The Infopad Project. ATM Backbone. InfoPad (A Portable Multimedia Terminal) - Video and Text/Graphics - Audio & Handwriting I/O

Transcription

1 The Infopad Project ATM Backbone NOW COMPUTE SERVER PICOCELL WIRELESS BASE STATION VIDEO SERVER SPEECH RECOGNITION HANDWRITING RECOGNITION 2 Mbit/sec CDMA Downlink Kb/s TDMA Uplink WWW INTERNET INFORMATION SERVER PAD SERVER CELL SERVER InfoPad (A Portable Multimedia Terminal) - Video and Text/Graphics - Audio & Handwriting I/O

2 7 Areas of the InfoPad Project Padgroup - Terminal and basestation CDMAgroup - Downlink system design RFgroup - Integrated CMOS RF circuit design Infonet - Network support for mobility Medley - Backbone network servers and protocols UIgroup - Applications and middleware Infodesign - System and low power design support

3 Computation/Communication Trade-offs MPP s Hardware Complexity Data Storage Workstations Computation Laptops Necessary Communications PDA s (Network Computers) Infopad

4 Argument for InfoPad as an I/0 device Reduces power, size and cost for the user to the maximum extent possible Information access is the primary application and the data is largely time critical or from large data bases Simplifies use by eliminating user support of system software The wireless link has high error rates that best supports robust, transient I/O data

5 InfoPad and Network Computers Network Computer A fixed network attached I/O device for viewing and manipulating information Low cost (<$500) Easy to use InfoPad All the above + Portability -> Low power and small size Wireless communications Network infrastructure to support mobility and Quality of Service

6 Goal of the Present Project To implement a wireless computing environment which will allow 50 users in a confined area to simultaneously access and communicate with multi-media network services using a low cost, light weight portable tablet.

7 November 1995 Demonstration Accomplishments COMPLETE SYSTEM WAS DEMONSTRATED (just-in-time). Connectivity from InfoPad aware applications, through the Infonet mobility support, basestation, radio link and finally to the hardware pad. 10 new pads have been built (4 at the time of the demo) 4 new base stations

8 Why Build an Entire System? Though demonstrations are important, the real purpose of the InfoPad project is the research that it facilitates. Joint research with other Universities and Industrial labs - (eg. new National Semiconductor Research Lab - transfer of software and hardware) Supporting over 40 student projects

9 Integration with other University Research Projects CMU MIT UCLA Stanford Coda Video Terminals Compress U of Colo Packaging SRI Glomo UCB InfoPad Daedelus Java Apps Medley

10 The Individual Research Projects ATM and Fast Ethernet Backbone BASE STATION and WIRED to WIRELESS BRIDGE PC/Ethernet Scalable Hardware Basestation Scheduling for Quality of Service and Capacity Optimization MAC Layer Protocols for Up and Down Links CDMA Power Control Algorithms Optimization of Modulation and SS Techniques for Interference Limited, Picocellular Channels User Interface based on Pen and Speech Input Network and InfoPad Aware Design Tools SPEECH AND HANDWRITING RECOGNITION WWW INTERNET INFORMATION SERVERS Support for Distributed Processing for the Mobile Network (the InfoNet) Low Power CMOS Radio Implementations - Mostly Digital Wideband Spread Spectrum, and Monolithic DECT (TDMA) Energy Optimized ARM Processor Core Migration of processing into Pad Under the RTOS Helios Low Power Signal Processing Accelerators Concurrent Electrical/Mechanical Casing Design Text/Graphics Decompression for Color Display System Level Power Analysis Tools DESIGN TOOL APPLICATIONS InfoNet AUDIO, PEN AND VIDEO TYPE SERVERS InfoNet NAMESERVER, CELL and PAD SERVERS

11 What s next? The 3 year initial project is finishing at the end of this year. Two new thrusts: Low power design for programmable components (both software and hardware programmable) Extension of the InfoPad System design to be more flexible and adaptable to varying wireless links and pad functionality with the lowest possible energy consumption

12 Future Directions Investigate the communications vs. computation tradeoff for communication links with bandwidths from 10kHz-1MHz Issues What kind of processing in the pad is most effective in reducing communication requirements How to retain advantages of the InfoPad system architecture, while increasing pad computation capability How to implement processing with the lowest possible energy, while providing high performance when required

13 Future Directions Implement a software infrastructure which allows seamless mobility over widely varying communication links, while optimizing the capacity and quality of service for all users Issues From the system viewpoint, where is QOS support required and what are the knobs that should be controlled What kind of support can be given in the radios and the pads for controlling the QOS What is the impact on the applications of an underlying variability in the bandwidth of the communication links

14 Future Directions Incorporate programmability into the wireless communications link at lowest possible energy cost Issues Extend Mostly Digital approach for wideband radio to other spread spectrum modulation schemes Extend CMOS DECT implemention to have multimodal radio capabilities supporting standard protocols (eg. GSM) Determine array structures for implementing low power, field programmable logic for flexible high rate baseband processing (eg. timing recovery)

15 Future Directions Demonstrate system level solution to low power design for the pad, while increasing its flexibility Issues Reduce total weight and power to 1-2 lbs and 1-2 watts even with a high resolution color display (2-3 years) Optimize energy reduction over all aspects of pad design through use of new tools for power analysis and optimization Implement a modular pad architecture (physical and electrical) to allow flexibility in pad capabilities

16 Conclusions Conventional wisdom has taken a step towards the InfoPad system design through the acceptance of the network computer Implementation of the complete InfoPad system has provided: An infrastructure within which to do research on mobile computing Provided a focus and motivation to try out new ideas in a variety of areas Given students managerial and project experience