Design and Evaluation of Artificial Intelligence and Deep Learning Applications for Communication Systems

Project: Research

Project Details

Description

Utilizing Deep Neural Networks (DNN) in engineering and sciences applications has grown significantly in recent years. Recently, a paradigm shift from model-based to end-to-end design and optimization of communication systems has been introduced using deep learning tools. In wireless communication system, the demand and exchange of information has significantly increased. New factors, such as low- latency requirements, mobility, changing channel conditions, or unknown channel models will benefit from deep learning applications. In addition, communicating over unlicensed spectrum, such as ISM band, became wide popular for wireless applications, such as WiFi and Bluetooth, which in turn increases the interference at the receiver. The plethora of both interference sources and mitigation algorithms along with the exponential growth of wireless systems necessitate designing an adaptive system that accounts for multiple interference sources without compromising the bandwidth of the signal-of-interest or increasing the computational complexity of the system. This issue requires an agile system with the ability to detect the interference and mitigate it without human intervention. The recent success in implementing supervised learning to classify modulation types by classification of different types of modulation using Convolutional Neural Networks (CNN), suggests that other problems akin to modulation classification would eventually benefit from that implementation. One of these problems is classifying the interference type added to a signal-of-interest, also known as interference classification. This project will design and evaluate key applications of DNN in communications systems physical layer, such as modulation and interference classifiers, MIMO Detection, power allocation, channel estimation, and encoding/ decoding. In addition, some classical communication blocks, such as Viterbi decoders, can benefit a lot from DNN assistance. Furthermore, deep reinforcement learning could be used to solve complex optimization problems in communication systems. Therefore, we will investigate the implications of such application on the design and performance of communication system
StatusFinished
Effective start/end date1/07/2131/12/22

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.