Pulse or digital communications – Receivers
Reexamination Certificate
1999-11-30
2002-12-17
Vo, Don N. (Department: 2731)
Pulse or digital communications
Receivers
C375S340000, C329S327000, C329S341000, C455S042000, C455S205000
Reexamination Certificate
active
06496547
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates generally to narrow-band digital frequency-modulated (FM) limiter-discriminator (LD) receivers and, more particularly, to a digital FM LD receiver that averages the outputs of two detectors to eliminate the effects of FM-clicks on bit error rates (BERs) for encoded digital signals.
2. Description of the Related Art
The FM-click-and-Gaussian noise model for frequency-modulation (FM) receivers is well-known in the electronic arts, having been introduced by Rice in a seminal 1963 paper (S. O. Rice, “Noise in FM receivers,”
Time Series Analysis,
M. Rosenblatt, Ed., Wiley, N.Y., 1963, pp. 395-422) and discussed in scores of works over the years. The Rice noise model assumes that the FM receiver accepts a signal in additive noise that consists essentially of a continuous Gaussian component occasionally interrupted by a click.
The performance of a narrow-band limiter-discriminator (LD) digital FM system was first described by Tjhung and Wittke (T. T. Tjhung et al., “Carrier transmission of binary data in a restricted band,”
IEEE Trans. Commun. Technol.,
vol. COM-18, pp. 295-304, August 1977) using the Rice noise model. Their approach was later simplified by Cartier (D. E. Cartier, “Limiter-discriminator detection performance of Manchester and NRZ coded FSK,”
IEEE Trans. Aerosp. Electron. Syst.,
vol. AES-13, pp. 92-70, January 1977) and subsequently a complete analytical solution for the bit error rate (BER) in a narrow-band digital FM receiver was described by Pawula (R. F. Pawula, “On the theory of error rates for narrow-band digital FM,”
IEEE Trans. Commun.,
vol. COM-29, pp. 1634-1643, November 1981) for certain regions of time-bandwidth product and frequency-deviation ratio. All of these early studies assume an integrate-and-dump (I&D) bit detector and rely on the Rice FM-click noise model.
The FM-click and the continuous Gaussian elements of the system noise were found to contribute equally to the BER for a frequency deviation ratio of h=0.7, and the clicks were found to dominate BER when h>0.7 and were found to be insignificant when h<0.7. Thus, the clicks were found to introduce a significant performance penalty in the optimum uncoded FM receiver using the integrate and dump (I&D) detector, for which the frequency deviation ratio, h=0.7 and the time-bandwidth product, BT=1.0.
In a later paper (R. F. Pawula, “Refinements to the theory of error rates for narrow-band digital FM,”
IEEE Trans. Commun.,
vol. 36, pp. 509-513, April 1988), Pawula describes the theoretical performance of an uncoded narrow-band digital FM receiver using an I&D detector that operates over a fraction of the bit-interval. As the bit-interval fraction is reduced to zero, the I&D detector becomes in the limit a sample and hold (S&H) detector. Pawula found that a S&H detector exacted a performance penalty of 0.5 dB when substituted for an I&D detector. Later experimental efforts have shown excellent agreement with Pawula's theoretical results for both the I&D and the S&H detector. Except as a limiting instance of a fractional-interval I&D detector, the S&H detector has been of little interest and has had little application in the art because of the 0.5 dB performance penalty.
A discussion of the Rice noise model by Bar-David and Shamai (I. Bar-David et al., “On the Rice model of noise in FM receivers,”
IEEE Trans. Inform. Theory,
vol. IT-34, pp. 1406-1419, November 1988) observes that practitioners in the art had long known that FM-clicks are the major limitation to improved BERs in LD threshold detectors, even at high signal-to-noise ratios (SNRs). Because the click is the culprit, the stochastic properties of FM-clicks had been widely investigated. The individual clicks had been shown to have a Poisson distribution at high SNRs, as would be expected. Individual clicks had been shown to be statistically independent of the Gaussian component at high SNRs.
As exemplified by the Bar-David and Shamai paper, the literature is replete with reported efforts to extend the BER performance of the uncoded digital FM receiver; including, for example, baseband pulse shaping, click estimation and cancellation, envelope compensation and sequence estimation. For many years, numerous practitioners have sought useful methods for detecting and eliminating the FM-click noise to improve the FM receiver noise threshold. Despite these many efforts to attain this well-known and long-sought objective in the FM receiver art, the problem of optimum click detection is still open.
Even more than uncoded digital FM reception, encoded digital FM reception also suffers from the limiting effects of FM-clicks. For instance, a seminal paper by Simon (Marvin K. Simon, “The impact of mismatch on the performance of coded narrow-band FM with limiter-discriminator detection,”
IEEE Trans. Commun.,
vol. COM-21, pp. 28-36, January 1983) explores the theoretical performance of convolutionally-encoded narrow-band FM with LD detection and Viterbi decoding, using the theoretical methods introduced by Pawula. Simon finds that FM-clicks are the direct cause of the failure of decoding techniques based on soft decisions that assume Gaussian statistics at the LD output. Specifically, Simon used a Chernoff bounding technique to decouple the coding and the modulation and obtained some surprising theoretical results. For instance, he found that the FM-clicks create a “mismatch” between the coding channel and the decoding metric peculiar to the digital FM modulation. So the Viterbi decoding scheme does not provide “maximal likelihood” (ML) decoding for this modulation format in the presence of clicks. Accordingly, the expected performance improvement of soft decision decoding over hard decision decoding is inverted so that the hard decision decoding is better. In fact, the soft decision decoding performance is so degraded by the click effects that it cannot match even the uncoded system performance. Using an analog-to-digital converter (ADC) saturation level, the best theoretical soft decision decoding performance found was only 0.3 db better than the hard decision decoding performance. These results provide a clear example of the heavy performance penalty exacted by click noise in encoded digital FM receivers.
With this in mind, Simon proceeded to describe a theoretical FM receiver for which all FM-clicks are summarily removed by a “genie-aided” click detector, for which he suggests no useful embodiment. Simon shows that a “genie-aided” click detector can improve hard decision decoder performance by 1.3 dB and soft decision decoder performance by 3.3 db over a digital FM receiver for which no genie is available. Although Simon clearly shows the value of a “pre-detector” for FM clicks, he neither teaches nor suggests any useful means for implementing such a mechanism for removing the FM-click noise component before decoding the digital FM signal at the receiver.
It is desirable to resolve this problem by providing a LD threshold detector that can eliminate the effects of FM-clicks on decoder BER performance. Until now, this has not been possible because of the well-known limitations discussed above. These unresolved problems and deficiencies are clearly felt in the art and are solved by this invention in the manner described below.
SUMMARY OF THE INVENTION
This invention arises from the unexpectedly advantageous observation that, when the same limiter-discriminator (LD) output signal is presented to a sample-and-hold (S&H) detector and to an integrate and dump (I&D) detector, an error in one does not necessarily imply an error in the other because the S&H and I&D detector outputs are offset in time by one-half bit and they are not quite correlated. To the extent that the two detector outputs are uncorrelated, comparing the two detector output signals provides information sufficient to identify bit error locations, thereby allowing bit error correction in a subsequent decoder. With convolutional coding and Viterbi decoding, threshold-compensat
Powell Robert F.
Sorgi Aladino D.
Cubic Defense Systems Inc.
Jester Michael H.
Vo Don N.
LandOfFree
Digital FM receiver employing combined sample-and-hold and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Digital FM receiver employing combined sample-and-hold and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital FM receiver employing combined sample-and-hold and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2942280