Operationalizing Multi-INT

With Labor Day behind us, many of us are turning our attention to fall activities and events. In the Defense and Intelligence world, the GEOINT Symposium is one of the significant fall events. The theme this year is “Operationalizing Intelligence for Global Missions”, which I’m sure will spark a wide variety of interesting discussions and presentations. It got me thinking about how one might take Multi-INT analysis operational on a global scale. 

Multi-INT is an abbreviation for multiple intelligence, and it refers to the fusion or correlation of different types of data into a more complete picture. The data can come from a variety of sensors from traditional space- and air-borne image collection to unstructured information from chat feeds and social media web sites like Twitter and Facebook. The objective is to provide the most complete and cohesive intelligence picture possible to support informed decisions.

Multi-INT requires the processing and analysis of multiple types of data before the data can be brought together to produce actionable results. For example, an analyst may want to see the results of a target detection produced from analysis of hyperspectral data fused with the tracks generated from an analysis of ground moving target indicator data overlaid on the latest image of the geographic area along with road data. Add the location and content of relevant tweets, and you have a powerful picture of what’s going on in an area of interest.    

This type of intelligence is only valid for a short period of time, as situations change rapidly.  How can we take what has been labor intensive work for skilled analysts in different intelligence domains and make it happen fast enough to be actionable? Automation and workflows in enterprise environments are the keys to success.

Traditionally, the analysis of each data type has been done at the desktop level, one data product at a time. Products are pushed out to consumers in non-standard ways. To do Multi-INT analysis, the analyst needs the derived products for each relevant intelligence type and may need the source data as well. With all of this as input, Multi-INT fusion, analysis and correlation can begin. Multi-INT products need to be disseminated to the appropriate stakeholders. All of this is time consuming, and the product generation and dissemination can be inconsistent. 

As an alternative, suppose the source data was pushed up into the enterprise, or the cloud. If the processing and analysis of the data could also be put into the cloud, there would be significant time savings with the elimination of data copies between various repositories and the desktop. Then, automated workflows to handle much of the processing and analysis would further speed product creation and also eliminate inconsistencies due to individuals applying different tools in different ways. Finally, derived products could also be stored in the cloud where users could pull them on demand, or subscribe to them to be notified when new products are generated. 

My example above is certainly optimistic in some ways. There are going to be cases that require an analyst’s review and input, and this step may need to be a routine part of the processing workflow. Developing a consistent automated or semi-automated workflow can be done with good planning and design. Requiring people and organizations from different domains to work together may be more challenging. However, we’re already seeing this kind of collaboration as many organizations begin to see that the concept of the cloud is being realized and can be used to help operationalize intelligence. Automation of Multi-INT products in the cloud is one of the topics I’m hoping to hear more about at GEOINT this October.  How about you?

Lastly, Multi-INT isn’t relevant just for Defense and Intelligence. Check out this article describing how Federal Agencies use Sonar, LiDAR, Optical Imagery to Preserve Seafloor Habitats: From Sensor to Sound Decisions.