FUZZY REASONING IN A MULTIAGENT SYSTEM OF SURVEILLANCE SENSORS TO MANAGE COOPERATIVELY THE SENSOR-TO-TASK ASSIGNMENT PROBLEM
Titel:
FUZZY REASONING IN A MULTIAGENT SYSTEM OF SURVEILLANCE SENSORS TO MANAGE COOPERATIVELY THE SENSOR-TO-TASK ASSIGNMENT PROBLEM
Auteur:
Molina, Jose M. Herrero, Jesus Garcia Jimenez, Franciso J. Casar, Jose R.
Verschenen in:
Applied artificial intelligence
Paginering:
Jaargang 18 (2004) nr. 8 pagina's 673-711
Jaar:
2004-09
Inhoud:
In this work, a surveillance network composed of a set of sensors and a fusion center is designed as a multiagent system. Negotiation among sensors (agents) is proposed to solve the task-to-sensor assignment problem (the allocation of tasks to sensors), addressing several aspects. First, the fusion center determines the tasks (system tasks) to be performed by the network at each management cycle. To do that, a fuzzy reasoning system determines the priorities of these system tasks by means of a symbolic inference process using the fused data received from all sensors. In addition, a fuzzy reasoning process, similar to that performed in the fusion center, is proposed to evaluate the priority of local tasks (sensor tasks) now executed by each sensor. The network coordination procedure will be based on the system-task priorities, computed in the fusion center, and on the local priorities evaluated in each sensor. Priority values for system and sensor tasks will be the basis to guide a negotiation process among sensors in the multiagent system. The validity of the fuzzy reasoning approach is supported by the fact that it has been able to manage environmental situations in a similar way as experienced human operators do. Included results illustrate how the negotiation scheme, based on task priority and measured through their time-variant priority, allows the adaption of sensor operation to changing situations.