Multi-tier wireless systems and cognitive radios.

The emergence and popularity of wireless data services require communication technologies and designs that increase system throughput while promoting efficient spectrum usage. This thesis studies two such emerging wireless technologies: tiered network architectures and cognitive radios.. Tiered wireless networks contain radios whose coverage areas can vary in their order of magnitude. The higher-tier radios have large coverage and may extend the reach of the overall network while the lower-tier radios focus on local traffic in hot spots. We study a two-tier cellular CDMA system in which a lower-tier base station, called a data access point (DAP), attracts only a small number of users and gives them high-speed data access while the umbrella higher-tier base station, called a macrocell, serves multiple simultaneous low-rate users, such as voice users. The DAP operates on the same frequency as the macrocell, implying that both cells experience cross-tier interference . The DAP users may adapt their spreading factor in accordance with interference conditions to achieve high-data rate, whereas the macrocell users have a fixed data rate. Three access schemes are studied for the DAP users: (1) the optimal scheme, which maximizes the total throughput at the expense of fairness; (2) the round robin scheme, which assures one slot per frame access; and (3) the proportional fair scheme, which takes advantage of channel fading to achieve higher throughput and fairness simultaneously. In practice, a macrocell may cover multiple hot spots, therefore, multiple DAP bases can be embedded in a macrocell coverage area. We thus extend the single-DAP system to a multi-DAP system. The access method for such system is essentially a cross-tier interference management scheme, in which those DAPs compete with each other, maintaining an adequate cross-tier interference level at the macro-cell base. The optimal access method for such multi-DAP system is similar to the single-DAP case: the user with the least cross-tier interference in each DAP cell is granted access. Cognitive radios can adapt their communication format and communicate with each other (or other radios) using unused frequency bands within a large spectrum. They autonomously detect the presence of primary users. That is, as long as authorized users of a spectrum band are inactive, the cognitive radios may use the spectrum to transmit. However, since there are uncertainties in the wireless channels, a node may be in a deep fade and cannot reliably detect primary users on its own. To improve the reliability of spectrum sensing, we study distributed collaborative sensing algorithms. Particularly, we study consensus algorithms that aid cognitive radios identify unused spectrum bands. In this algorithm, all users exchange their opinions until they reach a mutual agreement. As the final agreement combines multiple observations, the reliability improves. When the network is in good connection, this algorithm performs well. However, when the communication links experience errors, the algorithms perform quite differently. We study the convergence behavior of the consensus algorithm with errors. As networks are often tiered (for example, clustered sensor network), the conventional consensus algorithms need to incorporate such tiered architecture. Therefore, we propose two hierarchical consensus algorithms to let all nodes achieve consensus in a distributed manner. The hierarchical equal neighbor consensus algorithm aims at the simplest implementation to reach agreement, while the hierarchical average consensus algorithm aims at reaching the exact average consensus. The hierarchical consensus algorithms...