Analog voice privacy systems using TFSP scrambling: Full duplex and half duplex

In this paper we present possible full-duplex and half-duplex analog voice privacy systems that have been simulated over real channels. Previous papers have been concerned primarily with the issues of the strength of a system (i.e., unintelligibility to the casual eavesdropper and relative cryptanalytical strength for the sophisticated eavesdropper) and the amount of delay of a system. Well-known but not addressed have been the problems of decoding the scrambled signal in a real-channel environment. At the heart of the encryption systems proposed here is the sequential time and frequency segment permutation structure proposed by Jayant and Cox. This structure relies on digital processing to divide the signal into sub-bands and then to permute these bands in both time and frequency simultaneously to synthesize the scrambled analog signal. In discussing the decoding we address the issues of compensating for the properties of the channel, re-sampling the analog signal, and establishing and maintaining synchronization between the de-scrambler and the scrambler.