OPTIMIZING IoT CLOUD ARCHITECTURES FOR PIPELINING DATA THROUGH MACHINE LEARNING MODELS

摘要

Rapid advancements in Internet of Things enable meaningful solutions to several problems, which were never even thought of, until recently. Machine Learning, Deep Learning and Neural Networks play a vital role in this trend, and most of the IoT platforms, pipeline data through models, for a number of reasons, including but not limited to threat detection, real time analytics, disaster prediction, etc. These models require an enormous amount of computing resources, which typically requires a GPU, and integrating these models with the cloud poses a number of major challenges involving computing paradigm of the cloud. In this paper, we propose several optimized solutions to these problems faced while pipelining enormous amount data through models in real-time. These solutions addresses issues of scalability as well, when the platform provider needs to expand and will require increased number of pipelines in real time.