An information-reduced carrier-synchronization (IRCS) system has been proposed for use in a coded binary-phase-shift-keying (BPSK) radio-communication receiver subject to a low signal-to-noise ratio (SNR). The term "information-reduced" alludes to the use of an estimate of the instantaneous data symbol (and thus of the instantaneous phase modulation) to reduce the amount of randomness (and thus the amount of information) in the signal being processed in the carrier synchronizer.
In IRCS, the reduction of the amount of information is effected by attempting to convert the received modulated carrier to an unmodulated carrier (pure tone) before applying it to a phase-tracking loop, in the hope of improving performance. Traditional IRCS systems for synchronization with carrier signals modulated by BPSK include Costas loops, data-aided loops, and demodulation/remodulation loops. The traditional systems are designed to implement various approximations of a closed-loop structure that effects maximum a posteriori (MAP) estimation of phase. The degradation of tracking performance of such a loop in the case of BPSK is represented by a quantity called the "squaring loss," which is a measure of the degradation of the receiver signal-to-noise ratio and is associated with the mean-squared phase error of the loop. In the case of a conventional in-phase/quadrature (I-Q) carrier-tracking loop, the mean-square phase error is a result of signal and noise cross products that are generated in the effort to remove the data modulation from the loop error signal. At low symbol SNR, the squaring loss of an I-Q loop can be severe enough to prevent tracking.
If the data sequence and its timing were completely known, then a BPSK signal could be converted to a pure tone simply by multiplying the BPSK signal by the data waveform. One could then track the unmodulated carrier with improved performance by use of a phase-locked loop, which does not exhibit squaring loss. Short of complete knowledge of the data waveform and in the presence of noise, the best approximation of a pure tone could be obtained by feeding back an estimate of the data waveform corresponding to tentative decisions on the data symbols. Such feedback is called "decision feedback" for short.
Decision feedback is used within the traditional loops, but is not used to modify the loop structures.In the proposed IRCS system (see figure), decision feedback would be introduced at the input terminal of the loop; simultaneously, the structure of the loop would be modified (in the sense that its parameters would be modified) on the basis of the associated change in data-transition statistics in the input. The transition probability would be reduced from 1/2 (characteristic of BPSK signals in the absence of feedback) to a value closer to zero, so that the input signal would be converted to a close approximation of a pure tone, with a resultant improvement in carrier-tracking performance over conventional I-Q loops.
Although initially available data-waveform estimates are generally of low quality, they can be used to initiate the IRCS process by reducing the number of data transitions at the input. Once phase lock was achieved, the improved phase estimates could be fed back to the data detector, yielding improved symbol estimates for feedback, and thereby achieving even better phase tracking. This iterative process could eventually lead to virtual elimination of squaring loss, so that the performance of the system would approach that of a phase-locked loop operating on an unmodulated carrier signal.
This work was done by Victor Vilnrotter and Marvin Simon of Caltech for NASA's Jet Propulsion Laboratory. NPO-20261