Publications

Audio diarization is the process of annotating an input audio channel with information that attributes (possibly overlapping) temporal regions of signal energy to their specific sources. These sources can include particular speakers, music, background noise sources, and other signal source/channel characteristics. Diarization has utility in making automatic transcripts more readable and in searching and indexing audio archives. In this paper we describe the systems developed by MITLL and used in DARPA EARS Rich Transcription Fall 2004 (RT-04F) speaker diarization evaluation. The primary system is based on a new proxy speaker model approach and the secondary system follows a more standard BIC based clustering approach. We present experiments analyzing performance of the systems and present a cross-cluster recombination approach that significantly improves performance. In addition, we also present results applying our system to a telephone speech, summed channel speaker detection task.

This paper reports results from two recent psycholinguistic experiments that measure the readability of four types of speech transcripts for the DARPA EARS Program. The two key questions in these experiments are (1) how much speech transcript cleanup aids readability and (2) how much the type of cleanup matters. We employ two variants of the four-part figure of merit to measure readability defined at the RT02 workshop and described in our Eurospeech 2003 paper [4] namely: accuracy of answers to comprehension questions, reaction-time for passage reading, reaction-time for question answering and a subjective rating of passage difficulty. The first protocol employs a question-answering task under time pressure. The second employs a self-paced line-by-line paradigm. Both protocols yield similar results: all three types of clean-up in the experiment improve readability 5-10%, but the self-paced reading protocol needs far fewer subjects for statistical significance.

The continuing progress of Moore's law has enabled the development of radar systems that simultaneously transmit and receive multiple coded waveforms from multiple phase centers and to process them in ways that have been unavailable in the past. The signals available for processing from these Multiple-Input Multiple-Output (MIMO) radar systems appear as spatial samples corresponding to the convolution of the transmit and receive aperture phase centers. The samples provide the ability to excite and measure the channel that consists of the transmit/receive propagation paths, the target and incidental scattering or clutter. These signals may be processed and combined to form an adaptive coherent transmit beam, or to search an extended area with high resolution in a single dwell. Adaptively combining the received data provides the effect of adaptively controlling the transmit beamshape and the spatial extent provides improved track-while-scan accuracy. This paper describes the theory behind the improved surveillance radar performance and illustrates this with measurements from experimental MIMO radars.

RCM (Robust Collaborative Multicast) is a communication service designed to support collaborative applications operating in dynamic, mission-critical environments. RCM implements a set of well-specified message ordering and reliability properties that balance two conflicting goals: a)providing low-latency, highly-available, reliable communication service, and b) guaranteeing global consistency in how different participants perceive their communication. Both of these goals are important for collaborative applications. In this paper, we describe RCM, its modular and flexible design, and a collection of simple, light-weight protocols that implement it. We also report on several experiments with an RCM prototype in a test-bed environment.

Five modern static analysis tools (ARCHER, BOON, PolySpace C Verifier, Splint, and UNO) were evaluated using source code examples containing 14 exploitable buffer overflow vulnerabilities found in various versions of Sendmail, BIND, and WU-FTPD. Each code example included a "BAD" case with and a "OK" case without buffer overflows. Buffer overflows varied and included stack, heap, bss and data buffers; access above and below buffer bounds; access using pointers, indices, and functions; and scope differences between buffer creation and use. Detection rates for the "BAD" examples were low except for PolySpace and Splint which had average detection rates of 87% and 57%, respectively. However, average false alarm rates were high and roughly 50% for these two tools. On patched programs these two tools produce one warning for every 12 to 46 lines of source code and neither tool accurately distinguished between vulnerable and patched code.

Coherent solid-state optical sources based on Nd:YAG/Cr4+:YAG passively Q-switched microchip lasers cover the spectral range from 5000 to 200 nm, producing multikilohertz pulse trains with pulse durations as short as 100 ps and peak powers up to 1 MW. The wavelength diversity is achieved through harmonic conversion, parametric conversion, Raman conversion, and microchip-laser-pumped miniature gain-switched lasers. In all cases, the optical heads have been packaged in a volume of less than 0.5 liters. These compact, robust devices have the proven capability to take what were complicated laser-based experiments out of the laboratory and into the field, enabling applications in diverse areas. The short pulses are useful for high-precision ranging using time-of-flight techniques, with applications in 3-dimensional imaging, target identification, and robotics. The short pulse durations and ideal mode properties are also useful for material characterization. The high peak powers can be focused to photoablate material, with applications in laserinduced breakdown spectroscopy and micromachining. Ultraviolet systems have been used to perform fluorescence spectroscopy for applications including environmental monitoring and the detection of biological aerosols. Systems based on passively Q-switched microchip lasers, like the lasers themselves, are small, robust, and potentially low cost, making them ideally suited for field applications.

The integration of Remotely Piloted Vehicles (RF'Vs) into civil airspace will require new methods of ensuring aircraft separation. This paper discusses issues affecting requirements for RPV traffic avoidance systems and for performing the safety evaluations that will be necessary to certify such systems. The paper outlines current ways in which traffic avoidance is assured depending on the type of airspace and type of traffic that is encountered. Alternative methods for RPVs to perform traffic avoidance are discussed, including the potential use of new see-and-avoid sensors or the Traffic Alert and Collision Avoidance System (TCAS). Finally, the paper outlines an established safety evaluation process that can be adapted to assure regulatory authorities that RPVs meet level of safety requirements.

In this paper, we summarize contemporary approaches to quantifying convective weather delay reduction benefits. We outline a program to develop a significantly improved capability that can be used to assess benefits of specific systems. This program may potentially accomplish weather impact normalization for studies of National Airspace System (NAS) performance in handling convective weather. Benefits quantification and NAS performance assessment have become very important topics for the aviation weather community. In an era of significant federal government and airline budget austerity for civil aviation investments, it is essential to quantitatively demonstrate delay reduction benefits of improved weather decision support systems. Major FAA initiatives stress the importance of quantitative system performance metrics that are related to aviation weather. For example, the new FAA Air Traffic Organization (ATO) and the FAA Flight Plan 2004-08 both have quantitative performance metrics that are closely related to reducing convective weather delays. The Flight Plan metrics include: "Improving the percentage of all flights arriving within 15 minutes of schedule at the 35 OEP airports by 7%, as measured from the FY2000-02 baseline, through FY08," and "Maintaining average en route travel times among the eight major metropolitan areas." The ATO metrics include the percentage of on time gate arrivals and the fraction of departures that are delayed greater than 40 minutes. However, these metrics currently do not account for the differences in convective weather severity and changes in the NAS. The dramatic increase in convective season delays in 2004 (Figure 1) due to a combination of severe weather, increases in overall demand, and specific airport issues has demonstrated that one needs to consider these other factors. Approaches to delay reduction quantification that were viewed as successful and valid several years ago are no longer considered to be adequate by either by the FAA investment analysis branch or by the Office of Management and Budget (OMB). The paper proceeds as follows. We first discuss at some length the mechanisms by which convective weather delay occurs in the NAS and highlight challenges in delay reduction assessment. We consider this to be very important since one needs to understand how the system operates if one is to design an effective, accurate performance assessment system. We then consider benefits quantification based on feedback from experienced users of a system. Feedback on "average" benefits from a system at the end of a test period was used to generate delay reduction estimates for the Integrated Terminal Weather System (ITWS) and the Weather and Radar Processor (WARP). This end-of-season interview approach was not viable in highly congested en route airspace. Hence, a new approach was developed for Corridor Integrated Weather System (CIWS) benefits assessment that uses real time observations of product usage during convective weather events coupled with in depth analysis of specific cases. Next, we discuss the problems that arise when one attempts to quantify delay reduction benefits by comparing flight delays before and after the Integrated Terminal Weather System (ITWS) system was deployed at Atlanta Hartsfield International Airport (ATL). This seemingly simple approach has proven very difficult in practice because the convective weather events in the different time periods are virtually never identical and because other aspects of the NAS may also have changed (e.g., user demand, fleet mix, and other systems that impact convective weather delays). It has become clear that one needs a quantitative model for the NAS that would permit adjustment of measured delay data to account at least for the differences in convective weather and changes in user demand (i.e., flight scheduling). The paper concludes with recommendations for measuring near term benefits of various classes of convective weather decision support systems.

Major airlines and FAA Traffic Flow Managers alike would prefer to plan their flight routes around convective weather and thereby avoid the tactical maneuvering that results when unforecasted thunderstorms occur. Strategic planning takes place daily and 2-6 hr forecasts are utilized, but these early plans remain unaltered in only the most predictable of convective weather scenarios. More typically, the ATC System Command Center and the Air Route Traffic Control Centers together with airline dispatchers will help flights to utilize jet routes that remain available within regions of convection, or facilitate major reroutes around convection, according to the available "playbook" routes. For this tactical routing in the presence of convective weather to work, both a precise and timely shared picture of current weather is required as well as an accurate, reliable short term (0-2 hr) forecast. This is crucial to containing the system-wide and airport-specific delays that are so prevalent in the summer months (Figure 1), especially as traffic demands approach full capacity at the pacing airports. This paper describes the Tactical 0-2 hr Convective Weather Forecast (CWF) algorithm developed by the MIT Lincoln Laboratory for the FAA, principally sponsored by the Aviation Weather Research Program (AWRP). This CWF technology is currently being utilized in both the Integrated Terminal Weather System (ITWS; Wolfson et al., 2004) and the Corridor Integrated Weather System (CIWS; Evans et al., 2004) proof-of-concept demonstrations. Some of this technology is also being utilized in the National Convective Weather Forecast from the Aviation Weather Center (Megenhardt, 2004), the NCAR Autonowcaster (Saxen et al., 2004), and in various private-vendor forecast systems.

Wake vortices are a by-product of lift generated by aircraft. The vortices from the wings and other lift surfaces such as flaps spin off and trail behind an aircraft (see Figure 1). These vortices can be a hazard to other aircraft, especially lighter aircraft that are following at low altitude. For this reason, numerous air traffic control standards require increased aircraft separation when wake vortex avoidance is a concern. These separation standards provide the required safety: there has never been a fatal accident in the U.S. due to wake vortices when wake vortex separations were provided by air traffic controllers. Wake vortex behavior is strongly dependent on atmospheric conditions, giving rise to the possibility that wake behavior can be predicted with enough precision to allow reduced use of wake vortex avoidance separations. Because vortices can not be seen, and their location and strength are not currently known or predicted, separation standards and air traffic procedures are designed to account for the worst case wake behavior. Because of this, the imposed aircraft separations are larger than required much of the time, reducing terminal capacity and causing increased traffic delay. If procedures or technologies can be developed to reduce the use of wake avoidance separations, terminal area delay reduction may be achieved. A prototype wind dependent wake separation system is operating in Frankfurt, Germany for arrivals into closely spaced parallel runways. The system uses wind prediction at the surface to determine when separation for wake vortex avoidance must be used and when the extra separation does not need to be used [Konopka, 2001][Frech, et al., 2002]. This led the FAA to ask the question: does the wind prediction algorithm used in Frankfurt, or perhaps another algorithm, have sufficient performance to consider it for possible use in the US for a closely spaced parallel runway departure system? This paper reports on a research effort to answer that question. This is part of a larger FAA and NASA research effort [Lang et al., 2003].