Coordinating the Interplay between Physical Carrier Sense and Power Control in CSMA/CA Wireless Networks

The transmit power and the carrier sense threshold are two major MAC/PHY parameters in CSMA/CA wireless networks. The problem of transmit power control has been extensively studied in the context of graph-theoretic topology control and maintenance. What has not been studied in depth is the effect of signal to interference plus noise ratio (SINR) on the topology and the network capacity sustained under the physical model. In order to remedy the deficiency of conventional topology control (as a result of neglecting the physical SINR effect) and to further improve network performance, we present a joint control framework for determining the transmit power control and the carrier sense threshold. We identify that the joint control problem can be decoupled into two sub-problems: conventional topology control and dynamic control of carrier sense threshold. We then derive a desirable operating condition for the carrier sense threshold. We show that there exists an abrupt increase in the collision probability as the carrier sense threshold increases, and that this transition can be efficiently identified by observing the collision probability. Thus, to maximize the node throughput, each node may increase the carrier sense threshold as long as the collision probability is below a reasonable threshold. Based on the insight shed from our analysis, we devise a dynamic tuning algorithm for carrier sense thresholds, by which each node observes its local event (i.e., transmission collisions) and drives its carrier sense threshold towards the desirable operating point in a fully distributed manner. To demonstrate the utility of the proposed joint control framework, we equip a localized topology control algorithm, localized minimal spanning tree (LMST), with the capability of dynamic carrier sense adaptation (DCSA). Via simulation, we show that joint control of transmit power and carrier sense threshold is crucial for improving the network performance and that the resulting control algorithm LMST-DCSA significantly outperforms LMST alone.