We describe a digital mobile radio system in which users experiencing poor transmission reduce the number of bits per second used for speech coding from 32000 to 28000. They do so in a manner that greatly reduces the binary error rate. The reduced error rate more than compensates for increased quantization noise. We predict two statistics of user satisfaction in a mobile radio service area: POW, the percentage of users experiencing poor or worse transmission, and GOB, the percentage good or better. In a typical environment the maximum number of users that can operate within a 20 MHz band with POW ≤ 5 percent is 137, compared to 91 users when everyone transmits 32 kbits/s. The modulation technique is FH-FSK (frequency hopped, frequency shift keying) and the digital code is embedded ADPCM (adaptive differential pulse code modulation), in the mobile radio service area, signal strength depends on distance from mobile unit to a central base station. There is slow (shadow) fading due to terrain and building variations and rapid (Rayleigh) fading due to multipath propagation. To predict user satisfaction we have combined three different analyses: 1) a theory of FH-FSK, 2) a theory of the signal-to-noise ratio statistics in mobile radio service areas, and 3) a model, based on computer simulations and subjective testing data, of the dependence of subjective speech quality on ADPCM transmission rate and binary error rate.

This paper describes a new phase-shift-keyed/phasemodulation (PSK/PM) scheme, called Interplex, which improves the performance of multichannel telemetry systems by reducing the cross-modulation loss. In the case of two-channel systems, the improvement in data power efficiency relative to a conventional system can be as much as 3 dB. A major feature of Interplex is that it can be implemented with very minor modification of existing conventional PSK/PM systems.

In comparing different techniques for multiplexingNbinary data signals into a single channel, time-division multiplexing (TDM) is known to have a theoretic efficiency of 100 percent (neglecting sync power) and thus seems to outperform frequencydivision multiplexing systems (FDM). By considering more general FDM systems, we will show that both TDM and FDM are equivalent and have an efficiency of 100 percent. The difference between the systems is in the multiplexing and demultiplexing subsystems, but not in the performance or in the generated waveforms.

New methods of parallel and serial synchronization for time-multiplexed signals are presented. In the parallel case, i.e., when the sync code is transmitted on a separate channel,Msynchronization codes are generated by concatenating (in the sense used in concatenated codes) a common pn-like sequence of short length with words from a self-synchronizable code. The frame synchronizer can lock on any of theMcodes, which is particularly useful when frames of different lengths have to be transmitted. In the serial case, that is when each frame starts with an identical sync code, false sync due to replicas of the code, randomly generated by the data, is completely eliminated by transferring the sync code to the quadrature channel, while the data are transmitted on the inphase channel.

A method for the design of time-limited and effectively band-limited waveforms for analog communication systems is developed. The design problem is first formulated in theL_2 (0, T)space. Nonlinear modulation threshold effects are incorporated directly into the design by introducing constraints in the form of minimum distance between points on the locus. It is shown that all pertinent information can be expressed in terms of the inner product function generated by the locus and its cross derivative. The design is then transformed into a constrained minimization problem in a space of inner product functions. The cases of stationary locus andN-dimensional locus are further investigated.

An upper bound on the estimation error in the threshold region (probability of threshold effect and mean-square error) is obtained for nonlinear pulse modulation systems. The problem is viewed in anN-dimensional Euclidean space. The space of all received signals is divided into two regions, corresponding to the two types of error: weak-noise approximation and threshold effect. The threshold region is geometrically upper bounded by a larger region, and the estimation error is obtained as a sum of incomplete\Gammafunctions. The resulting bound on the mean-square error was found to be quite close for the cases calculated. An extension of the method to PPM system is also presented.