Precoder Matrix Detection: Description Primary User Cooperative Mobile Relay eNodeB Aim: Reception of MIMO signals by a secondary receiver that acts as a relay Detect unknown precoder CLSM relaying
Completed Work Aim: Design of receivers for LTE systems without PMI information at Cooperative Mobile Relay (CMR) Assumption: Single cell scenario, no interference Algorithms developed: Hypothesis testing Framework 1. Simplified Maximum Likelihood (ML) algorithm 2. Cluster variance algorithm
Completed Work
Massive MIMO
SNR needed at CMR vs. SNR at PU for 10% BLER 1 PU simulated using “ETU” channel model, moving at 60 Km/h with 2 receive antennas. 2 transmit antennas at eNodeB. CMR uses “EPA” channel model, is stationary. CMR has either 2 or 4 receive antennas in two scenarios. With added antennas at CMR, much more performance gain observed.
Precoder Detection using Temporal Correlations Simple ML and cluster variance algorithm neglect temporal correlations Additional performance gain by exploiting the correlations Aim: Identify scenarios with significant temporal correlations Propose a suitable model to represent the system ─ Verify the choice of model
Scenarios with Significant Correlations
Hidden Markov Model for Temporal Correlations We consider a hidden Markov model (HMM) Each state of HMM is given by a specific precoder matrix Observations: Set of log-likelihood ratios for each precoder, given received symbols and channel estimate ─ Depends only on current state Use HMM to compute conditional distribution of current precoder (i.e. state) given past observations and/or state Use above conditional distribution in a MAP decision rule
Validation of HMM for Temporal Correlations
Future Work