Method and system for modeling an electronic device under test (DUT) using a kernel method

A method of accelerating convergence of a kernel adaptive filtering for complex valued data used for modeling an electronic device under test (DUT) to generate a simulated response of the electronic DUT, the method comprising:

• determining a particular region on a complex plane;
• extracting a subsequence from complex valued training input series data as an intermediate training input vector, and extracting a corresponding single element from complex valued training output series data as an intermediate training output scalar;
• shifting an argument of each element of the intermediate training input vector by a first angle to provide a shifted training input vector, and shifting an argument of the intermediate training output scalar by the first angle to provide a shifted training output scalar, so that trajectories drawn by the shifted training input vector congregate around the particular region on the complex plane;
• extracting a subsequence from complex valued stimulus input data as an intermediate stimulus input vector;
• shifting an argument of each element of the intermediate stimulus input vector by a second angle to provide a shifted stimulus input vector, so that trajectories drawn by the shifted stimulus input vector congregate around the particular region on the complex plane;
• applying the kernel adaptive filtering to the shifted training input vector and the shifted training output scalar to calculate an output scalar corresponding to the shifted stimulus input vector;
• and
• shifting an argument of the calculated output scalar corresponding to the shifted stimulus input vector by the second degree, but in an opposite direction of the shifting of the argument of each element of the intermediate stimulus input vector, to provide a shifted output scalar.

A test system for modeling an electronic device under test (DUT) and generating a simulated response of the electronic DUT, the test system comprising:

• a signal generator configured to generate a stimulus signal;
• a divider configured divide the stimulus signal into a first portion and a second portion, the second portion of the stimulus signal being input to the electronic DUT;
• a first digitizer configured to receive and digitize the first portion of the stimulus signal;
• a second digitizer configured to receive and digitize a response signal output by the electronic DUT in response to the second portion of the stimulus signal;
• and
• a processing unit programmed to correct the digitized first portion of the stimulus signal and the digitized response signal, to measure training input series data of the digitized first portion of the stimulus signal and training output series data of the digitized response signal, and to utilize kernel adaptive filtering (i) to extract a device model from the training input series data and the training output series data, and (ii) to generate simulated responses of the electronic DUT in response to subsequent stimulus inputs, respectively.