VASCONCELOS, A. S.; http://lattes.cnpq.br/8740952344021265; VASCONCELOS, Alexandre Sales.
Résumé:
Wireless Sensor Networks (WSN) are being used in various kinds of application and one of the important aspects to consider is the running time, which is directly linked to energy consumption. Furthermore, there are other aspects just as important, like those related to Quality of Service (QoS), more specifically the Coverage Ratio of the region of interest in the application, the End-to-End Delay time (EED) and the Package Loss Rate (PLR) of a WSN. In this context, this paper proposes the use of asearch algorithm, based on Genetic Algorithms (GA), whose objective is to find the best configuration for deployment of sensor nodes in a WSN, using the residual energy metrics, coverage rate, EED and PLR. In order for the results obtained from the proposed approach to be the most realistic as possible, it was used simulation
modules that implement the presence of obstacles in the surroundings. The results obtained, initially without the use of obstacles, show that the use of GA provides a gain up to 78.0% for the sum of the residual energy, a reduction of almost 50.0% in the EED and 44.0% in the PLR, compared to an approach that uses random distribution of sensor nodes, called traditional approach. Another result shows that the use of the proposed approach had a success rate of 95.0% when searching for the deployment configuration of the sensor nodes to provide 100% coverage of the scene region, compared to a 10% success rate for traditional approach. There was also a significant difference between the results of the simulations with the module of the scenario obstacles activated and the ones with the module disabled. In the experiment to check the influence of the obstacles on the module over the PLR, it
was observed a reduction of the PLR with the module of the scenario obstacles off of 12.0%, whereas the reduction with the module activated was of more than 39.0%. The results of the experiments confirmed that the implemented model gives a better gain in the metrics the search, maximizing the coverage rate and time of life the WSN and minimizing the TAFF and TPP compared with the traditional model.