Model analog circuit to a higher level of abstraction using deep learning techniques to speed up testing simulations. For each circuit, the data was collected by sweeping all the combinations of inputs and by storing the obtained outputs. Then, different neural networks are trained and tuned using TensorFlow to find the best architecture. Next, we store and freeze the model to be used for hardware testing. By this, we were able to increase the testing speed to 65\% with 85\% accuracy.Keywords : Python, Tensorflow, Neural Networks, Verilog-A

-Worked closely with a software development team to contribute to the entire process.

Implementation of prediction method (RNN) to reduce energy consumption in a wireless sensors network : In this project, we want to implement a method for predicting engine vibrations. In fact, this work is intended for a coal mining company whose engines break down during their activities. And , As we know , this represent a waste of time so automatically a waste of money. So our project is intended for this company in order to predict the state of its engines using vibrations. Consequently, we can anticipate and intervene at the right moment. To do this, we will place an "Accelerometer" sensor in engines ,which sends the measurements to a monitoring center. But sending the data will consume a lot of energy and we will be obliged to change the sensors in a short time. Therefore, the prediction of the data will reduce this consumption.All that we have to do is send some of the measurements data and the monitoring central will run the implemented prediction algorithm (RNN).Keywords : Machine learning , Non-linear prediction , RNN , Wireless Sensors Network , Embedded SystemsTools : Pycharm , Python , Arduino , PanStamp

Global Navigation Satellite System (GNSS) is intended to give the position and the location of a moving object as well as his speed in any place of the world. So,in this internship, I focus on the mechanism of GNSS to calculate the position. And,the main task was to understand the GNSS-SDR library written in C/C++ and make the match with the theory. Finally, I run this code on a Cyclone V board which integrate the process Nios II. TOOLS: C/C++,Visual studio , Java, Eclipse