Repository to hold the code for "Channel Estimation using Bidirectional Recurrent Encoder and Attention
Channel Estimation has played a massive role in modernizing the Telecommunication Industry, unlocking doors to temporal characteristics associated with wireless channels. In this paper, we demonstrate using a Bidirectional Recurrent Encoder and Attention Module. Channel Estimation focuses on forecasting channel behaviors depending on pilot characteristics showcased by transmitters. BRE (Bidirectional Recurrent Encoder)is leveraged to capture past and future dependencies of the said characteristics, while the Attention module helps us single out relevant parts from the input sequence, increasing the robustness of the model significantly in comparison to other techniques available in the space. We demonstrate this application in a way that highlights the importance of advanced modeling techniques in the near future of Telecommunications.
- Data Generation
- Data Pre-Processing
We are going to be using the DeepMIMO Dataset, which gives us access to various data scenarios for channel estimation training. For our training we will be going with the following scenarios to check the robustness of our model.
- Data Preparation
- Bidirectional Recurrent Encoder
- Attention Model
- Model Training
- Evaluation (MMSE and BER)
- Write Paper
- Deployment (?)