Maximum Likelihood Estimation of Doubly-Selective Channels for Next-Generation 5G Distributed MIMO Systems with FANET Deployment

– 2021 Mitacs Globalink Internships – By the Wireless Lab at the EMT Centre of INRS (Project ID 24745)?

Introduction:

Traditional communication networks are primarily designed to serve low-mobility mobile users, i.e., with typical velocities below 120 km/h. At such velocities, time-varying channels can be assumed as locally constant, thereby making the design and optimization of wireless communication systems relatively easier. Future-generation multi-antenna systems 5G are expected to support reliable communications at Activities very high velocities reaching 500 Km/h (e.g., high-speed trains). It will also feature new types of ad-hoc networks such as Internet of vehicles (IoV) that allows exchange of data among vehicles and humans or the flying Internet of things (IoT) also known as FANET, a subset of VANETs that allows communication among UAVs or drones and humans. For such systems, classical assumptions of constant channels lead to severe performance losses. Therefore, highly innovative new channel estimation techniques specifically tailored to strong-mobility terminals need to be developed almost from scratch. This R&D project aims at developing a new channel estimator for distributed multiple input multiple output (MIMO) system that is completely aware of the channel variations in both the time and frequency domains during the estimation process. More specifically, the channel variations will be tracked via a two-dimensional (2D) polynomial fitting wherein the channel estimation task boils down to finding the optimal 2D polynomial coefficients.

The developed technique will be ultimately integrated into a smart cognitive transceiver (e.g., UAVs and small vehicles) that is able to instantly adapt to its environment conditions for maximum performance. To that end, the required polynomial coefficients must be estimated adaptively according to the Doppler spread of the channel, the signal-to-noise ratio (SNR) of the underlying communication link. The project will be carried out in close collaboration Huawei Technologies Canada, a world-leading manufacturer of wireless devices and TELUS, a key wireless service provider in Canada. 

Student Role:

The student will be provided with derivation guidelines of the intended channel estimation technique. In particular, he/she will start by assimilating the basic concept of another advanced technique that has been recently developed by our research group. This technique approximates the time-selective channels by a polynomial series expansion in the time domain. The candidate will then be called upon to generalize the whole concept to doubly selective channels (i.e., in time and frequency). Towards this goal and with the help of a senior member of our research group, the candidate will be in charge of developing an iterative maximum likelihood (ML) solution using the well-known expectation-maximization (EM) concept. An appropriate initialization procedure needs also to be designed so as to make the proposed iterative solution converge to the optimal polynomial coefficients that best approximate the channel variations in both the time and frequency domains. In order to make the proposed channel estimation algorithm a good candidate for the smart cognitive transceiver we are currently investigating, the size of the local approximation window and the order of the corresponding 2D polynomial must be carefully selected as function of the Doppler spread, SNR, and Rician K-factor, etc. In a second phase, the candidate needs to implement a python version of the developed technique that will be integrated in a FANET consisting of multiple micro-drones. This step will serve to showcase the proposed solution in real-time and over the air. The Wireless Lab has a long-lasting experience in hardware prototyping/integration and is endowed with cutting-edge equipment for this purpose. 

Required Skills:

The candidate should have an electrical engineering background. He/she should have basic knowledge of signal processing and communication theory and a solid background in linear algebra and probability theory. Prior exposure to statistical signal processing such as estimation and detection theory would be an asset. The candidate should be familiar with MATLAB and python programing language. The candidate should be dynamic, self-motivated, and team player; qualities strongly required for successful involvement in collaborative R&D projects carried out in close collaboration with our industrial partners from which the candidate can gain precious hands-on experience and soft skills with significantly increased potential.

Additional Info:

?? For more info, please click [here].