Passband equalizers with filter coefficients calculated from...

Details Passband equalizers with filter coefficients calculated from... Passband equalizers with filter coefficients calculated from...

1999-06-18

2003-03-18

Pham, Chi (Department: 2631)

Pulse or digital communications

Equalizers

Automatic

C455S164100, C348S614000

Reexamination Certificate

active

06535553

ABSTRACT:

The invention relates to the suppression of multipath distortion in radio receivers, such as those used for receiving television signals.
BACKGROUND OF THE INVENTION
Multipath reception conditions give rise to ghosts in NTSC television reception. Multipath signals that arrive at the receiver with enough time displacement from the principal signal as to cause discernible ghosts in a received television image are referred to as “macro-ghosts”. Multipath signals which arrive over a transmission path of lesser length than the strongest or “principal” signal reach the receiver earlier and are referred to as “pre-ghosts”; the ghost images they cause in a received television image appear to the left of the desired image. Pre-ghosts occurring in off-the-air reception can be displaced as much as six microseconds from the “principal” signal, but pre-ghosts preceding the principal signal by more than four microseconds are rare. Multipath signals which arrive over a transmission path of greater length than the strongest or “principal” signal reach the receiver later and are referred to as “post-ghosts”; the ghost images they cause in a received TV image appear to the right of the desired image. Typically, the range for post-ghosts extends to forty microseconds displacement from the “principal” signal, with most post-ghosts occurring in a sub-range that extends to ten microseconds displacement. Multipath signals that arrive at the receiver with too little time displacement from the principal signal as to cause discernible ghosts in a received television image, but which affect transient response, are referred to as “micro-ghosts”. Macro-ghosts are more common in over-the-air terrestrial broadcasts than cablecasting, in which micro-ghosts commonly occur because of reflections introduced by un-terminated or mis-terminated cables.
Ghosts are a problem in digital television (DTV) transmissions as well as in NTSC analog television transmissions, although the ghosts are not seen as such by the viewer of the image televised by DTV. Instead, the ghosts cause errors in the data-slicing procedures used to convert symbol coding to binary code groups. If these errors are too frequent, the error correction capabilities of the DTV receiver are overwhelmed, and there is catastrophic failure in the DTV television image. If such catastrophic failure occurs infrequently, it can be masked to some extent by freezing the last transmitted good DTV image, such masking being less satisfactory if the DTV images contain considerable motion content. The catastrophic failure in DTV image reception may be accompanied by loss of sound, which is harder to conceal than momentary loss of video. Loss or break-up of sound may occur by itself, also.
Filtering to suppress macro-ghosts is often referred to as “ghost-cancellation” filtering, with filtering to suppress micro-ghosts being referred to as “channel equalization”. For the sake of brevity, in this specification the term “equalizer” will be used generically to describe a filter that suppresses both micro-ghosts and macro-ghosts.
Baseband equalization of demodulated signals can be done with digital filters sampling at the Nyquist or symbol rate of the signal being equalized. Such equalization is called “synchronous equalization”, and equalization cannot be satisfactorily achieved at lower effective sampling rates. If adaptation of the coefficients of the digital filters is to be done by decision-feedback method, synchronous equalization is not satisfactory when multipath distortion is susceptible to change at appreciable rate. Such reception conditions are commonly referred to as “dynamic multipath conditions”, and the multipath distortion occurring under such reception conditions is commonly referred to as “dynamic multipath distortion”. The signal to be equalized must be oversampled to obtain the bandwidth in the feedback loop necessary to track changing multipath distortion. Equalization is done by a digital filter or filters having the delay between successive taps a proper fraction of that in the digital filter(s) used for synchronous equalization. Accordingly, baseband equalization of oversampled demodulated signal is termed “fractional equalization”.
Aside from considerations of sampling rate, two basic types of equalization have been employed in the prior art, namely, baseband equalization of demodulated signal and passband equalization for signal modulating a carrier wave. Adaptive digital lowpass filters called “baseband equalizers” are used in baseband equalization. An adaptive digital filter used as a baseband equalizer has weighting coefficients that are adjusted responsive to decision-feedback error signal that is extracted from the demodulated radio signal or to received training signal extracted from selected portions of the demodulated radio signal. Passband equalization as known in the art uses adaptive digital bandpass filters called “passband equalizers” for supplying equalized responses to modulated carrier waves. Typically, the modulated carrier wave is an intermediate-frequency signal derived from a transmitted radio-frequency signal that has been selected for reception. Since passband equalization is performed before demodulation, persons skilled in the art of digital communications radio receiver design have favored its use for radio-frequency signals using modulation resulting in the carrier being central to its sidebands. Examples of such modulation are double-sideband amplitude modulation (DSB AM), quadrature amplitude modulation (QAM), binary phase-shift keying (BPSK) and quadrature phase-shift keying (QPSK). Passband equalization is preferred because the demodulation results, being already equalized, are more reliably suitable for carrier synchronization.
In passband equalization as practiced in the prior art, a digital bandpass filter used as a passband equalizer has its weighting coefficients calculated in accordance with a baseband-to-bandpass transformation of the weighting coefficients that would obtain for an equivalent baseband equalizer. This is the case whether the equalizer has its weighting coefficients determined responsive to training signal extracted from the demodulated radio signal, or has its weighting coefficients adjusted responsive to decision-error feedback signal that is derived from the demodulated radio signal. That is, the signal is demodulated to generate baseband signals that are compared with ideal baseband signals as the basis for determining weighting coefficients for the equivalent baseband equalizer, using a training signal method, a decision-error feedback method or a combination of these two methods.
Passband equalization as known in the art is not particularly well suited for vestigial sideband amplitude-modulation (VSB) signals such as those specified by the 1995 ATSC Digital Television Standard. This is because, with carrier not being midband, baseband-to-bandpass transformation of the equivalent baseband equalizer weighting coefficients results in a passband equalizer having a bandwidth of nearly 12 MHz. If the VSB digitized I-F signal has its carrier in the lower-frequency portions thereof, the carrier must be offset nearly 6 MHz to avoid folding of the bandpass passband. The uppermost frequencies of the VSB digitized I-F signal will be nearly 12 MHz, pushing the sampling rate requirement upward. If the VSB digitized I-F signal has its carrier in the higher-frequency portions thereof, the sampling rate needed to support the 12 MHz bandwidth of the digital filter is still the same. The doubling of sampling rate required by the doubled-bandwidth passband equalizer makes analog-to-digital conversion and phase-splitter filtering considerably more difficult to implement in practice.
Even in passband equalization for signals that employ modulation with the carrier being central to its sidebands, there are previously unrecognized problems associated with using the baseband-to-bandpass transformation of the weighting coefficients of the equivalent baseband equalizer to generate the weighting coefficients for

Affiliated with

Also associated with

Rating

Passband equalizers with filter coefficients calculated from... does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Passband equalizers with filter coefficients calculated from..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Passband equalizers with filter coefficients calculated from... will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3000619