Title: National Science Foundation Stevens Center for Wireless Network Security
1National Science FoundationStevens Center for
Wireless Network Security
Energy Issues in Secure and Reliable Wireless
Multimedia Sensor Networking
Chetan NanjundaSatish BapatlaR.ChandramouliStev
en Institute of Technology
N. VijaykrishnanM.J. IrwinPenn State University
2Contents
- Energy Awareness
- Sensor Networks
- Security
- Error Control
- Optimizations
- Experimental results
3Energy Awareness
- Energy Sources
- Battery
- Solar (time variant)
- Energy consumption
- Computation
- Transmission
- Power consumption varies over time in wireless
sensor networks - Energy awareness spans several layers
4 5Constraints in Sensor Networks
- By design, sensors should be inexpensive
- Sensors are powered by a small battery, hence
limited power will be available - Battery replacement is an expensive process.
- sometimes not possible!!!
- The two basic resource constraints in a sensor
are - Memory
- Power
6Constraints in Sensor Networks
- Memory
- Operating system code space
- The network protocols requires the intervention
of operating system - Available code space
- Data processing, Security algorithms and Error
control should be run in this available code space
- Power
- Power consumption is in three domains
- Sensing
- Data processing
- Communication
7Information
8Avalanche Effect
- The bits in the ciphertext change with a
probability of one half whenever a single input
bit is complemented before encryption - This is one of the important criterion for
designing cryptographically good S-boxes
Example For a 128 bit AES encryption p1 ? AES ?
c1 p2 ? AES ? c2
if d(p1,p2) 1 ? d(c1,c2) 64
9Effect of Bit Errors
- Error Propagation Caused because of Avalanche
effect
if d(c1,c2) 1 ? d(p1,p2) 64
i.e if one bit in the ciphertext gets corrupted
the whole message is lost!!
- This triggers several retransmissions
- Energy inefficient
- Error control strategies are important
10 11Optimizations in Error Control
- Adaptive Error Control.
- Feedback loop allows the transmitter to adapt the
error coding according to the error rate observed
at the receiver - Example RCPC codes
12Optimizations in Error Control
- Viterbi Vs. List Viterbi
- Viterbi decoder
- Maximum likelihood decoder for convolution codes
- Has three basic parts
- Branch metric computation
- Path metric computation
- Trace back unit
- Probability of successful decoding is very high
for random bit error rates lt 10-2 (average
transmissions are below 1.1 at such BER)
13Optimizations in Error Control
- List Viterbi decoder
- Decodes M paths with ascending order of path
metric - The ith best path is the path with the ith least
minimal metric - This approach greatly reduces the number of
retransmissions for error rates 10-2 , 10-1 - Power asymmetry in sensor networks can be
exploited by list viterbi - One of the important design criteria is to
evaluate the tradeoff between computation and
communication
14Data Dependent Optimizations
- Layered Multimedia Encoding
- Different layers have different level of
priorities - Layered multimedia can be integrated with the
retransmission strategy to optimize transmission
power - The varying level of priority in these layers can
be exploited to provide varying levels of
security to optimize computational power
15Experimental Results
- Security
- AES (avalanche effect)
- Error Control
- List Viterbi (Distribution of Path search success
for different error rates) - Power Consumption
- Different Rounds of DES
- DES Vs. RC4
- Different Key Lengths of RC4
- Comparison of DES, IDEA, GOST
16Avalanche Effect
d(p1,p2) 1
P1-1
P2-1
Encryption
Encryption
A E S
P2-2
P1-2
A E S
P2-9
P1-9
one bit of plaintext was flipped at a random
position and the bit differences between original
cipher and the encrypted cipher was observed at
intermediate rounds
17Avalanche Effect In AES
one bit of plaintext was flipped at a random
position and the bit differences between original
cipher and the encrypted cipher was observed at
intermediate rounds
18Error Propagation
d(c1,c2) 1
C1-1
C2-1
Decryption
Decryption
A E S
A E S
C2-2
C1-2
C2-9
C1-9
- One bit in the cipher text was flipped in random
position and the bit differences between the
recovered plaintext and the ideal plaintext was
observed at intermediate rounds
19Error Propagation Caused by Avalanche
- One bit in the cipher text was flipped in random
position and the bit differences between the
recovered plaintext and the ideal plaintext was
observed at intermediate rounds
20Error Control Setup
DATA
CRC
RCPC
AES Enc
Puncturing matrix
Wireless Channel with Random Bit Error
Channel condition
List Viterbi
Received DATA
AES Dec
- Packet Size 200bits
- 16bit CRC
- Convolution mother code rate 1/2
21Distribution of Successful Decoding Over Path
Search Depth
10-2 BER
22Distribution of Successful Decoding Over Path
Search Depth
(0.2) 10-1 BER
23Distribution of Successful Decoding Over Path
Search Depth
(0.4) 10-1 BER
24Distribution of Successful Decoding Over Path
Search Depth
(0.6) 10-1 BER
25Distribution of Successful Decoding Over Path
Search Depth
(0.8) 10-1 BER
26Distribution of Successful Decoding Over Path
Search Depth
27Schematic Of H/W Setup
Read Voltage Current by GPIB
Labview
Vcc I cc
Power Supply
28DES Power Requirement for Different Rounds
power
(watts)
Number of Rounds
29Power Requirement for Different Key Lengths of RC4
power
(watts)
Key Length
30DES Vs. RC4 for Random Data
? DES
power
(watts)
? RC4
Random DATA
31Power Consumption Per Round
power
? IDEA
? DES
(watts)
?Gost
Key Length
32Overall Power Consumption
power
?Gost
? IDEA
(watts)
? DES
Key Length
33Questions??
34(No Transcript)
35(No Transcript)
36(No Transcript)