AbstractA cognitive radio (CR) is a transceiver that is aware, adaptive, and capable of learning. It may be visualized and realized as an intelligent software package (cognitive engine) controlling a software defined radio platform. The cognitive engine executes a set of nested loops constituting a cognition cycle, drawing on experience and stored knowledge to optimize a set of user‐chosen quality of service measures. The process relies on the effectiveness of a set of tools that are individually well known but rarely found together: multi‐objective genetic algorithms, case‐based decision theory, and neural networks. Practical implementation problems include passing environmental information from the radio to the cognitive engine, acting on that information, and performing real‐time control of the radio platform by the cognitive engine. In this paper, we discuss our approach to developing and implementing a CR. Copyright © 2007 John Wiley & Sons, Ltd.