Publications

Refine Results

(Filters Applied) Clear All

Tagged As

GraphChallenge.org sparse deep neural network performance [e-print]

September 22, 2020
  |
Conference Paper
Author:
Jeremy Kepner
…
Published in:
2020 IEEE High Performance Computing Conf., HPEC, 22-24 September 2020.
Topic:
analysis
R&D area:
R&D group:

Summary

The MIT/IEEE/Amazon GraphChallenge.org encourages community approaches to developing new solutions for analyzing graphs and sparse data. Sparse AI analytics present unique scalability difficulties. The Sparse Deep Neural Network (DNN) Challenge draws upon prior challenges from machine learning, high performance computing, and visual analytics to create a challenge that is reflective of emerging sparse AI systems. The sparse DNN challenge is based on a mathematically well-defined DNN inference computation and can be implemented in any programming environment. In 2019 several sparse DNN challenge submissions were received from a wide range of authors and organizations. This paper presents a performance analysis of the best performers of these submissions. These submissions show that their state-of-the-art sparse DNN execution time, TDNN, is a strong function of the number of DNN operations performed, Nop. The sparse DNN challenge provides a clear picture of current sparse DNN systems and underscores the need for new innovations to achieve high performance on very large sparse DNNs.
READ LESS

Summary

The MIT/IEEE/Amazon GraphChallenge.org encourages community approaches to developing new solutions for analyzing graphs and sparse data. Sparse AI analytics present unique scalability difficulties. The Sparse Deep Neural Network (DNN) Challenge draws upon prior challenges from machine learning, high performance computing, and visual analytics to create a challenge that is reflective...

READ MORE
GraphChallenge.org sparse deep neural network performance [e-print]

Multifunction phased array radar (MPAR) spectral usage analysis

July 24, 2012
  |
Project Report
Author:
John Y. Cho
…
Published in:
MIT Lincoln Laboratory Report ATC-395
Topic:
radar
R&D area:
R&D group:

Summary

This report addressed two technical risks associated with replacing current air traffic and weather surveillance radars with a single type of multifunction phased array radar (MPAR). The first risk is whether radio spectrum usage would increase with the MPAR network and whether the allocated band will have enough spectral space. This question is addressed in two steps. First, single-radar spectrum usage is estimated based on certain assumptions regarding the radar design. Second, locations based on a previous radar placement study are used together with a terrain-dependent propagation model to compute the number of frequency channels needed at each site. We conclude that the overall spectrum usage is likely to increase with MPAR, but that the targeted window in S band will be able to accommodate the occupancy at all sites. The second risk is whether self-interference will limit the ability of the MPAR to operate asynchronously and adaptively on different antenna faces. This question is addressed by employing a simple bistatic ground clutter model to characterize the interference between adjacent faces. We conclude that some interference is unavoidable, but it would likely only occur during times when a transmit beam is at its maximum off-broadside angle (~2% of the time).
READ LESS

Summary

This report addressed two technical risks associated with replacing current air traffic and weather surveillance radars with a single type of multifunction phased array radar (MPAR). The first risk is whether radio spectrum usage would increase with the MPAR network and whether the allocated band will have enough spectral space...

READ MORE
Multifunction phased array radar (MPAR) spectral usage analysis

An automated method for low level wind shear alert system (LLWAS) data quality analysis

May 26, 1994
  |
Project Report
Author:
David A. Clark
…
Published in:
MIT Lincoln Laboratory Report ATC-207
Topic:
wind shear
R&D area:
R&D group:

Summary

The Low Level Windshear Alert System (LLWAS) is an anemometer-based surface network used for detection of hazardous wind shear and acquisition of operational wind information in the airport terminal area. The quality of wind data provided by the LLWAS anemometers is important for the proper performance of the LLWAS wind shear detection algorithms. This report describes the development of an automated method for anemometer data quality (DQA). This method identifies potential data quality problems through comparison of wind data from each sensor within a network to the mean wind speed and direction of the entire network. The design approach and implementation are described, and results from testing using data from the demonstration Phase III LLWAS network in Orlando, FL are reported. Potential improvements to the automated DQA algorithm are presented based on experience gained during analysis of the Orlando data. These recommended improvements are provided to assist future development and refinement of the DQA methodology to be performed by the FAA Technical Center.
READ LESS

Summary

The Low Level Windshear Alert System (LLWAS) is an anemometer-based surface network used for detection of hazardous wind shear and acquisition of operational wind information in the airport terminal area. The quality of wind data provided by the LLWAS anemometers is important for the proper performance of the LLWAS wind...

READ MORE
An automated method for low level wind shear alert system (LLWAS) data quality analysis

Uplink ATCRBS environment measurements along the Boston-Washington corridor, volume 2: interrogator characteristics

February 28, 1979
  |
Project Report
Author:
Ferenc Nagy Jr
Published in:
MIT Lincoln Laboratory Report ATC-83,II
Topic:
surveillance
R&D area:
R&D group:

Summary

Airborne measurements of the Air Traffic Control Radar Beacon System (ATCRBS) 1030 MHz uplink environment due to ATC ground interrogators are described. The measurements were made using a special purpose airborne sensor-recorder during a flight from Boston to Washington (at 8500 feet) and back (at 17,500 feet) on 16 December 1977. Data were recorded at 24 locations between Boston and Washington. Volume 1 of this report summarized the measured distributions of pulse, interrogation, and suppression rates with received power. This volume presents specific characteristics of the sources of the interrogations observed from the air dUring this flight. These characteristics include the identities of 46 of the interrogators, the approximate locations of 22 unidentified interrogators, the interrogation repetition interval (PRI), scan period, and mode interlace of all interrogators, and the radiated P1, P2, and P3 levels in the form of antenna patterns for six selected interrogators. It was found that approximately 40% of the interrogators employ the standard AAC or 2ACA mode interlaces while 20% transmit no Mode C interrogations. The distributions of PRI's and scan periods were found to be relatively free of bunching which would cause undesired synchronization effects. The antenna plots measured from the air show relatively large scan to scan variations due to aircraft motion. In addition, some of the interrogator sites were found to suffer from reflections and sidelobe punch-through effects.
READ LESS

Summary

Airborne measurements of the Air Traffic Control Radar Beacon System (ATCRBS) 1030 MHz uplink environment due to ATC ground interrogators are described. The measurements were made using a special purpose airborne sensor-recorder during a flight from Boston to Washington (at 8500 feet) and back (at 17,500 feet) on 16 December...

READ MORE
Uplink ATCRBS environment measurements along the Boston-Washington corridor, volume 2: interrogator characteristics

Verification of DABS sensor surveillance performance (ATCRBS Mode) at typical ASR sites throughout CONUS

December 20, 1977
  |
Project Report
Author:
Walter I. Wells
Published in:
MIT Lincoln Laboratory Report ATC-79
Topic:
surveillance
R&D area:
R&D group:

Summary

A Transportable Measurement Facility (TMF) incorporating antenna, r-f, and reply processing elements of a Discrete Address Beacon System (DABS) sensor has been sited at, and in the vicinity of, several FAA terminal ASR's throughout the United States. Data collected at these sites have been thoroughly analyzed to verify the design of the DABS sensor and to establish the need for design refinements. This report presents the results that pertain to DABS and ATCRBS Mode range and azimuth accuracy and to the total ATCRBS Mode reply processing performance, It is shown that both range and azimuth accuracies for the DABS sensor are a factor of four to five better than those provided by existing ARTS (BI-4) interrogators, and that the average blip/scan ratio is 98% or better, dropping only a few percentage points in crossing track situations.
READ LESS

Summary

A Transportable Measurement Facility (TMF) incorporating antenna, r-f, and reply processing elements of a Discrete Address Beacon System (DABS) sensor has been sited at, and in the vicinity of, several FAA terminal ASR's throughout the United States. Data collected at these sites have been thoroughly analyzed to verify the design...

READ MORE
Verification of DABS sensor surveillance performance (ATCRBS Mode) at typical ASR sites throughout CONUS

Further studies of ATCRBS based on ARTS-III derived data

December 13, 1974
  |
Project Report
Author:
Alan G. Cameron
Published in:
MIT Lincoln Laboratory Report ATC-38
Topic:
surveillance
R&D area:
R&D group:

Summary

Lincoln Laboratory has been examining digital data extracted from operational ARTS-III radar beacon processing systems for the past eighteen months, in an attempt to determine the capabilities of today's civil Air Traffic Control Radar Beacon System. Our original report on the subject, "Empirical Assessment of ATCRBS," (Ref. [1]), covered data gathered during the first six months of the study from the Andrews AFB (ADW), Boston (BOS), and Las Vegas (LAS) ARTS-III eqUipment. This report discusses analysis of data received since the completion of that initial report, which was gathered at the Ontario, California (ONT), Albuquerque (ABQ), Minneapolis St. Paul (MSP), Milwaukee (MKE), and Boston (BOS) ARTS-III sites, and also the Suitland, Md. en route radar site. The major topics considered here are false targets caused by multipath reflections, and asynchronous interference (fruit). Four different mechanisms resulting in false targets were observed in the data and are discussed in detail in the report. In addition, levels of asynchronous interference as measured and analyzed, and two procedures are discussed by which the characteristics of the interrogators responsible for the fruit (including location) can be determined. One of these is described in detail, and shown to yield results that agree closely with known parameters. Several other topics, including improper reply decoding, are also discussed.
READ LESS

Summary

Lincoln Laboratory has been examining digital data extracted from operational ARTS-III radar beacon processing systems for the past eighteen months, in an attempt to determine the capabilities of today's civil Air Traffic Control Radar Beacon System. Our original report on the subject, "Empirical Assessment of ATCRBS," (Ref. [1]), covered data...

READ MORE
Further studies of ATCRBS based on ARTS-III derived data

Concept formulation studies of the surveillance apects of the fourth generation air traffic control system

September 21, 1971
  |
Project Report
Author:
Adrian E. Eckberg Jr
…
Published in:
MIT Lincoln Laboratory Report ATC-7
Topic:
surveillance
R&D area:
R&D group:

Summary

This report presents a first order feasibility study of four particular candidate surveillance systems for the fourth generation air traffic control system. No attempts has been made to compare these systems, rather we have chosen to examine in detail the most crucial aspects of each. This analysis has brought to light many of the features of these systems. These are detailed along with the research and development required to select the most attractive surveillance system.
READ LESS

Summary

This report presents a first order feasibility study of four particular candidate surveillance systems for the fourth generation air traffic control system. No attempts has been made to compare these systems, rather we have chosen to examine in detail the most crucial aspects of each. This analysis has brought to...

READ MORE
Concept formulation studies of the surveillance apects of the fourth generation air traffic control system
1-7 of 7