SBIR-STTR Award

FishEye Intelligent Dynamic Data Logging – Automated Metadata Integration
Award last edited on: 6/30/2023

Sponsored Program
SBIR
Awarding Agency
DOD : MDA
Total Award Amount
$2,634,389
Award Phase
2
Solicitation Topic Code
MDA17-003
Principal Investigator
William (Bill) Schley

Company Information

FishEye Software Inc

2 Mill And Main Place Suite 400
Maynard, MA 01754
   (978) 461-0100
   info@fisheyesoftware.com
   www.fisheyesoftware.com
Location: Single
Congr. District: 03
County: Middlesex

Phase I

Contract Number: HQ0147-18-C-7206
Start Date: 11/6/2017    Completed: 1/5/2019
Phase I year
2018
Phase I Amount
$149,927
Data logging of complex simulations can easily overwhelm computer resources or risk losing critical data. Knowing what data is most relevant while simultaneously minimizing collection resources, is extremely challenging.Intelligent Dynamic Data Logging (IDDL) senses run-time data volume, considers scenario objectives, and monitors system performance to record only the most relevant data. It maximizes the value of data collected while minimizing the data collection impact to processing, memory, network bandwidth, and persistent storage. This project establishes an IDDL framework that:1. Provides a process whereby users can prioritize data types with a dependency on objectives, specific conditions or scenarios;2. Defines and collects computer performance metrics that impact data logging capacity;3. Includes algorithms that intelligently prioritize data logging dynamically during the simulation and uses machine learning to predict and adapt data logging; and4. Architects, prototypes, and validates a data capture platform-as-a-service thato captures only the most important data at the right time while considering scenarios and real-time performance metrics ando adaptively and proactively captures unexpected high-priority data through machine learning.Intelligent Dynamic Data Logging increases scenario reliability while assuring the most relevant data is always available.Approved for Public Release | 17-MDA-9395(24 Oct 17)

Phase II

Contract Number: HQ0147-19-C-7126
Start Date: 3/14/2019    Completed: 11/4/2021
Phase II year
2019
(last award dollars: 2022)
Phase II Amount
$2,484,462

The Phase II project combines existing innovative machine data collection with machine learning to create an Intelligent Dynamic Data Logging system that enables MDA to increase the volume of simulation executions to tens of thousands or more annually.Data logging is a foundational component of missile defense simulation but applies to other industries and markets.Examples include self-driving vehicles that collect multisensory data and crash forensics; aircraft black boxes enabling analysis of unexpected events; and instrumentation of the Internet of Things across large numbers of small devices to understand overall performance. These numerous commercial and defense applications can all benefit from meaningful data collection and improve their overall system performance through Intelligent Dynamic Data logging.Approved for Public Release | 19-MDA-9932 (21 Feb 19) ---------- Connecting data silos and extracting information is a ubiquitous challenge for the Missile Defense Agency (MDA). MDA operates many subsystems from diverse suppliers built with multi-generation technology. These subsystems emit complex and wide-ranging data flows at petabyte scale that makes data conversion impractical. There is a desire for unified data access for analysis, and for Artificial Intelligence (AI) Machine Learning (ML) model development and insight. Impediments to these mission needs are constantly evolving data interfaces, diverse data formats, the need to move and analyze petabyte-scale-data, and difficulty in accessing data. The proposed approach automatically extracts meta-data from subsystems to dynamically link evolving data into a common data repository or “Data Lake” for routine and impromptu analysis. Metadata enables a process that maximizes performance by limiting data manipulation, and ensures integrity by keeping data in its original form. Metadata is used to create views into the repository so that the pools of heterogenous data appear homogeneous to a diverse analysis community and their custom and commercial tools. Approved for Public Release | 22-MDA-11339 (13 Dec 22)