Background
Plain/clear-text VS cipher-text
Security in Mobile Banking Apps - Problems ****
• Flaw in implementation, strong technique
• Drivers & Hardware
• Many mobile app developers are not trained in
computer science
• API (Abstraction)-> makes everything simple
Sensors (Medical Control Apps)
• Sensor/Actuator - Mobile App - Cloud Services
• Resource constraints in Sensors and Actuators
• Poor software development support for sensors
and actuators
• Real-time requirements of Health apps
Ways to jeopardize & secure
• APKs can be reversed into codes
Developer uses Unique User ID to secure. Can
also be jeopardize
• Security protocols introduce delays.
Service checking takes time.
• Criticality access control
What is network security
Confidentiality: only sender, encrypts msg, intended
receiver decrypts msg.
Authentication: sender & receiver want to confirm
identity of each other
Message integrity: sender, receiver want to ensure
msg not altered without detection
Access & Availability: services must be accessible
and available to users
Non-Reputation: source cannot deny the sent msg.
Principles of Cryptography
definition: a set of mathematical functions with a set
of nice properties
Properties of good encryption techniques
• encryption scheme depends not on secrecy of
algorithm but on parameter of algorithm
• extremely difficult for an intruder to determine the
encryption key
Cryptography algorithms:
• Symmetric key algorithm: one shared by a pair of
users used for both encryption & decryption
substitution cipher: substituting one thing for
another (m=a, n=b…)
use the frequency of words to break it
brute force (How to measure effort: worst
case is 26^26
DES: Data Encryption Standard - not secure -
56-bit symmetric key, 64-bit plaintext input
AES: Advanced Encryption Standard - process
data in 128bit blocks, 128/192/256bit keys -
secure
• Asymmetric or Public/private key algorithm:
public, & private key known only to individual
user
, “-” is the private key, “+” is
the public key
Given it should be impossible to compute
RSA algorithm
1. choose two large prime numbers p, q (large
computation-> set of large primes)
2. compute n = pq, z = (p-1)(q-1)
3. choose e (with e<n) that has no common
factors with z (e, z are relatively prime)
4. choose d such that ed-1 is exactly divisible
by z (ed mode z = 1)
5. public key is (n, e), private key is (n, d)
Encrypt bit pattern m: c = mod n
Decrypt bit pattern: m = mod n
**** m = ( mod n)^d mod n
Usage of cryptography for network security
Message integrity
• Definition: ensure message originally came from
A, not changed since sent by A
• Cryptographic hash: takes input m, produces fixed
length value H(m), computationally infeasible to
find two different messages x, y such that H(x) =
H(y). i.e., given m = H(x), cannot determine x
• MAC (Message Authentication Code)
Digital signature
• definition: (verifiable, non-forgeable) sender
digitally signs document, establishing he is
document owner
• Digital Signature = Signed Message Digest
apply hash function H to m, get fixed size
message digest, H(m)
End point authentication
Key Establishment
Symmetric solutions
• Deffie-Hellman Key Exchange: A & B have secret
integer a & b, A sends mod p to B, B computes
( mod p mod p, sends mod p, A computes (
mod p mod p. And they are equal.
No infrastructure support
computation load on users
• KDC (Key Distribution Center): server shares
different secret key with each registered user
need infrastructure support
single bottleneck, single point of failure
computation load centered at KDC
Public/private solution
• CA (Certification Authority): binds public key to
particular entity E, which registers its public key
with CA.
Security Problems
Authentication - user name & password
Identification - password
Data Provenance - how to ensure the source of data
Data Integrity
Data Injection
Data Alternation
Access Control
Characteristics - Idea: uses human body for
security
Biometrics - Some form of unchanging and personal
information - fingerprints, iris scan, facial scan
physiological Signals - difficult to match-> signal
process and make it better
• Definition: Some form of signals obtained from
the human body in free living condition without
intervention from the human user.
• Process: change Biometric Comparator to Signal
Extraction-> no ML, compare signal
• Differences from 1: time varying & highly random
& no permanent feature
Activity and Gestures - much difficult to match
• Depends on personal habits
• Human computer interface
• Process: change Biometric Comparator to Trained
Machine
EEG (脑电图) Based Authentication
Machine Learning Security
What is ML / What does it do: Draw a boundary in
the input feature space. Features are multi-
dimensional.