Design and performance analysis of quadratic-form receivers for fading channels

Abstract

Quadratic-form receivers (QFRs), which have quadratic-form decision metrics, are commonly used in various detections for fading channels. As an important type of QFRs, a quadratic receiver (QR) is employed in unitary space-time modulation (USTM) when sending additional training signals to estimate the channel gains is infeasible. In this dissertation, we first derive some tight bounds on the pairwise error probability of the QR, and discuss their implications to constellation design. Then to bridge the performance gap between the QR and the coherent receiver, we design three generalized quadratic receivers to incorporate channel estimation in USTM without the help of additional training signals. Since the first-order and the generalized Marcum Q-functions arise very often in the performance analysis of QFRs, we also derive some new representations and bounds for these Q-functions by using a new geometric approach, and apply the results obtained to the performance analysis of QFRs over fading channels.