This work was supported by the National Science Foundation under Grants No. ECS-0300130 and CCF-0430892.
The original capacity scaling results for MIMO systems assumed spatial independence between the fading affecting various transmitter/receiver pairs. In practice, the transmitter and/or receiver is often confined to a fixed area, and hence this spatial independence assumption is often violated for large numbers of antennas. We have considered MIMO capacity scaling results in such a scenario and accurately characterized the limiting behavior for a number of cases [Wei/Goeckel/Janaswamy, ICC 2003, IEEE TWireless 2005]. As expected, the linear scaling of the capacity observed in previous work is not evident for fixed-size arrays.
We have also considered the case of space-time coding over transmit antennas that are not necessarily co-located. For example, our motivating application is "police radio", where two single-antenna base stations can act as antennas in a standard space-time coding scheme to provide diversity on the broadcast downlink to the mobiles. In such systems, strict timing synchronization of the received signals is difficult. It must be enforced by the MAC, or, in broadcast scenarios, cannot be provided in any case. We have provided codes and algorithms such that the requirement for strict timing synchronization can be relaxed for both cellular architectures [Goeckel/Hao, CISS 2003] and for general ad hoc networks [Wei/Goeckel/Valenti, CISS 2004, IEEE TWireless 2006]. Since precise timing synchronization will likely be difficult in future cooperative networks, particularly those that run "connectionless", we anticipate that these algorithms or their extensions will be present in many applications.