The last decade has witnessed growing interest in dynamic spectrum access, which is motivated by the observation that a large portion of the radio spectrum has been licensed but remains highly underutilized while a few small unlicensed bands that are open to anyone are getting more crowded due to the explosive expansion of wireless services. To provide more flexible access to radio spectrum, dynamic spectrum access is introduced to enable unlicensed users to opportunistically utilize vacant spectrum chunks (known as spectrum holes) in licensed frequency bands.In dynamic spectrum access, non-contiguous orthogonal frequency division multiplexing (NC-OFDM) is widely adopted to efficiently utilize fragmented spectrum because it is convenient to keep silent on some spectrum fragments to avoid interference with licensed users. In NC-OFDM, a band of spectrum is divided into many orthogonal subcarriers and data are transmitted on a subset of them simultaneously. The subcarriers that interfere with the licensed users are deactivated. Because each subcarrier can be managed independently, this dissertation introduces a series of techniques that exploit the subcarriers to address problems in dynamic spectrum access.When unlicensed users called secondary users (SUs) are granted the permission to operate in the licensed bands, they must ensure that the interference caused by them to licensed users known as primary users (PUs) is within a limit. Even without such a requirement, SUs should avoid as many collisions as possible. To improve spectrum hole extraction rate and reduce collision rate, we propose a spectrum occupancy prediction model that helps estimate the spectrum availability. It measures a wide band of spectrum with OFDM and groups subcarriers to subchannels based on spectrum use activities. In each subchannel, frequent spectrum occupancy patterns are identified and used to predict future channel states (i.e., busy or idle). With the prediction, SUs are able to make use of spectrum holes more aggressively without introducing undue interference to PUs.In the spectrum holes discovered above, a mechanism is needed to coordinate medium access between devices. Because devices opportunistically utilize spectrum holes, a device may experience severe contentions with devices from various applications. We propose a collision detection and bitwise arbitration (CDBA) mechanism that quickly identifies the winner in a contention using combined information from both the time domain and the frequency domain. It enables collision detection in the frequency domain by selectively deactivating subcarriers at each transmitter.The CDBA assumes that all devices adopt the same channel width, but different radio technologies have different requirements on channel width. When heterogeneous radios coexist in a contention domain, wideband devices can hardly win medium access opportunities in contention with narrowband devices. To address the problem, we propose an adaptive channel bonding protocol in which a wideband device initiates transmission as long as there exist some idle narrow channels and gradually increases channel width during transmission whenever new narrow channels become available. To increase the chance to win some narrow channels, a wideband device contends on subcarriers of each narrow channel with a different priority.After the contention problem is addressed, we study how to increase the transmission speed when a device is granted the permission to transmit. As wireless networks move toward wider channel widths, it is common that different subcarriers experience different fade. To cope with the frequency-selective fading, modulation scheme for each subcarrier should be selected based on the subcarrier channel quality. We exploit the modulation diversity in our modulation scheme coding to improve network throughput.Besides unicast, broadcast is another fundamental mechanism in wireless networks. Because devices utilize spectrum fragments opportunistically, different receivers may have different vacant spectrum fragments at different locations. The broadcast is more challenging in dynamic spectrum access because the transmitter needs to consider the diversity of spectrum availability. We propose a spectrum fragment agile broadcast (SFAB) protocol to support broadcast under nonuniform spectrum availability. It encodes unique sequences on subcarriers of each spectrum fragment to achieve fast spectrum agreement between the transmitter and the receivers.
Read
