Pattern recognition based authentication in mobile and wireless system

1
Pattern recognition based authentication in
mobile and wireless system

• Yuan Jun

2
Presentation outline

• The security feature of modem wireless system and
  its flaws
• The Hidden Markov Model (HMM) theory
• The Hidden Markov Model used in hand writing and
  hand drawing pattern recognition
• Hand writing and drawing pattern recognition
  applied in the wireless system authentication
• conclusion

3
The authentication progress in GSM mobile system

• The authentication key Ki is 128 bit
• RAND is 128 bit
• The authentication algorithm A3 is used to
  generate 32 bit random number

4
The authentication progress in WLAN system

• The RC4 algorithm is used for the encryption
• Initialization vector is send in plain text
• The key KAB usually is 128 bits long
• CRC-32 is used for the protection of data
  integrity

5
The flaws in the wireless system

• The authentication key is not long enough
• The algorithm is published, even is not, it is
  still not complex enough to protect clients
• For the WLAN security feature-WEP
• - security algorithm RC4 is not secure in
  wireless system
• - the Initial Vector (IV) send with encryption
  data in plaintext
• - all the security feature has been published
  in the WLAN standard

6
The Hidden Markov Model and its application in
pattern recognition

• The Hidden Markov model is a statistical model
  which is doubly embedded stochastic process with
  an underlying process that can not been observed.
  But it can only be observed through another
  stochastic processes' observation sequence
• The following figure is showing us a Hidden
  Markov Model

7

• X- states
• Y- possible observations
• a- state transition probabilities
• b- Output probabilities

8
Hidden Markov Model used in pattern recognition

• Every characters has its own model
• Hand writing words usually processed in digital
  image
• Feature extraction is very important for the
  authentication
• Lexicon is used in the system for classification

9
The feature extraction in a hand writing

• We estimate a horizontal center line of the work
• Different positions in a word will account by the
  line, which including descender or ascender at
  that point
• The distance of the point to the line also will
  be measured

10
Hand drawing pattern recognition to create the
access control matrix
11
The access control matrix for a hand drawing
figures

• For the hand drawing figures, we should have two
  steps to process the pattern recognition
• First is to identify the elements contains in the
  figure, such as circle, triangle, square etc.
• Second, the relationship between elements should
  be clear.

12
HMM processing used for authentication

• The Hidden markov models processing can generate
  very complex matrix and data string which can be
  used for the system security protection
• These data string can be recognized strong enough
  for the privacy protection

13
conclusion

• Use pattern recognition process as a improvement
  of the wireless system is very efficient
• The unpredictable of the key can very effectively
  protect the transfer data
• Different from the traditional password
  protection, this can be used to protect system
  from dictionary attack.
• The user do not have to remember long and complex
  password