Metrics to Decide the Feedback Interval in Closed-Loop MIMO-OFDM Systems

In a closed-loop multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) system, the receiver estimates the channel and sends updates on the channel state back to the transmitter. The issue of deciding when to send updates is addressed in this paper and two metrics for making this decision are compared. One is obtained from the correlation matrix distance (CMD) which measures how far the MIMO channel has moved over a time interval, the other is a novel approach using mutual information (MI), whereby two calculations of MI are made, the first assuming that water-filling is employed at the transmitter and the second assuming that it is not. Using CMD, updates are sent when the correlation distance becomes greater than a given threshold. Using MI, updates are sent when the channel changes to the extent that the capacity with water-filling becomes less than that without. Matlab simulation results for a 2 * 2 MIMO system show that the two metrics result in a similar overhead, both being more efficient than constant interval. Extension to larger antenna arrays should be straightforward. MI, although computationally more intensive, can extend easily to decide when to switch between spatial multiplexing (SM) and diversity modes using space-time block coding (STBC) if the channel rank collapses.