Abstract Wireless environment monitoring is performed by spectrum sensors and network sniffing in wireless networks. By this, we will get a set of spatially distributed measurements. In this work, we show how machine learning algorithms can be used to generate probabilistic forecasts rather than point estimations using data from numerous sensing devices. In this respect, the QoS data from the pressure sensor is taken and based on that, univariate models were constructed. Feed Forward Neural Network (FFNN) and Gaussian Process (GP) artificial intelligence algorithms were used for the construction of the models. The algorithms are applied to model measurements data collected from a heterogeneous wireless testbed environment. To evaluate the constructed model, throughput has been predicted and examined for the Gas, Humidity, Pressure, and Temperature sensors. The results of various error metrics are used to assess the performance of models like NRMSE, MASE, sMAPE, WQL, and MAE, and based on the error metrics values, it is observed that both the FFNN and GP algorithms provide good results for the different sensors QoS values prediction, but the GP algorithms performs slightly better than FFNN. The results indicate that GP machine learning approach provides accurate results as compared to the FFNN approach for QoS values prediction in the wireless environment. The results will lead to secure message delivery in the network.
Alan : Mühendislik
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|