CAPACITY BOUNDS AND SCALING LAWS OF WIRELESS RELAY NETWORKS by

Driven by the fast-growing demand on wireless applications, cooperative relaying has recently received much attention. In this dissertation, cooperative relaying in MIMO systems and sensor networks is studied extensively, with focus on capacity bounds and scaling laws. The first thrust is on the capacity bounds of MIMO relay channels. First, a Gaussian MIMO relay channel model with fixed channel conditions is considered, and the corresponding upper bounds and lower bounds on the capacity are derived, which can be evaluated numerically. Next, an upper bound and a lower bound on the ergodic capacity are found when the channel links undergo Rayleigh fading. It is shown that the upper bound can meet the lower bound under certain conditions, indicating that the capacity can be characterized. Sufficient conditions for achieving the ergodic capacity are also investigated. The second thrust considers a wireless sensory relay network model that consists of one source node, one destination node and relay nodes. Different from existing work, the focus here is on the more realistic case where the relay nodes have no a priori knowledge of channel state information (CSI) for both the backward channels and the forward channels. An amplify-and-forward (AF) with network training relaying scheme is devised for cooperative relaying. The achievable rates and the scaling laws are characterized accordingly, in the joint asymptotic regime of the number of relay nodes n, the channel coherence interval L, and the bandwidth W . Due to frequency-selective fading, power allocation across the frequency subbands at relay nodes plays a critical role in the achievable rates. It is revealed that the scaling law can be characterized, if L/W is bounded below and W is sub-linear in n.