VIEWS: 2 PAGES: 1 POSTED ON: 4/27/2014
Efficient Mapping and Management of Applications onto Cyber-Physical Systems Prof. Margaret Martonosi, Princeton University and Prof. Pei Zhang, Carnegie Mellon University Motivation Testbed: Minimalistic Controlled Mobile Sensor Region-Scale Mobility Modeling § Model Human Mobility based on cellphone Call Detail § Cyber-Physical Systems: richly heterogeneous devices • Unreliable nodes: expected to fail, Records (CDRs) or similar aggregate info: (mobile devices, home electronics, taxis, robotic drones, § Sparse in time: Only when phone call in/out 0101 can crash or get etc.) that together gather sensor data, analyze it, and 010 § Coarse in space: Only to granularity of cell tower spacing. (Not 010 stuck coordinate large-scale actions in response to it. GPS) • Low weight, 010101 Challenge: How to program CP Systems? Low-cost prototype 01010 § 0010 • Adapt to changing 1 environment, reassign task based on node capabilities and localized Prior Approaches failures • Dynamic coverage estimation: § Our Approach offers accuracy sufficient for detailed regional § Simplifications to reduce complexity. by obtaining in-network relative studies of human and vehicular movements. § Assume homogeneous systems path signatures, then assign task to PLUGIN SENSOR MODULE: § Program for one particular deployment multiple nodes to compute 3-Axis Accelerometer 3-Axis Gyroscope MAIN BOARD: § Current approach leads to “brittle” systems. signature paths 3-Axis Magnetometer 16 MHz AVR AtMega128RFA1 16Kb SRAM 128Kb Flash Memory Mobile Offloading: An Initial Example § Deployments with multi-generation devices 2.4GHz 802.15.4 Radio § Multiple scenarios with different mobility capabilities § Select when to use 3G and when to use WiFi based on: Collaborative Coverage Without Location § Availability and bandwidth of 3G and WiFi 1. How to meet performance and accuracy goals while managing power § Delay tolerance of application and other scarce resources? • Sensing coverage for mobile § Cost of data transfers on each network 2. How to select which devices to use? From a static or dynamic pool of nodes in many § Mobility prediction of device resources. environments is hard to § Optimal MILP Scheduler Formulation 3. How to support dynamic adaptivity within a single deployment? How determine due to lack of to support portable operation across different deployments? known infrastructure or references for locations. • Sensing coverage estimation by obtaining relative motion Our Project path signatures. Dynamic task allocation allow nodes 1. Abstraction layer to allow CPS applications to express to more efficiently coordinate and predict the application needs current sensing coverage of § 8-10X cost reduction due to optimization and delay tolerance § Coverage and sensing requirements an area. 2. Device attribute catalog to summarize local nodes and their Publications capabilities Project Future Work § Sibren Isaacman et al. Human Mobility Modeling via Synthetic Call § Sensing capabilities, probability of success, accuracy, etc. Detail Records." 10th Intl. Conf. on Mobile System, Applications, and § Integrate WiFi offloading with mobility prediction Services (MobiSys 2012) 3. Model, Prediction and Control mobility of nodes § Broaden mapping problems from two-node offload to multi-node § Ozlem Bilgir Yetim and Margaret Martonosi. "Adaptive Usage of § Different types of motion (people, fixed sensors, robot, etc.) offload Cellular and WiFi Bandwidth: An Optimal Scheduling Formulation § Explore use of virtualization for mobile migration (Short Paper).” 7th ACM Intl. Workshop on Challenged Networks § Apply programming flow and dynamic adaptation to real-word (CHANTS 2012) Acknowledgments applications: § Frank Mokaya, Aveek Purohit, Pei Zhang, “Invited Paper: SensorFly: § First responder support Flying Sensor Network for Indoor Situational Awareness in a Disaster”, This work was supported by the National Science Foundation under The Wireless Personal Multimedia Communications Symposium collaborative grants CPS-1135874 and CPS-1135953. § Regional automotive traffic management (WPMC’12) Sep. 2012
"A Back-End Design Flow for Single Chip Radios - Princeton University"