Announcement and Call for Abstracts |
The Network Mapping and Measurement Conference (NMMC 2012) will be held March 19-20, 2012, at Arizona State University in Tempe, AZ (part of the Phoenix/Scottsdate metroplex). NMMC has previously been held at the Laboratory for Telecommunications Sciences (LTS) in College Park, Maryland (2008 and 2009), McGill University in Montreal, Quebec (2010), and the University of Wisconsin in Madison, Wisconsin (2011). It grew out of NetTomo workshops which were held in Adelaide, Australia (2007) and in Sunnyvale, California (2004-2006).
Network mapping and measurement plays a crucial role in the management, operation, and monitoring of network resources and infrastructure. Network structure and performance are not always directly accessible or measurable. Network mapping and measurement algorithms and systems enable efficient and intelligent inference and monitoring of large-scale complex systems. The focus of this conference is on mapping and measurement issues in networked systems (including probing and inference algorithms).
Authors are invited to submit abstracts describing original research or developments in theoretical and/or practical aspects of network mapping and measurement.
Topics of interest include (but are not limited to):
As well, we plan at least one session devoted to the Challenge Problems listed below.
- Network tomography
- Inference of link-level performance from end-to-end measurements
- Traffic matrix modeling and estimation
- Active and passive probing measurement and inference
- Social network analysis
- Inference of routing policies from path-level measurements
- Network anomaly detection
- Network and measurement visualization
- Bandwidth and capacity estimation
- Sparsity and adaptivity in network inference
- Network inference and monitoring in other domains (biology, chemistry, physics, power/water networks and other critical infrastructure)
- Network science
- Abstract deadline: February 1, 2012
- Registration deadline: March 1, 2012
- Conference dates: March 19-20, 2012
Abstracts are due by February 1, 2012. To submit an abstract for presentation, please e-mail a 1-2 page document summarizing the work you would like to submit to Violet Syrotiuk (firstname.lastname@example.org). Be sure to include your name and affiliation in the document.
Indicate whether the abstract addresses one of the challenge problems.
Registration and Local Arrangements
More information about conference registration and local arrangements will be posted on the conference website soon. There will be no registration fee to attend NMMC 2012.
Problem #1A: Developing a Control Theoretic Representation of Network Behavior
An elaborate theory of measurement and control has been developed over the past sixty years for linear electrical and mechanical systems, and its value has been proven by its broad and successful practical application. An important attribute of this theory is its ability to explain the behavior of complex systems and to provide simple mathematical tests which can determine the ability of a system operator to measure its performance and control its behavior. The last twenty years has seen the parallel growth of communications networks of ever growing complexity. The intent of this challenge problem is to examine the degree to which a theory of observability and controllability can be developed for these ever evolving and ever more complex communications networks. If so, then there would clearly be new opportunities for their efficient management using closed-loop control systems.
To develop this theory we desire to find suitable network-centric definitions for ----
While the terms used here are taken from linear systems control theory, and the concepts are roughly the same, there is certainly no assumption that the same mathematical constructs will apply. Thus the challenge is to develop an appropriate set of mathematical definitions which also have a reasonable intuitive meaning to engineers and practitioners.
- “Identification” – that is, the determination of the topology of a network,
- “Observability” – that is, the determination of traffic flows through the network, and
- “Controllability” -- the control of traffic flows through the network.
Problem 1B: Using Control Theoretic Concepts in the Limit of Restricted Measurements
In very few networks is it possible to make all the measurements that one would prefer, or even require, to have in order to assess the ability to identify, observe, and control a network. This portion of the challenge problem is to address just that. Specifically, the question is, how measurements, and of what type, are required to identify the topology and a networks, to determine its switching or routing behavior, and ultimate to control (manage) its flows.
In order to guide work on this challenge, it should be addressed within the following framework ----
An important implication of the last item is that it is quite possible that the network under examination is not “observable”, by whatever definition is developed for that in Part 1A. A key aspect of this challenge problem is to determine the effect of this nonobservability has on the other two of the attributes discussed above – namely identifiabiity and controllability.
- Activity measurements in the network will be made on the links, not the nodes.
- Timing accuracy of the activity measurements will be precise enough to judge simultaneity, but not to determine the sequence of links over which a call, packet or message passes.
- Transit times on the links are assumed to instantaneous (for now), thus removing propagation delay as a complicating issue.
- It should be assumed (for now) that no packets, message, or calls are lost, and that the probability of detection on a link is 100% and that there are no false alarms.
- But it should be assumed that it is not possible to measure activity on all links.
Note in passing that the question of identifiability, using a possibly insufficient number of link activity measurements, was the topic of the first few Network Measurement and Management Conferences (aka Network Tomography Workshops). Challenge Problems 1A and 1B expand on that question to frame a full theory of observability and controllability for networks.
Problem #2: Optimal methods for multistatic tracking of target reflectors
The availability of high speed signal process processing and sophisticated signal processing algorithms has made it possible to move beyond traditional radio ranging systems (“radars”), where a single transmitter and single receiver are co-located, and on to systems with multiple transmitters and multiple receivers, all of which are geographically dispersed. This challenge problem is to determine how to obtain the best performance from such a system. In order to constrain the problem a bit, we make the following assumptions ---
There are several important variations on this problem which should be considered.
- The objective is to locate and then track N >=1 potentially moving reflectors operating in field of view of the transmitters (“illuminators”) and the receivers.
- There are L, where L >=1, illuminators, whose static locations and technical characteristics are known, but which can’t be modified, either statically or in real time.
- There are M, when M >=1, receivers, whose locations are known, whose processing can be changed at will, and who can communicate freely with each other.
- The potential for moving the receivers, or, even the use of moving receivers.
- The constraints on performance imposed by the radio propagation behavior between each pair of illuminators and receivers. To constrain this a bit, consider two limiting cases –
- HF over-the-horizon radar – using linear FM to obtain good range resolution but propagating through time-varying, dispersive media, with the potential of needing to locatte and track reflectors over an area of tens of thousands of square km, and
- Wideband CDMA – using 5 MHz-wide direct-spread signals operating in the UHF band with very little atmospheric dispersion, with the potential of locating and tracking reflectors over an area of a few cells in a wireless system (nominally 10 square km).
- The advantages, if any, of using illuminators operating at substantially different frequencies.
The figure of merit for this challenge problem should be some function of the number of reflectors within the combined field of view which can be unambiguously detected and then tracked, and the spatial accuracy with which they can be tracked.
For more information please contact Violet Syrotiuk (email@example.com) or Charles Colbourn (firstname.lastname@example.org).