The Advanced Network Technologies Division (ANTD) of the Information Technology Laboratory (ITL) has four major project areas: Network Resilience, Cloud Computing, Internet of Things, and Future Network Technologies.
The ANTD’s projects include testing methodology of indoor localization and tracking systems, wireless networks, wireless networking specifically for smart manufacturing, network resilience, robust interdomain routing, high-assurance domains, measurement for complex systems, a distributed algorithm for suppressing epidemic spread in networks, the NIST cloud computing program, software-defined networks and virtual networks, and information-centric networking.
QUALITY OF THE RESEARCH
ANTD’s portfolio of projects is varied. Some projects focus on standardization and creation of International Organization for Standardization (ISO)1/NIST/Internet Engineering Task Force (IETF)2 standards. For example, the Cloud Computing initiative resulted in an ISO standard, and the Secure Border Gateway Protocol (BGPSec) resulted in an IETF standard. Some projects focus on the creation of data sets. For example, the Indoor Localization project resulted in the PerFloc data sets and competition.3 Others focus on simulation (e.g., the Complex Systems and the Inter-Domain Routing projects). Several focus on basic innovative research. Commendably, many projects have external collaborators, including universities, other ITL divisions, the NIST Engineering Laboratory, and contractor companies.
Indoor Localization Project
Network localization and navigation (NLN) is an area in which NIST can play an important role in creating databases and through roadmapping exercises. ANTD’s work in indoor localization is commendable. The ANTD has created a methodology to allow apples-to-apples comparisons of different smartphone-based NLN applications. ANTD’s database of measurements taken from representative buildings can be employed by ANTD personnel to evaluate the performance of proposed localization methodologies in future smartphones. This has enabled the testing of existing and new approaches to smartphone-based localization, demonstrating the important role of NIST in bringing together researchers
3 Further information is available at National Institute of Standard and Technology (NIST), “Rules of the PerfLoc Prize Competition,” updated May 17, 2018, https://perfloc.nist.gov/perfloc-competition-rule.php.
and developers from industry—increasingly based in Asia—and academia, and in facilitating the progress in this field. The current database, supporting measurements (readily available to NLN applications) from the sensor outputs of existing smartphones, and actual building data (e.g., the actual location coordinates of wireless access points in each of several buildings) are being used to support the development of the ITL-standard NLN roadmap, which is in progress. This project will, presumably, be further extended to create a richer database that includes measurements from sensors, suitable for NLN, but not available on existing smartphones, and measurements (not readily available to NLN applications) from sensors that are on existing smartphones. The ANTD might consider making this database publicly accessible for direct use by non-ANTD researchers, product developers, and systems engineers in the evaluation of prospective NLN technologies and methodologies.
There is also a need for a network-level device-to-device channel simulator (accounting for spatiotemporal consistency as well as correlation among transmitters and receivers) that developers from industry and academia can use as a common basis for comparison. A good example is the QuaDRiGa channel simulator,4 provided by the Fraunhofer Institute for Telecommunications, which offers a spatiotemporal consistency (at the transmitter side only). While this simulator is sufficient for cellular networks following the 3rd Generation Partnership Project (3GPP)5 standardized channel models, it would require extensions for NLN to account for consistency at both the transmitter and receiver sides and for device-to-device links in NLN. Providing such capability for NLN (beyond cellular networks) would put NIST on the map as one of the key players in the field.
Wireless Networking for Smart Manufacturing Project
Smart manufacturing is a growing and critical area of cyberphysical systems. Aligned with the mission of NIST, this project examines reliable wireless networking for smart manufacturing and the industrial Internet of Things (IoT) by investigating the radio frequency (RF) landscape, channel sounding and modeling, co-simulation, and scheduling. The project includes a plan to disseminate technology outcomes. Co-simulation and scheduling are heading in the right technological direction but need to focus more on flexible manufacturing scenarios referenced to industrial applications. Mechanisms such as consortia of industry and academia could be established as vehicles to support technological development and industrial deployment. The ANTD could proceed as a technology facilitator by utilizing the existing channel measurements and models, developed over the last several decades, and by identifying key technological components to form a framework and reference model, which would help industrial and academic developers to contribute technological innovations toward standards. This would help secure a leading role for the United States in smart manufacturing technologies.
Robust Interdomain Routing Project
The ANTD has been involved in the development of the IETF BGPSec standards. Development of these standards is both important and challenging. Border Gateway Protocol (BGP) configuration is difficult and error-prone, and global repercussions have occurred. Secure BGP will address these problems, but it will also add complexity and performance overhead. This effort is not obviously aligned with NIST’s unique niche in terms of framing problems and quantifying comparisons.
High-Assurance Domains Project
The IETF has defined several standards intended to solve problems such as SPAM and phishing. NIST is playing an important role by facilitating adoption of these standards. The ANTD is encouraging adoption of these standards within the government. This will help discover issues with these standards and will encourage the rest of the world to learn from NIST’s experience. Project staff have published in a top conference, Usenix LISA. This gives them an opportunity to directly impact industry, gain visibility, and improve recruiting opportunities.
Complex Systems Project
Complex systems are all around, and the ANTD has undertaken an ambitious effort to understand some of the emergent behaviors that pervade across a swath of complex systems that include social media and networks. The Network Resilience project, which includes the Complex Systems project, has collaborations with universities and a private company. There are two areas that the ANTD might revisit. First, simulation results need to be validated by real implementations in real systems. Concretely, several of the results on comparison of transmission control protocol (TCP) congestion control mechanisms, and their associated parameters, were done by simulation only, but they need to be strongly validated by actually implementing them in real networks like the Internet. Without this grounding, simulations are not convincing, regardless of the depth of the results from the simulator. (One organization that examines implementations in real networks is the Pantheon of Congestion Control.6) Second, ANTD presenters did not make clear the value of the external vendor who was being used to develop the simulator. Event-driven simulators (General Purpose Simulation System [GPSS]-based) are commonplace today, and existing open source simulators (e.g., ns-37) may scale beyond the sizes in use by the vendor. It is critically important that the project look deeply into making a wise choice between using outsourced simulators and building in-house simulators. Building one’s own simulator can give deep insight into the results, afford sufficient flexibility in repurposing the simulator for other aims, and afford NIST the opportunity to create an impact by standardizing its simulator for industrial and academic use. The choice needs to include consideration of whether the ITL can devote the appropriate level of resources.
Epidemics in networks can disclose the fundamental behavior of complex communication networks. The Susceptible-Infected-Susceptible (SIS) model has been initially applied to develop a distributed algorithm suppressing epidemic spread in a connected network. Among widely scattered research outcomes in this multidisciplinary research, the ultimate goal and targeting applications of this project need to be identified by the ANTD in more realistic networking structures and operating network protocols. This is necessary so that appropriate models and sufficient scale can be defined to obtain more meaningful engineering outcomes. It would be useful to insert proper epidemic or automata models into scenarios involving patching of threats, which tend to be binary. Evaluations with more realistic networks—for example, small world and power law—and even real network traces—for example, from the Stanford Network Analysis Platform (SNAP) repository—are essential to convince researchers that the idea works in practice.
Cloud Computing Project
Given the growing level of extramural activity in cloud computing, an expanding ANTD effort could increase NIST’s footprint and impact in the cloud computing area, including on the leading cloud providers. In particular, this effort needs to expand include development of benchmarks, data sets, and simulators for a broader focus area than Service Level Agreements (SLAs). The broadened focus area needs to include software offerings on public clouds (examples include Google BigQuery, AWS ElasticMapreduce, and Microsoft Azure; there are dozens of others); on new opportunistic modes of cloud computing (e.g., AWS Spot Instances and Google persistent instances); and on systems that run on these clouds (e.g., batch processing, real-time stream processing, and machine learning systems.). NIST is well positioned to do this work, and public clouds and data analytics systems are open resources with very little existing measurement and instrumentation studies. This is aligned with NIST’s mission, and it presents an exciting opportunity for NIST impact on industry and academia.
Software-Defined and Virtual Networks Project
The project on software-defined and virtual networks provides an excellent example of a well-articulated and well-contextualized research vision. The uMon project adheres to NIST’s mission statement of being a technology facilitator by creating a first step toward standardization. The development of network measurement technologies within an open source software-defined networking environment seems to be a very appropriate research direction for NIST. The project progress thus far is encouraging, with good publications and strong partners in the effort. It will be important for the group to develop and follow a detailed roadmap as it continues to pursue this research. Any resulting network monitoring and measurement tools need to be made open access for possible wide adoption. The future plan of the project to engage with open consortia and move toward standardization of the uMon project is commendable.
Information-Centric Networking Project
While the ANTD has been active in pursuing projects in the information-centric networking (ICN) area, the future importance of such architectures is not assured. ANTD’s role could beneficially be aligned with examining the fundamental networking problems that ICN is attempting to solve. The ANTD could also examine the efficacy of ICN as compared with traditional approaches. ITL needs to leverage current efforts to explore commercially promising technologies such as mobile edge computing.
Smart Grid Project
8 H. Gharavi and B. Hu, 2015, Scalable synchrophasors communication network design and implementation for real-time distributed generation grid, IEEE Transactions on Smart Grid 6(5):2539-2550.
9 H. Gharavi and B. Hu, 2017, Synchrophasor sensor networks for grid communication and protection, Proceedings of the IEEE 105(7):1408-1428.
The Smart Grid has been recognized as a key component of the nation’s energy strategic plan. NIST is one of the federal agencies with a statutory role in the Federal Smart Grid Task Force,10 led by the U.S. Department of Energy (DOE) Within the Smart Grid concept/architecture, the grid-to-end user interface is critical, and this has been recognized by the ANTD staff. This interface must provide phase matching between the end user’s electrical waveform and the main grid’s electrical waveform at the point of interface. It must protect the main grid from various types of damage that could be induced by connecting end users. It must participate in the distributed management of reactive power flow. It must protect end users from damage induced by transients or faults on the main grid. It must enable secure two-way communication between entities connected to the grid. This interface must provide the necessary observability and controllability to quickly identify problems and to quickly isolate problems that occur at the points where the end users connect to the main network. At the same time, this interface must not provide a means for adversaries to simultaneously disable or destroy large portions of the grid and its connected entities. This is, appropriately, a key focus for the ANTD, the ITL, and NIST. NIST participation in (or leadership of) the creation of a set of technical requirements (and possibly uniform national standards) for this critical interface would be beneficial. Another aspect of the emerging Smart Grid where the ANTD is making contributions, and in which the ANTD needs to play a major role, is in the design of the communication networks that will be critical for the maintenance, protection, and control of the grid-to-end user interfaces, and all of the other entities that comprise the Smart Grid.
A roadmap for what ANTD intends to accomplish in its Smart Grid project would be very helpful for evaluating the potential impact of this project.
Benchmarks, Simulator, and Data Sets
With respect to several of the research thrusts, including the Internet, clouds, IoT, wireless, and data analytics systems, the ANTD could have a much higher impact on industry and academia if it could provide new standardized benchmarks, simulators, and hosting/curation of data sets. In domains where developers have to choose among multiple systems or multiple protocols, it is useful to have standard benchmarks to help make this decision. In scenarios where new protocols were being developed (e.g., in IoT), standardized simulators can help validate the pros and cons of a protocol in a widely accepted way. Hosting of data sets is invaluable for validating new ideas, systems, and protocols using standard workload traces. Anonymization may be important; most current data sets are either maintained in academia or are occasionally released by companies. The PerFloc data set and competition, developed by the ANTD’s Indoor Localization project, is exemplary, and other ANTD projects need to consider hosting data set.
Overall, the ANTD’s future benchmarks and data sets in cloud computing and related emerging areas like data analytics systems could facilitate developers and startup companies choosing from among numerous open source systems and standards. NIST’s ability to convene stakeholders and assess pros and cons of various options makes this is an opportunity for real impact on both industry and academia.
RECOMMENDATION: The ITL should develop and publish benchmarks to be used for evaluating the performance of existing and proposed networks and network technologies in more areas and should develop simulators and make them available for researchers.
10 Further information is available at Department of Energy, “Federal Smart Grid Task Force,” https://www.energy.gov/oe/activities/technology-development/grid-modernization-and-smart-grid/federal-smartgrid-task-force.
Study of Real Systems
An important focus of research at NIST is measurement science for real systems. Today’s ecosystem of public clouds, Internet service providers (ISPs), and the Internet offer myriad opportunities for doing this. The ITL could profit from looking into studying data sets from ISPs, public clouds, and even private networks, perhaps improving instrumentation and measurement to facilitate data collection. In addition to using these for NIST research, the ANTD could help facilitate anonymization and other modifications to make these data sets into a form usable by industry and academia. NIST is positioned well to take advantage of its connections for this activity.
RECOMMENDATION: The ITL should work with Internet service providers, public clouds, and data centers to collect data sets needed for NIST researchers to perform evaluations of the performance of existing networking solutions. If possible, the ITL should make those data sets available to industry and academia.
Study of Internet Problems and Behaviors
One of ITL’s areas of expertise is the study of what problems need to be solved, rather than taking what is currently deployed and assuming that this was the right or only possible choice. Interdomain routing is an important area that could benefit from the ANTD’s expertise in framing problems. Although work on improving the specific protocol BGP is important, the ANTD need not assume that BGP was the only possible choice nor the right choice. An important area that has been largely ignored by the industry is studying what the “interdomain routing” problem is, outside the scope of a specific protocol. For instance, what policies are essential? Should some paths, even if physically there, be illegal or just discouraged? Should the source of the data be able to influence the path taken? What path-specific information might be secret?
RECOMMENDATION: In its role as a technology facilitator, the ITL should study Internet problems and behaviors, outside the assumptions inherent in deployed standards.
The ANTD is in a high-impact area of research and development (R&D). The size of the ANTD staff has shrunk since 2011, when it had 29 full-time staff. This can be largely attributed to transfer of 13 staff members from the wireless group to the NIST Communication Technology Laboratory (CTL) in 2014. The ANTD appears to have tactically addressed this via an increase in guest researchers, from 2 guests in 2014 to 24 currently.
ADEQUACY OF FACILITIES, EQUIPMENT, AND HUMAN RESOURCES
The division has an overall budget of $5 million and an addendum budget that adds $3 million for specified projects.
Networking and network-based technologies such as cloud computing and data analytics systems are a fast-growing market. As such, increased investment in the ANTD would bolster existing expertise and enable ANTD growth into new areas as the need and opportunity arises.
Considering the demonstrated expertise of the existing staff, the existing roadmap, and the track record, to date, of enlisting the participation of academia and industry, the Indoor Localization project has sufficient resources to accomplish its proposed goals.
DISSEMINATION OF OUTPUTS
Several ANTD projects focus on standardization and creation of ISO11/NIST/IETF12 standards. For example, the Cloud Computing initiative resulted in an ISO standard, and the BGPSec resulted in an IETF standard. ANTD staff in the High-Assurance Domains project are encouraging adoption of IETF standards within the government. ANTD’s cloud computing effort involves work on catalyzing standards for Service Level Agreements (SLAs) into clouds, including federated clouds. NIST’s involvement in the development of the ISO 19086 standard, which has since been adopted by Microsoft, is a commendable first step toward expansion of the cloud computing project in ANTD.
Other ANTD projects focus on the creation of data sets. For example, the Indoor Localization project resulted in the PerFloc data sets and competition.