Collaborative assignment
Kiam Tian Seow, Khee Yin How · 2002
A multiagent collaboration algorithm using the concepts of belief (B), desire (D) and intention (I) for the classical assignment problem is presented. The problem can be viewed as seeking a concurrent allocation of one different resource for every task. Existing sequential algorithms use a single agent to operate on all assignment values; each value indicates the application quality-of-service (A-QoS) of one resource for one task. However, for many network applications, it may be practically more effective or desirable to deploy multiple communicating agents to solve the problem. The proposed algorithm for collaborative assignment is motivated by this potential and the availability of multiagent technologies to implement it.Given an (N tasks and N resources) N × N assignment problem, the proposed algorithm has each different task represented by an agent possessing only (local) knowledge of A-QoS's of all resources for the task. The novelty of the algorithm lies in the generic BDI-based reasoning mechanism of task agents whose individual resource exchange intentions are arbitrated by a simple arbitration agent. The algorithm is examined both analytically and experimentally, and discussed, also in relation to existing multiagent work.