| Relaying in wireless networks wherein information is transferred through multiple hops and routes offers significant throughput enhancement and is widely considered as a promising candidate for next generation wireless networks. Relaying provides the performance advantages of a multi-antenna system without actually requiring the nodes to have multiple antennas. This thesis is devoted to the design and analysis of memoryless relay networks. A memoryless relay network is a relay network where, during each channel use, the signal transmitted by a relay depends only on the last received symbol at that relay. The key to increase throughput in a wireless relay network is to treat the entire network as a channel rather than viewing them as a set of links. Memoryless relaying enables this concept. Further, from a practical standpoint, memoryless relaying is very attractive since the complexity and cost of relaying is minimal for memoryless relay networks.;The thesis consists of five main results. First, for a two-hop single relay network, we show that estimate and forward (EF) where the relay transmits a scaled version of the minimum mean squared error (MMSE) estimate is SNR optimal for all modulation schemes and capacity optimal for binary-phase shift keying. We demonstrate that EF performs an optimal tradeoff between soft information and transmit power, and is well suited to both parallel and serial relay networks. Second, for a multiuser two-hop parallel network with amplify and forward (AF) relays, we establish a duality relationship between multiple access (MAC) and broadcast channel (BC). We show that the capacity region of a MAC with total user power PA and total relay power PB is equal to the capacity region of its reciprocal BC with base station power PB and total relay power PA. The duality extends to nodes with multiple antennas and more than two hops as well. Third, we show that, under a total network power constraint MAC-BC duality with proportional relay gains hold only for two cases. Fourth, we derive optimum relay strategies for a two-hop AF relay network with noise correlation, which may be due to common interference or noise propagation from preceding hops. Further, we show that noise correlation does not hurt even if the relays do not know the correlation structure. Fifth, we consider the relay optimization problem in a single user three-hop relay network. Based on the duality and noise correlation results, we construct an iterative algorithm that increases the capacity of the three-hop relay network during each iteration till convergence. Finally we conclude with examples of how memoryless relays can enable effective interference management in interference relay channels and two-way relay networks. |
