Wireless Sensor Networks – Ubiquitous Watchdogs of the Future?

By Dr Nigel Goh, Aug 2003

Wireless technology is not new. Electronic networks are old hat. So what’s the big deal? Actually, the big deal is that the sensors are no longer big! And they now talk to each other. Some of the problems that previously plagued miniaturizing such sensors and allowing them to be applied more widely have been overcome by technology and design. Advances in semiconductor research have allowed the sensors to be shrunk to the size of Panadol pills or smaller (Scientists are talking about 1mm³ sensors). However, the sensors need power to transmit signals and battery technology has not advanced to a stage that allows such signal transmission to be sustained, given the size limitations on the battery, Eveready Energizers included. Sensors are therefore designed to be periodically turned on, just long enough and using only the power required to transmit signals just far enough so that neighboring sensors can receive their signals; these in turn activate their neighbors until the signal is finally passed to the central base station. In this way, battery power is conserved.

Fig. 1: The Die Photo of the Micro Power Supply Chip

Early work on wireless distributed micro-sensor networks began at the University of California at Berkeley in 1997, led by Prof Kris Pister. This ambitious project aimed to develop a device the size of a sand grain containing sensors, computational ability, bi-directional wireless communications and an independent power supply. In addition, it had to be inexpensive enough to be deployed by the hundreds. The project was coined “Smart Dust”. Two years later, a complete operating system – ‘TinyOS” - for Smart Dust was developed by the same group, spawning numerous applications; they are still pushing the size barrier.

The potential applications of wireless sensor networks are limited only by the imagination. Clark Nguyen, a professor of engineering at the University of Michigan, on leave to work for the US Department of Defense’s Advanced Research Projects Agency (DARPA) was quoted in a Forbes.com article, “I like to be conservative about things, but in a way, [sensor networks] could be bigger than the internet”. The Pentagon is supporting wireless micro-sensor network research in order to obtain ‘hyperspectral’ data from a war zone. By spreading thousands of chips over a battlefield, it could be possible to detect troop movements, bioweapons and electromagnetic noise.

Besides the obvious military and security uses, micro-sensors have been used in monitoring oil pipelines near the Arctic Circle. These pipes have to be heated if their temperatures get too low to prevent the pipes from clogging and eventually bursting. Each wired temperature sensor currently being used could cost US$3,000-10,000 to install (just for wiring), and if a sensor is faulty or wrongly installed, each burst pipe could result in days of plant closure, at a cost of US$100,000 an hour. For less than this cost, 2 or 3 wireless sensors could replace each wired sensor, giving a more reliable reading in addition to the reduced installed cost. Scientists could also use micro-sensors to monitor microclimate changes within areas of rainforest rather than at specific points. Seismologists could use these networked sensors to monitor earth movements and provide early warning for earthquakes. Sensors could be put in bridges and buildings to warn of structural defects or weaknesses. Other types of sensors could be put in reservoirs to warn of hazardous chemicals. The possibilities are almost endless ….

At the National University of Singapore, a research team led by Associate Professor YC Liang at the Dept of Electrical & Computer Engineering has been looking into this area of research. An on-chip power management platform for micro-sensor applications has been developed (Fig 1). The circuit was designed to step up and stabilize voltage from a weak voltage source to a high voltage level for system operation. It consists of a PWM generator and a large NMOS switch, fabricated using 0.8um BiCMOS technology. The power management chip is integrated with a MEMS (micro-electro-mechanical systems) structure on another chip and a set of sensors for light and temperature detection. Work is now being carried out to develop an RF transmitter to be coupled with the integrated MEMS system to enable each unit to relay information between itself and other units within the network of wireless sensors. Assoc Prof Liang feels that this technology could be applied to monitor and provide intelligent feedback on light and temperature levels in buildings, sending signals to optimize the level of lighting and cooling required, leading to greater efficiency in building management. Further development on reduction of the size and cost is possible to make the wireless smart-dust more applicable in other surveillance applications.

One of the main problems faced by this technology is power management. Batteries need to be small enough to be incorporated within the micro-sensor itself. Related to this is the consumption of power. Power management in the miniaturized chips is crucial to enable useful application of these wireless sensor networks over a reasonable time period. Network bandwidth, given the billions of constant inputs, could be swamped. Researchers say none of these problems are insurmountable, and more and more such wireless networks, in one form or another, are likely to be implemented in the not-too-distant future.

And if you think that this technology will not really affect you, think again. In 2001, it is estimated that 7.5 billion embedded microcontrollers were sold. These are found in everything from toasters to cars, from smoke detectors to DVD controllers. All these could potentially be networked. “Why?” you may ask. We may not know yet, but there is a Law attributed to Metcalfe – a network’s value increases with the number of things connected to it. In an interview with MIT Technology Review in June, Ember (www.ember.com; one of the emerging leaders in embedded wireless networking) CTO Robert Poor described a scenario where sensors on lampposts in a city could not only monitor streetlamp function; they could also be connected to sensors on buses, sending signals on bus location, speed and accurate bus ETAs at bus stops – no more hoping and praying that your bus would come soon; you would know. Presumably, sensors on cars could also provide a record of your driving speeds on any particular roads …..

PS. Don’t tell our traffic police about that one!