Identifying optimal generic processors for biomedical implants

Abstract

The extremely limited resource budget available to medical implants makes it imperative that they are designed in the most optimal way possible. The limited resources include - but are not limited to - battery life, expected responsiveness of the system and chip area. We have already detailed the design of a design-space exploration (DSE) tool specifically geared towards finding the Pareto-optimal design front. In this paper, we choose processor configurations from the Pareto-optimal processor set found by the DSE using real implants as case studies. We find that even under the extremely biased constraints that we use, our processor(s) perform better than many of the real implants. This provides strong hints towards designing an implant processor that is generic enough to cover most, if not all, implant applications.