Answers are grounded in this paper and cite their sources.
We give reciprocity results for the channel capacity efficiency (i.e. the capacity/bandwidth – abbreviated here to capacity) of an N-transmit M-receive antenna wireless system, when the gain matrix is known to the receiver but not the transmitter. The channel may be non-Gaussian. Transmitter power or receiver noise may be variable. We consider the case when N or M is large but not both. Formulas are also given for the mean capacity when there is correlation at both ends. Capacity is approximately Gaussian. For a non-Gaussian channel the approximate mean capacity is not changed but the variance is increased by a certain factor. It is noted that reciprocity fails. The key to making it work is to introduce a ‘power factor’ that allows for the case when the power of each transmitter does not depend on N as well as for the usual case when the total power is fixed.
No takes yet. Share an insight, caveat, or question.
Withers et al. (2011) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: