Robust, automatic design of highly efficient, variable precision filters

ABSTRACT: The Numericus Group, LLC has demonstrated during Phase I that our new approach to digital filtering vastly outperforms existing algorithms. At the heart of this new approach are algorithms that are guaranteed to converge and do not require an expert to oversee their operation, unlike many existing algorithms commonly used today. The new approach allows us to automatically design robust, highly efficient filters that achieve any target specifications within a user-supplied accuracy. These new algorithms represent profound advances in the state of the art in two primary areas: applications that push the limits of existing technology, and applications where automatic filter design is critical. Importantly, both types of applications are crucial to military users. During Phase II, we will transition many of these algorithms from prototypes to commercial software packages, and release this software through signal processing suites widely used in military and non-military sectors. We will demonstrate the superiority of our new packages over existing packages by using them to solve two challenging filtering problems from a classified SIGINT program that processes data from a USAF collector.; BENEFIT: Products developed during Phase II of this project will be released through signal processing suites widely used by military and non-military users. Thus, users will benefit immediately from this new digital signal processing technology. In particular, our new technology can improve throughput, reduce latency, and increase accuracy in many existing time-critical military systems--a fact that we will demonstrate on a real-world cutting edge SIGINT program that processes USAF data. Products developed during Phase II of this project also lay the groundwork for additional products to be developed during Phase III. Because our algorithms are robust and guaranteed to converge without the need for an expert to guide the filter design process, it is possible to use our algorithms to compensate for device-specific phase distortions. This capability has numerous military applications, which we will pursue during Phase III.