Analog Voice Privacy Systems Using TFSP Scrambling: Full Duplex and Half Duplex

In this paper we discuss a possible analog voice encryption scheme. Although the channel signal is analog, all of the signal processing is done digitally. This gives us a flexibility that analog processing cannot give, and thereby allows us to do simultaneous time and frequency permutation. It also allows us to perform synchronization and equalization at the receiver. With the advent of the Digital Signal Processor (DSP),1 digital processing becomes economically competitive as well as feasible. * This work was performed while Mr Tribolet was on sabbatical leave from the Instituto Superior Tecnico, Lisbon, Portugal. 47 The systems that we will discuss in this paper are a full-duplex system and a half-duplex system. We have assumed either system would be used over all standard telephone channels with a bandwidth of 200 to 3500 Hz. Because of the delays involved in the signal processing, echoes could become a problem, so we chose to look at solutions that would avoid the echo problem. Alternatively, echoes might be cancelled using an appropriate algorithm, but this problem was not addressed in this study. One such solution to the echo problem is to choose a half-duplex system. Alternatively, for a full-duplex system we have divided the telephone band into three segments: 200 to 500 Hz is used for signalling and equalization by both users, 500 to 2000 Hz is allocated to the first user, and 2000 to 3500 Hz is allocated to the second user. In this way faulty echo cancellation need not impair the quality of the speech, since the two speakers use disjoint bands.