An Adaptive Reasoning and Learning Framework for Mobile Cognitive Radio Systems

Introduction

Today, most of us carry a portable device and use some services on wireless network over spectrum. However, the utilization of the wireless network spectrum is very low. A mechanism is thus needed to share the unused spectrum efficiently and adequately. Cognitive Radio (CR) was originally proposed to be a promising solution for the spectrum sharing problem. Joseph Mitola invented the basic CR concept in the 1990s (Mitola, 2000), (Mitola, 2001), (Mitola, 2006). The general definition of CR is a radio that can sense and adapt to its environment. For example, dynamically selecting channels to share spectrum and reuse is one of the applications of CR. To provide a better Quality of Service (QoS) such as maximizing the throughput for application over network, the design of CR may not only take into account radio aspects. Other environment related data such as geographical location, user preference, and system time should be considered because these contextual information data may actually alleviate the radio problems more efficiently.

The key idea of CR is a cognition cycle that implements the capabilities required of CR in a reactive sequence. There are six major stages of a cognition cycle as shown in Figure 1:

Figure 1.

The cognition cycle in a CR system

• •

  Observe: The CR observes its surrounding environment by parsing incoming stimulus streams such as radio and location raw data.

• •

  Orient: The Orient stage determines the significance of an observation by binding it to a previously known set of stimuli.

• •

  Plan: The Plan stage constructs corresponding models with dealt stimuli and makes plans including reasoning.

• •

  Decide: The Decide stage selects an appropriate decision among several candidate plans.

• •

  Act: The Act stage applies the most appropriate decision made in the Decide stage.

• •

  Learn: The Learn stage depends on observations, decisions, and actions. This stage learns the successful problem-solving record for the future.

In the infrastructure of cognitive radio architecture (CRA) (Mitola, 2006) that includes baseband hardware and software, digital hardware and software, and target applications, a key component is the reasoning and learning mechanism. Reasoning and learning allow radios to be not only aware and adaptive, but also cognitive. However, the diversity of technologies that must be integrated in CRA is very high, which includes geography, spectrum awareness, frequency occupancy, biometrics, time, spatial awareness, and software technologies such as policy engines, artificial intelligence (AI) techniques, signal processing, and network protocol (Fette, 2006). Thus, the complexity in reasoning and learning is too high to be handled by conventional methods. Cognitive radio (CR) calls for new methods that allow efficient reasoning and learning across multidisciplinary fields.

In this chapter we introduce the design and implementation of Adaptive Reasoning and Learning Framework (ARALF) for cognitive radios which will focus on how existing techniques in reasoning and in learning from multiple disciplines can be integrated such that intelligent choices and decisions are made efficiently. ARALF integrates two reasoning and learning methods with a tight coupling among four ontologies, namely user preference, time information, location awareness, and radio spectrum information.

The rest of this chapter is organized as follows. First, we will introduce the ARALF for cognitive radios. Second, we will introduce the sensor interface. Third, the objective-driven triggering will be introduced. Fourth, we will introduce multi-objective intellectual perceptron and case-based reasoning. Then we will introduce human machine interface, spectrum analysis and management. Final, we give evaluation and conclusion.