Integrated Design and Process Technology, IDPT-2003 Printed in the United States of America, September, 2003 2003 Society for Design and Process Science 1 TINYDB BY REMOTE Kevin Mayer Dr Ken Taylor Kevin.Mayer@anu.edu.au Faculty of Engineering and IT The Australian National University Acton, ACT 0200, AUSTRALIA Ken.Taylor@csiro.au CSIRO Mathematical and Information Sciences ANU Campus Acton, ACT 2601, AUSTRALIA ABSTRACT Wireless ad-hoc sensor networks (WASNs) are being developed and explored at a rapid rate by the research community around the world. Very soon, there will be many such networks deployed and used in commercial and other monitoring situations. Much work has focused on the development of energy efficient routing and aggregation algorithms, as well as hardware designs of small, integrated, and low powered boards, radios and sensors – all of which are crucial to a sensor network. This paper addresses the means of connecting the sensor network to the global Internet. We discuss the use of the mobile telephone network to achieve this as it provides an easily accessible, and almost ubiquitous means of connecting to the Internet. Issues to do with such deployment are briefly discussed, including a method for providing a gateway to the network, and issues to do with having multiple gateways – both to support sparsely deployed sensor networks, and for redundancy. ACRONYMS 1XRTT = Single carrier, radio transmission technology (an enhancement to CDMA offering high speed packet-switched data) CSIRO = Commonwealth Scientific and Industrial Research Organisation (Australia’s chief public scientific research body) GPRS = General Packet Radio Service (a packet-switched data service offered over GSM) NAT = Network Address Translation (a method that allows a privately addressed IP network to access the Internet via a gateway – done in accordance with RFC1918 (Rekhter et al., 1996)) SCADA = Supervisory Control and Data Acquisition INTRODUCTION Wireless sensor networks are an active area of research with rapid changes in hardware designs and software frameworks supporting them. A sensor network is comprised of many small micro-controllers, with attached sensors, working together in a wireless, ad-hoc network, to achieve a desired goal. The fact that the individual nodes in a sensor network work together, cooperatively, to reach a desired goal is important. Each device by itself can only do the task of a basic agent, however, together, the network performs something much greater. The size of these micro-controllers is what sets them apart from more traditional networks. MICA Motes, which are commercially available (Crossbow Technology Inc., 2002) have become very popular with researchers developing sensor network solutions. It is envisaged that a sensor network will be comprised of dozens of motes, each of which will only be millimetres in size. The most commonly used environment in mote networks is TinyOS (Hill, 2000) – a component based operating system which was written specifically for such networks, and is optimised for low powered processors with RAM in the order of a few kilobytes. Although the term ‘sensor network’ is used a lot, it refers to networks of devices that both sense, and act on the environment. Energy is one of the most important considerations when dealing with sensor networks. Given the large number of devices envisaged in each network, and given the wireless nature of the devices, they need to be self powered. This can either be via a battery, or a renewable energy resource. Either way, but in particular when dealing with fixed energy sources such as batteries, minimising energy use is the key to a successful sensor network. By definition, a key element of the sensor network is the data acquisition and interpretation. Due to the power constraints placed on nodes in a sensor network, in network aggregation and processing of data is encouraged in order to minimise radio communication, which is, by far the most power hungry operation (Hill et al., 2000). TinyDB (Madden and Hong, 2003) is an overlay application intended to run on all the motes in the sensor network, and provide an SQL like interface to the network which is presented as a table of streaming data. This paper discusses an ongoing project where we are integrating the concepts of SCADA, and sensor networks by using TinyDB to provide an interface to a remote sensor network. Some background information on TinyDB is given in the following section. This is followed by a description of the SmartLands (CSIRO,