FBO_18R0015_JA..pdf

Justification and Approval (J&A) for Other Than Full and Open Competition

Choose the funding level for this J&A Document:

< $700K >$700K and < $13.5M > $13.5M and < $93M > $93M

Contracting Activity: 374th CONS/LGCZ

Purchase Request / Local ID Number: W90GQZ-83340014

Towards designing complex networks resilient to stealthy attack and cascading failure Program/ Project(andPE, if applicable):

Program Type (PEO or Other Contracting): Other contracting

10 USC 2304 (c) (1) as implemented by FAR 6.302‐1, as "Only One Responsible Source.", required by FAR 6.303‐2(b)(4).

Estimated Contract Cost (including options): $ J&AType: Class lndividual

** The text in the signature blocks below is editable, including the title.

** To removea row, click the "X" next tothe signature block. CAUTION: Once removed, it can only be restored by downloading the BLANK FORM again.

X

APPROVAL (AFFARS 5306.304(a) ) ** The text in the signature blocks below is editable, including the title.

** To removea row, click the "X" next tothe signature block. CAUTION: Once removed, it can only be restored by downloading the BLANK FORM again.

Updated: 15 Mar 16 FOR OFFICIAL USE ONLY NOT APPLICABLE AFTER AWARD Page 1 of 6

NOTE: If a Justification and Approval was approved for the preceding acquisition, a copy of the approved J&A for the predecessor action must be included in the staff package for approval of the instant J&A. This applies to J&A staff packages that are submitted for review and approval ata level abovethecontracting officer. The predecessor J&Awill be used asa reference document bythe approving official.

COORDINATION (AFFARS 5306.304(a) )

Date Director, US Army lTC-PAC

Kenneth Evensen

AMSRD-AP / 262-9301

Signature

EVENSEN.KENNETH.C

Digitally signed by

EVENSEN.KENNETH.CHARLES.1086

HARLES.1086146807 146807

Date: 2018.04.11 15:35:40 +09'00'

Date Contracting Officer Thomas L. Reese

374 CONS/ PKSA / 225-7545

Signature

REESE.THOMAS.L Digitally signed by

REESE.THOMAS.LOUIS.1066817799

OUIS.1066817799 Date: 2018.06.28 08:37:45 +09'00'

Date

29 June 2018

Base Support Flight Commander Peter G. Forde, Capt, USAF

374 CONS/PKB / 225-9241

Signature

FORDE.PETER.GER Digitally signed by FORDE.PETER.GERALD.138530414

ALD.1385304141 1

Date: 2018.06.29 13:19:01 +09'00'

I. ContractingActivity.

Justification for Other than Full and Open Competition: U.S. Army International Technology Center Pacific (ITC-PAC) is the requiring activity. 374th CONS/LGCZ is the contracting activity. Document number W90GQZ-83340014for Research and Development Project "Towards designing complex networks resilient to stealthy attack and cascading failure".

II. Nature and/or description of the action being processed.

This determination provides the rationale for contracting by other than full and open competition for research and development project "Towards designing complex networks resilient to stealthy attack and cascading failure". This action is to be a new, firm-fixed-price contract with The University of Melbourne, Australia.

III. Description of supplies/services required to meet agency needs.

This requirement is to provide financing in support of the research project titled "Towards designing complex networks resilient to stealthy attack and cascading failure" led by

This project involves a collaboration between researchers from two

This is an ambitious program of research that will bring together researchers from communities which often operate independently: complex networks, complex systems, machine learning and data mining, and operations research. The team offers a unique and diverse combination of experience in the development of algorithms with proven efficacy in the analysis of real-world data sets in areas spanning complex materials and systems, road networks, telecommunications, cybersecurity, online social networks, mine routing, and biological regulatory networks.

This project builds on their combined previous work in three main areas: a) anomaly detection and explanation, b) adversarial learning in the presence of active adversaries, c) characterization and early prediction of vulnerabilities, bottlenecks and failure in node/link-capacitated networks.

For challenge 1.) a robust network is a network that maintains a certain degree of functionality after a limited number of failures or interruptions to links and nodes. Measuring robustness therefore entails first modeling network failure. In turn, network failure is defined in terms of network capacity, which is the degree to which a network can fulfill its function, even after the failure of individual nodes and links. Their aim in this challenge is to model network failure under a general framework that incorporates functionality, security and resilience. They will develop mathematical functions for calculating network capacity in terms of given properties of the nodes and links. They also aim to find a mathematical relationship between the capacity of network bottlenecks (vulnerable areas of the network) and the capacity of the network after the failure of a bottleneck. This will generalize the well-known max-flow/min-cut result which has been used with great success for finding bottlenecks under simpler definitions of robustness.

For challenge 2.) They will build on their previous work for evolving network anomaly detection that has considered offline scenarios [1], their previous work on algorithms for anomaly explanation, and their previous work on adversarial learning [2] for learning in the presence of adversarial corruption. The program of research will involve formulating an appropriate measure to score how anomalous a given network snapshot is, formulating an algorithm that explains why the snapshot is anomalous and ensuring the approach is suitable for deployment in real time as the network evolves. The robustness of the anomaly detection approach will be tested and enhanced by the use of adversarial learning methods applied to time-evolving data at varying scales.

For challenge 3.) They will build on their previous work in time series prediction and link prediction to develop methods for predicting properties of a network. They will investigate the extent to which algorithms based on deep learning (recurrent neural networks) are effective for early prediction of failure and whether there are signals in the evolving network which are correlated with cascading failure. The will identify and characterize endogenous mechanisms that govern the co-evolution of structural and functional connectivities in the lead up to global failure, and distinguish aspects or types of mechanisms that either promote or prohibit cascading failure. They will also improve their past methods for early prediction of cascading failure dynamics in networks by extending their algorithms to handle subtle transitions in nonstationary time series data.

The research proposed is complex with numerous unknowns that need to be determined by experiment.

Duration of this research and development is three years, composed of one year for the Basic Period and two 1-year for Option Periods.

The main tasks of this project include the following milestones.

Basic Period:

Milestone 1: Paper 1 on network failure (Various data sets on real capacitated networks)

Milestone 2: Paper 2 on network attack (Data from LANL[1])

Milestone 3: Master of Information Technology minor thesis student 1

Milestone 4: Master of Information Technology minor thesis student 2

Milestone 5: PhD student

[1] https://csr.lanl.gov/data/cyber1/ accessed Nov 21, 2017

Option 1 Period:

Milestone 1: Paper 3 on network failure

Milestone 2: Paper 4 on network attack

Milestone 3: Master of Information Technology minor thesis student 3

Milestone 4: Master of Information Technology minor thesis student 4

Milestone 5: PhD student h

Option 2 Period:

Milestone 1: Paper 5 on network failure

Milestone 2: Paper 6 on network attack

Milestone 3: Master of Information Technology minor thesis student 5

Milestone 4: Master of Information Technology minor thesis student 6

Milestone 5: PhD student

This acquisition will be a firm-fixed price purchase order in the amount for the basic period, for the option 1 period, and for the option 2 period, in the total amount of

IV. Statutoryauthority permitting Other than Full and Open Competition.

10 USC 2304 (c) (1) as implemented by FAR 6.302-1, as "Only One Responsible Source."

V. Demonstration that the contractor's unique qualifications or nature of the acquisition requires the use of the authority cited above (applicability of authority).

In evaluating the content of the Research and Development Program it has been determined that the research and development project will focus specifically on attainable means by which investigators from the Pacific Rim and the U.S. military research institutions may maximize collaborative effort in addressing the scientific challenges inherent in the topic.

Our endorsement and support for this project will help to attract attention to ITC-PAC's presence (and for our outreach objectives) as well as to demonstrate the interest.

ITC-PAC Director of Robotics & Autonomy, Dr. Robert Kania, found this project from a lead given to him by The U.S. Army Engineer Research and Development Center (ERDC) researchers. He searched for all sources that they identified as technically capable and promising. He found that have exceptional expertise in the field of robust static networks.he important max-flow/min-cut theorem has led to highly efficient algorithms for measuring the resilience of a network and for finding vulnerable regions (bottlenecks). To date, there are no analogous results for evolving networks. By producing a new modeling methodology for robustness in complex evolving networks, this project will generate a set of mathematical tools for analyzing and measuring resilience across a range of network applications. The output will be efficient and scalable algorithms for finding bottlenecks in large complex networks. This is a first crucial step in designing and managing networks that are resilient to the failure of nodes and links.

VI. Description of efforts made to ensure that offers are solicited from as many potential sources as practicable.

The ITC-PAC Director reviewed this research and development proposal and requests a contract with the primary/lead organization. ITC-PAC found the primary/lead organization of this technology is the Technical literature was reviewed and no other sources were identified. Further, consolation with experts at ERDC also concurred that the expertise and knowledge of the proposed contractor is uniquely suited for this project.

ITC-PAC Director of Robotics & Autonomy identified this project while he was searching areas of his expertise for potential sources of basic research. have exceptional expertise in the field of robust static networks. The important max-flow/min-cut theorem has led to highly efficient algorithms for measuring the resilience of a network and for finding vulnerable regions (bottlenecks). To date, there are no analogous results for evolving networks. By producing a new modeling methodology for robustness in complex evolving networks, this project will generate a set of mathematical tools for analyzing and measuring resilience across a range of network applications. The output will be efficient and scalable algorithms for finding bottlenecks in large complex networks. This is a first crucial step in designing and managing networks that are resilient to the failure of nodes and links. This research is the only source directly relevant and related to the current efforts being worked by ERDC as identified by their Subject Matter Expert (SME) cascading failure and attack is directly applicable to the mission. The main objective of this work is to develop efficient methods and scalable algorithms, oriented to implementation on high-performing computing platforms. This research has the potential to be directly beneficial to the . Nationwide, the geographically dispersed HPC networks. Military services and associates gain secure, robust access to supercomputers to collaborate with other HPC users in the defense community throughout the

The ITC-PAC Director reviewed this research and development proposal and requests a contract with the primary/lead organization. ITC-PAC found the primary/lead organization of this technology is the Technical literature was reviewed and no other sources were identified. Further, consolation with experts at ERDC also concurred that the expertise and knowledge of the proposed contractor is uniquely suited for this project.

VII. Determination by the Contracting Officer that the anticipated cost to the Government will be fair and reasonable.

This funding is for research and development project titled "Towards designing complex networks resilient to stealthy attack and cascading failure." Because this project will feature the latest scientific innovation in this field studied on unique lab equipment, there is no parallel value to be found in the commercial sector. ITC-PAC certifies the price of amount of for the basic period, the option 1 period, and for the option 2 period, in the total amount of . This certification is based on the following facts:

for

The funds will be used to support research and development efforts at the with collaborate with U.S. Army Engineer Research and Development Center (ERDC). The cost estimates are as follows:

Base Period:

- Equip. & Facility:

- Researcher Labor:

- Travel cost:

- University Overhead:

- Total:

Option 1 Period:

an enterprise network, against

Option 2 Period:

- Equip. & Facility:

- Researcher Labor:

- Travel cost:

- University Overhead:

- Total:

The above-mentioned costs are based on the following information. All costs are provided by the PI.

Personnel costs are consistent with typical Australian salary ranges posted on the salary-scale websites such as www.glassdoor.com. Equipment are priced within typical prices for regional suppliers. All costs herein are considered to be fair and reasonable.

• Equipment & Facilities are for:

Standard issue University Laptops [Details attachment in 1 and 2: Budget_Support_Atch1_currency.pdf, Budget_Support_Atch2_Computer.pdf] Total expenses Equipment & Facilities =

• Post-Doctoral Research Assistant Labor are for:

Post-doctoral scientist (Level B), accounting for AUD-USD conversion (inclusive of standard (ARC) Australian on-costs (x1.30)) [Details provided in attachment 3: Budget_Support_Atch3_Salary.pdf, ] Total expenses Post-Doctoral Research Assistant Labor =

• Travel costs are for:

PI and Post Doc will travel twice to the States for a 2 day visit to ERDC.

[Details provided in attachments 1, 5, 6, and 7:Budget_Support_Atch4_Flight.pdf , Budget_Support_Atch5_Hotel.pdf, Budget_Support_Atch6_Perdiem.pdf, and Budget_Support_Atch 7_RentalCar.pdf] Total cost for travel =

• University Overhead Costs are for:

The charges Indirect costs on salary components of x1.1 per standard discounted rates agreed for Defense contracts.

Total expenses for University Overhead are = These expenses for overhead are fair and reasonable and in-line with what other universities in Australia charge.

VIII. Description of the market research conducted and the results, or a statement of the reasons market research was not conducted.

ITC-PAC Director of Robotics & Autonomy identified this project while he was searching areas of his expertise for potential sources of basic research. have exceptional expertise in the field of robust static networks. The important max-flow/min-cut theorem has led to highly efficient algorithms for measuring the resilience of a network and for finding vulnerable regions (bottlenecks). To date, there are no analogous results for evolving networks. By producing a new modeling methodology for robustness in complex evolving networks, this project will generate a set of mathematical tools for analyzing and measuring resilience across a range of network applications. The output will be efficient and scalable algorithms for finding bottlenecks in large complex networks. This is a first crucial step in designing and managing networks that are resilient to the failure of nodes and links. This research is the only source directly relevant and related to the current efforts being worked by ERDC as identified by their Subject Matter Expert (SME) cascading failure and attack is directly applicable to the mission. The main objective of this work is to develop efficient methods and scalable algorithms, oriented to implementation on high-performing computing platforms. This research has the potential to be directly beneficial to Nationwide, Military services and associates gain secure, robust access to supercomputers to collaborate with other HPC users in the defense community throughout.

IX. Any other facts supporting the use of Other Than Full and Open Competition.

N/A

X. List of sources, if any, that expressed interest in the acquisition.

XI. A statement of the actions, if any, the agency may take to remove or overcome any barriers to competition before making subsequent acquisitions for the supplies or services required.

N/A

XII. Certificationbythe Contracting Officer.

As evidenced bymy signature above, I have determined this document tobeboth accurate and complete to the best of my knowledge and belief.

XIII. Certification by the technical/requirements personnel.

Asevidenced bymy(our) signature(s) above, I(we) certifythatanysupportingdatacontained herein, which is my (our) responsibility, is both accurate and complete.

The proposed contractor is:

, an enterprise network, against