A DisCSP solving algorithm based on sessions (original) (raw)
Related papers
A Complete Algorithm for DisCSP: Distributed Backtracking With Sessions (DBS)
Workshop OptMas in AAMAS, 2009
Many algorithms for Distributed Constraints Satisfaction Problem (DisCSP) resolution use additional links between variables not connected by constraints. This causes a higher needed memory space. In this paper, we propose an algorithm for DisCSP resolution, called Distributed Backtracking with Sessions (DBS) which does not use such additional links so that the initial problem's topology is respected. This algorithm is complete and requires a low space complexity. The main feature of this algorithm is to use the concept of sessions to provide a context for the exchanged messages.
A distributed algorithm solving CSPs with a low communication cost
Proceedings Eighth IEEE International Conference on Tools with Artificial Intelligence, 1996
We present a new distributed algorithm which nds all solutions of Constraint Satisfaction Problems. Based on the Backtrack algorithm, it spreads subtrees of the search tree over processes running in parallel. The work is optimally shared among the processes while the communication cost remains low. We show that the speedup of the resolution is asymp totically linear as the number of variables increases. Furthermore, we study the addition of Lookahead pruning techniques and Nogood Recording. Experimental results con rm the e ciency of the algorithm, even if the search tree is very unbalanced.
Asynchronous Inter-Level Forward-Checking for DisCSPs
2009
We propose two new asynchronous algorithms for solving Distributed Constraint Satisfaction Problems (DisCSPs). The first algorithm, AFC-ng, is a nogood-based version of Asynchronous Forward Checking (AFC). The second algorithm, Asynchronous Inter-Level Forward-Checking (AILFC), is based on the AFC-ng algorithm and is performed on a pseudo-tree ordering of the constraint graph. AFC-ng and AILFC only need polynomial space. We compare the performance of these algorithms with other DisCSP algorithms on random DisCSPs in two kinds of communication environments: Fast communication and slow communication. Our experiments show that AFC-ng improves on AFC and that AILFC outperforms all compared algorithms in communication load.
Algorithms for Distributed Constraint Satisfaction: A Review
Autonomous Agents and Multi-agent Systems, 2000
When multiple agents are in a shared environment, there usually exist constraints among the possible actions of these agents. A distributed constraint satisfaction problem (distributed CSP) is a problem to nd a consistent combination of actions that satises these inter-agent constraints. Various application problems in multi-agent systems can be formalized as distributed CSPs. This paper gives an overview of the existing research on distributed CSPs. First, we briey describe the problem formalization and algorithms of normal, centralized CSPs. Then, we show the problem formalization and several MAS application problems of distributed CSPs. Furthermore, we describe a series of algorithms for solving distributed CSPs, i.e., the asynchronous backtracking, the asynchronous weak-commitment search, the distributed breakout, and distributed consistency algorithms. Finally, w e show t wo extensions of the basic problem formalization of distributed CSPs, i.e., handling multiple local variables, and dealing with over-constrained problems.
Communication and Computation in Distributed CSP Algorithms
2002
We introduce SensorDCSP, a naturally distributed benchmark based on a real-world application that arises in the context of networked distributed systems. In order to study the performance of Distributed CSP (DisCSP) algorithms in a truly distributed setting, we use a discrete-event network simulator, which allows us to model the impact of different network traffic conditions on the performance of the algorithms. We consider two complete DisCSP algorithms: asynchronous backtracking (ABT) and asynchronous weak commitment search (AWC). In our study of different network traffic distributions, we found that, random delays, in some cases combined with a dynamic decentralized restart strategy, can improve the performance of DisCSP algorithms. More interestingly, we also found that the active introduction of message delays by agents can improve performance and robustness, while reducing the overall network load. Finally, our work confirms that AWC performs better than ABT on satisfiable instances. However, on unsatisfiable instances, the performance of AWC is considerably worse than ABT.
Distributed Breakout Algorithm for Solving Distributed Constraint Satisfaction Problems
1996
This paper presents a new algorithm for solving distributed constraint satisfaction problems (distributed CSPs) called the distributed breakout algorithm, which is inspired by the breakout algorithm for solving centraiized CSPs. In this algorithm, each agent tries to optimize its evaluation value (the number of constraint violations) by exchanging its current value and the possible amount of its improvement among neighboring agents. Instead of detecting the fact that agents as a whole are trapped in a local-minimum, each agent detects whether it is in a quasi-local-minimum, which is a weaker condition than a local-minimum, and changes the weights of constraint violations to escape from the quasi-local-minimum. Experimental evaluations show this algorithm to be much more efficient than existing algorithms for critically difficult problem instances of distributed graph-coloring problems.
2005
Constraint satisfaction problem (CSP) is a powerful formalism to represent and to solve many real-life NP-complete problems such as, planning, resource allocation, meeting scheduling, etc. The great success of this formalism is due essentially to its simplicity in expressing any real-world problem subject to constraints. A CSP is a triplet (X, D, C) composed of a finite set of n variables X, each of which is taking values in an associated finite domain D and a set of e constraints C between these variables. Solving a CSP consists in finding one or all-complete assignments of values to variables satisfying all the constraints. However, this task is hard and many efforts were devoted towards enhancing it by reducing the complexity of the original problem. Essentially, the complexity reduction in CSP formalism is achieved by integrating the local consistency property (LC) and its corresponding filtering techniques. Those techniques allow the simplification of the original problem by eliminating values or combination of values that cannot belong to any solution. Many levels of LC have been proposed in the literature, among them enforcing arc-consistency is the most preeminent one because of its low time and space complexities. Most efforts dealing with enforcing AC on any constraint network (CN) are centralized almost always limited to binary CN, i.e., where each constraint involves at most two variables. Non-binary CNs, where constraints involve more than two variables, are often strongly required to deal with hard applications. Nevertheless, there is very few works involving nonbinary constraints and they pertain only to the centralized framework. Recently, with the advent of distributed computing and networking technologies, especially with the omnipresence of naturally distributed real-world problems, the interest in enforcing LC property in naturally distributed manner and for both binary and non-binary CN has largely increased, but such techniques have not been widely studied yet. Moreover, solving real-life applications, mainly meeting scheduling problems, requires also more studies to cope with the new environment requirements. Our main target is i) to find solutions and build a novel generic system to enforce some levels of LC with reasonable cost on any CN and ii) to take this system to the real-life through one among the important combinatorial applications, meetings scheduling problems. Our study on CSP framework and its related research directions including, LC enforcement techniques, and especially ways of solving real applications, mainly meeting scheduling problems (MS) stir up our attention to do more investigations in this framework. Five main contributions of this thesis are the following. • The integration of LC enforcement techniques in a constraint solver reduces the exponential space, in the number of variables, of the search tree. This clear benefit coupled i
Distributed Constraint Satisfaction for Formalizing Distributed Problem Solving
1992
Viewing cooperative distributed p r oblem solving (CDPS) as distributed c onstraint satisfaction provides a useful formalism for characterizing CDPS techniques. In this paper, we describe this formalism and compare algorithms for solving distributed c onstraint satisfaction problems (DCSPs). In particular, we present our newly developed t e chnique called asynchronous backtracking that allows agents to act asynchronously and concurrently, in contrast to the traditional sequential backtracking techniques employed i n constraint satisfaction problems. Our experimental results show that solving DCSPs in a distributed fashion is worthwhile when the problems solved by individual agents are l o osely-coupled.
Stochastic Local Search for Distributed Constraint Satisfaction Problems
Stochastic Search Algorithms
Nowadays, many real problems can be solved using local search strategies. These algorithms incrementally alter inconsistency value assignments to all the variables using a repair or hill climbing metaphor to move towards more and more complete solutions. Furthermore, if the problem can be modeled as a distributed problem, the advantages can be even greater. This paper presents a distributed model for solving Constraint Satisfaction Problems (CSPs), in which agents are committed to sets of constraints. The problem constraints are ordered and partitioned, by a preprocessing step, so that the most restricted constraints are studied first. Thus, each agent solves a subproblem by means of a stochastic local search algorithm. This constraint ordering, as well as value and variable ordering, can improve efficiency because inconsistencies can be found earlier and the number of constraint checks can be significantly reduced.
The Distributed Constraint Satisfaction Problem: Formalization and Algorithms
IEEE Transactions on Knowledge and Data Engineering, 1998
We develop a formalism called a distributed constraint satisfaction problem (distributed CSP) and algorithms for solving distributed CSPs. A distributed CSP is a constraint satisfaction problem in which variables and constraints are distributed among multiple agents. Various application problems in distributed artificial intelligence can be formalized as distributed CSPs. We present our newly developed technique called asynchronous backtracking that allows agents to act asynchronously and concurrently without any global control, while guaranteeing the completeness of the algorithm. Furthermore, we describe how the asynchronous backtracking algorithm can be modified into a more efficient algorithm called an asynchronous weak-commitment search, which can revise a bad decision without exhaustive search by changing the priority order of agents dynamically. The experimental results on various example problems show that the asynchronous weak-commitment search algorithm is, by far more, efficient than the asynchronous backtracking algorithm and can solve fairly large-scale problems