Collaborators: Scientific Systems Company, Inc. (SSCI)

The overall objective of this research effort is to develop the SELENE network for specific usage in the Human Lunar Outpost. We are investigating a hybrid scheme combining Mobile Ad-Hoc Networks (MANETs) and Disruption Tolerant Networks (DTNs) for networking EVA radios in the Human Lunar Outpost.

Collaborators: Agile Communications, SPARTA, Shared Spectrum Company (SSC), PARC, UC Santa Cruz, U of Pennsylvania, Virginia Tech

The PIRANA project is developing scalable, adaptive, ad-hoc networks which exploit very inexpensive, yet flexible software radios. Innovative aspects of the program include the incorporation of DTN technology for disconnected operation, dynamic spectrum access for finding and exploiting any available frequencies rather than using pre-allocated frequencies, exploitation of multichannel MIMO, and the use of policy and reasoning techniques to make the right tradeoffs and adaptations for the particular mission. This program will require new ideas and innovations to combine previously isolated technologies in new and synergistic ways. We are working towards a 40 radio demonstration in Dec 2008, a 100 radio demonstration in Dec 2009, and a 1000 radio demonstration in Dec 2010.

Collaborators: SUNY-Stony Brook, Lehigh University, SPARTA, Michael Demmer, Terrance Swift

The SPINDLE II project is developing technologies that enable access to information when stable end-to-end paths do not exist and network infrastructure access cannot be assured. DTN technology makes use of persistence within network nodes, along with the opportunistic use of mobility, to overcome disruptions to connectivity.

Collaborators: MIT, UCLA, University of Kansas, and Blossom Research

The ADROIT project is building an open-source software-defined data radio, intended to be controlled by cognitive applications. The goal is to create a system that enables teams of radios, where each radio has its own cognitive controls and the ability to collaborate with other radios, to create cognitive radio teams that dynamically adapt in real-time to environmental conditions.

My dissertation investigated transport layer techniques that improve end-to-end fault tolerance and throughput. Often, access links (for both clients and servers) are a single point of failure for end-to-end reachability. Routing protocols (in particular, BGP) may also take a significant amount of time (often tens of minutes) to converge on a new route when a link failure is detected. Multihoming support at the transport layer addresses both types of failures by allowing a transport layer session to bind multiple IP addresses at each endpoint. This feature provides both endpoints with multiple paths with which to communicate, and thus the ability to failover to an alternate path when a path failure occurs. I investigated multihoming retransmission policies and failover thresholds with the Stream Control Transmission Protocol (SCTP).

I ported much of NETCICATS to a web environment. I developed a Netscape plugin that would fetch network-conscious images using our lab's Universal Transport Library (UTL) and display the images in the browser. UTL allows an application to use a common API for a variety of experimental transport protocols developed in our lab.

Using ns-2, I evaluated SCTP's congestion control algorithms for protocol correctness (with Sack TCP as a guide), and my tests concluded that the algorithms operated properly. I also studied PR-SCTP for correctness and efficiency, and found a few problems with the specification. The first problem was that the congestion window was being incorrectly credited for data that was abandoned. The second problem was that a single loss event could generate many Forward TSNs, which adds overhead and possibly contributes to congestion. The third problem was that the Forward TSNs did not include stream information, and thus often caused ambiguity at the receiver. This ambiguity restricted the receiver from actually abandoning data and caused head-of-line blocking between streams defeating the purpose of SCTP's multistreaming feature. I worked with the PR-SCTP authors to correct the specification.