Resumen en español

The reconfiguration, reprogramming, and deployment of new computational tasks in wireless sensor networks are complex and represent a problem satisfactorily unresolved at present. The aim of this paper is to propose the performance evaluation of the use of mobile intelligent agents as autonomous rescheduling mechanism in such networks. The method used for performance evaluation is done by measuring the energy consumption in the migration of mobile intelligent agents among the sensor nodes of the system and calculating the convergence time of the network, defined as the time it takes for the network to move from one state to another; in experiments it refers to the delay in changing the sampling time for the entire network. The most efficient solution, which was tested and evaluated in a network is composed of 40 nodes that detect in real time ammonia leaks, determined that the key issue is to reduce the expenditure of unnecessary energy in transmission from the wireless sensors to the base station, while avoiding unnecessary confirmations and transmissions of data and procedures among sensor nodes. This fact represents besides the reduction in the network energy consumption, a very significant saving for convergence time of the network.