UC San Diego

The popularity of wireless access will cause wireless ad- hoc networks to operate with higher node densities and increased levels of interference. Interference mitigation is therefore crucial in ensuring these networks operate efficiently. Often the lack of network planning and regulations for such networks require the targeted access strategy to be adaptive to network conditions and distributed. In this dissertation, we develop analytical models for a random ad-hoc wireless network using physical carrier sensing. The interference at any node aggregates contributions from all concurrent transmissions on the infinite plane, and is modeled as a shot noise process. A fixed-point equation for finding the intensity of actual transmission is formulated. Subsequently, the interference at the intended destination is defined as a conditional shot noise process given that the source transmits a packet. This modeled the inhibitory effect of nearby neighbors. Subsequently, we show that our model matches reasonably well with simulations. We show that network throughput is a function of the ratio of idle threshold to transmit power, as well as the ratio of packet detection threshold to transmit power. Transmit power is in fact a scaling factor for network throughput, i.e. we need not consider transmit power as an independent parameter to be optimized. Hence, we focus on the joint adaptation of idle and packet detection threshold in order to maximize network throughput. We found that the optimal target SINR is much smaller than what is currently used in practice. We also found the optimal success probability is smaller than one, which is in contrast to commonly made assumptions in the literature. Both conditions are dependent on the intensity of transmission attempts. We develop a heuristic to jointly adapt idle and packet detection threshold, and an algorithm that results in maximum network throughput. Physical carrier sensing is shown to be an effective medium access strategy for a secondary network, in the context of a cognitive radio network scenario. With simple enhancements, the proposed two-threshold protocol accounts for the degradation on each primary device due to the aggregate interference from secondary network. Using this, we show spectral efficiency can be significantly improved