A Spread Spectrum Approach to Next Generation Intrusion Detection
Abstract
The report describes a theoretical treatment of the tradeoffs among carrying capacity, distortion, and error recovery for spread spectrum solutions to message embedding approaches. For any given host image and (block) transform domain of interest signature vectors are derived that when used for spread-spectrum (SS) message embedding maximize the signal-to-interference-plus-noise ratio (SINR) at the output of the corresponding maximum SINR linear filter receiver. Under a (colored) Gaussian assumption on the transform domain host data, it is shown that these same signatures over minimum probability of error message recovery at any fixed host distortion level (or-conversely- minimize the host distortion for any probability of error target level) and maximize the Shannon capacity of the covert link. These results are further generalized to cover SS embedding in linearly processed transform domain host data with resulting orders of magnitude of demonstrated improvement over state of the practice SS steganographic practices. Optimized multi-signature/multi-message embedding in the same host data is addressed as well.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2005
- Accession Number
- ADA436185
Entities
People
- Dimitris A. Pados
- Michael J. Medley
- Stella N. Batalama
Organizations
- State University of New York