Queueing analysis for cross-layer design with adaptive modulation and coding

Abstract

With the development of wireless networks, Quality of Service (QoS) has become one of the most important mechanisms to improve the system performance such as loss, delay and throughput. Cross-layer design is seen as one of the main approaches to achieve QoS provisioned services in contrast to the well-adopted TCP/IP network model. This thesis focuses on the cross-layer design incorporating queueing effects and adaptive modulation and coding (AMC), which operates at both the data-link layer and the physical layer, to obtain the performance analyses on loss, delay and throughput using the matrix geometric method. More specifically, this thesis explores the potential to extend the cross-layer analysis, at the data-link and the physical layer respectively. At the data-link layer, since the traffic types such as voice, video and data are proven to be bursty, and the well-adopted Poisson arrivals fail to capture the burstiness of such traffic types, the bursty traffic models including ON-OFF and aggregated ON-OFF arrivals are introduced in the cross-layer analysis. This thesis investigates the impact of traffic models on performance analysis, identifying the importance of choosing the proper traffic model for cross-layer analysis. At the physical layer, IEEE 802.11ac standard is adopted for the cross-layer analysis. In order to meet the specifications of 802.11ac with higher-order Modulation and Coding Schemes (MCS), wider channel bandwidth and more spatial streams, the Signal-to-Noise Ratio (SNR) thresholds are re-determined for the AMC; in addition, a single user (SU) multiple in multiple out (MIMO) spatial multiplexing system with zero-forcing (ZF) detector is adopted for the cross-layer analysis. Furthermore, this thesis explores the impact of antenna correlations on the system performance. All of the work done in this thesis aims at obtaining more practical performance analysis on the cross-layer design incorporating queueing effects and AMC. The proposed cross-layer analysis is quite general, so that it’s ready to be applied to any QoS provisioned networks.