SUPPORTING COOPERATION AND CONSTRAINTS NEGOTIATION BETWEEN TIME AND RESOURCE MANAGERS
Jean-Pierre Camalot, Patrick Esquirol · 2007
This paper is concerned with the design of distributed decision support systems for organizations (companies, project consortiums, virtual enterprises), in which several decisionmakers have to cooperate. Typical situations can be found in the domain of the distributed management of time and resources (i.e. planning, scheduling, personal management, allocation of shared resources...). Each decision-maker has some latitude that enables to accommodate himself with unpredictable changes of his environment without disturbing the overall organization, making this last more reactive. Nevertheless this autonomy must be evaluated and controlled to ensure the global consistency of local decisions. To facilitate autonomy regulations between decision-makers, we propose to design tools supporting both local problem solving and cooperation. Our approach is based on a rigorous but non-deterministic use of CSP models and constraint propagation mechanisms ([16]). The goal is the improvement of the robustness of decisions each decisionmaker imposes to others, but also, in a reversible way, to evaluate the acceptability of external decisions he must comply with. We propose some protocols associated to asynchronous communications (via mailboxes) whereby decision-makers coordinate themselves and negotiate constraints. Managing interlaced conversations and several plausible solving scenarios is a complex task that needs first a good representation of the state of decision processes. We propose a tree-structure, in which nodes represent CSP models, and branching represent constraints addition derived from internal or external hypothesis. Our communication protocols and the evolution of such CSP trees are illustrated through several scenarios in the context of the resourceconstrained multi-pro...