithm. When additional Nogoods are sent, the efficiency of asynchronous backtracking is enhanced. This effect grows with the number of additional volunteered Nogoods. 1
DCOPolis: A Framework for Simulating and Deploying Distributed Constraint Optimization Algorithms
"... Abstract. A large class of problems in multiagent systems can be solved by distributed constraint optimization (DCOP). Until now DCOPs have exclusively been implemented in simulation—with each algorithm running in a different simulator. Furthermore, very few examples of realworld DCOP implementation ..."
Abstract

Cited by 22 (3 self)
 Add to MetaCart
Abstract. A large class of problems in multiagent systems can be solved by distributed constraint optimization (DCOP). Until now DCOPs have exclusively been implemented in simulation—with each algorithm running in a different simulator. Furthermore, very few examples of realworld DCOP implementation exists in the literature. This paper presents DCOPolis, a framework for comparing and deploying DCOP software in heterogeneous environments. DCOPolis makes three main contributions to the community. Different communications platforms, DCOP algorithms and problems can be plugged in for a truly comprehensive analysis of DCOP performance. DCOPolis contributes to comparative analysis of DCOP algorithms by allowing different stateoftheart algorithms to run in the same simulator under the same conditions or to be deployed on “real ” hardware in “real ” scenarios. Finally, DCOPolis introduces a new form of distributed algorithm simulation that shows promise of accurate prediction of realworld runtime. 1