Dynamic resource allocation in wireless fading channels

In mobile wireless networks, dynamic allocation of resources such as transmit power, bitrates, bandwidth, and antenna beams based on the channel-state information of mobile users is the general strategy to explore the time-varying nature of the mobile environment. This dissertation looks at the problem of dynamic resource allocation (DRA) in both single-user and multi-user wireless fading channels from different information-theoretic points of view. This dissertation investigates various key issues on practical implementation of DRA in wireless systems, such as delay constraint, algorithm complexity, feedback overhead, transceiver structure, and fairness between users, among others. This dissertation demonstrates the usefulness of convex optimization techniques for solving DRA problems in wireless fading channels.; The main research contributions are summarized as follows: First, this dissertation studies the capacity limits and power-control policies for the single-user fading channel under the assumption that the channel is known perfectly at both the transmitter and the receiver. This dissertation presents a class of novel power-control policies, referred to as the multi-target (MT) power control, for approaching the fading channel capacity limits under different transmission delay constraints. By exploiting the distribution information of the fading channel, MT power control employs various statistical "water-filling" algorithms for dynamic power and rate allocation. Consequently, MT power control reduces substantially the complexity of optimal policies that rely on instantaneous channel knowledge. Second, this dissertation considers the single-user fading channel for a more practical scenario where the channel is known perfectly at the receiver and only partial channel knowledge is sent back to the transmitter through a limited-rate feedback channel. For the multiple-input and multiple-output orthogonal frequency-division multiplexing (MIMO-OFDM) channel, this dissertation presents a closed-loop extension of the well-known vertical Bell Labs layered space-time (V-BLAST) architecture whereby the receiver jointly optimizes the power and rate assignments for transmit antennas and then returns them to the transmitter. Efficient algorithms for optimal feedback parameters of closed-loop V-BLAST are presented. Third, this dissertation investigates the information-theoretic limits of multi-user fading multiple-access channel (MAC) where multiple mobile users transmit simultaneously to a common receiver located at a bases station. Based on the users' channels at each fading state, the base station jointly optimizes the transmit powers, rates and signal covariance matrices for all users and then returns them to each corresponding transmitter. There are two commonly adopted measures for information-theoretic limits of multi-user communications networks, namely, capacity region and power region. A capacity region is the set of all achievable user rates given their individual power constraints, while the power region is the collection of all user powers under which a given set of rates is achievable. While characterization of capacity region for Gaussian MAC is thoroughly known, characterization of its power region is not yet fully understood. This dissertation presents the solution to this problem by drawing an interesting dual relationship between power and capacity regions of Gaussian MAC. Applications of power region to achieve efficient and fair power consumption among mobile users are presented.