This dissertation, "Optimal Data Dissemination in Stochastic and Arbitrary Wireless Networks" by Hongxing, Li, 李宏兴, was obtained from The University of Hong Kong (Pokfulam, Hong Kong) and is being sold pursuant to Creative Commons: Attribution 3.0 Hong Kong License. The content of this dissertation has not been altered in any way. We have altered the formatting in order to facilitate the ease of printing and reading of the dissertation. All rights not granted by the above license are retained by the author. Abstract: Data dissemination among wireless devices is an essential application in wireless networks. In contrast to its wired counterparts which have more stable network settings, wireless networks are subject to network dynamics, such as variable network topology, channel availability and capacity, which are due to user mobility, signal collision, random channel fading and scattering, etc. Network dynamics complicate the protocol design for optimal data disseminations. Although the topic has been intensively discussed for many years, existing solutions are still not completely satisfactory, especially for stochastic or arbitrary networks. In this thesis, we address optimal data dissemination in both stochastic and arbitrary wireless networks, using techniques of Lyapunov optimization, graph theory, network coding, multi-resolution coding and successive interference cancellation. We first discuss the maximization of time-averaged throughput utility over a long run for unicast and multirate multicast, respectively, in stochastic wireless networks without probing into the future. For multi-session unicast communications, a utility-maximizing cross-layer design, composed of joint end-to-end rate control, routing, and channel allocation, is proposed for cognitive radio networks with stochastic primary user occupations. Then, we study optimal multirate multicast to receivers with non-uniform receiving rates, also making dynamic cross-layer decisions, in a general wireless network with both a timevarying topology and random channel capacities, by utilizing random linear network coding and multi-resolution coding. In both solutions, we assume users are selfish and prefer only to relay data for others with strong social ties. Such social selfishness of users is a new constraint in network protocol design. Its impact on efficient data dissemination in wireless networks is largely unstudied, especially under stochastic settings. Lyapunov optimization is applied in our protocol design achieving close-to-optimal utilities. Next, we turn to latency-minimizing data aggregation in wireless sensor networks having arbitrary network topologies under the physical interference model. Different from our effort for stochastic networks where we target at time-averaged optimality over a long run, the objective here is to minimize the time-span to accomplish one round of aggregation scheduling for all sensors in an arbitrary topology. This problem is NP-hard, involving both aggregation tree construction and collision-free link scheduling. The current literature mostly considers the protocol interference model, which has been shown to be less practical than the physical interference model in characterizing the interference relations in the real world. A distributed solution under the physical interference model is challenging since cumulative interferences from all concurrently transmitting devices need to be well measured. In this thesis, we present a distributed aggregation protocol with an improved approximation ratio as compared with previous work. We then discuss the tradeoff between aggregation latency and energy consumption for arbitrary topologies when the successive interference cancellation technique is in force. Another distributed algorithm is introduced with asymptotic optimality in both aggregation latency and latency-energy tradeoff. Through theoretical analysis and