HEM: A hybrid evolutionary model for event detection in untrustworthy sensing

Since, event detection is open to many sensors in a participatory or in a crowd sensing network, one of the major challenges of this network to find the the authenticity of the reported events and the source nodes. If the nodes trustworthiness is unknown or the detected events truthfulness is also unknown, the event detection is a difficult task. In our paper, we study this challenge and observe that applying expectation maximization with genetic algorithm, trustworthy event detection is possible. We find the best local maximum points using Expectation Maximization and then gradually applying Genetic Algorithm we find the best value. We find our hybrid approach performs better since the best selection with the maximized expectation goes for evolving the new generation. In the long run, new generation becomes better and contributes faster. We do simulation study to support our model. We provide a comparative study among Genetic Algorithm and Expectation Maximization and the hybrid of the two above mentioned methods. We find that our proposed hybrid model provides better framework to find the trustworthy nodes, better convergence rate, more authenticated event detection.