MyAdvantech Registration

MyAdvantech is a personalized portal for Advantech customers. By becoming an Advantech member, you can receive latest product news, webinar invitations and special eStore offers.

Sign up today to get 24/7 quick access to your account information.

The Emergence of Wireless Thinking Networks



When examining Flame, a complex and sophisticated malware that appeared in 2012, investigators discovered an interesting feature: Flame could steal and transmit data from computers that had no Internet connections. 

It did this by using unsuspecting humans for bi-directional data transport. The process began by copying itself to every digital storage device that it encountered, including USB sticks and external hard drives. When humans hand-carried portable digital storage devices to unconnected PCs and desktops, Flame copied itself into the new computers and started stealing data. 

At this point, traditional spyware would have to upload the loot to a remote Internet server, but Flame was a bit more sophisticated. If there was no Ethernet connection, Flame just waited for the next external storage device to come along. When one did, Flame copied itself again and brought along a copy of the stolen data. It repeated the process until it found a computer with an Internet connection, and then started transmitting everything it had accumulated prior. 

Looked at a certain way, you could say that Flame used its human data mules as high-latency Ethernet connections. You could also argue that it is an early example of a promising tool in data communications: Disruption Tolerant Networking (DTN). 

New Approach to Networking 

The old network model assumed that the bulk of a network’s intelligence was located in controlling computers. If connections were lost, data would be lost as well. So, network designers tried to get as close as possible to what the Telco’s famously refer to as ‘Five Nines’ percent (99.999 %) uptime. The closer you got to perfection, the harder it was to achieve the next incremental improvement and the more expensive things would become. 

That model is changing, as downstream devices aren’t necessarily ‘dumb’ anymore. As integrated circuits become steadily smaller and more powerful, and software becomes more sophisticated, it is easier, more cost-effective and safer to distribute localized intelligence across the network. 

The University of Michigan, for example, has developed a low-power, smart sensor system that demonstrates many of the key principles that could be employed in the smarter, ‘thinking’ networks. At just nine cubic millimeters it is the size of a Vitamin C tablet, but is solar powered, has an internal battery and radio, and is equipped with its own processor called the Phoenix. The processor employs a unique power gating architecture and an extreme sleep mode to achieve ultra-low power consumption. 

Smart network nodes like the Phoenix system provide network designers with opportunities that have not been available to date. Smart nodes will require less bandwidth. Equipped with “situational awareness programming” that factors in parameters like power, network availability and the status of surrounding nodes, they can make independent decisions about whether there is any need to log in and report data. Smart network nodes will collect data, time stamp it, log it and — like Flame — report the data whenever a network connection becomes available. Even if the network is down for minutes or hours, this Disruption Tolerant Networking model can ensure that data is not lost, thus freeing designers from the need to pursue ‘Five Nines’ uptime. 

With their increased efficiency and intelligence, and their ability to hibernate when they have no useful information to report, smart network nodes require far less power than their ‘dumb’ predecessors. When combined with advances in power harvesting, as the tiny solar panel demonstrated on the Phoenix sensor system, this will give thinking networks the ability to extend the network edge to include locations and applications that previously had been completely inaccessible. These sophisticated, thinking networks will include network nodes that are completely independent of the power grid. 

Hardware for Thinking Networks 

Like Flame malware, with its human data mules, thinking networks can use multiple techniques to transmit and deliver information. Single vendor solutions and proprietary data communications protocols have already become a thing of the past. Users are demanding network-wide interoperability and the ability to make use of any data communication options that may be available. Legacy serial devices are being Ethernet-enabled with Wi-Fi connections using both embeddable and external Wi-Fi Access Points. Where a fiber optic buildout would be impractical, designers are deploying cellular routers that can establish network nodes anywhere that there is cellular telephone coverage. Thinking networks will continue the trend, using various combinations of copper cable, fiber optics, wireless, mesh networking and cellular data transmission. 

And, as network nodes get smarter, the methods used to transmit data will become increasingly irrelevant. Like Flame, thinking networks will find their way around obstacles and take advantage of whatever connection options are available, see to it that the data gets where it needs to go, and ensure that it arrives intact. 

None of this will make life any easier for network designers. Why? More and more network nodes will become independent of cable connections and the power grid. So, even though network designers will be able to invest less time and energy in pursuing perfect uptime, they will be forced to start thinking about power budgets. 

Recommendations for things like 30-day battery replacement cycles will be unacceptable. Network designs will have to use power efficiently to ensure that remote network nodes and devices never go dark. Traditional ‘always on’ communications schemes simply will not get the job done. 

Pros & Cons 

Distributed network intelligence and integrated communications infrastructures will be engines for increased efficiency and productivity. It is estimated that 50 billion devices will be network-enabled by 2020. Many of them will be M2M devices that communicate with one another to resolve problems with no need for human intervention, much the way ‘smart’ metering relieved utilities of the need to send employees out to site-visit and read meters in person. 

Thinking networks will ultimately be able to deliver just about any data, just about anywhere and the transmission methods involved will be completely transparent to the end user, whether that end user is a machine or a human being. 

But, the same hardware and software that help data travel across thinking networks will greatly complicate network security issues. Flame malware had already found a way to access the Internet from locations in which no Internet connection existed. We will soon live in a world where Internet connections are virtually everywhere. 

-- By Mike Fahrion, Chief Technology Officer, B+B SmartWorx/Advantech