This dissertation recommends system engineering designs that implement the latest technologies in OFDMA cellular and femtocellular networks, specifically in the area of resource allocation and interference coordination. These recommended designs guarantee good user experience for time-sensitive applications such as streaming video. While throughput is often the metric used to benchmark a system, field performance requires the system also guarantees a maximum service latency to satisfy users. This dissertation provides both intuitive and low-overhead schemes that are robust and practical for implementation. The novelty of this work is the application of stochastic control techniques that guarantee the Quality of Service (QoS) through proper buffer management. Guarantee of a non-empty user buffer for streaming applications prevents service interruption. The thesis considers both centralized and distributed topologies that result from either a single base-station serving many users, or many femtocell base-stations each serving a single user, respectively. This dissertation provides insight and solutions to the following question: Under the constraints of buffer management, how does a system engineer determine the transmission scheme, resource allocation algorithms, transmitter coordination, user feedback, and achievable QoS guarantees that maximize efficiency. A combination of theory, heuristics motivated in theory, and numerical simulations will justify the presented methods.