Abstract:
Most of voice channels used in voice terminals such as
mobile, telephone exchange are not ciphered. This makes them
vulnerable to eavesdropping and hacking. Present day securing voice
communication became an urgent need for civil and military
applications and especially for real time communication. This work
presents a real time microcontroller-based voice scrambling and
descrambling (full duplex) using 16KHz sampling frequency. Voice
scrambling / descrambling was done in frequency domain and time
domain. In frequency domain, a frequency inversion is implemented
using sinusoid wave multiplication and 8-order low pass Finite
Impulse Response filter. In time domain, a segment permutation
technique was implemented. When inputting, voice samples were
stored in memory sequentially. However, when playing the voice,
samples were read with a random sequence according to a look-uptable.
The scrambling / descrambling algorithms were verified using
MATLAB and implemented using ARDUINO DUE microcontroller.
The system was tested in real time by scrambling the voice input
from a microphone and output it through a speaker. Tests show that
the scrambled voice was not recognizable at all. Also, the
descrambled voice shows acceptable similarity with the original
voice.
