The past decade has witnessed a remarkable interest in wireless ad hoc networks that operate without any infrastructure support. The potential for deployment of such networks exists in many scenarios ranging from civil and construction engineering, disaster relief to sensor networks and military applications. The IEEE 802.11 Distributed Coordination Function (DCF) is the dominant MAC (layer) protocol for wireless ad hoc networks. Though the DCF scheme has survived the test of time, it still has its characteristic flaws namely the hidden and exposed terminal problems. Various protocols have been suggested to address these problems thereby leading to an overall increase in network throughput. Some of the popular protocols involve adjusting the carrier sensing threshold, replacing the omni directional antennas with directional antennas and intelligent calibration of transmission power. Studies have shown that directional antennas offer numerous advantages over omni directional antennas and thus has the potential to enhance the performance of ad hoc networks. For one, directional antennas have the ability to streamline their power in the direction of the receiver thereby offering longer transmission and reception ranges for the same amount of power. MAC layer protocols specifically designed to support such directional transmission and reception has been a current active area of research. Most of these protocols consider directional transmission under the assumption that the communication range of omni and directional antennas are the same. Apparently this assumption is not supported in any realistic situation. Moreover, the scientific community has not researched in depth the effect of topology on power consumption in ad hoc networks. In this thesis, we propose a cross-layer power optimization scheme for wireless ad hoc networks. The novelty of our work lies in the fact that we exploit the knowledge of the network topology to dynamically calibrate transmission power. The information about the network topology is retrieved from the Network Layer. Specifically, we consider three different network topologies -- (i) a sparse topology, (ii) a dense topology and a (iii) clusterbased topology. The MAC layer implements a dynamic power control scheme which regulates the transmission power of a transmitting node based on the node density at the receiver's end. In addition a transmitting node can adjust its own transmission power to the optimal level if the SINR value at the receiving node is known. Our cross-layer power control algorithm takes the above fact into account. Extensive simulation in NS2 shows that in all the three topologies our cross layer power control scheme (use of directional antennas with power control) leads to prolonged network lifetime and significantly increases the system throughput.