Cognitive Radios are able to dynamically use free spectrum in their vicinity, avoiding interference and congestion and thus providing robust communication links. Cognitive Radio Networks go one step further, as they do not only consider the link to the next node but focus on end-to-end optimization. Dynamic adaptations of the whole protocol stack are required, especially on the network layer. This paper analyzes networking technologies regarding their support for end-to-end optimization in tactical environments and proposes enhancements. Based on the findings, an architecture framework for Cognitive Radio Networks is proposed.