- Mon, 06/09/2021 - 10:00-12:00: public presentation PhD Piet Bronders
State of the art microwave measurements use a fixed sequence: they generate a stationary (mostly periodic) signal, measure the applied signals and responses, and extract features based on a device model (parametric or non-parametric). This way of working is perfect for single operating point measurements of simple devices. As components for the next-generation mobile devices will need to be more complex, more configurable, and more flexible, this classical way of working comes to a dead end. More device and signal parameters need to be swept (signal power, termination impedances, supply voltages, settings of the device …) over a potentially wide range of values. Therefore, measurement time increases in a combinatorial way: if 10 experiments are needed to scan 1 parameter, 10.000 experiments are required for as little as 4 such parameters. Scanning the complete space hence becomes impractical.
The instrument 2.0 aims at the characterisation over the complete parameter space while keeping the measurement time low and maintaining the accuracy to a user-selected level. Again, the introduction of additional prior device knowledge proves to be essential. Based on an initial ‘ideal’ behaviour of the component to test, we will investigate if and how extensive scanning of the space can be avoided though the use of space filling techniques.
The basis of these iterative techniques is simple: each iteration selects the most informative experiment to be performed. This experiment maximally improves the measurement accuracy over the selected exploration space. Experiments are then added until the user-specified accuracy is reached, or the allowed experiment duration is exceeded.
To design an optimal experiment trajectory, you will
- Assess the accuracy of a measurement based on stochastic variation of the measurement
- Assess the accuracy and the sensitivity over the parameter space
- Design an initial set of experiments based on an initial idealized model and a user-specified required accuracy
- Compare the methods for space filling available in the literature, apply them to the experiment design and select the most appropriate method
- Verify and refine the initial set based on measurements of the actual device
- Extend the methods for space filling to discrete variables (such as settings of the device)