1. Opportunistic Medical Monitoring Using Bluetooth P2P Networks Dae-Ki Cho, Seung-Hoon Lee, Alexander Chang, Tammara Massey, Chia-Wei Chang, Min-Hsieh Tsai, Majid Sarrafzadeh and Mario Gerla Department of Computer Science University of California, Los Angeles

2. Outline Motivation Introduction Bluetooth Overview Patient Monitoring and Record Management – NurseNet – BlueAlert Experiment Simulation Conclusion 2

3. Motivation Gaining popularity of ubiquitous computing in medical field Tiny, non-intrusive computing nodes, integrating sensors, actuators and radios can be applied to the human body to monitor vital signs. Limited bandwidth of GSM (e.g., EDGE), using Bluetooth or WiFi in medical monitoring is more appropriate. However, access to Bluetooth and WiFi access points may be intermittent The data can propagate hop-by-hop through the peers

4. Introduction To examine the feasibility and effectiveness of – Opportunistic ad hoc networking using Bluetooth – Data muling of medical records from patients to the Internet medical database. – BlueTorrent A P2P file sharing application based on ubiquitous Bluetooth-enabled devices The Bluetooth network may be used to alert a nurse that a patient has a medical emergency requiring immediate attention.

5. Bluetooth Overview Extended Inquiry Response (EIR) – Short, urgent message propagation – Supported by a new BT version – No connection delay

6. Patient Monitoring and Record Management Identifying two scenarios and propose two techniques NurseNet and BlueAlert NurseNet – Large scale techniques (e.g., WiFi, 3G) are not adequate or not allowed in some environments. BlueAlert – The conventional Bluetooth-based patient monitoring system is not suitable for emergencies due to Bluetooth connection establishment time (typically, 5-10 seconds).

7. Scenario I: NurseNet NurseNet is a patient data uploading system to the Central Database The patients device passes the medical data to a caregiver (say, nurse) device that then transfers the stored data over Bluetooth P2P to the database. Figure: NurseNet Architecture

8. Scenario II: BlueAlert BlueAlert is an emergency alarm protocol – For urgent messages, Bluetooth devices change their data propagation mode from BlueTorrent to BlueAlert. Use a new Bluetooth feature called EIR – EIR travels across a Bluetooth overlay without any connection delay

9. Experiment Environments Where: Parking lot Size: ~ 75 X 75 meters Open Space battlefield or disaster area When: Late Night Minimize interference Three Components Patient/Nurse/Doctor Used Kensington 33348 Bluetooth dongles (v2.0 EDR, Class 2, and Broadcom chipset) for nurse nodes, patient node and BT-AP node.

10. Experiment Setting Patient Moves inside a designated area of 10x10 meters Equip with two different body sensors – ECG and Pulse Oximeter Continuously transmit data to Gateway through Bluetooth Connection Nurse (Three nodes) – Collects data from patients – Moves around – Exchange data with other nurses Doctor – Data collector

11. Experiment Result All nurse nodes contribute to data dissemination Nurses exchange data, then upload data to AP As the number of hops increases, data delivery delay decreases

12. Simulation Setting: NurseNet Simulation Parameters – Area: 100 x 50 meters – Nodes: 50 patients + 5 nurses – Node Speed: 1 meter/seconds (Only nurse moves, Patients are static) – P: 20% (Returns to the main office.) At the office, the nurse uploads the data collected so far at one of the BT-APs. ………… Patient Nurse

13. Simulation Setting: NurseNet (cont) The setting is a Field Hospital. Each patient has body sensors and a Bluetooth-enabled gateway. The sensors keep generating medical data. Patient: static (50 nodes) Nurses: Mobile (5 nurses, average speed: 1m/s) – Stay with a patient for 5 minutes on average Examine two different Bluetooth Overlay mechanisms – P2N(Patient-to-Nurse): Nurse collects data from one patient – P2N + N2N(Nurses-to-Nurse): Nurses exchange data each other

14. Simulation Result: NurseNet Collection time – The collection delay is quite high. This is in part due to the assumption that the nurse does not collect data during the pause. Relaxing this constraint reduce the collection from 3000 to 100 seconds from annalistic model Uploading time – With N2N exchange, data upload rate to the hospital increases (and latency decreases) with number of nurses and with P. Collection timeUploading time

15. Simulation Setting: BlueAlert Patients move escorted by nurses Suddenly, one of patients needs an emergency care from a nurse. Emergency data is propagated by P2P using Bluetooth EIR Compare the emergency alarm propagation delay with BT2.0 and BT2.1(EIR) Patient Nurse Emergency

16. Simulation Result II Parameters: – Area: 100 x 100 meters – Nodes: {50, 100, 150 patients} + 5 nurses – Node Speed: 0.5 meter/seconds (Nodes are patients) – Delay measured until one of nurses receives an emergency message from a random patient. Analysis – The more number of hops helps the emergency data to reach to one of nurses – EIR: Even though the number of hops increases, propagation delay decreases because EIR does not take any connection delay among hops; the data is immediately delivered to the next hop. – But 2.0 takes connection setup delays along with each hop and it causes large delay. Number of nodes Number of Hops Delay(seconds)

17. Conclusions Examined Bluetooth based Patient monitoring system Established Feasibility and effectiveness of data muling by Bluetooth Overlays

18. Future Work Simulations – Update the simulation scenario so that the result of our simulation is more close to real world results. Experiments – Implement BlueAlert once Bluetooth v2.1 is available and merge BlueAlert and NurseNet together. – Brain storm on how to utilize EIR other then BlueAlert E.g., When NurseNet nodes exchange their file lists. (Reduce connection time)

19. Thank you Q&A