NVIDIA LAUNCHES GPU ACCELERATED GEOINT PLATFORM
By Yvonne Headington
A World First
NVIDIA, pioneers of visual computing, has unveiled a new accelerated Geospatial Intelligence (GEOINT) platform for defence and homeland security applications. Based on the company’s leadership in graphic processing unit (GPU) technology, the GPU accelerated system provides faster processing of high-resolution satellite imagery, video facial and object recognition and mission planning capabilities (using geographic information systems data).
The Company describes the accelerated platform as ‘the world’s first,’ enabling analysts to generate accurately actionable intelligence from petabytes (10005 bytes) of surveillance data, images and video. By using GPU accelerators, the GEOINT Accelerator analyzes data up to 10 times faster than systems using only central processing units (CPU). Although GPU accelerators are used in a range of defence and security applications, NVIDIA’s latest platform provides a complete set of applications required for the GEOINT analyst.
Too Much Information
During the launch on 16th July 2013 Dr Sumit Gupta, General Manager of the Tesla Accelerated Computing Group at NVIDIA, explained how the military is drowning in data although, “it’s more crucial than ever to extract critical information from video or images, especially in real time.”
Recent US operational experience has seen an explosion in available data. During 2007 the US Air Force (USAF) collected some eight year’s worth of video imagery from Iraq and Afghanistan; by 2009 the annual volume had tripled. According to a 2012 Rand study on The Future of Air Force Motion Imagery Exploitation, by 2015 the USAF, “could, in theory, require up to 117,000 personnel dedicated to motion imagery exploitation alone—one-third of the active-duty Air Force.”
Higher resolution video imagery and the addition of other sensors, such as light detection and ranging (Lidar) and synthetic aperture radar (SAR), compound the problem of data overload. “All this data needs to be integrated into a final simple action.” said Dr Gupta.
Companies such as MotionDSP, a Silicon Valley-based developer of real-time image processing and analytics software for video, already use NVIDIA GPUs for defence applications. The company’s Ikena ISR real-time image processing application runs on Windows-based workstations and servers with NVIDIA GPUs, providing improved image resolution and ‘big data’ management (which means that analysts can devote more time to exploiting intelligence rather than processing data). Customers include the USAF and the US Navy, as well as intelligence and civilian agencies.
US Company Luciad uses NVIDIA GPUs for its flagship LuciadLightspeed product, a next generation software solution that systems integrators and original equipment manufacturers (OEM) can use to develop C4ISR and air traffic management systems. The product provides data fusion, visualisation and analysis, covering static and moving data, maps, satellite imagery, and terrain elevation in a variety of formats. Luciad has extensive experience in supporting military C4ISR requirements, including those needed by NATO Forces in Afghanistan.
NVIDIA’s GPU acceleration is also being used to increase the level of autonomy for NATO’s MUSCLE (Minehunting UUV for Shallow water Covert Littoral Expeditions) autonomous underwater vehicle (AUV). The vehicle is operated by NATO’s Centre for Maritime Research and Experimentation (CMRE), based in La Spezia in Italy, for locating potentially dangerous First and Second World War mines.
AUV generally operate on pre-planned routes and record data for off-line processing. GPU acceleration increases the level of vehicle autonomy and MUSCLE has the flexibility to adapt to environmental conditions and sonar performance. In addition, MUSCLE is equipped with GPU accelerated high-resolution, high-frequency synthetic aperture sonar tec