Abstract
In mobile wireless networks, dynamic allocation of resources such as transmit powers, bit-rates and antenna beams based on the channel state information of mobile users is known to be the general strategy to explore the time-varying nature of the mobile environment. This talk looks at the problem of optimal resource allocation in wireless networks from an information-theoretic point of view. In particular, the multiple-access channel (MAC) with additive Gaussian noise and multiple transmit and receive antennas - or the so-called Gaussian multiple-input multiple-output MAC (MIMO-MAC) - is focused. There are two commonly adopted measures for information-theoretic limits of multiuser communications networks, namely, multiuser capacity region and multiuser power region. Capacity region is defined as the convex-hull of the union of all achievable rates for the users given their individual power constraints; and power region consists of all power-tuples of users with each, a given set of rate constraints is achievable. While characterization of capacity region for the Gaussian MIMO-MAC is thoroughly known in literature, characterization of its power region still remains not fully understood.

This talk provides a complete characterization of the power region for the Gaussian MIMO-MAC, with and without fading. Motivated by novel dual relationship between power region and capacity region, the characterization of power region is implemented via the Lagrange duality from convex optimization theory. Various means for characterizing the power region and their related applications in multiuser communications networks are presented. Practical constraints on delay requirements, feedback quality and receiver complexity are also addressed, and useful and insightful guidance is drawn for the design and analysis of wireless communications networks.