Michael P Card

US Patent:

20110219016, Sep 8, 2011

Filed:

Mar 4, 2010

Appl. No.:

12/717201

Inventors:

Michael P. Card - Manlius NY, US
Paul G. Springer - Manlius NY, US

Assignee:

SRC, INC. - North Syracuse NY

International Classification:

G06F 17/30

US Classification:

707769, 707802, 707E17005, 707E17014, 707E17044

Abstract:

A system and method for stream mining using state machines and high-dimensionality databases. After a data stream is digitized, a stream analyzer searches a high-dimensionality data structure containing state machine parameters to determine which state machines to activate and execute. A signal classifier creates the state machine parameters stored in the high-dimensionality data structure using a second high-dimensionality data structure programmed with information about signals of interest. If a state machine identifies a signal of interest, the system can optionally alert the user.

US Patent:

20110219037, Sep 8, 2011

Filed:

Mar 4, 2010

Appl. No.:

12/717380

Inventors:

Michael P. Card - Manlius NY, US

Assignee:

SRC, INC. - North Syracuse NY

International Classification:

G06F 17/00
G06F 7/00

US Classification:

707792, 707812, 707E17055, 707E17005

Abstract:

A high-performance object database wherein an application fetches an object from a database and the high-performance persistence framework constructs the object in memory. The database is programmed with a class that allows the application to selectively activate objects referenced by the class rather than automatically activate all referenced objects. The application selectively activates these referenced objects using read or write methods depending on whether the objects will be modified. Upon completion of the transaction, only those objects that were modified are written to disk. This high-performance persistence framework provides an object database capable of storing a large number of interconnected objects on disk and accessing them without having to activate all of the connected objects while simultaneously indexing the objects in a large number of independent dimensions all at once for fast data retrieval with complex queries.

US Patent:

20120280848, Nov 8, 2012

Filed:

May 3, 2011

Appl. No.:

13/099755

Inventors:

Michael Patrick Card - Syracuse NY, US
Paul Guy Springer - Manlius NY, US

Assignee:

SRC, INC. - North Syracuse NY

International Classification:

G01S 7/42

US Classification:

342 13

Abstract:

A method, a system and a program product for deinterleaving and classifying an incoming unknown radar pulse pattern uses a library of state machines that mesh with a library of known radar pulse patterns of a library of known radar emitter sources. A continuous sequence of radar pulse descriptor words is secured for the incoming unknown radar pulse pattern and a plurality of state machines that may mesh with the continuous sequence of radar pulse descriptor words is selected from the library of state machines. The plurality of state machines that may mesh with the incoming unknown radar pulse pattern is operated upon the incoming unknown radar pulse pattern. Radar emitter sources that correlate with state machines that actually mesh with the incoming unknown radar pulse pattern are designated as radar emitter sources that comprise the incoming unknown radar pulse pattern.