Web Services for Asynchronous, Distributed Optimization Using Conservative Signal Processing

TL;DR

This paper introduces a scalable, distributed web service framework for asynchronous optimization that leverages commodity database back-ends and signal processing tools, demonstrated through a Firebase-based implementation.

Contribution

It presents a novel approach to implementing nonlinear signal processing systems as distributed web services for asynchronous optimization using commodity databases.

Findings

Effective distributed optimization using Firebase as back-end.

Scalable approach leveraging existing database infrastructure.

Demonstrated solutions to common signal processing optimization problems.

Abstract

This paper presents a systematic approach for implementing a class of nonlinear signal processing systems as a distributed web service, which in turn is used to solve optimization problems in a distributed, asynchronous fashion. As opposed to requiring a specialized server, the presented approach requires only the use of a commodity database back-end as a central resource, as might typically be used to serve data for websites having large numbers of concurrent users. In this sense the presented approach leverages not only the scalability and robustness of various database systems in sharing variables asynchronously between workers, but also critically it leverages the tools of signal processing in determining how the optimization algorithm might be organized and distributed among various heterogeneous workers. A publicly-accessible implementation is also presented, utilizing Firebase as a back-end server, and illustrating the use of the presented approach in solving various optimization problems commonly arising in the context of signal processing.