Fuzzy Logic Modeling Of Digital I/O Buffers Based On The Use Of Prior Knowledge

With the increasing of the working frequency of integrated circuits and the reducing of circuits'size, signal and power integrity's analysis have becoming more and more important when we design high speed circuits and systems. How to build accurate and effective digital I/O buffer's macro models in order to simulate the circuits of the system level is an important question in the signal integrity analysis.At first, the paper introduce the basic process of the signal and power integrity analysis, especially in the importance of the digital I/O buffer model, summarize IBIS model used now and the being researchful modeling method of buffer, which is based the circuit's nonlinear dynamic character described strictly.The main job of the paper is coming up with the fuzzy logic modeling method of digital I/O buffer. Before building a question's model with fuzzy logic, we need to determine the structure of model composed of the number of rule, membership function and the form of premise and conclusion. Then we determine the parameters of model by building initial model and then learning and training. The paper introduces the simplification of TSK fuzzy logic system used in the situation. the complexity of the model simplified is reduced and the number of the parameters turns small, the number of rule and the parameters of the premise can be chosen by observing the DC characteristic of output port of driver, which simplify the process of model building greatly and make the initial model have better approximation precision and make the learning and training of model only need few iterations.On that basis, the paper make the way used in the power integrity modeling of digital buffer and the modeling of low voltage differential LVDS buffer. At last, by showing some real examples of the buffer modeling, we verify the validity of the method.