Implementation and Performance Analysis of Reliable and Secure End to End Voice Encryption over Public Mobile Network Based on Frequency Domain Using Dual Processor in FPGA Platform

This research aims in designing a method for implementing end-to-end voice encryption over mobile network such as GSM that independent to phone devices, operating system and network devices. The proposing method is an encryption system in domain frequency to keep the encryption signal remain within the range of sound frequencies that can be transmitted through GSM/2G/3G. This method differ compare to the method used in the previous studies, in which they use the time-based encryption method to get a low rate encrypted data so the results can be communicated via GSM network. We simulated the method using Matlab. The simulation results showed that the voice encryption on the frequency domain characteristics is very different from the original signal and can be passed completely through the GSM network. This method was implemented in the Xilinx FPGA using microblaze dual core processor on the Atlys board. The results of the signal reconstruction using FFT/IFFT module in FPGA Xilinx varied in a relatively small deviation, maximum -10.42 dB, compared to the original signal. Performance analysis of the prototype showed the sender speech can be reconstructed real time in the mobile handset of the receiver side, even in the unsynchronized condition. However, in a further prototype development, the precision level of the FFT/IFFT module needs to be improved, meanwhile the synchronization module and echo canceller need to be added. Keywords— Voice encryption, End-to-end mobile network, GSM/2G/3G, Dual core microblaze, Xilinx FPGA, FFT-IFFT, Domain frequency.