Towards mission-driven summarization for tactical networks

Lee, Jongdeog

Permalink

https://hdl.handle.net/2142/106303 Copy

Description

Title

Towards mission-driven summarization for tactical networks

Author(s)

Lee, Jongdeog

Issue Date

2019-08-05

Director of Research (if dissertation) or Advisor (if thesis)

Abdelzaher, Tarek

Doctoral Committee Chair(s)

Abdelzaher, Tarek

Committee Member(s)

• Nahrstedt, Klara
• Kravets, Robin Hilary
• Govindan, Ramesh

Department of Study

Computer Science

Discipline

Computer Science

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Tactical Networks
• Network Resource Managements
• Internet of Things

Abstract

In October 2017, the U.S. Army established six army modernization priorities to build a more lethal force. Considering the increasing importance of information superiority, it is not surprising that the Army Network has been selected as one of the six modernization priorities. One of the main reasons behind this decision is that, in the future, battlespace will become more contested by an adversary's sophisticated attacks on network infrastructure. Indeed, modern jamming threats are significantly different than in the 1980s, and the military is looking for a solution to support communication in this contested environment. There are two directions to address the problem. The first approach would be to make the network systems more robust and reliable, such as by adding in routing and security solutions. Another approach is more efficient network resource management that includes minimizing the transmitted data volume to save network bandwidth consumption. There are two representative techniques in this case: compression and summarization. Compression represents the original data with fewer bytes and is widely used for multi-media data. Summarization, or sampling, selects only a subset of the data from the whole set. In the current thesis, we propose a new mission-driven summarization layer that understands the application needs and optimizes the network resources using the sampling technique. This work is similar to the existing summarization efforts but is distinguished by three key factors. First, its nature is mission driven. This layer offers declarative application program interfaces to the applications to analyze the application needs so that the layer can flexibly manages network resources on behalf of applications. This is different than the traditional approach, such as time-driven and event-driven approaches where applications need to micromanage network resources. The second difference is generality. Rather than devising the optimized algorithm for a particular application as in state-of-the-art researches, this work is designed to provide summarization as a service. Hence, we focus on an easy-to-port feature across different application contexts. This generality calls for simplicity by observing a fundamental trade-off between generality and complexity. That is, more sophisticated solutions tend to have more assumptions and are more difficult to apply to the wider application domain. The proposed solution is designed to be simple and widely applicable yet effective.

Graduation Semester

2019-12

Type of Resource

text

Permalink

http://hdl.handle.net/2142/106303

Copyright and License Information

Copyright 2019 Jongdeog Lee

Owning Collections