SensorGrid architecture for distributed event classification

Abstract

Recent advances in electronic circuit miniaturisation and micro-electromechanical systems (MEMS) have led to the creation of small sensor nodes which integrate several kinds of sensors, a central processing unit (CPU), memory and a wireless transceiver. A collection of these sensor nodes forms a sensor network which is easily deployable to provide a high degree of visibility into real-world physical processes as they happen, thus benefiting a variety of applications such as environmental monitoring, surveillance and target tracking. Some of these sensor nodes may also incorporate actuators such as buzzers and switches which can affect the environment directly. We shall simply use the generic term sensor node to refer to these sensor-actuator nodes as well. A parallel development in the technology landscape is grid computing, which is essentially the federation of heterogeneous computational servers connected by high-speed network connections. Middleware technologies such as Globus (2006) and Gridbus (2005) enable secure and convenient sharing of resources such as CPU, memory, storage, content and databases by users and applications. This has caused grid computing to be referred to as computing on tap, utility computing and IBMs mantra, on demand computing. Many countries have recognised the importance of grid computing for eScience and the grid has a number of success stories from the fields of bioinformatics, drug design, engineering design, business, manufacturing, and logistics. There is some existing work on the intersecting fields of sensor networks and grid computing which can be broadly grouped into three categories: (1) sensorwebs, (2) sensors to grid, and (3) sensor networks to grid. In the first category of sensorwebs, many different kinds of sensors are connected together through middleware to enable timely and secure access to sensor readings. Examples are the SensorNet effort by the Oak Ridge National Laboratories (ORNL) and NIST (USA) which aims to collect sensor data from different places to facilitate the operations of the emergency services; the IrisNet effort by Intel to create the seeing Internet by enabling data collection and storage, and the support of rich queries over the Internet; and the Department of Defense (USA) ForceNet which integrates many kinds of sensor data to support air, sea and land military operations. In the second category of sensors to grid, the aim is to connect sensors and instruments to the grid to facilitate collaborative scientific research and visualisation (Chiu and Frey 2005). Examples are the efforts by the Internet2 and eScience communities in areas such as the collaborative design of aircraft engines and environment monitoring; DiscoveryNet (Imperial College UK) which aims to perform knowledge discovery and air pollution monitoring; and the earthquake science efforts by the CrisisGrid team (Indiana University USA) and iSERVO (International Solid Earth Research Virtual Observatory). Finally, in the third category of sensor networks to grid, the aim is mainly to use grid web services to integrate sensor networks and enable queries on live data. Examples are efforts by Gaynor et al (2004) to facilitate quicker medical response and supply chain management; and the SPRING framework proposed by Lim et al (2005). Our focus and approach, which we refer to as sensor-grid computing executing on an integrated sensor-grid architecture or simply SensorGrid for short (Tham and Buyya 2005), see Fig. 5.1 builds on the three categories of existing work described above and aims to achieve more by exploiting the complementary strengths and characteristics of sensor networks and grid computing. Sensor-grid computing combines the real-time acquisition and processing of data about the environment by sensor networks with intensive distributed computations on the grid. This enables the construction of real-time models and databases of the environment and physical processes as they unfold, from which high-value computations such as analytics, data mining, decision-making, optimisation and prediction can be carried out to generate on-the-fly results. This powerful combination would enable, for example, effective early warning of threats (such as missiles) and natural disasters, and real-time business process optimisation. One other key aspect of sensor-grid computing is the emphasis on distributed and hierarchical in-network processing at several levels of the SensorGrid architecture. As will be explained later, the sensor-grid computing approach is more robust and scalable compared to other approaches in which computations are mainly performed on the grid itself. The sensor-grid computing approach together with the SensorGrid architecture enable more timely responses to be achieved and useful results to be available even in the presence of failures in some parts of the architecture. The organisation of this chapter is as follows. In Section 2, we describe a simple centralised approach to realise sensor-grid computing. We then point out its weaknesses and propose a distributed approach. In Section 3, we describe two applications of distributed sensor-grid computing which we have implemented. In Section 4, several challenges and research issues related to sensorgrid computing are discussed. Finally, we conclude in Section 5. © Springer-Verlag Berlin Heidelberg 2007.