The Wide Angle: Hack-Proofing Body Sensor Networks

Here in the United States, solutions for curtailing the rising cost of healthcare as well as solutions for providing universal healthcare are being hotly debated. We think that coupling smart phones with networks of wearable sensors and actuators, called body area sensor networks, can help in several ways. But for such pervasive healthcare technologies to become prevalent, much progress is needed to make them safe, secure, and dependable.

Smart phones already work as small, sophisticated computers. They have multiple communication capabilities such as Wi-Fi and cellular, innovative user interfaces, such as multi-touch screens, and they have advanced computation and sensing capabilities. Researchers like us are looking at ways to couple smart phones with wearable sensors as well as in-vivo physiological sensors and actuators, including electrocardiograms and insulin pumps.

Imagine your phone helping you to assess your activity levels, keeping track of your food intake, reminding you to take your medication or sending alerts when there is some abnormal function. Integrating social networking technology with smart phones and these networks will further open up a vast possibility for personalized healthcare.

The Risks
Just like a computer virus can infect a personal computer, it could infect a body area sensor network. And many of the compelling technologies -- wireless, for example -- used for pervasive healthcare also make it prone to security and privacy issues. Anyone with a receiver can tune into the appropriate frequencies. Given enough time, the eavesdropper can listen in and know all about the health of a patient, a serious violation of her privacy.

More capable attackers can even meddle with the function of a body area sensor network, by pretending to be sensors and forwarding wrong information to the doctor. It's also possible for the attackers to modify the data being received by the base station from a legitimate sensor. If done carefully, attackers can prevent important health information from being reported to the doctor. Even more serious is the capability of the attackers to cause physical harm by initiating treatment when it's not needed.

Lightweight Solutions are Needed
We believe that to protect against such attacks, designers have to consider security maintenance as one of the primary pillars of pervasive healthcare. Techniques have to be developed that can encrypt data in such a way that it can be un-encrypted only by the intended recipient.

However, this is not as straightforward as it seems because individual sensors have very limited capabilities. The encryption and other cryptographic techniques used today were developed for more capable machines such as PCs and laptops. Faced with the task, sensor might become overwhelmed, both from processing and energy consumption perspective, rendering them useless.

Therefore a whole new approach is needed to develop security solutions for body sensor networks that are extremely lightweight in terms of the energy required to run them. Further, security should be enabled by default and should be largely transparent so that it doesn't become a hassle. Users should not be expected to type passwords, pins or scan their fingerprints each time the system has to function. Such requirements will reduce the usability of these networks, causing many users to simply disable the security features or reduce the usage of the system.

One solution, physiological value based security, relies on the sensed physiological signals such as those from an EKG or pulse oximeter, to secure data automatically.

Power Problems
Another major issue is power. Sensors are devices that require power and changing and discarding batteries is not only problematic for the user (inconvenient) but also for the environment (pollutes).

Researchers are already working on developing green powering solutions for body area sensor networks, utilizing the enormous potential of the human body. For example, power could be generated by limb movement, body heat or by vibration (see Workshop on Body Sensor Networks, BSN'09 for more about this). This means that security solutions developed for these networks need be simple enough to work with the so-called scavenging technologies.

Increase Awareness
Pervasive health care promise solutions that can fundamentally change the way healthcare is provided in the United States and around the world. However, in our zeal to develop technology we should not forget the potential security issues. The biggest threat to any system comes from the lack of awareness of the users. No encryption technique can protect against a user voluntarily releasing her medical information, on a phone to a hacker pretending to be her doctor. We believe that security issues with respect to pervasive healthcare have to be addressed in a two-pronged manner -- technologically and socially.

User acceptance of mobile and pervasive technologies doesn't seem to be much of a problem as long as these technologies are safe, secure, and affordable. Ultimately, for pervasive healthcare technologies to be widely adapted they have to be encouraged by health professional, insurance and federal agencies, and regulations -- and this is where the great challenges lies for pervasive healthcare.

Sandeep K.S. Gupta is a professor of computer science at Arizona State University, Tempe, Arizona. He directs the IMPACT Lab, which focuses on researching pervasive autonomic computing technologies for healthcare and cyber-physical systems. He has co-authored the book "Fundamentals of Mobile and Pervasive Computing." Krishna K. Venkatasubramanian is a Ph.D. student researcher in the IMPACT lab working on security solutions for pervasive healthcare.