IP concerns drove encryption fortransistor-level models in recent years

1
The Problem

• IP concerns drove encryption for transistor-leve
  l models in recent years
• Netlists, process technologies are highly
  sensitive
• Even with behavioral modeling, IP protection is a
  concern
• Data-processing algorithms, filter designs, etc.
  are just as sensitive as process technologies
• IBIS tables obscure many design details -AMS
  equations may highlight some design aspects
• -AMS adoption may be hampered by lack of
  consistent IP protection
• Usage Model for Encryption Today is Mixed
• Model encryption today is tool-specific,
  non-standard

Without IP Protection, Equation-based
Behavioral Buffer Modeling May Not Succeed
2
A Proposal

• An industry-wide effort to provide a standard
  for model encryption is hereby proposed
• Principles
• Encryption is necessary for improved IP
  protection
• Encryption does not threaten behavioral
  modeling, so long as speed and tool-neutrality
  advantages are preserved
• Requirements A standard for model encryption
  must
• Permit decryption ONLY within the simulation tool
• Support public/private key AND tool-embedded-key
  formats
• Permit regular strength upgrades, as technology
  advances
• Support tool-neutrality encrypt in one tool,
  decrypt in another

The IBIS Open Forum (Futures Task Group) Should
Take the First Steps Toward a Standard