Signal Processing for Cognitive Radios

16MACHINE LEARNING IN COGNITIVE RADIOS

16.1 INTRODUCTION

In Chapter 2, we defined a cognitive radio as a software-defined radio-based communications device that is cognitive and intelligent. Being cognitive requires the radio to be able to acquire knowledge and self-comprehend. To be considered as intelligent, it must not only be able to acquire knowledge and comprehend but also be able to apply acquired knowledge. Hence, acquiring knowledge and comprehension is at the heart of a cognitive radio’s identity. However, acquiring knowledge and comprehension is learning. In the case of cognition, such learning necessarily needs to include conscious self-learning, while in the case of intelligence, it can be any form of learning. Regardless, it is the ability of learning that is at the heart of a cognitive radio.

Referring to Figure 2.2, learning is the process that converts information into actual knowledge. An entity that has the ability to gather information will not be an intelligent device if it does not have the ability to convert that information into knowledge (i.e., learn from information) and then apply that knowledge in making reasoned decisions, as shown in Figure 2.2. As such, a cognitive radio is expected to learn knowledge by processing information gathered through spectrum sensing and by executing actions.

The basic operation of a cognitive radio can be abstracted in a so-called cognitive cycle that highlights this process as shown in Figure 16.1: Already learned ...