An adaptive equalizer comprising a computing unit which receives a known signal sequence to estimate transmission channel characteristics and effect compensatory control of tap coefficients by use of a first algorithm that has fast convergence property, and a tap coefficient computing unit for making compensation for relatively slow changes in a random data input after the compensation for the transmission channel characteristics, which either employs an algorithm that involves a relatively low computational complexity or intermittently executes computation which contains an interpolation of consecutive sets of tap coefficients between intermittent intervals.
RELATED APPLICATIONS
The present application is a continuation-in-part of U.S. application Ser. No. 604,940, filed Oct. 29, 1990, now U.S. Pat. No. 5,068,873, entitled "Equalizer", issued on Nov. 26, 1991.
An adaptive receiver apparatus including a demodulating circuit for demodulating a reception signal, an adaptive equalizer for equalizing a demodulated output to output a demodulated signal, a difference signal upon equalization and a channel impulse response estimate estimated upon equalization, a noise level detector for detecting a noise level based on the difference signal upon equalization, a channel impulse response variation magnitude detector for detecting a variation in the magnitude of the impulse response estimate at a predetermined level, and a coefficient controller for adjusting a coefficient determining a tracking property of an adaptive algorithm of the adaptive equalizer depending on the variation magnitude of the channel impulse response estimate detected by the channel impulse response variation magnitude detector and the noise level detected by the noise level detector.
An automatic equalizer for a data transmission channel capable of removing intersymbol interference and phase fluctuations accurately without lowering the convergence rate even when the frequency characteristic of the channel sharply changes. A summation output to be fed to a decision unit has the phase thereof so rotated as to compensate for an amount of phase fluctuation. The output of the decision unit to be applied to a feedback filter, which follows the decision unit, and an error to be applied to a tap coefficient control section each has the phase thereof rotated in such a manner as to add phase rotation in matching relation to the amount of phase rotation. Even when the tap coefficient control section is implemented with an RLS (Recursive Least Square) alogorithm, it can update tap coefficients by the same procedure with no regard to phase fluctuations.
A data burst is modulated on an IF carrier, and a spectral null is artificially created in the center frequency region of the spectrum of the modulated data burst by taking the difference between delayed and non-delayed versions of the IF data burst. A sync burst is modulated on the IF carrier so that its spectrum corresponds to the center frequency region of the data burst and superimposed on the IF data burst and transmitted. At a receive site, the IF sync burst is recovered by passing the superimposed IF signal through a bandpass filter having a passband corresponding to the spectrum of the IF sync burst and the IF carrier is recovered from the recovered sync burst. The sync burst at baseband frequency is recovered from the IF sync burst by a demodulator using the recovered carrier. Using the recovered carrier, a baseband sync burst and a baseband data burst having a spectral null are synchronously detected from the superimposed IF signal and fed into a decision feedback equalizer where the original data burst is recovered.
In a channel impulse response estimator (12), a local preamble generator (35) generates a local preamble signal identical with a transmission preamble signal in a received signal. A preamble detector (37) detects the transmission preamble signal in an estimated sequence signal to produce a preamble detection signal. Responsive to the preamble detection signal, a sequence selector (39) selects one of the local preamble signal and the estimated sequence signal as a selected sequence signal which is supplied to a re-modulator (14). Responsive to the preamble detection signal, a step-size generator (41) generates one of first and second step-size signals as a selected step-size signal which is supplied to a tap coefficient producer (26).
The detection of symbols in digital burst transmissions is improved by an equalizer (30) that initializes the values of only a subset of its filter coefficients (C, D), and thereafter begins to decode the symbols using an iterative, decision directed algorithm that determines all of its filter coefficients. This can advantageously provide an acceptable trade-off between computational complexity and equalizer performance.