EMBEDDED SYSTEMS ACTIVITIES @IISc, 2010 - 11
Main activities • Research in various departments • Development support for products • Entrepreneurship activities • Curriculum development and training
Research activities 2010 onwards
Research areas • System design – Design of embedded systems – Verification and testing – SOC – Low power DSP blocks – Reconfigurable machines for communication systems • Circuit and sensors – Low power circuits for embedded applications – Design of MEMs sensors and sensor network components – Energy harvesting systems
Departments and areas • CEDT: System design (7) • CSA: Software for ES (2) • ECE: Low power SOC and MEMs (2) • ME: MEMs sensors (2) • SERC: SOC and verification (2) • Nano Science and Technology: (8)
Some statistics for 2010 • Number of PhD (2010): 8 • Number of MSc (Engg): 10 • Number of ongoing projects (industry and government support): more than 20
Typical research topics • Verification of mixed signal circuits • Reconfigurable DSP blocks for specific applications • Reconfigurable decoders for error control • New arithmetic for low power digital filters • Strategies for interference mitigation in coexistence situations • Low power radio for embedded systems • DRM • Energy neutral sensors: protocols and base energy saving • MEMS and CMOS integration on SOC
Directed research and development 2010 onwards
Capabilities: Technology space Embedded Networking & Telecom Systems Domain labs: India centric apps: Energy generation, management, healthcare, infotainment 18-Mar-11
Current work …1 1. Solar energy for lighting • Considerable energy used for lighting in buildings during the day • Significant saving if solar energy is used • No storage • Connectivity from the grid for back up • Embedded system wireless control for the entire floor 2. Sensor networks for diverse applications – Embedded sensors distributed over a geographic area – Customizable sensors: cameras, microphones, weather monitors, pollution sensors ...... – Integrated into a social networking site (custom built) – Immediate applications: disaster warning, traffic control 18-Mar-11
Current work …2 3. Delay Tolerant Networks for rural applications a. Large delays (up to one day) tolerated b. Embedded system with multiple connectivity c. Low seed connectivity used for queries and “telegrams” and high speed for data transfers d. Public information system and email delivery applications 4. Energy neutral sensors a. Energy harvesting techniques b. Optimal usage of energy for longest possible NW life time Protocols, circuit techniques, network “intelligence”..... 18-Mar-11
Mule on a Bicycle
The system
Embedded System
DATA MONITORING AND CONTROL PLATFORM An open source Project
PLATFORM REAL TIME INFORMATION MANAGEMENT REAL TIME Reception REAL TIME Processing REAL TIME information distribution
APPLICATION AREAS DISASTER MANAGEMENT TRANSPORT INFORMATION SYSTEM SECURITY ALERT SYSTEM Applications WEATHER MONITORING SYSTEM
TRANSPORT INFORMATION SYSTEM Real time vehicle tracking Vehicle status maintained in a central server Status analyzed and advice sent to user as SMS Assist information management in personal vehicles, fleet of trucks, service vehicles
BLOCK DIAGRAM Disaster Information Processing Gather Process Distribute
TARGET SPECIFICATIONS Data gathering SMS: 250 messages per hour Pictures: Embedded camera system with Flikr upload interface Twitter: sensor data with embedded sensors with Twitter upload interface Cell phone interfaces: Currently android Data aggregation Open Source tool : Swift River Aggregation of 6, 000 messages/day, SMS, Twitter, email, flikr …… Processing and analytics, inference engine, User alerts and messages Dedicated system (LAMP architecture)
Person Recognition Based on Automatic Learning For security applications
Need Statement To monitor continuously presence of a person or persons in an enclosure (like seat in a Cockpit) and to alert if the person changes in the enclosure Requirements and Assumptions : ◦ Meant for aircraft safety ◦ Persons authenticated for first ten minutes in the beginning when the enclosure is first occupied (say, during take-off) ◦ Applicable to all seats inside cockpit. ◦ Non Cooperative Process: the occupants don’t participate in any authentication process ◦ Possible Events : Absence of Pilot from his seat. Presence of unknown person in the pilot’s seat.
BLOCK OVERVIEW OF THE SYSTEM Remote Server (for configuration) Configurations Event (Flight Settings) updates RECOGNITION SYSTEM CAMERA ARRAY Neural Network Inside cockpit Event Registration MIC Array Pattern Analyzer PRESSURE SENSOR ARAY • Face On control wheel • Audio • Pressure Pattern
Face Detection • By skin color - red chrominance • By motion – Consecutive frame subtraction, detect blinking • Model-based • Combination
Speaker Detection • Consists of : • Feature Extraction • Feature Matching • Pitch is not a reliable feature • Perceived frequency depends upon amplitude, receiver, timbre • Mel-Frequency Cepstrum Coefficients • 19 Mel Cepstra, 19 delta Cepstra • Obtained by DCT of log magnitude FFT of sampled audio. • Model speaker’s voice (GMM -UBM) • Gaussian Mixture Model (GMM) for speaker • Universal Background Model (UBM)
Wish Specifications • Reliability of recognition ~ 99%. • Maximum time delay between event and its registration ~5 sec. • Detection and tracking at 15 fps. • Max. video resolution 640*480 pixels (VGA), preferable 320 X 288 • No. of cameras : 2-6. • Find a method to measure reliability.
Energy Neutral Sensors PV and Thermal energy based
Specifications • No external power source • Harvested from PV and Thermal sources (Peltier block) • At least 4 sensors • Energy storage with super capacitors • Transmission and storage strategies • Network based data gathering exploiting dense sensor deployment
Current status • Energy neutral system with temp, insolation, humidity, soil moisture and one other optional parameter functional • Harvesting with a 20 cm 2 solar panel, 5 cm 2 Peltier block • Tests underway for throughput and speed • Platforms available for experiments
Entrepreneurial activities 2010 onwards
Some Observations 1. Risk taking is almost nonexistent among students and faculty members 2. They are technically sound 3. May develop product for nearly production level 4. IISc focus in on research and NOT much on innovation and product development 5. IISc incubation is largely open only to students, alumnae, and faculty… counter to point 1 18-Mar-11
IISc Incubation Partner Vision, Mission, Objectives, Areas of interest Idea reception from Ideas from faculty labs in IISc and external members, students, sources (web based) incubation Centre Research and innovation Committee for cycle evaluation of ideas and Faculty research preliminary selection Final Specifications, Timelines, Business models Quick prototyping, IISc Digital Product Lab demonstration, first Product development, level user feedback, and Incubation idea selection Product development cycle Pilot production, Finished product Testing and evaluation, Customer contact and Release of product field testing, IP portfolio documents management, Investor contact.... Product Launch
Digital Product Lab (DPL) • Set up at IISc consisting of – Embedded labs (with others like Intel, Freescale, Xilinx...) – ID Lab (CEDT and CPDM) – Rapid prototyping (CPDM) – Testing lab with all test facilities – Wireless communication systems lab...
Use of the DPL • For developing products requested from the selection process • For developing ideas of IISc researchers • For developing products suggested by others and selected by the selection process
Curriculum activities 2010 onwards
Embedded Curriculum Initiative: Objectives • To encourage embedded systems innovation in India with university as the fulcrum and the epicentre from which the innovation propagates to the industry, academia, and startups. – Developing appropriate curricula – Setting up state-of-the-art labs – Integrating innovation as well as incubation with the curricula in universities
Key elements • Embedded Systems Curriculum • Lab : Set up, Lab exercises and mini projects • Student Internships • Curriculum Development Workshops • Faculty Training • Consortium of participating institutes
Phases of Lab exercises • Level 1: Familiarization exercises: For those not yet familiar with Embedded Systems, a set of small exercises would be provided that would help familiarize them with a few basic concepts. • Level 2: Preset Exercises: As a part of the regular course that is being recommended, a set of very small exercises/projects would be provided that could be completed in a couple of lab sessions. • Level 3: Mini Projects: These exercises would train the candidate in depth in one of the applications of the embedded system
Recommend
More recommend
