A robust secure speech communication system using ITU-T G.723.1 and TMS320C6711 DSP

Secure speech communication plays a dominant role in civil and military communication systems. Speech scrambling and descrambling, particularly in time-domain, has been a key component in these systems. Removal of redundant information is very significance in secure speech communication systems. In addition to security, fast implementation of security algorithms is also very important for real-time applications. Thus after the penetration of emerging technologies like DSP (Digital Signal Processors) this importance has increased even more. Traditional schemes and methods used for speech scrambling either do not remove the redundancy of the signal or pay less attention to this parameter, which thereby provides an opportunity to an interceptor to descramble the signal with convenience. So far most of the implementations of scrambling techniques based on DSP have been carried out either on 16-bit fixed-point DSPs or on multiprocessors, which offer low accuracy and comparatively slower speed and thus are not suitable for real-time applications. In this research we have focused on these three parameters simultaneously and proposed a novel redundancies-free, high speed, real-time secure speech communication system. This system is based on time-domain speech scrambling/descrambling, ITU-T dual rate speech codec G.723.1 and Texas Instrument's 32-bit floating point DSP TMS320 C6711. Real-time speech signal is captured through DSP port and redundancy is removed by compressing the original signal to 5.3 or 6.3 kb/s. Subsequently, compressed signal is scrambled using hopping window and sliding window techniques using linear congruent pseudo-random generator. The results show that our system is redundancies-free, fast, secure and more suitable for real-time applications.