Wireless ad hoc networks have been an important area of research since past decade. They are self-organized, rapidly deployable and require no fixed infrastructure. They are quickly becoming an important part of our modern communication systems with a wide range of applications like personal communications, gaming, military, disaster relief operations, environment monitoring and more. Transmission of data like voice and video has become increasingly popular for such networks. These applications require strict end-to-end delay constraints on data delivery, requiring speedy transmissions by wireless nodes. Various transmission schemes have been proposed to deal with the delay parameter by implementing smart routing techniques, using more power to guarantee delivery, etc. But for battery powered wireless nodes, minimizing energy consumption is of supreme importance. In absence of strategic transmission schemes the energy consumption of such nodes transmitting delay critical data is very high. Minimizing energy consumption without affecting communication activities is crucial to prolong the lifetime and improve robustness of the network. The power-delay trade-off is hence an important factor in designing the transmission schemes for such networks. Most of the techniques proposed to deal with this trade-off follow the layered approach of the OSI model, which was primarily created with a wired architecture in mind. Characteristics such as absence of a central authority, links subject to multiple access, fading, interference and limited power availability exacerbate the variance of network conditions and bring greater challenges for multimedia delivery. This situation causes interdependencies across the network layers, unlike their wire-line counterparts. Recent studies indicate that power management for ad-hoc networks can be significantly improved using a cross-layer design in which different layers of the OSI stack are jointly optimized for performance. Our approach discusses a cross layer collaboration achieved by a synergy between Application, MAC and PHY layers to serve delay critical applications energy efficiently. We propose a data transmission scheme based upon the channel condition, the power constraint of an individual node and the delay constraint of the data that is being transmitted. We formulate an optimization problem and solve it using dynamic programming techniques for calculating the optimal transmission policy. We provide Matlab simulation results along with simulations performed on the Ns2 network simulator to study the energy-delay tradeoffs in a delay constrained ad hoc network environment. Additionally, we compare the benefits of our transmission policy to a scheme without the benefits of cross-layer network parameter information.