Nirmal C. Warke

A 12.5Gb/s SerDes in 65nm CMOS Using a Baud-Rate ADC with Digital Receiver Equalization and Clock Recovery

A DSP-based low-power 12.5Gb/s SerDes using a baud-rate ADC and a digital data-path is developed for backplane data communication. A digital 2-tap FFE and a 5-tap DFE in the RX provide channel compensation. A BER of <10-15 is measured over legacy backplanes with 24dB loss at Nyquist. The power consumption and die area are 330mW and 0.45mm2 per TX/RX pair

Modulation classification in unknown dispersive environments

The problem of distinguishing reliably between signaling formats in the presence of noise, interference, unknown dispersive channel conditions, as well as timing and frequency mismatches is addressed. Methods based on a combination of blind equalization and universal classification are presented and their performance is assessed through simulations.

Optimum codec companding for high-speed PCM data transmission in telephone networks

Codecs used in telephone networks in the United States have a /spl mu/-law companding characteristic solely designed for the purpose of transmitting voice signals. These /spl mu/-law codecs are not well suited for the latest generation of high-speed digital voice-band modems requiring PCM data transmission at the quantization levels of the codec. In this paper, we have shown that using a linear codec in the telephone network is optimal for PCM transmission (in terms of a minimum symbol error rate) for the range of SNRs observed over most modern telephone lines. We demonstrate that with a linear codec we get a dramatic improvement in the performance of these PCM modems without any increase in the complexity of the modem. In practice, it would be necessary and relatively simple to implement an additional feature in the /spl mu/-law codec to detect a voice-band modem during the initial phase of training and switch to a linear companding characteristic.

Performance comparison of TEQ design techniques for FDM ADSL

TEQ design to maximize the data rate for FDM ADSL channels is challenging due to the bandsplit filtering. In this paper, a performance comparison of several TEQ design techniques is presented for typical FDM channels. It is observed that (a) no single TEQ design criterion achieves the maximum performance, and (b) the performance is limited by the resulting TEQ design producing large variations in the passband region and/or emphasizing the stopband region. It is then shown that in many of these cases, better performance can be obtained by applying a spectral flatness constraint to the TEQ design. Additionally, we demonstrate that the dual TEQ and per tone equalizer are well suited to equalizing FDM ADSL systems.

Impact of DFE Error Propagation on FEC-Based High-Speed I/O Links

Modern state-of-the-art I/O links today rely exclusively upon a high SNR channel and an equalization-based inner transceiver to achieve a BER of 10-15. The equalizer typically consists of a transmit pre-emphasis driver for pre-cursor equalization and a receive DFE for post-cursor cancellation. Recently, forward error-correction (FEC) coding has been proposed to improve the BER and reduce power in high-speed I/O links. However, error-propagation in the DFE is a significant issue affecting code performance. The link performance is also tied to FEC implementation parameters like degree of parallelism. This paper presents a framework for analyzing the impact of DFE burst errors and implementation parameters on end-to-end link performance. For 10.3125Gb/s transmission through a channel with 19dB loss at Nyquist rate and serial FEC implementation, we find that a code rate r = 0.8 gives the best ISI penalty vs coding gain trade-off, and a codeword length of 750 bits is necessary to meet target performance. Further, it is observed that the performance of burst error correction codes does not necessarily improve with codeword length, i.e., there is an BER-optimal block length at a given code rate.

Multidimensional signal constellation design for periodic channel impairments

We address the problem of designing a multidimensional signal constellation for the efficient transmission of symbols over periodic time-varying channels. The PCM telephone network is a classic example of such a channel. We describe a computationally efficient algorithm for designing the multidimensional signal constellation to maximize the data transmission rate over the channel. The algorithm takes into account constraints on the average error rate as well as the average transmit energy in designing the signal constellation. We provide simulation results on the performance of the proposed algorithm for PCM transmission over typical telephone channels.

Impact of statistical mismatch on both universal classifiers and likelihood ratio tests

In this work we demonstrate the robustness of the universal classifier proposed by Ziv and Gutman (with the LRT being a special case) to general source and model order mismatch.