Bandwidth scavenging for device coexistence in wireless networks

The objective of this thesis is to develop a wireless channel access framework that allows secondary user (SU) devices to coexist with primary user (PU) devices via utilizing unused spectrum or whitespace found between the PUs' transmissions. Those whitespaces typically last for short durations (i.e. on the order of milliseconds) in between the active data transmissions by the PUs. The key design objectives for an SU access strategy are to "scavenge" the maximum amount of spatio-temporally fragmented whitespace while limiting the amount of disruptions caused to the primary users (PU). These conflicting goals become particularly challenging without deterministic prior knowledge about the future occurrences and durations of the whitespaces. Our approach to address this problem is to develop stochastic whitespace access mechanisms based on previously observed statistical model of the channel whitespace. The key contributions of this thesis are as follows. First, it provides an extensive statistical analysis of the whitespace characteristics using simulations and experiments on a prototype testbed, in the presence of various primary user traffic scenarios. The simulated network allows explorations of the effects of various traffic and topology conditions, and the experimental testbed provides insights into real world whitespace measurements in the presence of various hardware and operating system related limitations. The second contribution is an opportunistic access strategy for the secondary users that is developed based on the measurement and modeling of whitespace resulted from ad hoc mode 802.11 PU traffic. This opportunistic channel access, or scavenging, during ultra-short and non-deterministic 802.11 whitespace is then evaluated for functionality and performance through analytical modeling, network simulation, and testbed experiments. It is demonstrated that the proposed strategy is able to consistently scavenge above 90% of the available whitespace capacity, while keeping the primary users disruption less than 5%. The third major contribution is to generalize the above bandwidth scavenging approach by extending the proposed technique for arbitrary whitespace distributions. This generalization has resulted in a new access strategy that can be applied to non-802.11 primary traffic and is able to handle whitespace durations with multi-modal density functions. Effectiveness under arbitrary whitespace profiles is achieved by introducing a new concept of transmission opportunities within a given whitespace, and then developing a knapsack optimization based transmission probability shuffling mechanism across the transmission opportunities with the whitespace. Combining the above three components, the thesis offers a framework for SU support within a PU network with arbitrary topology and traffic profiles.