An approach to distributed collaboration problem with conflictive tasks

Abstract

Distributed collaboration problem (DCP) is a kind of distributed resource allocation problem. Previous works give the solution of DCP with non-conflictive tasks. In this paper, we present a completely distributed algorithm (CDA) to solve DCP with conflictive tasks where each task requires exactly two resources. We call this the measurement collaboration problem (MCP). We prove the aliveness and correctness of CDA and give the run-time performance of CDA by simulation. To provide a baseline of performance parameters, we design a simple single-wait-state algorithm (SWSA). The simulation indicates CDA gradually has higher efficiency than SWSA, especially when the tasks are highly conflictive. CDA can also be utilized in many aspects, such as disaster rescue, distributed agent collaboration, objects position and so on. © Springer-Verlag 2004.