This paper presents an improved decision feedforward equalizer (DFFE) for high speed receivers operating on highly dispersive channels. The DFFE has been recently proposed for multigigabit communication receivers, where the use of parallel processing is required. Well-known parallel architectures for the traditional decision feedback equalizer (DFE) have a complexity which grows exponentially with the channel memory. Instead, the new DFFE avoids that exponential increase in complexity by using tentative decisions to iteratively cancel intersymbol interference (ISI). Additional complexity reduction can be achieved by improving the reliability of the initial tentative decisions. To provide more reliable initial tentative decisions to the DFFE, a simple reduced-state Viterbi algorithm (VA) is proposed in this work. Computer simulations demonstrate that the combination of DFFE and VA not only allows a similar performance to the typical DFE to be achieved, but it also results in a significant complexity reduction.

This paper presents a non-concatenated forward error correction (FEC) code suitable for applications in 100Gb/s optical transport networks (OTN). A typical requirement in this application is a net coding gain (NCG) >;10 dB at a bit error rate (BER) of 10-15with an overhead (OH) of ~20%. As discussed in [1], non-concatenated codes are the ultimate frontier in terms of performance for OTN applications, because of their superior performance, lower latency, and lower overhead than concatenated codes. However, a major stumbling block for the use of these codes has been the existence of BER floors at levels significantly higher than the required 10-15(typically 10-10). In this paper we present a new coding scheme based on a low density parity check (LDPC) code with an expected net coding gain of 11.30dB at 10-15, 20% OH, and a block size of 24576 bits. This represents a significant improvement over the previous state of the art [2], based on a concatenated code with a block size of 74844 bits and 20.5% OH. The code is designed to minimize the BER floor while simultaneously reducing the memory requirements and the interconnection complexity of the iterative decoder [3]. Experimental results obtained with an FPGA-based hardware emulator demonstrate an NCG of 10.70 dB at a BER of 10-13and no error floors. These experimental results are extrapolated to 10-15using importance sampling techniques, resulting in the expected performance stated above. Moreover, we find that fixed-point implementation is the main cause of error floors below 10-13Based on this finding, we introduce a new low complexity postprocessing technique to push BER floors down to 10-15.

This paper introduces the decision feedforward equalizer (DFFE), a new low complexity iterative equalization architecture for high speed receivers used on channels with large intersymbol interference (ISI). The decision feedback equalizer (DFE) is one of the preferred receivers for high ISI channels. The high data and symbol rate of current communications systems often require parallel processing receiver implementations [1]. Unfortunately, the complexity of parallel architectures of the DFE grows exponentially with the channel memory [2] and limits their application to low ISI channels. The DFFE avoids the exponential growth of the DFE by using tentative decisions to iteratively cancel the ISI. We show that the DFFE can achieve performance similar to the DFE. These benefits make it an excellent choice for high-speed receivers required to operate over highly dispersive channels.

In this paper we present the architecture and experimental evaluation of a single-chip CMOS maximum-likelihood sequence detection (MLSD) transceiver for electronic dispersion compensation (EDC) of multimode optical fibers (MMF) at 10 Gb/s data rate. The primary application of this transceiver is the 10 GBASE-LRM standard for 10 Gb/s Ethernet over MMF [1]. Application to single-mode fiber (SMF) channels and/or other standards is also possible. Both optical module and line card applications are discussed. The architecture of the receiver is discussed in considerable depth, including implementation details. Extensive experimental results on several channels are presented. For the precursor, postcursor, and symmetric stressors specified by the 10 GBASE-LRM standard we demonstrate receiver sensitivities of -14.84 dBm, -14.37 dBm, and -13.68 dBm, respectively, in a line card application with 6 inches of electrical interconnect These sensitivities exceed the requirements of the standard by 8.34 dBo, 7.87 dBo, and 7.68 dBo', respectively. Measured results on industry defined worst-case fibers and on nonstationary channels such as the dynamic stressor defined by the 10 GBASE-LRM standard are also presented.

In this paper we propose a multiple-input, multiple-output (MIMO) maximum-likelihood sequence estimation (MLSE) receiver for electronic dispersion compensation (EDC) of multimode optical fibers at 10 Gb/s data rate. As the primary example of the effectiveness of the techniques introduced here we present a receiver for the emergent 10 GBASE-LRM standard for 10 Gb/s Ethernet over multimode optical fibers [1]. We show that an MLSE receiver provides an advantage of at least 2 dB over other receivers, such as the decision-feedback equalizer (DFE), on 99% of the local area network (LAN) multimode fiber population, as modeled by a database proposed by the IEEE 802.3aq Task Force [2]. The paper deals with both theoretical and practical aspects of the architecture. To enable the complete integration of this receiver as a single chip in current 90 nm CMOS technology, we propose the parallel-processing digital implementation of a feedforward equalizer (FFE) and a Viterbi decoder together with an interleaved front-end analog to digital converter (ADC). The MIMO structure jointly compensates for fundamental channel impairments such as multimode dispersion, and implementation-related effects such as the limitations of the analog front end (AFE), particularly the one known as fixed-pattern noise (FPN) [3]. As a result of the high speed operation, FPN has a major impact on performance. We show that, without compensation, FPN could result in a degradation of 7.86 dB. With the compensation proposed here, the latter is reduced to 0.23 dB.

In this paper, we investigate the effect of mismatches in the frequency responses of the track-and-hold (T&H) amplifiers of an interleaved array of analog-to-digital converters (ADC) used as the front-end of a high-speed communications receiver. Furthermore, we introduce digital signal processing (DSP) techniques to compensate the performance loss caused by these mismatches. These techniques take advantage of the specific application of the ADC as a front-end of a digital communication receiver, assumed to be based on a parallel processing implementation of a decision-feedback equalizer (DFE) (Kasturia and Winters, 1991; Parhi, 1999). With a relatively simple modification, the DFE can be transformed into a multiple-input, multiple-output (MIMO) equalizer. The latter effectively compensates the effect of the T&H frequency response mismatch. The performance measure used in this paper is the signal to noise ratio (SNR) at the slicer. An SNR loss of 2.5 dB or more could result from the T/H mismatch. Most of this loss is compensated by the MIMO equalizer. Our conclusions can be easily extended to receiver architectures other than the DFE, for example maximum-likelihood sequence estimation (MLSE) (Agazzi, 2005). As an example of the effectiveness of the techniques introduced here, we present an optical receiver with electronic dispersion compensation (EDC) for the emergent IEEE 802.3aq standard for 10 Gb/s Ethernet over multimode fibers

In this paper we investigate maximum likelihood sequence estimation (MLSE) receivers operating on intensity modulated direct detection optical channels. Our study focuses on long haul or metro links spanning several hundred kilometers of single mode fiber with optical amplifiers. We describe the structure of MLSE-based optical receivers operating in the presence of dispersion and amplified spontaneous emission (ASE) noise, and we develop a theory of the error rate of these receivers. Computer simulations show a close agreement between the predictions of the theory and simulation results. We also address some important implementation issues. Optical channels suffer from impairments that set them apart from other channels and therefore they need a special investigation. Among these impairments are the facts that the optical channel is nonlinear, and the dominant source of noise is often ASE noise, which is distributed according to a noncentral chi-square probability density function (pdf). Additionally, optical fibers suffer from chromatic and polarization mode dispersion (PMD). Although the use of MLSE in optical channels has been discussed in earlier literature (J. H. Winter and R. D. Githin, Sept. 1990) (H.F. Haunstein et. al., 2001) no detailed analysis of optical receivers using this technique has been reported so far. This motivates the study reported in this paper.

A new piecewise linear modulation model of handwriting is proposed. In this model, the velocity of handwriting trajectory is modeled as the impulse response of a time varying second order system. For mathematical tractability, the entire trajectory is segmented into several non-overlapping frames, while the natural frequency and the damping factor of the system are assumed to vary linearly with time in each frame and are continuous along the entire trajectory. In other words, handwriting is regarded as an oscillation modulated by a continuous and piecewise linear signal. The parameters of this model are estimated by Powell's optimization algorithm which does not require the computation of the first order derivative. The number and lengths of the frames are decided by a modified binary search algorithm along with the estimation of parameters. This model has achieved very high data compression rate as well as accurate reproduction of real handwriting.

We develop a theory of the error rate of receivers that use tentative decisions for adaptive cancellation of intersymbol interference (ISI). We formulate precise conditions under which tentative decisions can be effectively used to cancel linear or nonlinear ISI, and verify the predictions of the theory by Monte Carlo simulations on four typical channels. In the case of magnetic recording channels, we show that the conditions for effective cancellation are satisfied by the precursor interference generated by inductive read heads, resulting in an improvement of the noise margin of 2 dB at the output of a symbol-by-symbol detector or 1 dB at the output of a Viterbi decoder at a bit error rate of 10/sup -9/. This technique has been incorporated in an experimental VLSI partial response maximum-likelihood (PRML) receiver, and the results confirmed by laboratory measurements. We also show that no improvement is obtained when this technique is used to compensate nonlinear distortion, which is another common impairment of magnetic recording channels, and we establish fundamental limits for the improvement achievable by any nonlinear equalization technique in magnetic recording.

Documents need to be sampled at different rates for different output devices: 300-600 dpi for laser printers, 100-200 dpi for fax, and 75-100 dpi for bitmap terminals. To output a high resolution document on a low resolution device, it may be necessary to introduce downsampling. Standard signal processing techniques such as linear filtering and decimation don't work very well at low resolutions. Better results are obtained by a nonlinear filtering technique we introduce in this paper, called nonlinear document equalization. Even better results are obtained by taking advantage of fonts designed specifically for bitmap terminals and other low resolution devices. However, character-level information is required to make use of fonts. This information is not always available; OCR is not 100% accurate. We propose a hybrid approach: downsample by font substitution when possible, and decimate when necessary. Unfortunately, the result tends to look like a "ransom note". Equalization is used to blend the two cases together so that gaps in the OCR analysis become almost unnoticeable.

An algorithm for robust machine recognition of keywords embedded in a poorly printed document is presented. For each keyword, two statistical models, called pseudo 2-D hidden Markov models, are created for representing the actual keyword and all the other extraneous words, respectively. Dynamic programming is then used for matching an unknown input word with the two models and for making a maximum likelihood decision. Although the models are pseudo 2-D in the sense that they are not fully connected 2-D networks, they are shown to be general enough in characterizing printed words efficiently. These models facilitate a nice "elastic matching" property in both horizontal and vertical directions, which makes the recognizer not only independent of size and slant but also tolerant of highly deformed and noisy words. The system is evaluated on a synthetically created database that contains about 26000 words. Currently, the authors achieve a recognition accuracy of 99% when words in testing and training sets are of the same font size, and 96% when they are in different sizes. In the latter case, the conventional 1-D HMM achieves only a 70% accuracy rate.>

An algorithm for robust machine recognition of keywords embedded in a poorly printed document is presented. For each keyword, two statistical models, named hidden Markov models (HMMs), are created for representing the actual keyword and all the other extraneous words, respectively. Dynamic programming is then used for matching an unknown input word with the two models and making a maximum likelihood decision. Both the 1D and pseudo-2D HMM approaches are proposed and tested. The 2D models are shown to be general enough in characterizing printed words efficiently. These pseudo-2D HMMs facilitate an elastic matching property in both the horizontal and vertical directions, which makes the recognizer not only independent of size and slant but also tolerant of highly deformed and noisy words. The system is evaluated on a synthetically created database. Recognition accuracy of 99% is achieved when words in testing and training sets are in the same font size, and 96% is achieved when they are in different sizes. In the latter case, the 1D HMM achieves only a 70% accuracy rate.>

The authors introduce a method to render optical character recognition algorithms based on pseudo two-dimensional hidden Markov models (PHMMs) independent of image transformations such as scaling, translations, slant, vertical and/or horizontal stretching, etc. Estimation of transformation parameters and image normalization are performed simultaneously with recognition. When combined with a previous method for joint segmentation and recognition of connected and degraded text, this method can be used to recognize extremely degraded documents that include characters affected by various geometric transformations. Experiments with isolated characters where scaling, slant angle, and translation are varied over ranges of 4: 1, 0/spl deg/ to 45/spl deg/, and 0 to 40 pixels respectively, are presented. Also presented are experiments with connected text where images have been affected both by geometric and stochastic distortions of various degrees, that show the high effectiveness of this technique.>

The authors discuss major issues related to algorithms and architecture of the transceiver system and be the signal processor that they designed and implemented in a single VLSI chip. The functions of each block are described in the context of the overall system. The ideas behind their algorithms and architectural design are summarized. While the main focus of this study is on the signal processor, a companion analog chip in the transceiver performs certain front-end processing and is briefly summarized. The transmitter is entirely an analog implementation, except for scrambling. It consists of a D/A converter and a second-order Butterworth filter, followed by a line driver that feeds line pulses to the hybrid. A/D conversion at the receiver is done by a sigma-delta modulator with a double integration loop, which places the quantization noise at high frequencies mostly outside the baseband.>

Tradeoffs in the design of the timing recovery functions in a subscriber loop receiver are analyzed. The techniques considered are applicable to both the echo cancellation (EC) and time compression multiplexing (TCM) methods of full duplex transmission. Emphasis is on those techniques that lend themselves to implementation in MOSLSI technology, where the objective requirement is that timing recovery be implemented on a sampled-data signal (with the minimum possible sampling rate where EC is used). The wave difference method (WDM) for timing recovery appears to be the best candidate. A detailed study of its performance is carried out analytically and by computer simulation for the case of binary and alternate mark-inversion (AMI) line coding. A closed form expression describing the binary jitter performance of the WDM and its continuous time counterpart, the spectral line technique, is used to compare the two techniques. Analytical and simulation results for recovered phase and jitter are presented for various cable pulse responses carefully chosen to represent worst-case or nearly worst-case conditions. Two methods for including frequency detection in the WDM, the quadricorrelator and the rotational detector, are also simulated.

This paper reports on studies concerning the feasibility of large-scale integrated realization of the circuits needed to provide hybrid-mode full-duplex digital transmission at 80 kbits/s or higher rates over standard local telephone loops. Alternative means of achieving the required 60 dB or so of echo cancellation have been studied in detail. The conclusion is that a combination of analog and digital circuit techniques permit practical MOSLSI realization of the complete modem, including filters, echo canceller, timing recovery, and A/D and D/A converters, without need for external circuit elements, trimming, or adjustments. The preferred system configuration has been evaluated by means of analysis, simulation, and laboratory and field measurements. A complete full duplex system, including an experimental NMOS integrated circuit echo canceller, was built and tested. Measurements showed a bit error rate lower thanl0^{-8}with line attenuation up to 40 dB, operating at 80 kbits/s. We conclude that a fully integrated MOSLSI ciruit to implement all functions for a hybrid-mode digital local loop is entirely feasible.

This paper describes a new technique for implementing an echo canceller for full-duplex data transmission (such as in digital subscriber loops and volceband data sets). The canceller can operate in spite of time-invariant nonlinearities in the echo channel or in the implementation of the echo canceller itself (such as in the D/A converters). The basic structure of the linear echo canceller is not changed, but taps are simply added to account for the nonlinearity. The number of taps which must be added depends on the degree of nonlinearity which must be compensated. Numerical results based on computer simulation are given which show that typical nonlinearities encountered in MOS D/A converters can be compensated by a relatively small number of taps added to the linear echo canceller, and substantial improvement in the cancellation results.