1
Energy-Efficient and Multi-modal Body Area Sensing System for Remote Diabetes Management System Architecture Sensor Data Doctor Suggestion •Data collector aggregates different types of data into a single record. Local storage stores user configuration, analyzed data and raw data in the external storage. • Some analysis are done on phone while some analysis are sent to the remote server to save battery life. After the analysis is done on server side, online doctor suggestion are sent back to the mobile client. Qiumin Xu, Ling Hu, Sangwon Lee, Murali Annavaram, Farnoush Banaei-Kashani Integrated Media Systems Center University of Southern California Challenges •Energy consumption are critical in real-time sensing systems. There are many choices at each Wireless Body Area Network (WBAN) design stage. • Initial Development: Efficiency vs. Programming simplicity • Sense: Sampling rate vs. accuracy • Transmit: Signal quality vs. decoding complexity & encryption • Local computation vs. remote computation. •Each choice has dramatic impact on power consumption, however designer has little knowledge of the energy impact of their choice. • Energy impact varies dynamically • Signal quality for data transmissioin, Indoor / Outdoor GPS, Compression factors. • Dramatic battery drain reduces 200 hours standby to 5 hours. Even without external sensors the in-built sensors also drain battery. Related Work • Energy comparison between 3 WBAN functions, 3 languages and 3 models QDA: QRS Detection, AES: Encryption, ZIP: 10 min data (180KB) • PyS60 Energy >> J2ME > C++ This difference is due to runtime environment overheads and memory management. Active Energy API • Provide a set of API for designers to obtain system services at the lowest energy cost, such as GPS, data transmission • API automatically selects the best implementation • Implementation relies on Active Energy Profiling framework Experiments • Runtime energy profiling using active energy API • Results from Active Energy API (AEP) • AEP has a shallow slope • Energy savings increase with time. • After 30 minutes • Local: 773 Joules • Remote: 487 • AEP: 416 Joules Introduction •The alarming rise in diabetes rates requires thorough understanding in biological reasons, social and environment impact. •A real time multi-modal body area sensing system can provide a powerful database for medical research, as well as remote diabetes management. •Battery life in mobile device is critical in continuously sensing system, therefore an energy-efficient framework is essential. Conclusion • Energy efficiency must be dealt with in all aspects of the WBAN design, selecting from programming language to sensor sampling rate. Energy efficiency also plays major role in robustness. • An Active Energy API is developed to automatically select the optimized implementation. Experiment shows that around 2X battery improvement is obtained. iCampus iWatch CT Database Analysis Website 13.83 0.07 1.24 13.82 0.06 0.79 3.52 35.33 0.54 25.97 38.76 0.42 12.65 0.04 517.00 147.88 2.42 377.19 91.76 1.74 0.61 0 1 10 100 1000 QDA AES Gzip QDA AES Gzip Q A G Nokia N95 Nokia E75 iPhone 3GS Execution Time (Seconds) C++ J2ME PyS60 AEP START Transmit Data using Selected Networks END Get GPS Coordinate Collect Sensing Data MakeDecision(Data, Length, Address) GetPosition() Position Information Get Activity Data Using Local Computation Local or Remote Get Activity Data Using Remote Computation Local? Yes No SendData(Data, Length, Address) Results of Remote Computation SendData(Data, Length, Address) Response from Server Execute Make Decision() Onetime Initialization With Sample Data Initialize(Data, Length, *SA) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Power(mW) Time(Minutes) Bluetooth Communcation with Two Alive Heart Reate Monitors 3G Upload data w/ Gzip Bluetooth Communcation with Two Alive Heart Reate Monitors Positioning Local Computation & Gzip (Initialization) WiFi Upload data w/o Gzip Energy Profiling Energy Profiling 0 100 200 300 400 500 600 700 800 900 0 5 10 15 20 25 Energy(Joules) Time(Minutes) Local with 3G Remote with 3G AEP
