Wireless medical devices offer increased patient comfort and unprecedented monitoring possibilities. But first, it’s important to optimise power consumption.
By: E. Parton, imec, Leuven, Belgium, and H. de Groot, Holst Centre/imec, Eindhoven, Netherlands
Wireless sensor nodes
Wireless devices have transformed our professional and leisure environments—they are about to do the same to our healthcare settings. One fascinating application is the wireless sensor node, which can monitor heart rates, brain waves, body temperature, blood pressure and other vital signs.
| Figure 1: A sleep monitoring system comprises a headband with three sensor nodes measuring two EEG channels (electroencephalogram) to monitor brain activity, two electro-oculogram channels to monitor eye activity and 1 electromyogram channel to monitor the chin muscle activity. |
 |
Body sensors currently are used in the intensive care unit to monitor patients and to alert doctors if there is a change in the patient’s condition. Cables link the sensors to a computer. While this is not a fundamental problem in this instance, it’s easy to see how wireless sensor nodes could benefit ambulatory patients in the hospital and elsewhere.
Patients undergoing sleep monitoring tests curently must wear sensors that are connected to a computer by strings of wires, which can be uncomfortable. Research institute imec and the Holst Centre have developed a headband-like wireless monitoring system with five integrated sensors that capture and transmit all of the data needed to perform effective sleep tests (Figure 1).
This device will improve patient comfort during sleep tests conducted in hospitals and it will eliminate cable artifacts. Tests done in association with sleeping disorder center Kempenhaeghe in the Netherlands have shown that the wireless monitoring system operates just as effectively as a wired system. In the not-too-distant future, one can imagine a patient going to the doctor to get a sleep test cap and performing the test at home. Results would then be analysed remotely by a sleep disorder professional.
Once wireless sensor nodes become mainstream, a multitude of new applications will emerge. The sensor nodes can be integrated into an array of products including blankets, car seats and clothing. A baseball cap with integrated EEG sensors could continuously measure the brain activity of an epilepsy patient, for example (Figure 2). A shirt could double as an ECG monitor. The possibilities are endless.
| Figure 2: Rendering of a cap with integrated EEG sensors for epilepsy patients. |
 |
To achieve a wireless sensor node as we envision it, a great deal of research and development is still needed. One obstacle to its fulfilment is related to power consumption. Since the node is not wired to the electrical grid, a battery is needed. This battery should be as small as possible to fit within miniaturised systems that will be integrated into clothing. If small dimensions are not a priority, the reduced power consumption will enable longer autonomy—a long battery life is indispensable when it comes to implantable sensor nodes—or additional functionalities.
An ECG belt that is being developed by imec and the Holst Centre is worn like a conventional heart-monitor belt used by athletes. Instead of monitoring just the heartbeat, however, the belt can record and transmit a full ECG. This device can be very handy for outdoors enthusiasts with a heart condition or for use in athletic events (the belt was tested during the Brussels Marathon). The challenge is to miniaturise this system and to achieve a sufficient level of autonomy. Depending on the application, this can range from days to forever. The ultimate vision is to integrate this electronic heart specialist into a small box that can be attached to a belt (Figure 4) or, better yet, be embedded in a shirt.
Power consumption
The power budget of the ECG sensor node was analysed by examining the sensor’s individual building blocks: the sensing and read-out unit, wireless communication radio, digital signal processor (DSP) and energy supply unit. It became clear that the biggest power consumer in the node was the radio chip, which is responsible for the wireless transmission of sensor data (Figure 5). Typically, the wireless functionality consumes 50% to 85% of the total power budget.
| Challenging elements of wireless sensor nodes |
A medical/sports wireless sensor node consists of:
- a sensor to measure body parameters (to check glucose levels, for example)
- an actuator to perform an action (to inject insulin, for example)
- an analogue interface that converts the analogue sensor’s data into digital signals
- a digital signal processor (DSP), which is, in fact, a minicomputer that collects all data, performs some calculations and makes decisions, if required (for example, if the glucose level is greater than x, then inject y µl of insulin)
| Figure 3: The building blocks of a sensor node. |
 |
- a radio chip that fulfils the wireless transfer of data to a mobile phone or the doctor’s laptop
- a power supply that comprises a battery and power management circuit. For some applications, an energy-harvesting device such as a solar cell or vibration scavenger can be added.
|
|
What do you do when you find out that your television is responsible for the lion’s share of your energy bill? You buy a power-efficient LED TV or cut down on your viewing time. Imec and Holst Centre researchers adopted a similar strategy for the sensor node radio.
Ultra-low power radio
Imec and the Holst Centre developed several ultra-low power (ULP) radios, each one suited for a different application. The three radio architectures, ranging from high to low data rate capability, are described below.
The first technology that was developed—a pulse-based ultra wideband (UWB) ULP radio is a unique combination of low power and medium data rate (100 kb/sec to 20 Mb/sec). This is suited for applications where sensor data are combined with media streams: a heartbeat belt that communicates with an MP3 player (to queue up less rhythmically intensive music as the heart rate increases during a jog) or a hearing aid that wirelessly communicates with an MP3 player are two examples. If the ULP UWB radio is incorporated in the belt, hearing aid and MP3 player, the power consumption is less than 5 mW with excellent interference resistance. This is approximately five times less than when a commercial low-power radio is used with medium data rates. UWB radio operates in the 6 to 10 GHz radio band. It has fewer interference issues than a competing Bluetooth device operating in the crowded 2.45-GHz ISM band.
UWB radio has another advantage: localisation. The radio signal is very broad, and the device’s location can be determined in a radar-like manner. Being able to localise a device without infrastructure or triangulation (i.e., requiring at least three devices for an accurate position) is a unique feature in many application scenarios. Currently, inside localisation can only be achieved using multiple technologies.
The third alternative—the wake-up radio—was developed for very low data rates and ultra low power consumption (60 µW in continuous operation). This radio can be operated in parallel with a conventional radio that switches it on when data needs to be received or transmitted. In this way, power is conserved. For example, the radio on a cell phone with Bluetooth functionality is continuously searching for Bluetooth devices. This quest consumes a significant amount of power. By combining the Bluetooth radio with a wake-up radio, the latter could wake up the Bluetooth radio when it needs to connect to another Bluetooth device. A potential medical application is an implantable sensor that needs to transmit data periodically to a doctor’s computer.
Reducing power consumption
Another way to reduce the power consumption of wireless sensor nodes is to reduce the amount of data that has to be transmitted to a central device on the body or to a laptop. This can be achieved by processing some data locally within the sensor nodes and sending a small amount of processed data instead of a large amount of raw sensor data. An additional advantage is that immediate feedback to the patient is possible.
Figure 8: By doing local processing in the
sensor node, the radio’s power consumption decreased. This is offset, however, by the increased power consumption of a general purpose microcontroller. This problem was solved by the development of a low-power DSP optimised for processing biological parameters. For example, system power consumption in an ECG patch with an off-the shelf Zigbee-like radio dropped by a factor of 10 when the ECG processing is done locally with imec’s BioDSP. |
 |
Contrary to expectations, power consumption of the sensor node increased when local processing was initially performed on a general-purpose microcontroller. Radio power consumption decreased, because there was less data to send, but the power of the commercial microcontroller spiked as it is was not optimised for this kind of processing. For this reason, imec and Holst Centre developed a dedicated ultra-low power DSP optimised for processing of bio-parameters such as EEG, ECG, EOG and EMG.
A new future with ultra low power
The power consumption of wireless medical devices can be dramatically reduced by incorporating the right ULP building blocks. Two strategies were outlined in this article:
1. Adding a ULP radio (UWB, BAN or wake-up radio for medium to low and ultra low data rates) can reduce power consumption by a factor of 10.
2. Adding a ULP DSP reduces power consumption by a factor of 10 when some local processing is done in the device or sensor node.
Combining both the ULP radio and ULP DSP strategies would result in an 18 X reduction in power consumption in an ECG patch, bringing us one step closer to ubiquitous body-worn autonomous sensors.
Els Parton
is Scientific Editor at imec, Kapeldreef 75B-3001, Leuven, Belgium
tel. +32 16 281467
Harmke de Groot
is Program Director, Ultra Low Power Wireless and DSP, at Holst Centre/imec Netherlands, High Tech Campus 31, 5656 AE Eindhoven, Netherlands
tel. +31 40 4020453
Login or
register to post comments
