In future, detailed and fully up-to-date personal information, including data from real-time body sensors, will be accessible for widespread use by multiple services and applications. This information can come from sources as diverse as certified medical exams carried out by qualified professionals, or mass-market on-body accelerometers casually used for computer gaming. The use of body sensor data can start early in people’s life, when baby phones measure the emotional state to gently guide the newborn into a peaceful sleep. Late in people’s life, body sensing can play an important role in assisting seniors to age in dignity in a safe natural home environment.

During the decades in between, sensor data can be used in applications ranging from sport coaching to intensive computer gaming and medical– preferably ambulatory–monitoring during periods of illness, or chronic need of attention. Yet the widespread use of body data, not only including people’s physiology but also their emotions, raises societal issues. It also comes with interesting scientific challenges, particularly because the technical constraints of node power consumption and wireless links and networks have to be taken into account. The VITRUVIUS project aims at exploring the underlying key consequences for the architecture of Body Sensor Networks (BSN) and the handling of information about the individual’s body coming from power-constrained wireless sensor nodes

The handling of medical information about the human is migrating from a hospital-centric to a patient-centric approach. Body data is not only used in healthcare but also in lifestyle and entertainment services, where the person-centric thinking already is pervasive. Ownership and exercising access rights of personal privacy-sensitive data is a delicate question, and has to be studied at architecture level, in the context of severe limitations in ultra low power sensors nodes with unreliable and bandwidth-limited sensor modules.

The proposed project VITRUVIUS addresses some of the underlying key questions for the architecture of BSNs.

• Is the envisaged system architecture is viable for entertainment, lifestyle and social services simultaneously?

• How can services providers easily develop new services and applications, without rolling out their own body sensor hardware, but reusing and coexisting with each other? This aspect is a key focus question in this IOP GenCom Call.

• How to partition (automatically or by computer aiding) body-centric and infrastructure centric processing and storage of data?

• Can a service developer adequately handle the technical constraints that govern the BSN?


Participants and consortium


The Space4U and Trust4All ITEA projects have defined middleware and procedures for the trustworthy operation of low-resource devices. TU/e-SAN has been partner in these projects and brings both experience and implementations to the project. The results of these projects have served as a reference for the multimedia middleware ISO standard in the Mpeg consortium. The TU/e SPS group has received top ratings in the latest national research review, has a long history of successful collaboration with many of the industrial, clinical and university partners of the Vitruvius project, has also been successfully involved (together with the SAN group) in the Space4U project, and has a longstanding experience in ontology-based real-time decision-support systems which, among other highlights, has led to the spin-off of MEDECS. TU/e-SAN participates in the FP6 project WASP on the topic of programming and configuring networks of sensors. Tooling from this project becomes available for Vitruvius.

WMC - Twente Institute of Wireless and Mobile Communications

WMC is a partner in the Awareness project and in Smart Surroundings. Both projects have a medical dimension in combination with wireless sensor networks. WMC has hands-on experience with both the proposed software architecture (which in VITRUVIUS is similar to the one chosen in Awareness), the application of sensors networks and medical decision support. With respect to existing implementations, input can be taken from projects MAGNET and MAGNET Beyond, as well as PNP2008 where WMC is involved in. Awareness scored highest in the Freeband review and produced 3 patents and 2 spin-offs.


MEDECS is with their Gaston system market leader in The Netherlands and came out best in benchmarks. They represent years of experience in medical decision support.

Kempenhaeghe - Kempenhaeghe epilepsy and sleep Center, Heeze

Kempenhaeghe has a strong clinical-experimental infrastructure and expertise, with a clinical population that does not depend critically on this product, and a long-standing experience in using and maturing experimental technology. The Netherlands have a unique infrastructure in epilepsy research. Very few countries have specialized epilepsy centers, practically none of them have a focus on monitoring technology. Kempenhaeghe is well known for this focus and their experience in the home monitoring of patients; they were nominated the Brainport prize for this work. Kempenhaeghe developed an electronic patient record system that is now used by all epilepsy centers in The Netherlands. Their spin-off “Hobo” is a center of expertise for seizure detection and sleeping disorders based on their patent portfolio. Within the context of epilepsy and sleep-wake disorders, there is already a lasting cooperation with TU/e-SPS, the signal-processing group of TU/e. Kempenhaeghe is internationally visible, in part through a yearly international Epilepsy and Sleep symposium (300-350 doctors every year).


Goals of the Project

The VITRUVIUS project aims at exploring the underlying key consequences for the architecture of Body Sensor Networks (BSN) and the handling of information about the individual’s body. The connection of the privacy-sensitive “body state” to a rapidly evolving landscape of services, is the key theme of the project.


VITRUVIUS System Architecture

At the heart of the project is the architecture that provides essential body information to service providers in an accurate, reliable, robust and trusted privacy-preserving manner. It relieves the application developer from the need to understand the details of the underlying sensor system.

At or near the body, a number of sensor nodes extract information from the body via dedicated sensors, sensor single signal processing modules which pre-process the data (through calibration, artefact rejection, signal validation and compression), and wireless (e.g. Zigbee) links that convey the information securely to a body hub. The sensor nodes may also contain actuators so that the interface between body hub and sensor nodes will, in general, be bi-directional.

Based on the sensor signals, the body hub estimates key physiological parameters (e.g. heart rate, velocity, temperature) by means of multi-sensor signal processing, and subsequently extracts key diagnostic information by means of a hub local decision-support engine. This processing is controlled by application-specific components uploaded into the hub.  A wireless PDA can serve as hub node, interacting with the user (e.g. perceived pain level, system feedback, privacy policy settings, authenticating trusted sensors, etc).

Diagnostic information is conveyed by a wireless link, preferably using existing data formats such as the Zigbee Medical Profile to the ‘fixed world’ where it is integrated with other information (e.g. medication status, electronic patient record, ‘fixed-world’ sensors) into a compound decision-support scheme that forms the heart of the application, and that is connected to the various caregivers. Again, the information that is send out is typically controlled by uploaded components (see below).

Since the sensor nodes and body hub will, in general, be battery-operated, it is essential to minimize the amount of information that is conveyed wirelessly, even if this requires many signal-processing and decision-support computations. This is so because Moore’s law causes these computations to become ever more energy-efficient, whereas wireless links are governed by fixed energy limits. A direct implication is that signal-processing and decision-support functions, which both act to reduce the amount of data, should be shifted as deeply as possible into the sensor nodes and hub. This efficiency consideration is perfectly consistent with our view of the user as being the owner of his/her body information, with no more information released to the outside world than strictly needed.

Signal-processing algorithms, but even more so, decision-support schemes are specific to an application or service, and hence need to be configurable and uploadable in a secure and trust-preserving manner. To this end, the hub architecture contains a secure upload and configuration manager, trust and ownership monitor, and security interface. The first module provides means for run-time upload and installation of application-specific components. For instance, a service may want to use its own local decision support component specific only to this application, to get access to specifically tailored information. In this case, the application or service can request installation of the component on the hub.  The secure upload and configuration manager checks the component certification details, verifies the future system integrity and installs it on a system. The trust and ownership monitor not only constantly examines the current security, performance and reliability properties of the system, but also predicts and verifies these properties at the system reconfiguration time, thereby authorizing configuration changes. The Security interface serves as a “body firewall” that shields the hub from the outside world, blocks body spy ware, and provides only limited privacy-sensitive data to parties that are not entitled full access. For example, the user or his social care givers receive different information than doctors or professional sport coaches. Some parties may only be interested in whether medically certified products are used according to professional protocols (liability), others may be interested in physiological data (Sp02 saturation, respiration details), while others are best served with interpreted data (grandma has fallen). If necessary the hub can make the data “anonymous” for certain applications. For instance, heart rhythm and blood pressure data should, in case of monitoring a cardiac patient, be routed to a health monitoring application to be supervised by a hospital cardiologist. The same signals, however, can be used for sports or fitness monitoring and coaching. So, different applications, using data of the same sensors, (can) have different supervisors and/or service providers.

Within the hub, the raw incoming signals as well as the temporal evolution of the hub’s state can be collected in a storage memory for inspection in special cases (e.g. liability disputes), or at special moments (e.g. when the hub is close to the fixed-world transceiver, permitting wireless communication at very low power).
To facilitate application development, it is essential that the medical or domain specialists can take the helm in the development of the decision-support scheme, without requiring knowledge engineers as go-betweens. This implies that an ontology-based approach is needed, where specialists can freely define the ontologies that are relevant to the application at hand, along with guidelines and decision rules in terms of these ontologies. For this reason a state-of-the-art ontology-based decision-support development framework will be taken as a starting point [‎14, ‎15, ‎16], and a separate back-end will be developed to split the resulting decision-support rules into executable ‘fixed-world’ and ‘hub-based’ modules. In this manner, service providers can quickly develop new applications and services based on high-level body information, without requiring knowledge about specific sensor technology.

For the sensor nodes we re-use the existing sensor node system “AquisGrain”. It is based on the IEEE 802.15.4 standard for wireless personal area networks, including the physical and networking layer, a commonly used chipset from Texas Instruments (ChipCon 2420), an MSP430 microprocessor and FreeRTOS.  The hub will be implemented as a PDA with Zigbee 15.4 radio and a WiFi link to a wireless access point.

The software platform developed in the Trust4all project will serve as a starting point for the software architecture of the hub. The Trust4All developed a middleware software architecture specifically for embedded systems that require a defined level of trust, due to the nature of the services they provide. The middleware software architecture offers a certain level of dependability as required in home medicare, home security, domotica, and on-the-move applications. The Trust4All middleware has been the foundation for the ISO MPEG standard for multimedia middleware in low resource devices.

For communication between application components we develop a service oriented architecture that admits loose coupling and reuse.