Cross-layer design and analysis for wireless networks

We develop novel analytical models for radio link level protocol analysis and design for point-to-point transmissions in multi-rate wireless networks. Specifically, error control and scheduling, which are two primary components of any radio link level protocol, are analyzed. Multi-rate transmission is assumed to be achieved by the implementation of adaptive modulation and coding (AMC) in the physical layer. The delay statistics for Go-Back-N and Selective Repeat automatic repeat request (ARQ)-based error control protocols with non-zero feedback delay are derived analytically. We develop queueing models to calculate delay statistics and throughput performances for two classes of scheduling policies, namely, weighted round-robin (WRR) scheduling and opportunistic scheduling schemes. Here, multiple users are assumed to share one channel in a time multiplexing manner. The analytical model for the channel-quality-based scheduling can be applied to any scheduling schemes as long as the evolution of joint service/vacation and channel processes can be determined. As an example, we analyze the max-rate (MR) scheduling scheme. Applications of the analytical models for cross-layer design and packet-level admission control under statistical delay constraints are illustrated. The proposed analytical models provide frameworks to fairly compare different scheduling policies considering different traffic, system and channel parameters.; Besides link level protocol design and analysis in single hop wireless networks, end-to-end protocol design issues for multi-hop wireless networks pose significant research challenges. For the end-to-end transmission scenario, both single-path QoS routing and optimal multi-path routing protocols are developed. In particular, we propose both exact and approximate decomposition approaches to solve a tandem queueing problem considering the implementation of AMC in the physical layer and ARQ-error recovery in the link layer. The decomposition approach is then employed to develop a single-path QoS routing protocol incorporating all important QoS measures, namely, end-to-end bandwidth; average delay, loss rate and statistical delay requirements. For the optimal multi-path routing problem, we employ the dual decomposition approach from convex optimization to develop cross-layer optimization frameworks for multi-hop wireless networks using decode-and-forward cooperative diversity. Specifically, we propose two distributed algorithms capturing functionalities in different layers of the protocols stack, namely the relay selection and power allocation in the physical layer, routing in the network layer and congestion control in the transport layer. The convergence of the proposed algorithms and the significant gains in terms of power consumption and transmission rates due to the cooperative diversity are illustrated through typical numerical results.; The proposed analytical models and protocols in this dissertation provide important frameworks for cross-layer design and optimization in wireless networks. These frameworks exploit, different network degrees of freedom such as link adaptation, multiuser diversity, cooperative diversity, etc. They also solve research challenges due to the decentralized architecture of future wireless networks.