Network Embedded Systems Sensor Networks Fall 2013 Introduction Marcus Chang, mchang@cs.jhu.edu 1
Embedded System An embedded system is a computer system designed to do one or a few dedicated and/or specific functions It is embedded as part of a complete device often including hardware and mechanical parts By contrast, a general-purpose computer, such as a personal computer (PC), is designed to be flexible and to meet a wide range of end-user needs - Wikipedia 2
Embedded System Example 1 Watch Convert crystal oscillations to date and time GPS Read/Store/Show location Hear Rate Monitor Measure HR Transmit data to watch 3 Image: Garmin
Embedded System Example 2 Movement: Measure direction and speed Selection: Detect clicks Communication: Transmit data to PC 4 Image: Razer
Course Outcomes Learn to program embedded systems Direct hardware access Learn to program mobile devices Low-power system design Learn to program wireless networks Radio protocols 5
Embedded Programming Normal applications vs. embedded systems Normal applications Embedded Systems 6
Hardware Comparison Smartphone JHU Node CPU 1.9 GHz (4C) 20 MHz (1C) RAM 2 GiB 4 KiB Program/Data Storage 16 / 64 GiB 32 KiB / 8 MiB Radio Bandwidth 100 Mbps 500 Kbps Standby Current 8 ma 3 μa 7 Image: Samsung
Trade Offs Processing/Memory/Storage vs. Power Consumption Embedded systems can run for years on battery General Purpose vs. Dedicated Functionality Minimal Hardware Lower cost: $10 8
Course Prerequisites Networking course, Operating Systems course, systems programming Must feel comfortable programming in C Interested in hardware and electronics Grad student attitude (self-directed, deal with open ended problems/assignments, highly motivated, critical, dedicated) 9
What will this class be like? Tuesdays Lecture, topic of the week Thursdays Seminar, 1-2 students present a paper and lead discussion Coding exercise (hand in), discuss small code snippets 2 programming assignments Midterm Final project (no exam) 10
Programming Coding Exercises Expose problems commonly found in embedded systems Programming Assignments Larger, open-ended Written report Final Project Tie in with assignments Design decisions 11
12
Wireless Sensor Networks Sensors, computation, and communication feasible at small packages and large numbers Ability to monitor phenomena at fine spatial AND temporal granularities Vision: Embedded Networked Sensing will reveal previously unobservable phenomena A new type of scientific instrument: macroscope 13
Wireless Sensor Networks for Soil Science A. Terzis, R. Musaloiu-E., J. Cogan, K. Szlavecz, A. Szalay, J. Gray, S. Ozer, C.-J. M. Liang, J. Gupchup, R. Burns 2010 14
Habitat Monitoring Box Turtles lay eggs in the soil where solar radiation provides the heat for incubation Gender is determined by the incubation temperature Male box turtles hatchlings develop @ 27-28 C, whereas females develop @ 29-30 C Global warming has the potential of raising the summer soil temperatures where turtles nest and thus potentially lead to a bias in hatchling sex 15
Koala network architecture Requirements Reliable data collection Long network lifetime Robustness to failures Flexibility Design outline Motes collect measurements to local flash Sleep most of the time (> 99%) to conserve energy Gateway periodically wakes up the network to retrieve mote measurements 16
Soil Temperature and Moisture 17
Four months of data Soil Temperature ( 0 C) Soil Moisture (GWC) Forest Edge Open Grassland Parking Lot This is the first study to demonstrate box turtle nesting conditions in situ. 18
RACNet: A High-Fidelity Data Center Sensing Network C.-J. M. Liang, J. Liu, L. Luo, A. Terzis, F. Zhao., SenSys 2009 19
Data Centers (DC) US data centers consumed enough power for 5.8 million average households in 2006. Roughly 50% is used in supporting equipment (power distribution and cooling) Assumes the worst-case. Power is over provisioned. Devices are over cooled. 20
Data Center Operation Cost Trend Equipment is cheap, management is not 21 http://communities.intel.com/openport/blogs/server/2008/02/20/ datacenter-power-management-power-consumption-trend
DCGenome Project Goal: Reduce the DC operation cost and improve DC efficiency. Profile DC dynamics, such as temperature, humidity, and power. Correlate the data with rack power consumption, cooling system, and server workload. Understand how heat is generated and transferred, and how systems are cooled. 22
Design Space Requirements Reliable data collection Dense network (input/output temperature for each server) Design Outline: Master node with radio Slave nodes connected to master One master per server rack Power: USB scavenging 23
Temperature Contour Map Front and back of server rack 24
MEDISN: Medical Emergency Detection in Sensor Networks. JeongGil Ko, Jong Hyun Lim, Yin Chen, Razvan Musaloiu-E., Andreas Terzis, Gerald Masson, Tia Gao, Walt Destler, Leo Selavo 2009 25
Physiological Monitoring Motivation: track vital signs in structured and unstructured clinical settings Emergency room General ward During transport Emergency response 26
Design Space Requirements Reliable data collection High data rate High mobility Deployable in disaster scenarios 27
MEDiSN Design Outline Physiological Monitors (PMs): Sample medical data, compress, encrypt and transmit over wireless Relay Points (RPs): Forward medical data to Back-end server and management commands from back-end server to individual PMs. Back-end server: persistently stores medical data and drives GUI clients 28
Schedule Week 1: Introduction and Hardware Week 2: Embedded Programming Week 3: Medium Access Control Week 4: Link Estimation and Tree Routing Week 5: IP Networking Week 6: Near Field Communication Week 7: (seminar, no lecture) Week 8: Energy Management Week 9: Review and Midterm Week 10: Time Synchronization Week 11: Localization Week 12: Energy Harvesting Week 13: (seminar, no lecture) Week 14: TBD 29