Machine Learning via Serverless Cloud Computing in AWS Lambda

This work focuses on a novel approach to machine learning for efficient big data processing within a system dominated by mobile devices.  The proposed solution achieves this by utilizes custom, lean, and modular lambda functions to perform machine learning.  His solution:

  1. Employs a lean design philosophy such that serverless cloud computing resources are used efficiently.

  2. Interfaces lambda functions directly with cloud storage to efficiently access large amounts of data.

  3. Emphasizes code modularity and reliance on microservices to maximize scalability and reusability.

  4. Interacts with mobile device applications seamlessly as gateways for user interaction.

​Further information is in this publication.

Smart Wireless Network for Automation of Residential and Commercial Loads to Facilitate Participation in Demand Response Initiatives

In this work, researchers study a smart wireless network for automation of residential and commercial loads that would facilitate their participation in system-wide demand response initiatives.  Primary research objectives include: 1) developing a cost-effective and ultra-low power meshed network, 2) developing a learning-based optimization algorithm for load automation, and 3) applying this optimization to demand response.  Further information is available in this publication.

Low-Cost Solar-Powered USB Charger with Internal Lithium-Ion Battery Storage

In this work, researchers develop a low-cost solar-powered USB charger (5Vdc) with internal lithium-ion battery storage.  The design employs the LT3652 charge controller and TPS61032 power electronic boost converter.

Design and Testing of Custom FPAA Hardware with Improved Scalability for Emulation of Smart Grids

In this work, researchers present work in the field of analog emulation of electric power system dynamics via field-programmable analog array (FPAA) technology.  Specifically, they discuss development of a new emulation tool with increased modularity and operator density for analysis of larger power systems and smart grids.  One innovative aspect of this work is the fact that it employs custom FPAA boards developed by re-searchers at The College of New Jersey and Drexel University.  The work places emphasis on decreased prototype size, increased density of computational analog blocks (CAB’s), more effective FPAA interconnection scheme, and batch-mode FPAA configuration.  Further information is available in this publication.

Comparative Study of Accuracy and Computation Time for Optimal Network Reconfiguration Techniques via Simulation 

The objective of this work is to examine the effect of load flow analysis type, horizon length and discount factor, as well as switch ordering on the accuracy and speed of dynamic programming-based optimal network reconfiguration studies for mini-mizing loss.  The researchers employ simulation of a 118-bus power system to obtain results.  These results demonstrate that strong relationships exist between the parameters of an optimal network reconfiguration study and its performance.  The researchers discuss how these relationships may be used to predict performance of future studies.  Further information is available in this publication.