Safe and Dependable Bio-Sensor Networking for Pervasive Healthcare

1. Safe and Dependable Bio-Sensor Networking for Pervasive Healthcare SANDEEP K. S. GUPTA Department of Computer Science and Engineering (Affiliated with BMI, BME, EE) School of Computing and Informatics Ira A. Fulton School of Engineering Arizona State University Tempe, Arizona sandeep.gupta@asu.edu

2. IMPACT (Intelligent Mobile Pervasive Autonomic Computing & Technologies) LAB MISSION Creating Humane Technologies for an Ever Changing World. La Famiglia Capo di Tutti Capi:Sandeep Gupta Consigliere:Georgios Varsamopoulos (Post Doc) Soldatto: • Ayan Bannerjee (PhD) • Ken Bannister (MS) • Guofeng Deng (PhD) • Gianni Giorgetti (PhD) • Michael Jonas (PhD) • Su-Jin Kim (PhD) • Tridib Mukherjee (PhD) • Qinghui Tang (PhD) • Krishna Venkatasubramanian (PhD) Picciotto: • Edward Raleigh (Undergrad) • Tom Murphy-Hoye (High School Junior, – Desert Vista) http://impact.asu.edu

3. Thermal Management for Data Centers Pervasive Health Monitoring Criticality Aware-Systems Mobile Ad-hoc Networks ID Assurance • Goal: • Increasing computing capacity for datacenters • Energy efficiency • Features: • Online thermal evaluation • Thermal Aware Scheduling • Sponsor: • Goal: • Protect people’s identity & consumer computing from viral threats • Features: • PKI based • Non-tamperable, non-programmable personal authenticator • Hardware and VM based trust management • Sponsor: • Goal: • Container Monitoring for Homeland Security • Dynamic Supply Chain Management • Features: • Integration of RFID and environmental sensors • Energy management • Communication security • Sponsor: • Goal: • Protocols for mobile ad-hoc networks • Features: • Energy efficiency • Increased lifetime • Data aggregation • Localization • Caching • Multicasting • Sponsor: Intelligent Container IMPACT: Research Use-inspired research in pervasive computing & wireless sensor networking • Goal: • Pervasive Health monitoring • Evaluation of medical applications • Features: • Secure, Dependable and Reliable data collection, storage and communication • Sponsor: • Goal: • Evaluation of crisis response management • Features: • Theoretical model • Performance evaluation • Access control for crisis management • Sponsor: Medical Devices, Mobile Pervasive Embedded Sensor Networks BOOK: Fundamentals of Mobile and Pervasive Computing, Publisher: McGraw-Hill  Dec. 2004

4. Pervasive Healthcare

5. Holy Grail of Pervasive Healthcare “ …cheap enough technology will enable early detection and a lot less folks will have a heart attack, stroke or cancer going forward. “ “.. A nanochip will search your blood for cancer – five years before they grow uncontrollably ….” “Your doctors may not be certain about what is going on inside your body; but technology will.”

6. Detect symptoms Diagnosis Treatment Visit medical facility Healthcare Today Source: Wikipedia Definition • Healthcare, is the prevention, treatment, and management of illness and the preservation of mental and physical well being through the services offered by the medical, nursing, and allied health professions [Wikipedia]. Provision • In most places healthcare is provided in organized manner through a system of medical facilities • Provided through public sector, private or both Presence of Universal Healthcare Primary Delivery Model

7. Imaging & Visualization Communication Medical Devices Region-wise Percentage of population over 60 An Aging World • Department of Health projects by 2050 over 20% of the population in U.S. will be above 65. Consequences • Acute shortage of trained medical professionals. • Bureau of Labor Statistics estimates 18,000 job openings for physicians created annually by physician retirements. • Reduced healthcare delivery coverage . • Increase in the medical costs. Solution • Extend traditional delivery model, with: • Communication infrastructure • Devices facilitating remote checkup, surgery • Visualization and imaging tools • Result: Remote Monitoring models • Telemedicine model • Centralized monitoring model Source: World Population Prospects, The 1998 Revision, United Nations Secretariat Region-wise World population by 2050 Extended Delivery Model Source: International Energy Agency (IEA)

8. Features • Extends remote monitoring model by enabling: • Physical presence of caregivers required only during emergencies • Improved coverage and ease of monitoring • Utilize in-vivo and in-vitro medical sensors • Mobile patients. No time & space restrictions for health monitoring • Better quality of care and reduced medical errors • Early detection of ailments and actuation through automated health data analysis Camera EEG EKG BP SpO2 GPS Mp3 PDA/phone Gateway Lifeshirt non-invasive monitoring Developed @ Vivometrics Medical Tele-sensor Can measure and transmit Body temperature Developed @ Oak Ridge National Laboratory Nano-scale Blood Glucose level detector Developed @ UIUC Motion Sensor Pervasive Healthcare Use Pervasive Computing for day-to-day healthcare management to enable real-time, continuous patient monitoring Body Area Network Applications Home-based Care Sports Health Management Disaster Relief Management Medical Facility Management GOAL:Enable independent living, general wellness and disease management.

9. Differences & Advantages • Pervasive Healthcare • Continuous Patient Monitoring. • Automated diagnosis and treatment. • Utilizing medical facilities only if condition very serious. • Current Healthcare • Detect symptoms • Go to medical facilities (professionals) • Medical professional performs diagnosis and treatment. • Automated • Real-time • Inexpensive • Very efficient • Manual • Slow • Costly • In-efficient Pervasive Healthcare Technology is Necessary to Meet Future Needs

10. Pervasive Healthcare - Conceptual Overview Feedback for Adaptation Medical Sensor Plane Management Plane Knowledge Generation Plane Doctor Actuation Patient Data Collection Knowledge • Collect Medical & contextual data • Local Processing • Medical Actuation • Important Issues: • Accuracy of knowledge generation, minimize false positives • Ensure data does not overwhelm • Focus doctor’s attention on important data • Summarize data carefully Generate Knowledge • Storage Management • Sensor Management • Generate Context

11. Regulations • Encompasses both hardware and software components • Authority: • Food & Drug Administration • Legislation: • Health Insurance Portability & Accountability Act • Standardization • Hardware-Software interfaces • Software-Software interfaces • Data communication • Data storage Issues in Pervasive Healthcare Systems Hardware • Low power • Minimal form factor • Energy scavenging • Packaging • Material Constraints • Long-term operation Software • Reliability • Energy-efficiency • Security • Fault-tolerance • Context-awareness • Actuation • Organization & Reimbursement • Who will deploy and control such systems: doctor/hospitals/third-party ? • Who will cover the costs of installation, management and repair ? • Will insurance cover such systems ? Agood pervasive healthcare solution is the one which takes into account: The limitations/requirements imposed by the hardware & software components Regulatory constraints Personalized requirements of each patient in its design.

12. Talk Overview Minimize heat generated by health monitoring sensors Improve patient privacy by using efficient data security protocols. Provide automated emergency handling capability Improve energy-efficiency of the medical sensors used. Context awareness for adapting to change in the patient’s health Provide feedback to patients based on their health data trends

13. Ayushman

14. Ayushman*: A Pervasive Healthcare System * Sanskrit for long life Environmental Sensors (Temperature etc) • Project @ IMPACT Lab, Arizona State University • To provide a dependable, non-intrusive, secure, real-time automated health monitoring. • Should be scalable and flexible enough to be used in diverse scenarios from home based monitoring to disaster relief, with minimal customization. Internet Stargate Gateway External Gateway Central Server Medical Sensors (EKG, BP) controlled By Mica2 motes Medical Professional Home/Ward Based Intelligence Body Based Intelligence Medical Facility Based Intelligence Vision • To provide a realistic environment (test-bed) for testing communication • protocols and systems for medical applications.  K. Venkatasubramanian, G. Deng, T. Mukherjee, J. Quintero, V Annamalai and S. K. S. Gupta, "Ayushman: A Wireless Sensor Network Based Health Monitoring Infrastructure and Testbed", In Proc. of IEEE DCOSS June 2005

15. Medical Data Management Architecture

16. Current Setup database RS232 Base Station Oximeter 802.11 ZigBee Central Server Blood Pressure Environmental Data (accelerometer, Temperature, humidity, Light) Bluetooth Internet Body Area Network • Properties • Hardware and software based architecture • Multi-tiered organization • Real-time, continuous data collection • Query support (past, current data) • Remote monitoring capability through the Internet • Simple alarm generation Remote Clients

17. Enabling Technologies Iris TOS v.1.x-2.0 MicaZ Imote2 Mica2Dot Ad-hoc Networking TelosB Mica2 + Commercially available sensor boards Open source OS with support for ad hoc networking

18. Phone to WSN Interface • Design Principles: • To minimize the changes to the existing WSN architecture (required to maintain backward compatibility with previous apps.) • To leverage COTS hardware and existing software solutions (to minimize the development time). • Issues to address: • Phone to sensors interface • Data handling on the cell phone Monitoring and Control Software

19. Phone to Sensor Interface: We evaluated three options: Direct Connection WAN Connection Bluetooth

20. Standard Data Flow Transparent Data Forwarding Transparent Data Forwarding Data Data Data RADIO TX MCU Data Radio MCU UART UART Cmd RX Cmd Cmd Cmd TinyOS Pkt Data TinyOS Pkt Data 0101101100110 RS232 / USB Processing &Data Storage WSN Gateway TinyOS Pkt Cmd TinyOS Cmd 0101101100110

21. Bluetooth Gateway Bluetooth BlueRadios BR-C40A With Serial Port Profile BT RFCOMM TelosB with the Bluetooth Module Cmd Data RS232 / USB Data Data Radio MCU UART Cmd Cmd

22. Software on the Phone Bluetooth RFCOMM • Phone Software: • Establishes the connection to the BT gateway • Handles the TinyOS packet structure • Coverts the raw sensor data in engineering units • Provides the users with a GUI • Dispatch control commands to the WSN

23. The Personal Sensor Network

24. Desktop Screen Shot Patient Details Current Sensor Value Sensor Values Trend Query Result: Archived Data Location of Server

25. VIDEO DEMO – Using CellPhonefor Sensor Monitoring

26. Other Similar Projects Proactive Health Project @ Intel Developing sensor network based pervasive computing systems Managing daily health and wellness of people at homes Proactively anticipate patient’s need and improve quality of life. Code Blue Project Sensor network based health monitoring @ Harvard Developing sensor network based medical applications for: Emergency Care Disaster Management Stroke patient rehabilitation AMON Project @ ETH, Zurich Developing multi-functional wearable health monitor E.g.: BP, pulse, SpO2, ECG, Temperature Aware Project @ the Center Pervasive Healthcare, University of Aarhus, Denmark. Applying context aware computing to hospital scenarios Developing context aware hospital bed, pill box which is aware of its patients. SMART: Scalable Medical Alert and Response Technology PROJECT @ MIT Help patients and providers Allow dynamic monitoring of simple physiologic signs and location Improve communication of patient data http://www.intel.com/research/prohealth/ http://www.eecs.harvard.edu/~mdw/proj/codeblue/ http://www2.wearable.ethz.ch/amon.0.html http://www.pervasive-interaction.org/Aware/ http://www.csail.mit.edu/events/news/2006/smart.html

27. Third International Conference on Body Area Networks Tempe, Arizona, USA March 13 – March 15, 2008 • Synopsis: • Recent advances in the field of wireless sensor networks have moved them beyond their traditional areas of application in monitoring of remote and mobile environments. Sensor networks are increasingly being deployed within and at the surface of the human body to form Body Area Networks(BodyNets). They can be utilized in diverse applications such as: • Physiological monitoring • Human computer interactions • Education and entertainment through interactive games. • BodyNets 2008 aims to establish a forum to bring together research professionals from diverse fields including computer science, biomedical engineering and medicine in both academia and industry to address the technical, social, systems and application issues related to BodyNets. Organizing Committee: General Chair: Sethuraman (Panch) Panchanathan, ASU, USA TPC Chair: Sandeep K. S. Gupta, ASU, USA TPC Co-Chairs: Daniel Siewiorek, CMU, USA Timothy Buchman, WUSTL, USA Loren Schwiebert, WSU, USA Industry Track Co-Chairs: Carlos Cordeiro, Intel Research, USA Mary-Murphy Hoye, Intel Corp., USA Conference Coordinator: Zita Rozsa, ICST, Europe Publicity Chair & Webmaster: Krishna Venkatasubramanian, ASU, USA Local Arrangement Chair: Georgios Varsamopoulos, ASU, USA Publications Chair: Leif Hanlen, NICTA, Australia Sponsorship Chair: Gianni Giorgetti, ASU, USA Steering Committee Chair: Imrich Chlamtac, Create-Net, Italy General Vice-Chair: David Tacconi, Create-Net, Italy Address: Life Sciences A-wing, Room 190 Arizona State University (Tempe Campus) Tempe, AZ 85287USA http://www.bodynets.org/

28. Technical Program Committee: Oleg Medvedev Moscow State University, Russia Alex Mihailidis University of Toronto, Canada Joe Paradiso MIT Media Labs, USA Nissanka Priyantha Microsoft Research, USA Vimal Patel BMI, ASU Chiara Petrioli University of Rome 'La Sapienza', ItalyMajid Sarrafzadeh UCLA, USA Sarma Vrudhula ASU, USA Matt Welsh Harvard University, USA Paul Wright UC Berkeley, USA Eric Yeatman Imperial College, U.K. Lin Zhong Rice University, USA Ionut Aaron IBM T. J. Watson Research Labs, USA Metin Akay ASU, USA Mostafa H. Ammar Georgia Tech, USA Tracy Camp Colorado School of Mines, USA Carlos Cordeiro Intel Corp., USA Chaitali Chakrabarti ASU, USA Roozbeh Jafari UT Dallas, USAWillian Kaiser UCLA, USATom Martin Virgina Tech, USA Sal Mastroianni Motorola Labs, USA Deep Medhi University of Missouri-Kansas City, USA BodyNets 2008 Keynote Speakers: Title: Artificial Intelligence on the Body, in the Home, and Beyond Speaker: Dr. Diane Cook, Washington State University Title: On Innovation, Quality of Life and Technology of BodyNets Speaker: Dr. Sundaresan Jayaraman, Georgia Institute of Technology • Tutorial Speakers: • Title: • Body Sensor Networks for Health-care Monitoring: Premises, Challenges and Prospective • Speaker: • Dr. Roozbeh Jafari, University of Texas at Dallas • Title: • Energy-efficient Design for Mobile Phone-Centered Wireless Body Area Networks • Speaker: • Dr. Lin Zhong, Rice University

29. BodyNets 2008:Technical Program Communication Techniques Effect of quantization on beamforming in binaural hearing aids Sriram Srinivasan , Ashish Pandharipande , Kees Janse Investigation of Wireless Data Transmission between Hearing Aids Crista L. Malick, Steven J. Franke, Qi Xie, Jennifer T. Bernhard, Mitesh Parikh, Douglas L. Jones, and Francois Callias. Body-Coupled Communication for Body Sensor Networks Adam Barth, Stephen Wilson, Mark Hanson, Harry Powell , Dincer Unluer, John Lach Analysis of Body Sensor Network Using Human Body as the Channel Jerald Yoo, Namjun Cho, Hoi-Jun Yoo Software Technology & Platforms Distributed Pervasive Services using Group Service communication supporting Body Area Networks Christopher Foley, Sasitharan Balasubramaniam, Dimitri Botvich, WilliamDonnelly, Stefan Michaelis, Jens Schmutzer, Thomas Stair Service Discovery and Composition in Body Area Networks Matteo Coloberti, Clemens Lombriser , Daniel Roggen, Gerhard Troester, Renata Guarneri, Daniele Riboni A Framework for Creating Healthcare Monitoring Applications Using Wireless Body Sensor Networks Sameer Iyengar , Filippo Tempia , Raffaele Gravina , Antonio Guerrieri , Giancarlo Fortino , Alberto Sangiovanni-Vincentelli CareNet: An Integrated Wireless Sensor Networking Environment for Remote Healthcare Yuan Xue , Stephen Wicker , Philip J Kuryloski , Shanshan Jiang, Roozbeh Jafari, Ruzena Bajcsy , Yanchuan Cao , Sameer Iyengar Communication Protocols On the performance of Bluetooth and IEEE 802.15.4 radios in a body area network Rahul C. Shah, Lama Nachman and Chieh-yih Wan IEEE body area network and medical implant communication Bin Zhen, Huan-Bang Li, Ryuji Kohno ZigBee-Based Wireless Sensor Network for Real-Time Transmission of Wavelet Compression of ECG Signals Shuo-Jen Hsu, Shih-Wei Chen, Wan-Ya Chen, Hsin-Hsien Wu, You-Yin Chen Novel QoS Scheduling and Energy-saving MAC protocol for Body Sensor Networks Optimization Begonya Otal, Luis Alonso, Christos Verikoukis Powering and Energy Adapting Radio Transmit Power in Wireless Body Area Sensor Networks Shuo Xiao, Vijay Sivaraman, and Alison Burdett Approaches to Self-Powered Biochemical Sensors for In-Vivo Application Eric M. Yeatman, Danny O’Hare, Cate Dobson, Eleni Bitziou Joint Encryption/Multiple Access for Body Area Sensor Networks Walter D. Leon-Salas, Deep Medhi, Yugi Lee

30. BodyNets 2008:Technical Program HCI/Wearable Computing SMASH: A Distributed Sensing and Processing Garment for the Classification of Upper Body Postures Holger Harms, Oliver Amft, Daniel Roggen, Gerhard Troester Modeling of EOG and Electrode Position Optimization for Human-Computer Interface Niina Nojd, Jari Hyttinen An Architecture for Smart Textiles Mark T Jones, Thomas Martin, Braden Sawyer Activity and Signal Classification ECG Segmentation in a Body Sensor Network Using Adaptive Hidden Markov Models Huaming Li, Jindong Tan Body Posture Identification using Hidden Markov Model with Wearable Sensor Networks Muhannad Quwaider, Subir Biswas Classifying Wheelchair Propulsion Patterns with a Wrist Mounted Accelerometer Brian French, Asim Smailagic, Dan Siewiorek, Vishnu Ambur, Divya Tyamagundlu Analysis of human performance using physiological data streams Gaurav Pradhan, Balakrishnan Prabhakaran Medical Applications The SmartCane System: An Assistive Device for Geriatrics Winston Wu, Lawrence Au, Brett Jordan, Thanos Stathopoulos, Maxim Batalin, William Kaiser, Alireza Vahdatpour, Majid Sarrafzadeh, Meika Fang, Joshua Chodosh Preliminary Studies for the development of a Ubiquitous Computing and Health-monitoring System for Wheelchair Users Jongbae Kim A wireless platform for fall and mobility monitoring in health care Pepijn W J Van de Ven, Alan Bourke, John Nelson, Gearaid O’Laighin Physiological Signal Monitoring in the Waiting Areas of an Emergency Room Dorothy Curtis, Jason Waterman, Jacob Bailey, Eugene Shih, Thomas Stair, John Guttag, Robert Greenes, Lucila Ohno-Machado Non-Medical Applications The Speckled Golfer D K Arvind A Wearable Wireless RFID System for Accessible Shopping Environments Sreekar Krishna, Vineeth Balasubramanian , Narayanan Chatapuram Krishnan , Terri Hedgpeth SerPens -- A Tool for Semantically Enriched Location Information on Personal Devices Sourav Bhattacharya, Joonas Kukkonen, Petteri Nurmi, Patrik Floreen

31. Medical Sensor Safety

32. Tissue Heating • Medical sensors implanted/worn by human need to be safe. • Sensor activity causes heating in the tissue. • Heating caused by RF inductive powering • Radiation from wireless communication • Power dissipation of circuitry • Goal: minimize tissue heating. • Two solutions: • Communication scheduling for minimizing thermal effects: • Rotate cluster leader – balance energy usage + distribute heat dissipation • Thermal aware routing: route around thermal hotspots Tissue Blow-up Heating Zone Cluster leader

33. Communication Scheduling System Model • Consider only one cluster • 2D Model • Rotate cluster head to distribute energy consumption & reduce heating Requirements • FCC Regulation • Antenna vs. Freq trade-off SAR = σ E2 / ρ (W/kg) E = induced Electric Field p = tissue density σ = electric conductivity of tissue IEEE Requirement (1g Tissue) Temperature Rise: Pennes Bio-heat Equation Whole Body Average SAR = 0.4W/Kg Peak Local SAR = 8W/Kg CE Whole Body Average SAR = .08W/Kg Peak Local SAR = 1.6W/Kg UCE Heat by power dissipation Heat by metabolism Heat by radiation Heat accumulated Heat transfer by convection Heat transfer by conduction Solution • Random selection may lead to higher temperature rise • Similar to Traveling salesman problem but with dynamic metric • TIP Heuristic: Leader selection based on sensor location, rotation history Results FDTD + enumeration Optimal 720960 hrs (est.) FDTD + Genetic Algorithm 100 hrs (est.) Near Optimal TIP +enumeration Optimal 7.6 hrs Near Optimal TIP + Genetic Algorithm 5 min Four Approaches • FDTD + enumeration • FDTD + Genetic Algorithm • TIP + enumeration • TIP +Genetic Algorithm Temperature Temp rise in sensor surroundings Comparative Result Coordinate y Coordinate x Q. Tang, N. Tummala, S. K. S. Gupta, and L. Schwiebert, Communication scheduling to minimize thermal effects of implanted biosensor networks in homogeneous tissue, Proc of IEEE Transactions of Biomedical Engineering

34. Solution Modeling EM radiation and power dissipation of sensors Identifying hotspot area Withdrawal strategy to avoid overheated area Averaging power consumption and heat dissipation Slight degradation of delay Thermal Aware Routing Area Hotspot • In vivo environment maybe sensitive to the heating of power dissipation and radiation of Implanted sensors • Energy/load balancing is not equal to heating balancing: large time scale vs. short time scale Link Hotspot Temperature distribution of TARA Q. Tang, N. Tummala, S. K. S. Gupta, and L. Schwiebert, TARA: Thermal-Aware Routing Algorithm for Implanted Sensor Networks, Intl Conference on Distributed Computing in Sensor Systems, 2005

35. Information Security

36. Security in Pervasive Healthcare Context • Patient data is transmitted wirelessly by low capability sensors • Patient data is therefore easy to eavesdrop on • Security schemes utilized may not be strong enough for cryptanalysis • Patient data is stored in electronic format and is available through the Internet • Makes it easy to access from around the world and easy to copy • Data can be moved across administrative boundaries easily bypassing legal issues. • Electronic health records store more and more sensitive information such as psych reports and HIV status • Preserving patient’s privacy is a legal requirement (HIPAA) Excruciating Factors • Wireless connectivity is always on • No clear understanding of: • Trusted parties • Security policies for medical environment • Devices are heterogeneous with limitedcapabilities • Traditional schemes too expensive for long term usage

37. Security Related Issues Technology • Efficientcryptographic primitives • Cheaper encryption, hash functions • Better sensorhardware design • Cheap, tamper-resistant sensor hardware • Better communication protocol design • Better techniques for controlling access to patient EHR New Attacks • Fake emergency warnings. • Legitimate emergency warningsprevented from being reported in times. • Unnecessary communication by malicious entity with sensors can cause: • Battery power depletion • Tissue heating Legislation • Health Information Privacy and Accountability Act (HIPAA) • Passed in 1995 • Provides necessary privacy protection for health data • Developed in response to public concern over abuse of privacy in health information • Establishes categories of health information which may be used or disclosed Requirements • Integrity - Ensure that information is accurate, complete, and has not been altered in any way. • Confidentiality - Ensure that information is only disclosed to those who are authorized to see it. • Authentication – Ensure correctness of claimed identity. • Authorization – Ensure permissions granted for actions performed by entity.

38. ECG, Heart/Pulse Rate + Blood Pressure Properties Blood Glucose • Universal: Should be measurable in everyone • Distinctivelycollectable: Should be measurable in an unambiguous manner • Random: To prevent brute-force attacks • Timevariant: If broken, the next set of values should not be guessable. Value Time  Physiological Value based Security Aim • Use of the physiological values (PV) from the body to exchange the keys. • Possible Examples: • Simple • Blood Pressure, Heart Rate, Glucose level • Complex • Temporal variations in different PVs. • Combination of multiple PV Advantages • Easier and safer key generation • Reduced Deployment Costs • Plug-n-Play like capability with Body Sensor Networks GOODCHOICES: Inter-Pulse-Interval FIND OTHERS … Sriram Cherukuri, Krishna K. Venkatasubramanian, Sandeep K. S. Gupta, “BioSec: A Biometric Based Approach for Securing Communication in Wireless Networks of Biosensors Implanted in the Human Body”, in Proc of IEEE ICPP Workshops, 2003

39. Inter-Pulse-Interval (V’1) Inter-Pulse-Interval (V1)  EKG EKG Inter-Pulse-Interval (V2) = = PPG Inter-Pulse-Interval (V’2) PPG  Time Domain Analysis Measurement • Measure Inter-pulse intervals (IPI) from two sources: EKG and PPG • Sampling Frequency: 1000Hz, 14 health patients, 85 older and sick patients Quantization • 67 consecutive IPIs from a person quantized into 2, 128 bit binary streams (keys) • Comparison criteria False Rejection (FR) and False Acceptance (FA) Result • The value of two keys, but close for same person. • Comparison criteria: Hamming Distance • A threshold of ~40 bits (older patients) minimizes FR & FA • Results for two PPG data series based IPI collection yielded better result First Proposed: C. C. Y. Poon and Yuan-Ting Zhang and Shu-Di Bao, “A Novel Biometrics Method To Secure Wireless Body Area Sensor Networks for Telemedicine And M-Health”, IEEE Communications Magazine, 44(4), 2006, pp 73-81. Has been verified by us.

40. IPI values 1 16 256 Window FFT Signal sample FFT Coefficient Table (FCT) 16 (windows) 8 (coefficients) Quantize (4 bits) Minimum Hamming Distance EKG X Key1 Block 1 Block 16 X C Q H I P 16 EKG Blocks Block 1 Block 16 Distance Matrix P A C J L Realign & Choose PPG 32 bits 16 PPG Blocks 8 coefficients * 4 bits = 32 bits Key2 Frequency Domain Analysis Steps • Collect 256 IPI values (EKG & PPG) • MIT Physio Bank Database • For both: • Divide them into windows of 16 and perform 16 point FFT • Drop latter half coefficients & generate FCT • For each row in FCT: • Quantize into 4 bits • Yielding 16 blocks, 32 bits each. • Pair wise compute hamming distance of each of the 16 blocks for EKG with PPG • Find the closest blocks and realign them • Choose first 4 blocks for 128 bit key Results • Same person: keys 10% apart, different person: keys 30-40% apart • For EKG, keys identical for same person

41. PV Based Data Security: Protocol Measure Pre-defined PV @ Sender PVs & Receiver PVr Generate Random Key @ sender KeyRand Encrypt message with Key Rand C = EKeyRand(Message) γ = PVs  KeyRand Hide KeyRand using PV Send encrypted message Receiver encrypted message KeyRand= PVr  γ Unhide KeyRand using PVr Message’ = DPVr(C) Decrypt message with Key Rand K. Venkatasubramanian, and S.K.S. Gupta, "Security For Pervasive Health Monitoring Sensor Applications", To Appear in Proc of 4th Intl. Conf. on Intelligent Sensing and Information Processing (ICISIP), December 2006.

42. Criticality-Awareness

43. Critical Events Cannot be responded to, using the routine set of capabilities of the subjects. Requirements Request based context evaluation is inadequate. Continuous context monitoring is required. Emergency Management Fire Medical Emergency Hurricane Flooding Criticality • Consequences of critical events characterized by urgency for taking remedial (response) actions • Responseactions are usually exceptional in nature • Usually happen in groups (earthquake + severely hurt people) Exceptional Actions Normal Situation Criticality Criticality Awareness improves System DEPENDABILITY

44. Important Properties of Criticality Responsiveness • Measures the speed with which the system initiate detection of criticalities Correctness • Determines the accuracy and confidence of the detection process. Window of Opportunity (Wo) • Time within which all mitigative actions should ideally be taken • Value of Wo is criticality dependent. • Example • 90 Sec (Data Center, cooling failure) • 5 Min (Tornado) • 1 Hour (Heart Attack) • 30 Days (Disaster Recovery) D + Ta ≤ Wo Time for Initiating mitigative actions Time to take mitigative actions

45. Criticality Mitigation Process Detection (Humans, sensors etc) Evaluation Planning/ Scheduling Enabling Actions Planning & Scheduling Execution of Actions (Humans, Agents etc) Control Access to Resources

46. Criticality and Access Control & • Provide medical information (EHR) automatically through: • Patient medical sensors/ PDA /cell -phone directly • Preserve patient privacy as per HIPAA disclose EHR only to associated doctors • Rescue doctors don’t get access as per HIPAA • How to make it work Hurricane (Natural Disasters) Destruction and Flooding Rescue FURTHER • Traditional access to EHR is REACTIVE • Initiated by medical professional after observing the patient • Slow response • How to speed it up • Proactive system monitoring. • Facilitates reaction within a window-of-opportunity. • Provides privileges for non traditional accesses for criticality mitigation. • Properties: • Proactive – takes access control decisions independently of specific user request • Alternate Privilege Provision – provide any privileges for mitigation, • Wo-aware – rescind privileges after Wo expires • Dynamicity – not adhere to any assumpitons regarding criticality or its behavior • Non-Repudiability – maintain detailed records of actions taken during criticality Criticality Aware Access Control S.K.S Gupta, T. Mukherjee, K. Venkatasubramanian, “Criticality Aware Access Control Models for Pervasive Applications”, In Proc of IEEE Pervasive Computing, 2006.

47. Locate people, & provide aid, based on ailment Refer EHR for informed diagnosis and treatment Criticality Aware Access Control Detect Criticality Plan/Schedule Actions Execute Actions Grant Privileges For each trapped + hurt person , their pervasive health monitoring system: • Obtain info on doctors in vicinity. • Check if A1, A2 and A3 allowed for them • If not present, generate privileges • P1. View past health info • P2. View current health info • P3 . View allergy information • Assign privileges to doctors simultaneously. • Record actions, if taken • People Injured in the aftermath of a natural disaster • Check periodically for new criticalities • New plan & schedule if Wo expire or new criticality • Reset all previous privilege assignment Proactive • Obtain health information • Compute type of ailment, possible treatment, Wo. • Generate list of actions to facilitate treatment A1. Provide past health info A2. Provide current health info A3. Provide allergy information Alternate Privilege Provision Wo-aware Non-Repudiability • Whole process carried out by Pervasive healthcare system • Actions generated may sometimes contradict, such cases may mandate sequential assignment of privileges • Role-Privilege model used for implementation, where doctor’s role changed for assigning privileges. • Privileges provided for actions generated and not predetermined for different criticalities • Detection process done periodically interval system dependent • Carried out by doctors Dynamicity Comments

48. Context Awareness Acutation Other Issues Energy-Efficiency

49. Energy Efficiency Solutions Need • Sensors have very small battery source. • Sensors need to be active for long time durations. • For implantable sensors, it is not possible to replace battery at short intervals. Challenge • Battery power not increasing at same rate as processing power. • Small size (hence less energy) of the batteries in sensors. Better Battery Solar Energy Vibration Body Thermal Power

50. ‘k’ Bits ‘n’ Bits Info Source ME coding Modulator RF Transmitter 1 0 0 1 0 1 …… 0 0 0 1 0 0 0 0 0 1 0 0 0 0 …… k -source bits n - code bits Energy-Efficiency Source Coding Biosensor Communication M Symbols = 2k Need • Sensors have • low data rate. • Short range of operation. • Demands low power and low complexity at both circuit and system level. Solution • Minimum Energy Coding • Sources with unknown statistics. • Minimum energy codes considered • More energy efficient. • Only one bit-1 per code • Achieves • Lesser number of bit-1 in the transmitted code • Safely assign to source symbols of any probability of occurrence. • Code Rate = (k / n) = (k / 2k-1) System Model Y. Prakash, S.K.S Gupta, Energy Efficient Source Coding and Modulation for Wireless Applications, IEEE Wireless Communications and Networking Conference, 2003. WCNC 2003. Volume: 1, 16-20 March 2003, Page(s): 212-217.