My research and development interests span many aspects of systems -- I like designing and building real systems that do cool stuff! As a masters student at Utah, I focused on mobile, wireless sensor networks and applications and tools for embedded systems. As a staff researcher, I get to explore my other interests, including wired and wireless network protocols, network management, distributed systems of many sorts, virtualization, etc.

• A3/VMI
For the past year (as of October 2012), I've been working on a multifaceted DARPA-funded project (in the CRASH program); our prototype is the Advanced Adaptive Application Environment. My role in this project is to create a system for virtual machine introspection (VMI). We've written a pretty nice debugging library that supports multiple backends (a Linux-guest-aware Xen VM backend, and a Linux ptrace implementation for userspace stuff), and supports symbol information from DWARF and ELF. You can do regular debugger stuff, plus some extra stuff :). It's still evolving... see our git repo. The primary idea of this part of the A3 project is to tie our VMI system to our Xen-based deterministic replay system (implemented by my colleague, Anton Burtsev). By applying VMI to a deterministic replay of recording of a VM's execution, you can do all kinds of powerful performance and security analyses (that may be impossible at runtime). With this system, we can both debug and prevent problems at runtime, or for future runtimes -- hence our relevance to CRASH.

• NCR (National Cyber Range)
I did a bunch of work helping to rethink and redesign testbed software for better extensibility and scalability (of course, in the Emulab context). I helped implement those ideas in the Utah Emulab, and within the prototype system Utah and John's Hopkins Applied Physics Lab built. The primary challenges the NCR imposed on tested design involved secure isolation and scale of experimentation; the first is tricky to do while maintaining full automation; the latter is a small amount of programming :). Anyway, it was interesting.

• Emulab
I do design and hacking (and a little sysadmin stuff) within the Emulab software system, and the Utah Emulab testbed. It can run the gamut from kernel hacking through refactoring to feature addition... and bugfixing (although never within my code, of course!).

• ProtoGENI
(not so) Recently, I've been working on wide area and virtual node software support for ProtoGENI, a new testbed that will be a real prototype of the proposed GENI facility.

(ancient) (really) Old stuff:

• Mobile Emulab: A Mobile, Wireless Sensor Network Testbed
Although network simulation continues to be the dominant method of experimental evaluation in wireless networking, much research has established the failure of simulator models to adequately express wireless signal propagation. These shortcomings can lead to incomplete evaluation of routing protocols and applications. When mobility is added to the research, real evaluation becomes still more difficult due to the difficulty of creating and controlling mobile nodes in a real environment.

In response to these problems, and with the hope of encouraging real mobile wireless research, we created a remotely-accessible mobile wireless network testbed. The testbed consists of several robots, each with a small computer and small wireless devices called motes, manueverable in an area surrounded by fixed motes. Through a variety of interfaces, remote researchers can control these robots interactively over the Web. An overhead tracking system localizes the robots to within 1 cm, providing repeatable positioning and valuable knowledge to researchers studying how signal propagation affects their experiments.

• mobiQ: Improving Quality of WSNs Through Mobility
mobiQ is an WSN application designed to improve quality of a sensor network using mobile base stations as sinks. In a sensor network, quality consists of primarily network lifetime and quality of sensed data. End of life of a deployed sensor network occurs when enough nodes have failed so that useful sensing is no longer possible. With some routing protocols, some nodes may end up routing significantly more packets than others, and thus fail sooner than others, due to battery expiration. Furthermore, if nodes die, fail due to hardware, or expire due to environmental effects, the network's sensing ability is decreased.

mobiQ was designed to use mobile sinks to equalize the number of messages sent by each node, investigate event hotspots, and detect fixed node failures and obtain event data on their behalf when possible. Each mobile sink is restricted to its own sector of the network, but fixed nodes may connect to whichever sink is "closest" (in terms of routing). Mobile sinks autonomously move based on the output of several estimators, which use a combination of current fixed node state and location data.

• SNAP-M
I developed SNAP-M to provide a generic interface for communicating with WSNs from a PC. SNAP-M provides a GUI, plugin-based interface through which users interface with WSNs. Several plugins are already supplied, and should be generally useful. They provide support for connecting to Emulab motes and TinyOS SerialForwarder sources, monitoring incoming packets, sending packets and scripting sends, and managing mobility in Mobile Emulab. A set of libraries parses user-supplied Java Active Message wrappers (generated by mig from C structures) and helps users easily construct and view packets by transforming message fields into human-readable formats instead of raw bytes. Users may add their own Java plugins for any specific application control required.

• David Johnson. Design and Implementation of a Mobile Wireless Sensor Network Testbed. Masters Thesis, University of Utah, May 2010.
• David Johnson, Daniel Gebhardt, Jay Lepreau. Towards a High Quality Path-oriented Network Measurement and Storage System. In Proc. of the Ninth Passive and Active Measurement Conference (PAM 2008), Cleveland, OH, April 2008.
• D. Johnson, T. Stack, R. Fish, D. Flickinger, L. Stoller, R. Ricci, J. Lepreau. Mobile Emulab: A Robotic Wireless and Sensor Network Testbed. In Proc. of IEEE INFOCOM 2006, Barcelona, Spain, April 2006.
• D. Johnson, D. Flickinger, T. Stack, R. Ricci, L. Stoller, R. Fish, K. Webb, M. Minor, J. Lepreau. Poster Abstract: Robot Couriers: Precise Mobility in a Wireless Network Testbed (also Demo Abstract: Emulab's Wireless Sensor Net Testbed: True Mobility, Location Precision, and Remote Access). In Proc. of ACM SenSys 2005. San Diego, CA, Nov 2-4 2005.
• A. Tripathi, M. Koka, S. Karanth, I. Osipkov, H. Talkad, T. Ahmed, D. Johnson. Robustness and Security in a Mobile-Agent based Network Monitoring System. University of Minnesota Tech Report, May 2004.

• D. Johnson, T. Stack, R. Fish, D. Flickinger, L. Stoller, R. Ricci, J. Lepreau. Mobile Emulab: A Robotic Wireless and Sensor Network Testbed. IEEE INFOCOM 2006, Barcelona, April 2006. [ ppt, 6-up pdf ]
• Mobile Emulab Testbed Demo. TinyOS Tech Exchange III, Stanford University. Feb 2006.
• Mobile Emulab Testbed Demo/Poster. ACM SenSys 2005. San Diego, CA, Nov 2-4 2005.
• Mobile Emulab Testbed Demo. IEEE SECON 2005. Santa Clara, CA, Sep 2005.
• Mobile Emulab Poster Presentation. ACM MobiSys, Seattle, WA, June 6-8 2005.

I like hacking, listening to psychedelic rock (and humming and whistling, according to colleagues!), and reading (especially anthropology novels). I also enjoy exploring remote areas, especially ghost towns, when I have time.

This summer (2009), I took up road biking in the Wasatch mountains. At first I focused on time trialing my way up single canyons, but that pretty much ended (for now!) after the Snowbird Hillclimb. Now I'm trying longer rides with more elevation to build up for next year. You can see some of the GPS data and elevation plots here. Now (2012) I've both improved and backslidden... but the mountain roads still call me, and I respond :).

I've taken lots of pictures around Utah and on conference trips; view the gallery.