Page 10 - GCN, August/September 2017

[BrieFing]
Does
DARPA tunes machine learning to radio signals
FedRAMP
make TIC
BY SUSAN MILLER
The Defense Advanced Research Proj- ects Agency wants to apply the same kind of machine learning to the radio spectrum that is used by advanced systems for applications that include voice recognition and management of autonomous vehicles.
DARPA has issued a broad agency announcement for a new Radio Frequency Machine Learning Systems program that would address the need for enhanced situational awareness re- garding the ever-changing composition of RF signals for the internet of things and spectrum sharing.
Machine learning is widely used to manage data and images, but similar work in the radio spectrum offers unique challenges.
“What I am imagining is the ability of an RF machine learning system to see
and understand the composition of the radio frequency spectrum — the kinds of signals occupying it [and] differentiat- ing those that are ‘important’ from the background and identifying those
that don’t follow the rules,” said Paul Tilghman, a program manager in DARPA’s Microsystems Technology Office.
The technology would be able to discern subtle differences in the RF signals among identical, mass-man- ufactured IoT devices and recognize signals intended to spoof or hack into those devices.
“If we get this right, we will have RF systems with the ability to discern and characterize signals in the ever more crowded spectrum,” Tilghman said. “And that will give emerging automated systems and the military commanders that rely on them much- needed information to understand the landscape of the wireless domain.”•
redundant?
BY MATT LEONARD
International Trade Administration CIO Joe Paiva said agencies should not have to use a trusted internet connection to access cloud environments that have already been approved by the Federal Risk and Authorization Management Program.
Joe Paiva
Connecting to vendors’ cloud services is no different from a teleworker ac- cessing the office network or agen- cies connecting to data centers, Paiva said at the
Advanced Technology Academic Re- search Center’s Federal Cloud and Data Center Summit in August.
Paiva told GCN that on-site employ- ees use Transport Layer Security when they connect to the agency’s Amazon cloud, so their data is “never passing through the untamed wild.” Cloud ser- vices that are FedRAMP-approved and have full authority to operate “run in my environment; they’re not external.”
All the traffic to the public internet still goes through a trusted internet connection, he added.
The Office of Management and Budget has mandated the use of trusted internet connections since 2007 to re- duce the number of connections to the public internet. When asked if his setup ignores the mandate, Paiva said, “I just interpret it the way I think it makes sense to be interpreted.”
He added that his concerns about TIC have to do with cost. “If you don’t know this, TIC is egregiously expen- sive,” he said. •
Creating a cloud secure enough for classified workloads
10 GCN AUGUST/SEPTEMBER 2017 • GCN.COM
BY SUSAN MILLER
Cloud computing might be saving money for many federal agencies, but the intel- ligence community is finding that it is in- creasingly expensive to buy and maintain private infrastructure secure enough for classified or sensitive workloads.
To make it easier for infrastructure- as-a-service providers to offer public clouds that can securely handle such workloads, the Intelligence Advanced Research Projects Activity is exploring a concept it calls classified as a service.
According to IARPA’s request for in- formation, ClaaS would be “a classified private enclave encompassing multiple public cloud nodes in multiple loca- tions to accommodate general-purpose,
classified workloads elastically based on demand.”
It would eliminate the security issues related to IaaS, which could be vulner- able to side-channel attacks due to shared resources.
The idea behind ClaaS is based on bare metal-as-a-service offerings that give commercial cloud clients exclusive use of cloud server hardware for preset periods of time.
IARPA is considering developing new technologies that would give public cloud operators a way to provide secure classified, general-purpose processing
to the government by replicating the properties of air-gapped private enclaves within the public cloud for finite periods of time. •