what is behavioral modeling,what kind of job it is.

You sit around and try to come up with mathematical models of various components such as transistors, inductors, and even things as seemingly simple as vias. (At high enough frequencies, all of these have highly non-trivial equivalent circuit models.) Historically this was done by building your behavioral model out of a network of ideal components (e.g,. for a physical capacitor, you might start with a capacitor that has an inductor to model lead inductor, a resistor across it to model loss, etc.), although these days often 'black box' models that represent behavior in terms of a rational polynomial or even just a fancy table are often used. (The model you use depends largely on what sort of simulator you're planning to use the model in...)

In many cases, a significant part of the job entails making device measurements in the first place, and this is also somewhat more challenging than you might initially think, e.g., due to correctly de-embedding fixturing effects from, say., an 0201 resistor at 5GHz. Alternatively, initial measurements might come from a field solver, which is highly accurate but slow (and hence not practical use for rapid simulation by circuit designers).

Sure, it's "promising" -- for the right type of individual. It is a vital part of modern IC design -- circuit designers can't correctly simulate their designs without good models -- so you do get to share in the glory of an IC coming back from a fab and working correctly on the 1st try because all your models matched up with reality.

From an abstract point of view, the job is trying to 'twiddle the knobs' (and occasionally add and remove knobs) on your model such that they match known data.

---Joel Kolstad

This group seems to have died lately. No one seems to post to it any more. :-(

--

"What is now proved was once only imagin\'d" - William Blake

If your original electromagnetic model is linear, then nowaday you can simplify your life by means of formal model reduction. You may have a look at our paper where we have tried to compare compact modeling and model reduction

J. Lienemann, E. B. Rudnyi and J. G. Korvink. MST MEMS model order reduction: Requirements and Benchmarks. Linear Algebra and its Applications, 2005. Corrected proof is available on-line at

(if you do not have access to ScienceDirect, please contact me directly for a reprint)

Best wishes,

Evgenii Rudnyi

--

Half-day course before EuroSimE, Grand Hotel di Como, Sunday April

23rd, 2006. Automatic Compact Modelling for MEMS: Applications, Methods and Tools.