Small autonomous wireless sensors, linked into a network, can be used in a variety of applications ranging from health and lifestyle, automotive, smart building, and predictive maintenance to smart packaging. The miniature sensor nodes have their own energy supply consisting of energy harvesting and energy storage devices; a low-power wireless connection to the other sensor nodes within the network; and some built-in intelligence to carry out basic data-processing tasks.
This article discusses the benefits and technological challenges of using such wireless sensor nodes in three different application areas. First, for healthcare and lifestyle, sensor nodes can constitute a body area network (BAN) that may enhance existing health monitoring systems and enable new personal health applications. Early technology deployment has led to the identification of key technology challenges that need to be addressed to enable a widespread use of BANs. Second, micropower and ultra-low-power sensing technologies can play a prominent role in condition monitoring of industrial machinery, particularly when installing wired sensors is neither feasible nor technically attractive. Third, wireless sensors integrated in a thin, flexible foil can be used in smart packaging to allow sensing, location-aware, and communicating products in a broad range of applications. Research in these areas is being carried out in the framework of the Wireless Autonomous Transducer Solutions (WATS) program at Holst Centre/Imec.
Sensor Node Basics
Wireless sensor networks (WSNs) consisting of small nodes with sensing, processing, and wireless communications capabilities are becoming widely used in our society. They are being used for therapeutic and diagnostic purposes, as well as for monitoring industrial processes, in active RFID tags, and in automotive applications. At present, the majority of WSN nodes rely on batteries for operation. The nodes we'll discuss are autonomous—relying on energy harvesting for power—and thus require low-power electronics and sophisticated energy management.
Figure 1 shows the major components of a typical node. Besides sensor and actuator functionality, the node contains a front end, microprocessor (µP), digital signal processor (DSP), and a radio for wireless communication. The node is powered by a micropwoer module, which consists of an energy harvester, a voltage converter (either AC/DC or DC/DC), and an energy buffer for temporary storage of the harvested energy (either a small, rechargeable battery or a supercapacitor). Developing such a node typically requires a combined expertise in wireless ultra-low-power communication, packaging and 3D integration, sensors and actuators, low-power design, and energy harvesting technologies. The latter are needed to make the products truly autonomous.
Figure 1. Schematic representation of the major components of a WSN
A number of energy harvesting principles are currently under development and the first commercial systems are entering the market. The main technologies are based on vibrational, thermal, photovoltaic, or RF harvesting and it is expected that they can supply energy in the 10 µW to 1 mW range. Another requirement for general use and interoperability of WSNs is the availability of a standard protocol for communication. Emerging communications standards such as IEEE 802.15.4 are becoming available for WSNs.
The specific requirements and technology challenges for WSNs obviously depend on the application they will serve, on the effect to be sensed, and on the data rate of the transmitted data. Consider power consumption as an example. On the one hand, 90 µW seems enough to power a pulse oximeter, to process data, and to transmit them at intervals of 15 s. On the other hand, 10 µW turns out to be sufficient to measure and transmit temperature readings every 5 s. In general, 100 µW is considered to be sufficient for relatively complex autonomous WSN nodes operating at relatively high data rates. If one considers MEMS-based energy harvesters, 100 µW/cm2 is considered to be a value that is attainable but challenging. Importantly, MEMS technology offers a route towards cost-effective harvester fabrication.
Now let's examine autonomous WSN use in BANs, condition monitoring of industrial machinery, and smart packaging.
BANs for Personal Health
The use of wireless sensor nodes is technologically most advanced in the healthcare and lifestyle sector. It is expected that WSN technology will soon enable people to carry their personal BAN that provides medical, lifestyle, wellness, assisted living, sports, and entertainment functions for the user. The network would comprise a series of miniature sensor/actuator nodes implanted or located at the body surface. Each node will have its own energy supply consisting of energy storage and energy harvesting devices. Each node has enough intelligence to carry out its own tasks and is able to communicate with other sensor nodes or with a gateway node worn on the body. The gateway node communicates with the outside world using a standard telecommunication infrastructure such as a wireless local area or cellular phone network, allowing experts to provide services to the individual wearing the BAN.
Early adopters of BAN technology are seen in the sports and lifestyle area, using a wireless sensor node to measure one or a few body parameters—e.g., heart rate or physical activity—and communicate the data wirelessly to a cell phone, usually via Bluetooth. These examples have important limitations, mainly in terms of power consumption. The current state-of-the-art in low-power electronics enables significant increases in the lifetime of these systems while reducing their size—ultimately these may become autonomous systems. Early prototypes based on this technology are now becoming available.
The availability of such BANs will play an important role in future healthcare, which will evolve from disease-centric to patient-centric care, transitioning the point of care from the hospital to the home, and shifting the emphasis to prevention rather than cure. BAN technology will benefit some existing applications, reducing the monitoring burden and enhancing the patient's comfort, e.g., a wireless ECG patch for ambulatory monitoring of cardiac activity. Moreover, BANs also have the potential to enable applications that were impossible before this technology was introduced.
In 2008, Imec reported the development of a first-generation wireless ECG patch (Figure 2). Low-power and high-performance ECG monitoring is achieved through the use of a proprietary single-channel ASIC for bipotential read-out. The patch also integrates a low-power microcontroller, low-power radio, antenna, battery, and optimized power management circuitry. Depending on the application, the ECG patch either continuously streams the single-channel ECG data to a receiver within a 10 m range or performs local analysis on the data to extract R-peak and other fiducial points. Power consumption of <2 mW is achieved in both modes. The wireless ECG patch has been tested in ambulatory settings, to evaluate how physical activity affects the quality of ECG recordings. Results have shown that the patch provides signals with excellent quality in resting conditions but the quality of the signal degrades as the level of activity increases. For activity levels up to those that correspond to running on a treadmill at 7.5 km/h, the quality of the signal is maintained at a usable level for further analysis.
Figure 2. Example of a BAN: a wireless ECG patch, developed at Holst Centre/Imec
Technology evaluation in various application environments has led to the identification of key technology challenges that need to be addressed to enable a widespread deployment of BANs and to meet the ultimate target of developing wireless body sensor nodes consuming 100 µW power on average. The challenges are related to:
- Ultra-low-power technologies. Recent advances in the design of ultra-low-power front ends can already enable drastic reductions in the node's power budget. To meet the target of 100 µW per body sensor node, further research is needed on ultra-low-power analog interfaces, sensors, DSP, and radios.
- Autonomous systems. Today's prototypes can run for a few days at full functionality. Breakthroughs in ultra-low-power technologies will enable months or years of autonomy. Harvesting energy from the environment during the operation of the system will eventually allow the system to run perpetually with a small rechargeable battery or a supercapacitor acting as a temporary energy buffer.
- Multi-parameter sensors. Extending the range of functionality to include new sensing modalities will be crucial in fostering research in personal health applications, leading to new discoveries.
- Increasing functionality. Low-complexity and real-time algorithms are required to enable intelligent autonomous systems.
- Dry electrodes. These are required to enable simple setup of the system by the user.
- Integration technology. Electronic integration in bidimensional flexible and stretchable foils will enable unobtrusive body sensor nodes that are integrated in patches, clothes, or even fashion accessories.
Industrial Condition Monitoring
The use of wireless sensing technologies for condition monitoring is in a more exploratory phase. Condition monitoring in general provides a toolbox for reducing unplanned machinery downtime, increasing equipment availability, and enabling state-dependent maintenance rather than preventative maintenance and has consequently gained in popularity over the last two decades. Condition monitoring is best applied when failures are inherently random and develop slowly, and thus won't be identified during planned maintenance. In addition to the sensors, condition monitoring requires mechanisms for objective data assessment, data processing, storage, and review facilities. The system should also yield an increase in profits. For these reasons, at the present time condition monitoring has been limited mostly to large, expensive systems.
Condition monitoring tools include a variety of methods such as vibration analysis (Figure 3), oil analysis, performance and current monitoring, thermal and corrosion monitoring, and visual, auditory, and manual inspections. Sensors play a prominent role in these tools. The sensing quality depends strongly on the distance between the spot where the fault is occurring and the sensor's location. This is especially true for vibration sensors, where having one or more sensors in proximity of the fault region greatly increases the failure detection capability, increasing the time between detection and final failure. During our investigations of bearing-induced vibrations on a rotating shaft we have found that, even in the temporary absence of line-of-sight, we could still retrieve vibration data from the sensor fixed to the rotating frame. Future experiments will need to focus on determining signal quality and creating events, as well as generating event-based rules for automated diagnosis systems.
In some particular cases, the use of wireless micropower and ultra-low-power sensing technologies offers clear advantages compared to wired solutions—when wired sensors cannot easily be used, e.g., when parts are attached to a moving frame, or when measurement accessibility is limited in certain hazardous environments. Wireless sensors can be mounted on moving parts and in difficult-to-access locations allowing for earlier failure detection. Sensors operating in industrial environments require high reliability, long lifetimes (including low-power), and constant performance. Besides developing sensing functionality, one of the major challenges is the radio/network element. New network protocols need to be developed and implemented to comply with the harsh environments present. Among other qualities, the protocol would need to (re)transmit data in a complex, dynamic environment and energy-saving features will play an important role. In addition, the protocol should implement node initialization and calibration features.
Wind turbines are an excellent example of a situation where condition monitoring and early fault diagnosis offer huge advantages. Wind turbines tend to be located in remote areas, especially those in offshore wind parks, and access to the site and the turbines is difficult. In addition, wind parks involve the presence of similar or identical systems working together in a cluster, continuously varying system parameters (such as wind speed and direction, shaft load, etc), huge installation and operation costs, and have an estimated lifetime of about twenty years. A suitable condition monitoring system needs to integrate a variety of sensors to measure acceleration, velocity, position/direction, fluid quality and level, temperature, viscosity, and pressure, as well as wind direction and speed. The sensors are subject to a harsh environment and must be extremely reliable and well calibrated to prevent false alarms or lost events. Setting up a wireless sensor network for such an application is very challenging, as it requires the creation of a common platform for these varied sensor types, research into signal and noise levels to ensure useful data, considerations of power consumption, as well as multinode networks. Once the hardware issues are addressed, the system needs to incorporate robust condition monitoring rules and appropriately defined alarm levels.
Wireless sensor networks for condition monitoring have been implemented for structural monitoring for buildings, bridges, roads, dams, and other civil infrastructures as well as for energy usage evaluation.
Today, a package is designed to protect its contents and to support advertisement and information. Tomorrow, a package may well become a smart product on its own terms. It could sense freshness and release odors, talk and interact with individual customers, monitor and store the history of its transportation, actively combat theft and counterfeiting, and will take an active role in the supply chain. Smart packaging has a huge market potential because it can address some of the megatrends in our society including globalization (by tracking and tracing), the aging population, individualization, population growth, and food safety.
Smart packages already exist, but mass commercialization is currently hampered by the costs associated with this type of packaging technology. One typical example is the 'smart blister'—developed by the Holst Centre and the Compliers Group—that converts a blister pack of drugs into a device that records which tablet was removed and when this occurred (Figure 4). This improves the integrity of data in drug trials, but is still too expensive for introduction to the mass market.
Figure 4. Concept demonstrator of the smart blister
A major challenge, therefore, is cost reduction. We believe that only printable electronics can deliver sophisticated capabilities to packaging at a price that makes its adoption economically viable. The technologies that will enable such smart packages at the right cost derive from currently used color printing as well as from advanced technologies and packaging concepts of microelectronics. Technology challenges include power management to create low-power electronics and battery integration; the development of sensors to measure parameters such as temperature, humidity, shock, and the presence of ethylene gas to measure the ripeness of fruit; thin-film barriers and indicators, preferably displays or thermochromic inks; and foil handling and lamination. These developments will pave the way toward cost-effective wireless sensor platforms on thin, flexible foils.
Wireless sensor nodes are emerging in diverse application fields, where they all face some common challenges when it comes to a widespread acceptance: autonomous operation, functionality, intelligence, miniaturization, and manufacturing cost. The specific requirements depend on the target application. In healthcare and lifestyle, the benefits of using WSNs have already been demonstrated, and an early deployment of BAN technology has led to the identification of key technological challenges for further improvement. When applied to condition monitoring, wireless solutions are particularly interesting for situations when installing wired sensors is either not feasible or technically unattractive. In smart packaging, the success of using WSNs is very much dependent on the ability of making cost-effective solutions.
ABOUT THE AUTHORS
Ruud Vullers is Principal Researcher and Program Manager of the Micropower program at Holst Centre/Imec. He holds an MSc in Physics from the Radboud University Nijmegen and obtained a PhD. in Physics from the Katholieke Universiteit Leuven, Belgium.
Julien Penders is Program Manager at Holst Centre/Imec, where he leads the activities on System Integration and Body Area Networks. He holds a M.Sc. degree in Systems Engineering from the University of Liege, Belgium, and a M.Sc. degree in Biomedical Engineering from Boston University, MA.
Mihai Patrascu, PhD., MSc, has finished his M.Sc. at the Faculty of Electrical Engineering, Delft University, The Netherlands. He obtained a PhD. on control of MEMS actuators at the University of Twente, The Netherlands. He has been a researcher at the Holst Centre since 2006.