Here is the abstract you requested from the IMAPS_2010 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.
|Accurate and Efficient BER Calculation by Statistical Simulation Based on Physical Transmit Jitter Model|
|Keywords: statistical simulation, bit-error-rate, jitter|
|Accurate bit-error-rate (BER) measurement is critical in high speed serial link designs. The traditional Monte Carlo method is impractical to calculate extremely low BER values specified in most standards as trillions of bits are required in simulations. Statistical analysis provides an efficient alternative to the Monte Carlo approach for low BER calculation by computing eye probabilities with statistical method. Transmitter (Tx) jitter posts a huge challenge in statistical simulation due to its pattern- and time-dependent nature and the resulting computational complexity. This paper introduces a novel approach to calculate Tx jitter in statistical simulation based on physical models of various jitter components. Fast yet rigorous calculation of Tx jitter effects on channel performance is made possible by a highly efficient NP-complete algorithm designed to reduce the computational complexity. The approach accurately capture effects of uncorrelated random jitters present at transmitter output, jitter amplification introduced by the channel, frequency dependency of periodic jitter and data duty-cycle-distortion (DCD) caused by asymmetric rising and falling edges of transmitted bit sequence. The simulation results from the proposed statistical techniques provide excellent correlation with the brute force Monte Carlo approach. The paper demonstrates that the conventional post-processing approach for Tx jitter is based on overly simplified assumption and therefore fails to accurately model the physical jitter behavior. The proposed method is applied to serial link simulations with Algorithm Model Interface (AMI) SERDES models.|
|Fangyi Rao, SW Expert
Santa Clara, CA