One of the greatest challenges facing the U.S. defense and intelligence community remains the ability to consume, analyze and produce actionable intelligence, surveillance and reconnaissance. An increasing array of sensor systems and data types, demands for real-time analytics, and finding the signal in the noise all point to paradigm shifts in how we exploit data in motion.
Recently, IBM released a solution to many of the issues facing organizations seeking to extract intelligence from voluminous data channels. IBM’s Infosphere Streams, the product of a long collaboration between IBM and the U.S. government, provides a significant advancement in parallelization, data stream “database” functions, scalability and analytics against voluminous data streams.
In short, Infosphere Streams enables complex analytical functions with extremely low latencies. Infosphere Streams supports high-volume, structured and unstructured streaming data sources such as images, audio, voice, VoIP, full-motion video, TV, financial information, radio, e-mail traffic, chat, location information, satellite and airborne imagery, biometric information and many more forms of data.
TerraEchos Inc., a provider of a digital acoustic, fiber optic sensor array, recently chose Infosphere Streams as an embedded execution platform and services tier for its Adelos S4™ product. As the first commercial licensee of Infosphere Streams in the world, Terra-Echos integrated the product into the core Adelos S4 compute stack to manage, analyze and perform advanced computational signal processing calculations to meet specific U.S. government requirements and performance objectives. Adopting and standardizing on the Infosphere Streams paradigm allows TerraEchos to meet and exceed its data processing, analysis and classification objectives as part of an overarching set of sensor exploitation requirements.
Within the Adelos S4 data processing environment, the fiber optic sensor array produces a voluminous amount of sensor data streams, each with its own unique binary properties. Depending on the individual configuration and installation design, Adelos’ opto-electronics (S1) produce more than 2,000MB of raw binary digital data per second. An embedded digital signal processor (S2) performs initial calculations of sensor signals and converts the data streams into digital acoustic signals or live audio streams.
During the initial phases of a recent U.S. government program, the pre-eminent hurdle was to find a novel platform to process, analyze and produce reliable intelligence from the thousands of acoustic data channels. Moreover, the compute challenge required the ability to cross-correlate statistical and other classification algorithms against individual or groups of acoustic streams in network-real-time. With the sensor system collecting information at the speed of light, it wasn’t permissible to conduct complex analysis out of process; rather, database, statistical functions and other complex mathematical functions needed to be performed as the data streams entered the processors.
As an embedded computational engine, the Streams tier (S3) has met and exceeded initial requirements in a variety of operational field tests. Moreover, increasing attention has been placed on the ability to cross-correlate digital acoustic information with other forms of real-time, structured-unstructured data streams—or, conversely, reach back into non-real-time databases containing e-mail traffic, imagery or biometrics and consume these data as additional, parallel operations within the Streams application tier. This allows analysts to cross-correlate key metadata attributes such as time, space and ontology to increase the probability of target classification and, by extension, push these analytics to other applications for visualization, exploitation, fire-control, string analysis and pattern recognition. These iterative, cross-correlative loop mechanisms, leveraging embedded Streams, address many existing and future U.S. government requirements associated with data fusion in a manageable, practical, affordable and scalable manner.