Countermeasures: Persistent surveillance data analysis no easy task | ADM Sep 2010
In the counter-IED battle, prying useful information from recorded surveillance data for indications of covert activities by insurgents – including IED placement – is a time consuming task with heavy demand on scarce analytical resources.
Tom Muir | Canberra
In Afghanistan, where the Improvised Explosive Device (IED) is the number one threat to military personnel and the local civilian population, the ADF’s Counter IED Task Force wants the analysis of surveillance data from its forward bases automated to speed the flow of intelligence within its own networks and to those of its allies.
But that’s no easy task and the RPDE organisation was co-opted to develop such a system.
In ADM’s June 2010 issue our previous report, ADF seeks persistent Forward Operating Base (FOB) surveillance, described the RPDE task aimed at developing a FOB persistent surveillance system enabling pattern-of-life analysis.
Sponsored by the Counter IED Task Force, its purpose is to reduce troop-led surveillance operations and provide the means for improved detection and analysis of insurgency attacks and networks leading to significantly improved situational awareness.
Surveillance is crucial in counter insurgency (COIN) operations, to counter IED attacks, improve force protection and detect change.
Currently, the ADF surveillance capability is restricted, due to the extent of terrain to be covered and the limited assets deployed.
Defence believes that extending the capability to persistent and expansive surveillance will improve the detection and analysis of insurgency attacks and networks, and provide significantly improved situational awareness (SA).
Currently, the ADF relies on obtaining intelligence through reconnaissance missions – Special Forces patrols, UAV surveillance (now including RAAF Heron and soon the RB-Q7 Shadow 200 UAS) and forward operations.
The requirement for the ‘FOB Eye’ as it was dubbed, is to address a capability shortfall in conducting surveillance over multiple areas of interest, within an area of influence, in support of a FOB.
Such a capability would enhance the ADF’s ISR operations, reduce troop-led surveillance operations and facilitate post-incident event analysis.
In August last year the Defence/Industry RPDE program initiated a Quicklook to identify industry solutions for persistent surveillance over a localised area to protect ADF FOBs on operations.
The proposed system is expected to provide continuous (all weather, day and night) systematic watch over a defined battlespace to provide timely information for combat intelligence purposes.
Airborne or ground based persistent surveillance systems, such as the widely employed Rapid Aerostat Initial Deployment (RAID) system, use a variety of sensors tethered from an aerostat, and later evolving to other platforms, including fixed towers and relocated masts.
They commonly use high resolution cameras and electro-optic (EO) and/or infra-red (IR) sensors characterised by large fields-of-view recorded over long periods of time.
The analysis of data from such systems found early favour with the US Army’s Constant Hawk image analysis system that applied pattern analysis to video images of a particular locality to find useful patterns of changing behaviour.
When changes were noted and checked more closely the system led to the early detection of roadside bombs and insurgent ambushes.
Reportedly this largely eliminated roadside bomb attacks on some supply convoys, which travel the same routes under airborne surveillance.
In pattern analysis, events of interest can include starting points and destinations of tracks and nodes for related entities within the persistent field of view.
They can also include activity and event-based normalcy and anomaly detections such as unique driving behaviors occurring before the detonation of a suicidal vehicle.
The advent of newer and more capable surveillance sensor systems has led to improved SA for the commander on the ground through real-time transfer of data, although only the most simplistic use of the imagery is possible due to the time-consuming and manual nature of the available analysis tools.
Archived data is used mainly for post-event analysis or to perform network analysis of a facility of interest, but here again, due to the manual techniques used, these analyses typically take many hours to many days to complete, and the end-product is text reports with simple graphics that are not machine readable.
The ADF lacks the resources for the time-consuming analysis of surveillance information and hopes, with the help of RPDE, to come up with an automated pattern-of life analysis capability for the very advanced persistent surveillance systems they plan to employ in the vicinity of their own FOBs, in place of current coalition-provide RAID towers and similar systems.
It is anticipated that for the surveillance capability the RPDE organisation will be seeking an advanced surveillance tower system that will combine various sensor packages such as ground surveillance radar, EO and thermal camera suites, and acoustic and flash detection systems.
As with most advanced systems, sensor data streams will be fused, enabling automatic tracking of multiple individual targets at rapid update rates.
This will include the ability to slave one sensor to another, for example, allowing a camera to follow a specific target which has been identified by the system, utilising data received from say a radar unit.
But is RPDE Task 34 FOB Eye—the melding of advanced surveillance capabilities with clever automated pattern-of-life analysis—verging on the reinvention of the wheel?
The answer is no – ADM believes RPDE is setting out to develop a unique capability that will automate the pattern-of-life analysis of imagery and other surveillance data.
These may include track-to-activity associations, correlation of activity patterns and so on, in this way assisting off-site ADF analysts to glean worthwhile intelligence from material recorded by the networked surveillance towers.
It seems that the potential ADF surveillance solution could well complement the US RAID capability through the provision of a limited footprint solution for small, remote FOBs not capable of sustaining or warranting the presence of the US RAID capability and the 20 plus contractors that operate each sensor node.
Indeed, RPDE pattern of life technology could enhance the effectiveness of current US RAID systems.
The final product ADM suspects will be something of an Aussie-Lite version of the major US surveillance and analysis system described below, highly effective in its protection of smaller and remote FOBs, but not designed for the much broader goals of DARPA’s PerSEAS project.
Last year the US Defense Advanced Research Projects Agency (DARPA) sought industry research proposals for a Persistent Stare Exploitation and Analysis System (PerSEAS) that can automatically and interactively discover intelligence from optical or infra-red devices in the air on drones, for example, or spread over urban, suburban, and rural environments.
PerSEAS is a software systems development and demonstration effort for automatically and interactively discovering actionable intelligence from Wide Area Motion Imagery (WAMI) of complex urban, suburban, and rural environments.
Used in a forensic mode, the system will exploit hours and days of WAMI data to identify threat activities and the underlying threat indicators.
Used in a near real time mode, the system will alert the user to developing threat activities in time to interdict.
In addition to the electro-optical/infrared (EO/IR) data available from WAMI sensors, PerSEAS will receive/send information from/to other intelligence sources.
The envisioned PerSEAS will significantly reduce the time required to perform many current exploitation tasks and greatly enhance an analyst’s ability to exploit the burgeoning volume of WAMI data.
The major thrust of the program is the subsequent processing of the low-level tracks and other extracted features to yield events of interest, which in turn would be linked to form activities and then integrated to discover potential threat patterns.
The discovery and identification of the potential threat patterns would then produce alerts and cues.