Data-to-decision: Fueling netcentric defense solutions with the IIoT and fog computing

3In today's defense arena, traditional tactics using large, fixed systems no longer suffice. A deployed naval strike group might find itself needing to quickly integrate its data systems with those of on-site allies - some of which may be many years older - and process data with these combined platforms in real time. We are now in the era of the industrial internet of things (IIoT), and those who make the most flexible, intelligent use of the data streaming in from a dizzying array of clients have the advantage.

The principle of , of ­gathering massive amounts of diverse, potentially disparate information and merging it into a real-time set of actionable conclusions, governs much of the military’s current and future technology adoption policies.

From individual soldiers to joint command, computing and intelligent client networks offer the ability to pursue these goals more effectively. Manufacturers are now moving to provide robust platforms, connectivity, and complete solutions that dovetail with the military’s core technology priorities.

Top-level flexibility: DDS

Data Distribution Service (DDS) is an open-source middleware standard that enables the sending and receiving of data, commands, and events between network nodes regardless of location, host operating system, programming language, or host hardware platform. DDS acts as an intermediary – a translator of sorts – between different systems to facilitate interoperable data exchanges across networks. (Figure 1.)

DDS features automatic discovery and enables high-performance data-centric communication. The technology is low-latency, secure, fault-tolerant, and highly scalable. Since DDS’s arrival in 2004, the technology has been adopted and mandated in standards including the United Kingdom’s Generic Vehicle Architecture (DEF STAN 23-09), NATO GVA (STANAG 4754), and the U.S. Future Airborne Capability Environment (FACE).

21
Figure 1: DDS is a key enabler of the global information grid (GIG).

Just as operating systems such as and build value on top of the open-source kernel, PrismTech – an early DDS proponent founded in 1992 and acquired by in late 2015 – extended DDS with its own Vortex DDS Intelligent Data Sharing Platform. Vortex as a whole enables real-time data sharing across platforms from sensors to smartphones to servers, using a wide variety of programming languages, operating systems, browsers, and more. Vortex is deployable across a range of public, private, and hybrid cloud configurations, unicast and multicast networks, and many third-party or legacy applications in military platforms such as naval combat-management systems.

Ground-level flexibility: Cloudlet architecture

DDS platforms enable analysis and distribution of data across a wide variety of data sources and targets, including cloudlets. A cloudlet is effectively a small, mobile designed to emulate cloud computing at the internet’s edge – so-called . Cloudlets enable applications in which connectivity to the internet may be limited, network latency must be low, and/or interactive programs demand local processing beyond what clients can or should provide.

An example of cloudlet application is augmented reality (AR), in which graphical data overlays real world images. One AR example is the DARPA-funded ARC4 system designed with experts at Applied Research Associates (ARA), which uses a helmet-attached display to add information such as enemy locations, satellite footage, route conditions, and mission objectives into the user’s field of view. To be practical, AR systems must have well below 30 ms of latency between controller input (head-turning, for example) and display output. Otherwise, the experience becomes unnatural and frustrating. Pulling overlay data from , especially from remote areas, often introduces too much lag. Near-proximity cloudlets can solve this problem.

Cloudlet clients can be used in many different environments. A local server may manage a variety of semi-autonomous and autonomous systems, including wearable clients, robots, and unmanned air and ground vehicles. Each of these can interconnect in numerous ways to reduce resource demands on the local server. The local server, in turn, has more connectivity options to the wide-area network.

Conversely, cloudlet servers often assist clients by taking on some of their processing loads. Consider speech recognition, a vital technology able to help soldiers keep their hands free for noncomputing tasks. If there’s limited or no cloud connectivity, pushing the processing load back onto a wearable system might result in either inaccurate/unusable application results or impairment of other applications running on the system. Cloudlet servers can absorb these resource demands to keep clients running optimally; such offloading also helps extend client battery life.

Elements of a strong cloudlet server

Until recently, establishing a mobile command post often meant loading a five-ton truck with gear, satellite dishes, boxes of cabling, racks of servers, power supplies, generators, and more, all of which had to be assembled and cabled together. New directives emphasizing greater capability and much improved mobility would rather see groups pull up in a Humvee carrying an already-running cloudlet server.

A server, or network of servers, stands at the heart of every cloudlet. As with every other type of server, the desirable attributes of a cloudlet server will depend on the specific circumstances and applications. Effective cloudlet servers must contain ­-friendly processors, be extremely rugged, and be upgradeable/scalable.

22
Figure 2: The 12-core VPX3010 can be part of a system to handle highly parallelized compute-intensive jobs in the field.

A size-constrained server, ADLINK’s VPX3010 is a 3U blade measuring approximately 7.5 by 4 inches – roughly the size of an open hand. (Figure 2.) Its size means that a shoe box-size enclosure can house two or three VPX3010 blades to make up a fully integrated .2-compliant system.

Roy Keeler is senior product and business development manager, aerospace and , for ADLINK Technology. He has spent 30 years in the embedded computing, , software-defined radio, and IoT spaces within the aerospace and defense markets. Roy served in the United States Marine Corps before attending George Mason University, where he earned a BS in computer science and electrical engineering.

ADLINK Technology www.adlinktech.com