Research Projects

Analog and mixed-signal layout automation has been attempted for years with little traction in academia and industry. Although the completely manual approach is time-consuming, it has remained the de facto. The primary reason for automation not catching up is the inability to provide fine-grain control over the layout. Unlike digital layouts, it is often considered an artistic endeavour to generate analog counterparts. To resolve this fundamental issue, an assisted automation approach is proposed here. The analog designer/layout/mask engineer is central to this process and the proposed tool aids in accelerating the layout generation. In this proposal, specific questions that will be addressed are as follows: • Can a real-time layout template be generated using the designers’ schematic input? • Will a first-cut layout parasitic estimate during design with simplistic design rules help reduce the design-layout turnaround time? • Can constraint programming and satisfiability module theory solvers be used to accelerate AMS layout placement and routing? • Does the assisted output match the performance and layout cleanliness expected by designers? • Do layout techniques such as symmetry, common-centroid, and alignment help in reality? • Can learning algorithms be used to aid designers and can they be extended to auto-generate schematics and layouts?