Cross-layer Resource Allocation for Multi-user Communication Systems a Dissertation Submitted to the Department of Electrical Engineering and the Committee on Graduate Studies of Stanford University in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

Satisfaction of different QoS demands for various broadband services in wireless networks requires that multi-user packet scheduling intelligently use both channel state information (CSI) and queue state information (QSI). A combination of queuechannel-aware scheduling and power/rate allocation on each transmit dimension is known as “cross-layer resource allocation”. This thesis investigates various aspects of cross-layer resource allocation to illustrate its important role in multi-user communication systems. Of particular interest are downlink and uplink wireless systems that use orthogonal frequency division multiplexing (OFDM) modulation and multi-input multi-output (MIMO) transmission with multiple antennas. There are four major contributions in this thesis: First, queue-proportional scheduling (QPS) is presented and is shown to exhibit superior throughput, delay, and fairness properties. QPS provides a capability that can arbitrarily scale each user’s average queueing delay relative to others, which makes QPS suitable for networks driven by heterogeneous traffic. Second, geometric programming (GP) is applied for cross-layer resource allocation in multi-user OFDM systems with CSI, where GP formulations lead to numerical efficiency and strong scalability. Third, efficient power/rate optimization algorithms are developed by using Lagrange dual decomposition for multiuser MIMO-OFDMA (Orthogonal Frequency Division Multiple Access) systems with CSI. Finally, cross-layer resource allocation in multi-user MIMO-OFDMA systems with channel distribution information (CDI) is addressed. It is shown that outage rate region for scheduling can be efficiently characterized by using a Gaussian approximation of mutual information along with a successive feasibility check method. This efficient approach is directly applicable to finding power/rate allocation for QPS as