Xiaolin Zhu

Engaged in innovative research on the combination of 5G wireless communication technology and artificial intelligence, including spectrum modeling, prediction and model upgrading based on Pytorch for building, optimizing and adjusting models; Used RSSI data of WIFI spectrum from Shenzhen Airport and Shenzhen University laboratory to train the time series model (Multilayer Perceptron model with ResNet network structure adopted), and predicted MCS (modulation and coding strategy) on different WIFI channels to make decisions for packet sending strategy. Optimized model by using cross-entropy loss and customized loss functions, and evaluated the model based on accuracy and packet loss rates; Improved the 5-classes MCS classifier to a more refined 10-classes classifier, and adjusted the real-time prediction to a non-real-time prediction.

Engaged in innovative research on the combination of 5G wireless (mainly for WIFI) communication technology and artificial intelligence, including the use of Python for spectrum networks simulation, wireless data acquisition, and the design of supervised learning networks for spectrum modeling and prediction; Studied the topology theory and the demand principle of WIFI and used NS3 network system simulation platform to acquire wireless training data. Handled data pre-processing and designed LSTM (Long Short-Term Memory), CNN (Convolutional Neural Network) and other composite neural networks for spectrum modeling and prediction. Used the RMSE (root mean square error), packet loss rate and channel efficiency measurements to optimize model prediction, in terms of improving the throughput of WIFI spectrum.

Supported all programmes hosted by IDO; Assisted international students of UIC and supported international visitors and visiting students; Received international exchange students and school partner representatives and assisted in organizing activities and the promotion of the summer exchange program.