Exploring Multicore-based Hardware/Software Architectures for Mobile Edge Computing Device

1. Exploring Multicore-based Hardware/Software Architectures for Mobile Edge Computing Device IMPACT Lab Arizona State University

2. Outline • Mobile edge computing, mobile edge computing devices (MECD) • Wireless sensor network (WSN) applications • Desirable MECD features • Explore multi-core architectures for MECD

3. Wireless Sensor Network Hierarchy Back-end servers MECD Mobile edge computing device Networked Sensors

4. WSN Applications • Botanical garden (Ken) • Ayushman (Krishna) • Smart container (Guofeng) • Kids network (Su) Pay attention to: • Structure hierarchy • Potential term project topic

5. Botanical Garden

6. Physical layer impact • High temperatures reduce transmission range. 8 dB at 65 C. • No WiFi farther out. Extension requires self-powered nodes. Solar power [1,2] • Node power consumption. How to measure? [1] http://www.ee.ucla.edu/~kansal/papers/sensys_hsu_05.pdf [2] http://camalie.com/WirelessSensing/WirelessSensors.htm

7. Operating system projects • TinyOS vs. Contiki comparison • Both run on Tmote • Contiki adds protothreads and dynamic program swapping • TinyOS documentation • Hardware abstraction: MSP430, AVR128L; CC1000, CC2420

8. Organization Environmental Sensors (Temperature, Humidity) Patient Internet Local Gateway External Gateway Central Server Medical Professional Body Based Intelligence Home/Ward Based Intelligence Biomedical Sensors (EKG, BP) Medical Facility Based Intelligence Ayushman Vision Rationale: • Aging Population • Increasing healthcare cost • Shortage of medical personnel Goals: • Remote health monitoring (HM) • Test-bed for HM systems • Employ off-the-shelf components • Wireless biosensors • Wearable/in-vivo Desirable Properties: • Self-configuring • Real-time • Scalable Challenges: • Integration of diverse technologies • Minimize data loss • Reliability • Maintaining safety & security Status: • System development and Integration

9. Kids Networks Su Jin Kim

10. Social Science Project • How children’s social interactions (especially preschool) relate to their school success • Observation: • For 10 seconds, observe a target child • Identify the peers that he or she is interacting with • Collect data about interactions (e.g. positive emotions, negative emotions, aggressive behavior)

11. KidNet Project • Motivation • Apply it to older children who may not stay in the same classroom all day • Goals • Record the peers and duration of interacting • Interacting: within some small distance (2-3 ft.) • Track students’ location for safety and security • Advantages • Automatic, Real-time, Scalable

12. Proximity & Localization • Wearable Proximity Sensors • Detection of proximity • Duration of proximity • Localization using fixed nodes • Location of each child

13. Challenges • Accuracy • Detecting an object within 2-3 ft. • Energy • Should operate at least 10 hours • Wearable and Safe Devices • Should not be heavy and hurt kids • Reliable Communication • Indoor: reflection, blockage etc. • Scalability • Need to be expand to an entire school

14. Smart Shipping Container • Rationale • government needs • business needs • Goals • RFID, environmental sensing, communication, event detection, … • Challenges • mobile, large number, non-technical issues, …

15. RFID Reader MICAz mote TelosB mote TelosB mote ML Cargo Tag Sensors Sensors MICAz mote MICAz mote MICAz mote Container: architecture INTER-Container TelosB mote Attached to nearby containers. Proximity motes form an ad hoc (multi-hop) inter-container network. GPS Receiver 1 MICAz mote Container(s) External Hosts Stargate Internal Wireless Sensor Networks USB Memory Card MICAz mote 2.4 GHz 2.4 GHz USB 51-pin Stargate Managing Internal network (hardware, power and security); data processing, & routing outgoing packets to external interface. Ethernet Mobile Computing Computers at point of work (Handhelds) & at the Data Center. Held by custom officers and load/unload workers. Querying current and historical data and DB downloading from the logging systems. Enterprise Servers: Computers at the Data Center. Collecting real-time data from containers, managing DB & responding to critical events reported by containers. 802.11 RS232 PCMCIA Compact Flash GPRS PCMCIA Modem 802.11 Compact Flash card Cellular Network

16. Container: pictures RFID Reader + MicaZ Mote Stargate TeloB MicaZ

17. Mobile Edge Computing Device (MECD) • Back-end servers • High computing power • Global decision/policy maker • Interface to users • Physically fixed • MECD • Mobile • Unmanned • Comm. with server & sensors • via multiple types of networks • Dealing with large amount of • sensors • Networked Sensors • Large number • Mobile • Small form factor • Sensing and limited wireless • comm. capability Scalable reliable Low system cost, flexible

18. Desirable MECD Features • High processing power • Localized data processing • Database management • Event detection • Alert generation • Distributed infrastructure management • Security • Reliability • Real-time • Power efficiency • Network management • self-configurable, self-diagnostic, self-healing • ZigBee, WiFi, WiMAX, Bluetooth, GPRS and Ethernet

19. Desirable MECD Features (cont’d) • Low power consumption • Mobile & unmanned • Virtualization • Integrating various types of sensors from different vendors • MultiOS • Ease of development • Low cost

20. Exploring Multi-core Architectures • High processing power • Low power consumption • Low cost

21. Multi-core processor: high processing power • Homogenous (symmetric): • Symmetric multiprocessing (SMP) • Heterogeneous: Dedicated cores and diverse special purpose cores for hardware acceleration • Data processing • Distributed management • Network protocol • VPRO®

22. Multi-core processor: low power consumption • Reduced dynamic power • Each processor core can be individually turned on or off • Each processor core can run at its own optimized supply voltage and frequency • Fine-grain & ultra fine-grain power management and dynamic voltage and frequency scaling • Dynamic task assignment

23. Multi-core processor: low cost • Reduced hardware • SDR (software defined radio) enabled by a multi-core processor

24. Approach & Deliverable • Approach • Design & analysis to improve the understanding of multi-core processor’s application to MECDs • Deliverable We will answer the following fundamental questions: • A set of feasible multi-core based architectural designs that addresses the emerging requirements for MECDs • An optimal multi-core based architecture (in terms of both computing and communication addressing multiple types of networks and topology) for MECDs • Challenges and restrictions of using multi-core processors in MECDs

25. RA Opportunity • Motivated graduate student • Strong problem solving skills Talk to Dr. Gupta