Self-organizing Mobile Sensor Network: Distributed Topology Control Framework

1. Introduction

Today the diversities and capabilities of mobile network-enabled systems (e.g., mobile robotic sensors) have been increasing steadily. With the emergence of Ambient Intelligence (AmI) (Remagnino & Foresti, 2005) and the growth of robotic technology, much attention has been paid to increase the availability and use of large-scale robotic sensors with relatively simple capability in the field of robotics. These progresses based on mobile robotic systems have made intelligent environments adopting AmI (i.e., AmI environments) possible, and are gradually becoming more and more pervasive (Dressler, 2007). The robotic sensors allowed to disperse themselves into an unknown area of interest can be used for a wide variety of applications, such as disaster relief searching, environmental or habitat monitoring, surveillance, investigation of water and sewage pipes, etc (Cortes et al., 2004; Choset, 2001; Zecca1 et al., 2009).

Could we provide our societies and many users with smart mobile robotic sensors deploying autonomously that would satisfy their needs, and possibly establish AmI environments composed of those sensors? Technologies to achieve this are already at hand. Advances in building AmI environments have been already achieved for Java programming, for XML (eXtensible Markup Language) documents, and for wireless communications such as WiFi, Bluetooth, RFID (Radio Frequency IDentification), etc. However, such advances are not enough to disperse the mobile robotic sensors into an unknown area of interest, to fill the area, and to establish AmI environmental infrastructures. What we lack is a self-organizable coordination of autonomous robotic sensors that have the capacity to address problems they encounter. This scenario is closely related to the concept of AmI, which hypothesizes a future where people are surrounded by smart mobile sensors. The scenario is not restricted to daily life environment such as homes, schools, offices, or stores. In order to allow a better support to societies and users, AmI environments can be expanded to include airport facilities, sewage disposal plants, seashores, the Polar Regions, the moon, etc. The ideal scenario we imagine in this chapter considers a nearby future where mobile sensors autonomously establish smart environments, in the sense that they are able to be aware of the needs of societies or users and to provide desirable services satisfying the needs.

The mobile robotic sensors are regarded as a swarm of wireless sensors mounted on spatially distributed autonomous mobile robots (Sahin, 2005). To enable mobile robotic sensors to successfully perform the aforementioned tasks, it is usually required to configure their swarm network and/or maintain their network adapting to geographically constrained environments. To make the scenario of AmI environment, a reality, our goal is to develop the computational basis of unified coordination for supporting various purpose applications of mobile robotic sensors running the same algorithm. For their purpose, this chapter presents the deployment control architecture and algorithms needed to coordinate movements of robotic sensors within their swarm. Specifically, deployment control includes such functions as self-configuration, flocking, and adapting to the environment under the self-organizable framework. Moreover, we try to delineate the requirements that such coordination should satisfy.

The rest of this chapter is organized as follows. In Section 2 the state-of-the-art is summarized and commented in order to provide potential readers with the current trends for the coordination of mobile robot swarms. Section 3 presents the computational robot model and the formal definitions of problems to address. As a solution approach to the unified self-organizable framework, Section 4 describes the local interactions among three neighboring robots and their convergence property. In Sections 5 and 6, based on their framework, we present two kinds of coordination approaches, the adaptive self-configuration and adaptive flocking algorithms, and show the results of simulations. Conclusions are drawn in Section 7.