| PhD Seminar

Name of the Speaker: Mr. Sai Praneeth (EE21D200)
Guide: Prof. RadhaKrishna Ganti
Venue: ESB-244 (Seminar Hall)
Online meeting link: https://meet.google.com/uyx-sasc-dqd
Date/Time: 27th February 2024 (Thursday), 3:30 PM
Title: Near-Pilotless Decoding in Single Carrier Systems using Matrix Decomposition

Abstract :

Multiple Input-Multiple Output (MIMO) is a key enabler of higher data rates in the next generation wireless communications. However in MIMO systems, channel estimation and equalization are challenging particularly in the presence of rapidly changing channels. The high pilot overhead required for channel estimation can reduce the system throughput for large antenna configuration. So, through our work, we provide an iterative matrix decomposition algorithm for near-pilotless or blind decoding, in an uplink single carrier system with frequency domain equalization. This novel approach replaces the standard equalization and estimates both the transmitted data and the channel without the knowledge of any prior distributions, up to one pilot (for scaling correction). Our simulations demonstrate improved performance, in terms of error rates, compared to the more widely used pilot-based Maximal Ratio Combining (MRC) method. The work emphasises this idea to a more simpler case of SIMO first.

One other secondary problem that we are working on is the overhead reduction technique for downlink CSI compression. Higher DL data rates are achieved by effective implementation of spatial multiplexing and beamforming which is subject to availability of DL Channel State Information (CSI) at the base station. For Frequency Division Duplexing (FDD) systems, the DL CSI has to be transmitted by the User Equipment (UE) to the gNB and it constitutes a significant overhead which scales with the number of transmitter antennas and the granularity of the CSI. In this work, we investigate a more traditional dimensionality reduction method that uses Principal Component Analysis (PCA) and therefore does not suffer from the challenges posed by neural networks like CSINet and EVCSINet.