/pdf/embedded-electronic-circuits-for-cryptography-hardware-a9deldw3sq.pdf

Embedded electronic circuits for cryptography, hardware security and true random number generation: an overview

In the stringent timing requirements of high-speed systems, the effect of jitter on the bit-error rate and overall timing error can no longer be overlooked. As jitter timing budgets are normally specified in the form of individual jitter components, a fast and reliable jitter decomposition technique plays an integral part in the design and verification of high-speed data links. This article presents a machine learning approach to extract the random and deterministic jitter components from the timing histograms of the eye diagram. Training data are generated using the dual-Dirac model, and a preprocessing technique is proposed using vector fitting to create a rational function representation of the jitter probability distribution function. Numerical results indicate that the proposed method is able to perform jitter decomposition accurately and generalize well under different circumstances. In addition, results show that the preprocessing step is able to accelerate the training process and improve the overall performance of the neural networks significantly.

Jitter Decomposition of High-Speed Data Signals From Jitter Histograms With a Pole–Residue Representation Using Multilayer Perceptron Neural Networks

An improved method for decomposition of random and deterministic jitter measurements using tail fitting of the probability density function of total jitter is proposed. The currently employed techniques assume that the tails of the probability density function tend to a Gaussian function. We show that to be inaccurate by deriving analytical expressions for the accurate asymptotic form of the probability density and cumulative distribution functions. In doing so, we prove that the tails approach a Gaussian function multiplied by a term that is inversely proportional to the total jitter. Monte Carlo simulations of jitter consisting of a combination of random and deterministic components are performed and nonlinear least squares fits to the derived asymptotic forms are used to estimate the root mean square of the random component and the bounds of the deterministic component for a few examples.

Improved Jitter Distribution Tail-Fitting Algorithm for Decomposition of Random and Deterministic Jitter

A small amount of jitter can quickly eat up timing budgets and create timing issues. Precise characterization of deterministic and crosstalk-induced jitter can help isolate and solve issues within high-speed links. Characterizing deterministic and crosstalk-induced jitter is challenging, however, because many types of jitter work together to create the overall jitter profile. Methods are presented in this paper to characterize the deterministic and crosstalk-induced jitter from measurements of total jitter. An improved tail-fit deconvolution method is proposed for characterizing the impact of deterministic jitter in the presence of random jitter. The contribution of random jitter to total jitter is found first, and then that contribution is accounted for to find deterministic jitter. A Wiener filter deconvolution method is also presented for extracting the characteristics of crosstalk-induced jitter from measurements of total jitter made when the crosstalk sources were and were not present. The Wiener filter allows for accurate deconvolution of the measured histograms for total jitter even in the presence of the measurement noise. The proposed techniques are shown to work well both in simulations and in measurements of a high-speed link.

New Methods to Characterize Deterministic Jitter and Crosstalk-Induced Jitter From Measurements

The performance of ultra-high frequency (UHF) sensors has a significant impact on the accuracy of UHF partial discharge (PD) detection. In this paper, a method is presented for the time-domain characterization of UHF sensors to provide a comprehensive evaluation of their performance in the time domain. The peak envelope value and the envelope width are adopted in the method as two characterization parameters for predicting the output waveforms of UHF sensors. However, due to the ill-posedness of deconvolution, the measurement of these parameters is highly sensitive to noise. Therefore, an automated optimization scheme is devised based on the multi-parameter filter technique to solve the problem. Two UHF sensors are then investigated using the proposed method. The measured characterization parameters are found to be dependent on the orientation of the sensors, which is discussed from the perspective of sensor geometry. Further, PD experiments are carried out using the same UHF sensors to validate the method. The actual performances of the sensors are in good agreement with their time-domain characterization parameters. The proposed method would be complementary to existing frequency-domain characterization methods and is useful for the comparison of different UHF sensors, especially when their output waveforms are utilized for PD detection.

A time-domain characterization method for UHF partial discharge sensors

Although decreasing IC feature size and increasing I/O speed enable better system capability and performance, they also introduce technological challenges. One of the most important challenges is as I/O speed increases: jitter should decrease accordingly to ensure a reasonable bit error rate (BER) for a link system. Precise jitter characterization of signals at critical internal nodes provides valuable information for hardware fault diagnosis and next generation design. Understanding the separate contributions to jitter is challenging in this high speed industrial world. In this paper, Bounded Uncorrelated Jitter (BUJ), Random Jitter (RJ) and Periodic Jitter (PJ) are examined. A method is proposed to find BUJ in the presence of RJ, by first estimating RJ and then finding BUJ through de-convolution. A new tail-fit method is proposed to estimate the probability distribution for RJ. This new tail fit method is applicable for the general signal histogram, while the traditional tail fit method is only valid for the specific or ideal signal histogram.

Ting Zhu

Papers

New Methods to Characterize Deterministic Jitter and Crosstalk-Induced Jitter From Measurements

A new tail-fit method design for jitter decomposition