A transmitter antenna array has the ability to direct data simultaneously to multiple
receivers within a wireless network, creating potential for a more integrated view of
algorithmic system components. In this thesis, such a perspective informs the design
of two system tasks: the scheduling of packets from a number of data streams into
groups; and the subsequent spatial multiplexing and encoding of these groups using
array processing. We demonstrate how good system designs can help these two tasks
reinforce one another, or alternatively enable tradeoffs in complexity between the two.
Moreover, scheduling and array processing each benefit from a further awareness of
both the fading channel state and certain properties of the data, providing information
about key flexibilities, constraints and goals.
Our development focuses on techniques that lead to high performance even with
very low-complexity receivers. We first consider spatial precoding under simple
scheduling and propose several extensions for implementation, such as a unified timedomain
precoder that compensates for both cross-channel and intersymbol interference.
We then show how more sophisticated, channel-aware scheduling can reduce the
complexity requirements of the array processing. The scheduling algorithms presented
are based on the receivers’ fading channel realizations and the delay tolerances of the
data streams. Finally, we address the multicasting of common data streams in terms
of opportunities for reduced redundancy as well as the conflicting objectives inherent
in sending to multiple receivers. Our channel-aware extensions of space-time codes for
multicasting gain several dB over traditional versions that do not incorporate channel
knowledge.
Grant number: CCR-9979363
NSF, HP/MIT Alliance,