Linear Cancellation Research Articles

Hyper Radial Basis Function Neural Networks for Interference Cancellation with Nonlinear Processing of Reference Signal

Vorobyov, S. A., and Cichocki, A., Hyper Radial Basis Function Neural Networks for Interference Cancellation with Nonlinear Processing of Reference Signal, Digital Signal Processing 11 (2001) 204–221 Efficient interference cancellation often requires nonlinear processing of a reference signal. In this paper, hyper radial basis function (HRBF) neural networks for adaptive interference cancellation is developed. We show that the HRBF networks, with an appropriate learning algorithm, is able to approximate the interference signal more efficiently than standard radial basis function (RBF) networks. The HRBF network-based canceller achieves better results for interference cancellation. This is due to the capabilities of the HRBF networks to approximate arbitrary multidimensional nonlinear functions and better flexibility in comparison to RBF networks. Simulation examples and comparisons to the FIR-based linear canceller and the RBFN-based canceller demonstrate the usefulness and effectiveness of the HRBFN based canceller.

Cochannel interference suppression through time/space diversity

Wireless systems are subject to a time-varying and unknown a priori combination of cochannel interference, fading, and Gaussian noise. It is well known that multiple antennas can provide diversity in space that allows system tradeoffs between interference suppression and mitigation of fading. This paper describes how to achieve these same tradeoffs through diversity in time provided by channel coding. The mathematical description of time diversity is identical to that of space diversity, and what emerges is a unified framework for signal processing. Decoding algorithms are provided for repetition codes, rate 1/n convolutional codes, first-order Reed-Muller codes, and a new class of linear combination codes that provide cochannel interference suppression. In all cases it is possible to trade performance for complexity by choosing between joint estimation and a novel low-complexity linear canceler structure that treats interference as noise. This means that a single code can be used in a variety of system environments just by changing the processing in the receiver.

Limited linear cancellation of multiuser interference in DS/CDMA asynchronous systems

This paper presents a linear cancellation detector for the Gaussian channel, operating over limited time intervals of the received signal, in a similar way to a multiinput multioutput finite-impulse response (FIR) filter. The parameters defining the detector become time invariant, and the conditions to be met by the signatures are stated. A bound for the multiuser interference due to the limited time correlation, and upper and lower bounds for the error probability are obtained. The theoretical bounds and numerical results show that the detector is adequate for systems intended for many users whose amplitudes can be restricted to a given range.

Transient Evoked Otoacoustic Emission with Unexpectedly Short Latency

Two alternative approaches for studying short-latency click-evoked otoacoustic emission (OAE) in normal-hearing subjects were employed. Growth of a click-evoked “ear canal response” with stimulus increase became progressively more non-linear and saturated when the latency of the analyzed segment of response increased. This relation between latency and shape of the response input/output function was observed even after linear component cancellation, indicating that it could be an intrinsic property of OAE. Hence, the existence of an essentially linear short-latency OAE component which is probably eliminated by commonly used artifact cancellation technique is suggested. Taking into account the fact that transient evoked otoacoustic emission (TEOAE) may be completely suppressed by simultaneously presented noise, a “true” artifact cancellation was performed by subtracting the ear canal response in the presence of a masker from the conventional click-evoked OAE recording. An additional TEOAE component with a latency of 2.5-5 ms was found. Its growth with stimulus intensity was indeed more linear than that of later components. However, latency and frequency of this TEOAE component, being specific for each subject, can hardly be explained by both a commonly assumed latency-frequency relationship of TEOAE and a generally used estimation of TEOAE latency as the sum of the forward and backward traveling wave propagation times.

Electronic crosstalk cancellation in high densityWDM systems using linear cancellation

Electronic crosstalk cancellation in high density wavelength division multiplexed (WDM) systems is demonstrated for the first time for a two channel 2.5 Gbit/s WDM system using linear cancellation. For a crosstalk level of -5 dB, the system transmission power penalty is reduced from 4.3 to 2.6 dB, corresponding to an improvement of 1.7 dB in receiver sensitivity.

New detector for 1,k codes equalized to class II partial response

A modified linear canceller is described which approached maximum-likelihood performance for a 1,k RLL code on a PR2 (class II partial response) channel. With white noise, this detector gains 1.8 dB over simple threshold detection. Performance is analyzed both theoretically and through bit-by-bit simulations for a variety of noise colorations and misequalization conditions. The technique is applicable wherever the channel is equalized to a low-pass characteristic as in instrumentation and video tape recording with Miller-squared code or optical recording with a 1,7 code and pulse-width modulation. >

Adaptive Equalization Techniques for Digital Video Recording Systems

An adaptive automatic equalizer applicable to high-rate data recording systems such as digital VTRs is proposed. This equalizer is constructed on the nonlinear equalization technique called linear cancellation and on probabilistic decoding techniques. Two particular advantages are noted: that equalization with logical error correction is performed and that the variable part of the equalizer can be composed solely of digital circuits. It has been confirmed that sufficient eye-pattern opening can be achieved using this equalizer at an operation speed of 46 Mbits/sec.

Automatic equalizer for digital magnetic recording systems

This paper describes a new adaptive automatic equalizer based on both Linear Cancellation and probablistic decoding techniques for eliminating Intersymbol Interference in digital magnetic recording systems. It is confirmed that sufficient eye opening and error rate improvements can be achieved using this equalizer at an operation speed of 46 Mbps.

A Simplified Two-Stage Equalizer with a Reduced Number of Multiplications for Data Transmission over Voiceband Telephone Links

The paper presents an equalization scheme based on the two-stage equalizer originally introduced by Proakis. Mostly additions are required for implementation of the equalizer structure and the adaptation algorithm. The equalizer can be applied to data transmission Systems which use four-phase modulation, especially in 1200/2400 bit/s modems, thus replacing the fixed equalizer recommended by CCITT. The results presented in the paper deal with the analog and digital implementation of the proposed equalizer. Its performance is compared to the performance of the conventional transversal equalizer, a decision-feedback equalizer, an ideal linear canceller, and an ideal QPSK system.

The effect of bias on the linear canceller

The linear canceller has been shown to be effective in mitigating intersymbol interference in the presence of slope distortion. However, even moderate amounts of bias can affect the stability and performance of the canceller. We infer the eigenvalues associated with the autocorrelation matrix of the jointly adaptive infinitely long linear canceller structure with the aid of a theorem on asymptotic convergence. As in the case of the fractionally spaced equalizer, these eigenvalues are spread over a wide range; hence, the linear canceller taps can grow large when a bias exists in the updating error signal. This occurs even though the canceller is only symbol-space d, in contrast to the phenomenon in an equalizer. One solution is to apply the tap-leakage algorithm to both jointly adapting filters constituting the canceller.

A Unified Theory of Data-Aided Equalization

A unified theory is presented for data-aided equalization of digital data signals passed through noisy linear dispersive channels. The theory assumes that some past and/or future transmitted data symbols are perfectly detected. We use this hypothesis to derive the minimum mean-square error receiver. The optimum structure consists of a matched filter in cascade with a transversal filter combined with a linear intersymbol interference canceler which uses the ideally detected data symbols. The main result is an expression for the optimized mean-square error as a function of the number and location of the canceler coefficients, the s/n, and the channel transfer function. When the number of canceler coefficients is zero, we get the well-known result for linear equalization. When the causal or postcursor canceler approaches infinite length, we obtain the well-known decision feedback result. When both the precursor and postcursor cancelers become infinite, we obtain the very best result possible, namely, the matched-filter bound dictated from fundamental theoretical considerations. Neither the decision feedback nor the matched-filter results can be achieved in practice since their implementation requires infinite memory and storage. Our theory can be used to calculate the rate of approach to these ideals with finite cancelers.

Adaptive Cancellation of Intersymbol Interference for Data Transmission

In this paper, we analyze a technique for accurately detecting transmitted data symbols contained in a modulated signal that has been degraded by a linear dispersive channel and additive Gaussian noise. The approach uses an adaptive equalizer which provides preliminary decisions to an adaptive canceller. The canceller output is used to remove the interference from an adaptive matching filter, resulting in the desired signal. Channel equalization attempts to invert the channel transfer function, while avoiding excessive noise enhancement. However, cancellation (as used in echo cancellers), attempts to generate a replica of the interfering signal and subtract it from the actual received signal containing the sum of the desired signal and interference. The cancellation approach, unlike equalization, offers the possibility of removing interference without enhancing the level of noise already present in the received waveform. Simulation results for transmission over practical channels show significant improvement of linear cancellation over both linear forward and decision-feedback equalization.

Linear cancellation technique for suppressing impulse noise

At the present time, known techniques for suppressing the effects of impulsive disturbances can be identified with one of three general approaches: (a) The use of nonlinear saturating elements or noise clippers; (b) Disturbance-triggered, gating-out schemes; (c) The use of special signal design such as special coding, or linear premodulation waveform alteration and postdemodulation correction of the message. In this paper a new technique that is fun&- mentally simple and hear is described; it is effective even when the i-f filter responses to the various impulses overlap, as long as the number of overlapping responses is not great. The only important assumption made is the one that is fundamental in all impulse-noise studies - that the individual bursts have durations that are substantially shorter than the duration of the shortest important message element. This technique (which we call the linear-cancellation technique") makes use of an important property of the impulse response of a linear filter - that its maximum value is directly proportional to the filter bandwidth. This technique is applicable to the reception of all types of modulation, even to pulsed transmissions, subject, of course, to the fundamental assumption about the duration of the input noise pulse relative to the duration of the shortest important message element.