Flexible Spectrum Assignment for Local Wireless Networks

摘要: In this dissertation, we consider the problem of assigning spectrum to wireless local-area networks (WLANs). line with recent IEEE 802.11 amendments and newer hardware capabilities, situations where nodes have ability adapt not only their channel center-frequency but also width. This capability brings an important additional degree freedom, which adds more granularity potentially enables efficient assignments. However, it comes new challenges; when consuming a varying amount spectrum, should seek reduce interference, they efficiently fill available spectrum. Furthermore, performances obtained in practice are especially difficult predict employ variable bandwidths. We first propose algorithm that acts decentralized way, no communication among neighboring access points (APs). Despite being decentralized, is self-organizing solves explicit tradeoff between interference mitigation usage. order for APs continuously conditions while using one network interface, relies on kind measurement, during monitor surrounding short durations. implement testbed observe drastic performance gains compared default assignments, or assignments fixed Next, procedure explicitly achievable practice, operate arbitrary configurations, traffic intensities, transmit powers, etc. notoriously difficult, as requires capturing several complex interactions take place at MAC PHY layers. Rather than trying find model acting level generality, explore different point design space. Using limited number real-world measurements, use supervised machine-learning techniques learn implicit models. these models largely outperform other measurement-based based SINR, perform well, even used contexts very from context prevailing initial set measurements learning. then build second uses above-mentioned learned assign distributed collaborative, meaning exchange control traffic. It utility-optimal -- feature enabled both by presence predicting collaboratively decisions. testbed, simple scheme discover themselves collaboration wired backbone network. possible effectively gear towards objectives (in terms efficiency and/or fairness), depending utility functions optimized nodes. Finally, study scheduling packets time frequency domains. Such ways been made progress system design, make dynamically tune negotiate band [...]