A general framework for reordering agents asynchronously in distributed CSPs (original) (raw)

Asynchronous Coordination Under Preferences and Constraints

Structural Information and Communication Complexity, 2016

Adaptive renaming can be viewed as a coordination task involving a set of asynchronous agents, each aiming at grabbing a single resource out of a set of resources totally ordered by their desirability. Similarly, musical chairs is also defined as a coordination task involving a set of asynchronous agents, each aiming at picking one of a set of available resources, where every agent comes with an a priori preference for some resource. We foresee instances in which some combinations of resources are allowed, while others are disallowed. We model these constraints, i.e., the restrictions on the ability to use some combinations of resources, as an undirected graph whose nodes represent the resources, and an edge between two resources indicates that these two resources cannot be used simultaneously. In other words, the sets of resources that are allowed are those which form independent sets in the graph. E.g., renaming and musical chairs are specific cases where the graph is stable (i.e., it the empty graph containing no edges). As for musical chairs, we assume that each agent comes with an a priori preference for some resource. If an agent's preference is not in conflict with the preferences of the other agents, then this preference can be grabbed by the agent. Otherwise, the agents must coordinate to resolve their conflicts, and potentially choose non preferred resources. We investigate the following problem: given a graph, what is the maximum number of agents that can be accommodated subject to non-altruistic behaviors of early arriving agents? We entirely solve this problem under the restriction that agents which cannot grab their preferred resources must then choose a resource among the nodes of a predefined independent set. However, the general case, where agents which cannot grab their preferred resource are then free to choose any resource, is shown to be far more complex. In particular, just for cyclic constraints, the problem is surprisingly difficult. Indeed, we show that, intriguingly, the natural algorithm inspired from optimal solutions to adaptive renaming or musical chairs is sub-optimal for cycles, but proven to be at most 1 to the optimal. The main message of this paper is that finding optimal solutions to the coordination with constraints and preferences task requires to design "dynamic" algorithms, that is, algorithms of a completely different nature than the "static" algorithms used for, e.g., renaming.

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

The Asynchronous Backtracking Family

2003

In the last years, the AI community has shown an increasing interest in distributed problem solving. In the scope of distributed constraint reasoning, several asynchronous backtracking procedures have been proposed for nding solutions in a constraint network distributed among several computers. They dier in the way they store failing combinations of values (nogoods), and in the way they check the

Nogood-based asynchronous forward checking algorithms

Constraints - An International Journal, 2013

We propose two new algorithms for solving Distributed Constraint Satisfaction Problems (DisCSPs). The first algorithm, AFC-ng, is a nogood-based version of Asynchronous Forward Checking (AFC). Besides its use of nogoods as justification of value removals, AFC-ng allows simultaneous backtracks going from different agents to different destinations. The second algorithm, Asynchronous Forward Checking Tree (AFC-tree), is based on the AFC-ng algorithm and is performed on a pseudo-tree ordering of the constraint graph. AFC-tree runs simultaneous search processes in disjoint problem subtrees and exploits the parallelism inherent in the problem. We prove that AFC-ng and AFC-tree only need polynomial space. We compare the performance of these algorithms with other DisCSP algorithms on random DisCSPs and instances from real benchmarks: sensor networks and distributed meeting scheduling. Our experiments show that AFC-ng improves on AFC and that AFC-tree outperforms all compared algorithms, particularly on sparse problems.

ADOPT-ing: unifying asynchronous distributed optimization with asynchronous backtracking

Autonomous Agents and Multi-Agent Systems, 2008

This article presents an asynchronous algorithm for solving Distributed Constraint Optimization problems (DCOPs). The proposed technique unifies asynchronous backtracking (ABT) and asynchronous distributed optimization (ADOPT) where valued nogoods enable more flexible reasoning and more opportunities for communication, leading to an important speed-up. While feedback can be sent in ADOPT by COST messages only to one predefined predecessor, our extension allows for sending such information to any relevant agent. The concept of valued nogood is an extension by Dago and Verfaille of the concept of classic nogood that associates the list of conflicting assignments with a cost and, optionally, with a set of references to culprit constraints. DCOPs have been shown to have very elegant distributed solutions, such as ADOPT, distributed asynchronous overlay (DisAO), or DPOP. These algorithms are typically tuned to minimize the longest causal chain of messages as a measure of how the algorithms will scale for systems with remote agents (with large latency in communication). ADOPT has the property of maintaining the initial distribution of the problem. To be efficient, ADOPT needs a preprocessing step consisting of computing a Depth-First Search (DFS) tree on the constraint graph. Valued nogoods allow for automatically detecting and exploiting the best DFS tree compatible with the current ordering. To exploit such DFS trees it is now sufficient to ensure that they exist. Also, the inference rules available for valued nogoods help to exploit schemes of communication where more feedback is sent to higher priority agents. Together they result in an order of magnitude improvement. 1. DCOP definitions could also include it to help specify branch and bound solvers. 10. Because the corresponding constraint increases for the first time the cost of the computed nogood. 11. Assuming no mechanism is used to block immediate retransmission of nogoods, such as our lastSent structure. 12. Assignments having the same value are considered identical, even if their tag differs (allowing for re-using old nogoods). 13. Note that with the first scheme (i), where assignments are not tagged with counters, ADOPTing should not delete old nogoods from lr (which is done with the second scheme), but checks them when ok? messages are received.

Dynamic Prioritization of Complex Agents in Distributed Constraint Satisfaction Problems

International Joint Conference on Artificial Intelligence, 1997

Cooperative distributed problem solving (CDPS) loosely-coupled agents can be effectively modeled as a distributed constraint satisfaction problem (DCSP) where each agent has multiple local variables. DCSP protocols typically impose (partial) orders on agents ensure systematic exploration of the search space, but the ordering decisions can have a dramatic effect on the overall problem-solving effort. In this paper, we examine several

The Effects of Agent Synchronization in Asynchronous Search Algorithms

2007

The asynchronous searching techniques are characterized by the fact that each agent instantiates its variables in a concurrent way. Then, it sends the values of its variables to other agents directly connected to it by using messages. These asynchronous techniques have different behaviors in case of delays in sending messages. This article depicts the opportunity for synchronizing agents’ execution in case of asynchronous techniques. It investigates and compares the behaviors of several asynchronous techniques in two cases: agents process the received messages asynchronously (the real situation from practice) and the synchronous case, when a synchronization of the agents’ execution is done i.e. the agents perform a computing cycle in which they process a message from a message queue. After that, the synchronization is done by waiting for the other agents to finalize the processing of their messages. The experiments show that the synchronization of the agents’ execution leads to lower costs in searching for solution. A solution for synchronizing the agents’ execution is proposed for the analyzed asynchronous techniques.

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.

DFS ordering in Nogood-based Asynchronous Distributed Optimization (ADOPT-ng)

2006

This work proposes an asynchronous algorithm for solving Distributed Constraint Optimization problems (DCOPs) using a generalized kind of nogoods, called valued nogoods. The proposed technique is an extension of the asynchronous distributed optimization (ADOPT) where valued nogoods enable more flexible reasoning, leading to important speed-up. Valued nogoods are an extension of classic nogoods that associates each nogood with a threshold and optionally with a set of references to culprit constraints. ADOPT has the property of maintaining the initial distribution of the problem. ADOPT needs a preprocessing step consisting of computing a depth first search (DFS) tree on the agent graph. We show that besides bringing significant speed-up, valued nogoods allow for automatically detecting and exploiting DFS trees compatible with the current ordering since independent subproblems are now dynamically detected and exploited (DFS trees do not need to be specified/computed explicitly). However, not all possible orderings on variables are compatible with good DFS trees, and we find that on randomly ordered problems ADOPT-ng runs orders of magnitude slower than on orderings that are known to be compatible with short DFS trees. Being an extension of ABT, ADOPT-ng can also profit of the dynamic ordering heuristics enabled by Asynchronous Backtracking with Reordering (ABTR). However, our experiments imply that efficient dynamic ordering heuristics for ADOPT-ng will have to maintain compatibility with some DFS tree (e.g., to be decided by rebuilding DFS trees based on current search state). Experiments comparing ADOPT-ng with Valued Dynamic Backtracking show that ADOPT-ng also brings significant improvements over the old valued nogood-based algorithm.

Synchronization of multi-agent plans

1982

Consider an intelligent agent constructing a plan to be executed by several other agents; correct plan execution will often require that actions bc taken in a specific sequence. Therefore, the planner cannot simply tell each agent what action to perform; explicit mechanisms must exist for maintaining the execution sequence. This paper outlines such mechanisms. A framework for multiple-agent planning is devclopcd, consisting of several parts.