Towards a platform for ambient intelligence

1
Towards a platform for ambient intelligence

• ICT DRC AdviesRaad 12 jan 2006

2
Project example (1)

• BioSens (STW TUD/UM, Medtronic, Vitatron, Twente
  Medical Systems, Maastricht Instruments,
  Systematic Design, Weijand Consultancy)
• Biomedical signal processing platform for
  low-power real-time sensing of cardiac signals
• Mathematical modeling of cardiac signals and
  pathologies
• Design of WT-based algorithms for sensing and
  feature extraction
• Low-power analog integrated circuits that
  implement the WT and artifact detection
• Platform
• Database of relevant (recorded and synthesized)
  cardiac signals, annotated and clustered
• Database of WT-based algorithms, specified,
  tested,
• ECG player (to generate test signals)
• Various dedicated IC designs, specified, tested,

3
Project example (2)

• UWB for the process industry (Senter
  Utellus/TUD/Shell/WIB)
• For TUD a natural follow-up on the AIRLINK
  project
• Platform
• UWB test bed
• Protocols
• Application layer designs
• Measurement data
• User experiments
• Modular UWB radio
• Analog front-end ICs
• Antennas
• Modulation and signaling techniques
• Signal processing algorithms
• Routing and synchronization algorithms
• Embedded software
• Interfacing products
• Standards (?)
• Really new applications not possible with
  conventional technology
• UWB based solutions and products

4
Project example (3)

• SiMS (smart implantable medical systems)
• TUD Stryker, Medtronics, Vitatron, Twente
  Medical Systems, VUMC, AMC, EUR, Telemetronics,
  TNO-FEL, Philips Medical Systems, Philips
  Research (several groups), SystematIC, Maastricht
  University (math dept.) etc.
• Platform
• Sensors
• Digital architecture (processing/controlling
  unit)
• Compiler (tool)
• On-chip antennas
• (On-chip) battery
• Wireless modules

5
What?

• Hard IP
• Hard IP is optimized for power, size, and
  performance, and mapped to a specific technology.
  
• Mask layouts
• Library cells
• Test chips
• DSP
• PCB
• The best aspect of hard IP is guaranteed
  performance, be it speed, power consumption, and
  so on.
• Soft IP
• Soft IP is delivered in a synthesizable form,
  which must then be implemented into the target
  technology
• Netlist
• Hardware Description Language (HDL)
• C/C code
• MatLab/SimuLink code
• Java
• Soft IP is easy to retarget and configure.
• Meta IP
• User scenarios
• Use cases

6
what?

• Design techniques
• Design methodologies
• Proofs of concept
• Best practice
• Interface definitions (I/O, API)

7
Why? (University)

• Validate our research
• Focus our research
• Prove our research
• You do not count if you cannot present your
  realizations
• Marketing
• PR
• Acquisition
• Students
• Projects
• Develop multi-disciplinary expertise

8
Why? (Industry)

• IP gap
• IC complexity is growing at a rate of 58 per
  year, but design productivity is increasing at
  only 21 per year.
• Most of the people we train will end up in
  industry (B.Sc.s, M.Scs, PD Eng and Ph.Ds)

9
Why not? (University)

• Will only pay off on the longer term
• Additional work (burden)
• Less papers
• Not appreciated

10
Actions up to now

• Discussions with and proposal for a framework
  from SystematIC Design
• Proposal in NIRICT for a 'Platform Lab' in the
  context of the NIRICT Innovation action
• Pilot to explore interested parties in the
  context of the Delft Innovation Lab (also part of
  the 3TU Program) - contract with Westfield BV

11
We need

• Your comments
• Your suggestions
• Your ideas
• Your input
• Your contribution