Estimation of data-dependent power voltage variations of FPGA by equivalent circuit modeling from on-board measurements

Abstract

An equivalent circuit model was evaluated in simulating data-dependent power voltage variations of a field-programmable gate array (FPGA). The equivalent circuit model was Linear Equivalent Circuit and Current Source (LECCS) model representing dynamic switching current inside the FPGA with an equivalent current source. The current source was supposed to depend on input data for the FPGA on which a cryptographic circuit was implemented. Model identification was based on the procedure of LECCS model identification from on-board measurements and the current source was identified for all values of input data used in this work. The identified current source was investigated in accordance with the operation process of the cryptographic circuit and found an excellent correlation to the operation process. The identified LECCS model was combined with an equivalent circuit of the power distribution network for the FPGA core circuit to simulate power voltage variations for the 1,000 input texts. The simulated variation waveforms were compared to the corresponding measured ones to evaluate the LECCS model. Results indicated that the simulated and measured power variations matched excellently for all input data with high cross-correlation coefficients from 0.7 to 0.9. LECCS model is, therefore, able to predict the data-dependent power voltage variation by combining a PDN equivalent circuit.