Network control architectures in wireless communication and mobile computing: Power control and quality of service issues.

摘要

Network control is key for efficient wireless networking and mobile computing. This thesis formulates and investigates selected core models of wireless network control elements, advancing the state of the art in this admittedly very broad subject. In particular, joint control opportunities across multiple network layers are identified, modeled and analyzed, leading to the design of efficient control algorithms.; Beginning with plain transmitter power control, novel integrated schemes are introduced and investigated, resolving the coexistence problem of QoS-protected streaming traffic and bursty delay-tolerant data traffic, sharing a common wireless channel.; The problem of joint control of transmitter power and relay switching is formulated and studied for a two-hop wireless packet communication. The core issue is whether intermediate relay nodes should receive or transmit at each time slot and what power level transmitters should use to efficiently forward packet traffic.; Data prefetching and caching over fluctuating wireless channels is then studied jointly with transmitter power control. The main issue is the trade-off between data access latency at the receiver and interference stress induced on the wireless channel by the transmission power. The introduced problem formulation couples receiver buffer control (application layer) with transmitter power control (MAC layer).; The problem of joint terminal-to-base task migration (over the wireless channel) and terminal processor speed management is then formulated and studied for reduced energy drain on battery-limited devices. Terminal computation tasks can potentially be uploaded to a base station server, get executed there fast, and have the results downloaded back to the terminal. The risk is that the 'soft' connectivity induced by the wireless channel (due to mobility, etc.) may increase the resulting delivery latency.; The focus is on formulating simple---yet insightful---models that capture fundamental performance trade-offs and leverage powerful control and optimization methodologies for exploring the nature and structure of core optimal network controls. Based on the latter, appropriate approximations and justified heuristics are introduced for complexity reduction and design of efficient operational algorithms.