Lanterna Rally

Author: Sheu-Sheu Tan

Publisher:

Published: 2013

Total Pages: 160

ISBN-13: 9781267995414

We develop channel aware scheduling and resource allocation schemes with cross-layer optimization for several problems in multiuser wireless networks. We consider problems of distributed opportunistic scheduling, where multiple users contend to access the same set of channels. Instead of scheduling users to the earliest available idle channels, we also take the instantaneous channel quality into consideration and schedule the users only when the channel quality is sufficiently high. This can lead to significant gains in throughput compared to system where PHY and MAC layers are designed separately and the wireless fading channels are abstracted as time invariant, fixed rate channels for scheduling purposes. We first consider opportunistic spectrum access in a cognitive radio network, where a secondary user (SU) share the spectrum opportunistically with incumbent primary users (PUs). Similar to earlier works on distributed opportunistic scheduling (DOS), we maximize the throughput of SU by formulating the channel access problem as a maximum rate-of-return problem in the optimal stopping theory framework. We show that the optimal channel access strategy is a pure threshold policy, namely the SU decides to use or skip transmission opportunities by comparing the channel qualities to a fixed threshold. We further increase the spectrum utilization by interleaving SU's packets with periodic sensing to detect PU's return. We jointly optimize the rate threshold and the packet transmission time to maximize the average throughput of SU, while limiting interference to PU. Next, we develop channel-aware opportunistic spectrum access strategies in a more general cognitive radio network with multiple SUs. Here, we additionally take into account the collisions and complex interaction between SUs and sharing of resources between them. We derive strategies for both cooperative settings where SUs maximize their sum total of throughputs, as well as non-cooperative game theoretic settings, where each SU tries to maximize its own throughput. We show that the optimal schemes for both scenarios are pure threshold policies. In the non-cooperative case, we establish the existence of Nash equilibrium and develop best response strategies that can converge to equilibria, with SUs relying only on their local observations. We study the trade-off between maximal throughput in the cooperative setting and fairness in the non-cooperative setting, and schemes based on utility functions and pricing that mitigate this tradeoff. In addition to maximizing throughput and fair sharing of resources, it is important to consider network/scheduling delays for QoS performance of delay-sensitive applications. We study DOS under both network-wide and user-specific average delay constraints. We take a stochastic Lagrangian approach and characterize the corresponding optimal scheduling policies accordingly, and show that they have a pure threshold structure. Next, we consider the use of different types of channel quality information, i.e., channel state information (CSI) and channel distribution information (CDI) in the opportunistic scheduling design for MIMO ad hoc networks. CSI is highly dynamic in nature and provides time diversity in the wireless channel, but is difficult to track. CDI offers temporal stability, but is incapable of capturing the instantaneous channel conditions. We design a new class of cross-layer opportunistic channel access scheduling framework for MIMO networks where CDI is used in the network context to group the simultaneous transmission links for spatial channel access and CSI is used in the link context to decide when and which link group should transmit based on a pre designed threshold. We thereby reap the benefits of both the temporal stability of CDI and the time diversity of CSI. Finally, we consider a novel application of cross layer optimization for communication of progressive coded images over OFDM wireless fading channels. We first consider adaptive modulation based on the instantaneous channel state information. An algorithm is proposed to allocate power and constellation size at each subchannel by maximizing the throughput. We next consider both the variance and the average of the throughput when deciding the constellation size for adaptive modulation. Simulation results confirm that cross-layer optimization with adaptive modulation enhances system performance.