- “OPNET Modeling and Simulation of Large Scale MPLS Computer Networks”
Accipiter Systems
This program will analyze and complete three worst case, large scale MPLS sample simulation scenarios that were selected from the extra samples identified in Accipiter Systems’s current PDG program. This proposed research leverages the hardware and network models developed in the current project as well as the tool limitations that were found. The tool of choice to perform these simulations is the tool suite from OPNET Technologies. OPNET offers a modeling/simulation product for network R&D. This tool enables designers to evaluate how networking equipment, technologies and protocols will perform under simulated network conditions. The specific tool being used is OPNET Modeler. OPNET Modeler is the industry's leading network technology development environment, allowing for the design and study of communication networks, devices, protocols, and applications. The world's most prestigious organizations use Modeler to accelerate R&D.
MPLS (Multiprotocol Labeled Switching) will be the protocol of focus for this research. MPLS combines the properties of virtual circuits common in switch architectures with IP datagrams common in router architectures. MPLS combines the concepts of switching and routing. MPLS-enabled routers forward packets using short, fixed-length labels, which have local significance, through a forwarding paradigm called label swapping.
The deliverables include a Simulation Design Report for the scenarios to be simulated. The Simulation Design Report will describe the practicality of building the simulations with the quantities of entities required, simulation implementation strategies, and the OPNET tool support for the various required networking mechanisms. Other deliverables are the network simulation models and a Final Report.
The commercialization potential for network modeling and simulation is far reaching and growing in importance. These newly developed models will be useful for commercial companies and government contractors performing network simulations. This work has been leveraged by Accipiter Systems, a newly formed, Pittsburgh-based company, in research being performed for Lockheed Martin resulting in local job creation and an Area of Excellence recognized by DARPA, PSU, ITT, AFRL and others. Lockheed is involved in several networking related programs with government agencies where simulation is required. These programs include Optical RF Combined Link Experiment (ORCLE), Airborne Lasercom Terminal (ALT) and Transformation Communications MILSATCOM (TCM). Future work is being pursued with Lockheed, CSC and other DOD contractors.
- "Embedded Hardware Face Detection"
Advanced Interfaces, Inc. and The Pennsylvania State University
Face detection is a vital task in surveillance and security applications and is the precursor to face recognition. Further, next generation video conferencing and market research applications rely on face detection and classification to personalize user experience.
Advanced Interfaces (AI) has developed software implementations for face detection and classification as part of its VideoMining™ suite of solutions. There have been a few attempts at mapping face detection on hardware platforms. However, these solutions suffer either from the poor quality of algorithms used or from the lack of a compact solution as they use multiple boards. Both these drawbacks are serious limitations to integration of robust face detection capability into compact embedded devices. The goal of this proposal is to circumvent these difficulties.
The proposed solution would address the needs of high-growth digital video surveillance market that is estimated to experience a compounded annual growth rate of 55% to reach $7.4 billion by 2007. If this project is successful, we anticipate to help diversify our company products to new applications such as automated video surveillance monitoring, video conferencing and video monitoring for intelligence and market research at retail stores. TTC member companies such as Tyco, Marconi and Seegrid can potentially benefit from this technology by integrating it with their products. Companies like OKI will benefit from the validation of their SoC platforms for Video Processing applications. The value proposition includes localized faces for face recognition and biometrics, automated security triggers and alarms, personalized video conferencing, reduction in overall system cost, increased performance and better video encoding.
Our experience working in commercial environments has yielded robust algorithms and technology for face detection that we can port to OKI SoC uPLAT7C platform and Xilinx ML 310 Embedded Development Platform. The embedded platform will be directly attached to the video source, such as a security, robot or video conferencing camera and will take video images as input to deliver face location as output. The implementation will be tested and validated using our proprietary face image database, which contains over 30,000 annotated images.
The research issues involved in this project include evaluating multiple face detection algorithms from a hardware implementation perspective, increasing parallelism at the algorithm level to exploit hardware speedup, studying the impact of bit precision on hardware cost and partitioning the algorithm to determine parts that can be mapped on to the processor core and the FPGA fabric. The project will result in the design of an architecture that will support face detection for different image/video sizes and will be reconfigurable to support tradeoff between desired decision quality and speed.
- "Low-Cost Single-Chip Radar"
IC Mechanics, Inc. and Carnegie Mellon University
There are many commercial and military applications for small, low-cost, low-power, high resolution directional RF front end + antenna arrays. One economically interesting application is to build small, low-cost radar systems for a variety of tasks; e.g., collision avoidance and navigation in vehicles, and airport and building security screening of people. Phased-array RF front ends + antennas can generate an RF signal that is highly directional (e.g., the beam width for a 10 x 10 element phased antenna array is roughly 7°), enabling very high performance short range radar systems. The goal of this research proposal is to demonstrate the key component of single-chip radar systems: a single-chip, low-cost, low-power, high-resolution directional RF front end + antenna array.
To date, phase-array radars have been extremely expensive bulky systems only suited for military applications. By exploiting the extremely high fT and fMAX that can be achieved in leading edge IC processes, RF circuits operating in the 50-100GHz range can be build in a cost effective manner. This opens up the novel possibility of integrating a low-cost two-dimensional array of phased antenna elements and the associated RF transmit and receive circuitry on a single IC substrate. We propose to demonstrate that leading edge commercial IC process technology can be used to implement low-cost single-chip RF front end + phased-array antenna elements.
- "Ultra-Fast 3D Imaging Camera for Demanding Price/Performance Applications"
Intrigue Technologies, Inc.
Information about visual 3D shapes is increasingly important in robot navigation, obstacle avoidance, object recognition, object modeling, industrial inspection, industrial robot guidance, computer graphics, human-computer interaction and virtual reality systems. Although range imaging has been an active area for several decades, today there are no accurate and inexpensive ranging image sensors that would stream range maps at video rate, or higher. Intrigue Technologies, Inc. will development a 3D imaging camera based on our patented custom CMOS range image sensor for object modeling and other industrial applications at close and intermediate distances (0.05-10m). Our method and sensor leverage a dynamic triangulation range imaging principle, which is the extension of well-known conventional triangulation. For surfaces with reasonable reflective properties, the sensor will be able to provide about 500 frames per second of 64x64 range pixels each, resulting in more than 2,000,000 range point measurements per second. The range accuracy for a stand off distance of 0.85 meters and the baseline of 70 millimeters is estimated to be around 200um. In addition to the high speed and accuracy, we are also able to leverage temporal signal signatures created by the sweeping laser, thus providing a level of immunity to ambient illumination and surface reflectance properties. Not only will our sensor be able to do demanding 3D imaging applications at lower cost, but due to its speed it could enable new possibilities, such as capturing rapid 3D events, which to our knowledge none of the existing systems can do.
- "Processible, Conductive Polymer Films for Touch Screen and other Electronic Applications"
Plextronics, Inc.
Plextronics, Inc., a two-year old start-up based in Pittsburgh, PA that is pioneering the commercialization of Inherently Conductive Polymers, is submitting this proposal in collaboration with Elo TouchSystems, a wholly owned subsidiary of Tyco Electronics and North America’s largest touch screen manufacturer. The goal of this project is to produce conductive polymer systems that will be developed and optimized for use in touch screens. As a result, we will deliver materials that are highly conductive, stable, and transparent – an ideal solution for Elo’s touchscreen application – and suitable for several additional electronic applications such as printed circuit boards, flexible displays, passive components, and plastic electronics.
Touch screen technology has revolutionized the way users input data in industrial, medical, point-of-sale, kiosk, retail, and gaming applications. Touch screens enable people to use computers instantly without training, virtually eliminate operator errors, eliminate keyboards and mice, stand up to harsh environments, and provide fast access to all types of digital media. The market for touch input products is forecasted to reach $2.5 billion by 2005 with an annual growth rate in units sold of 20%.
Today, vacuum deposited optically transparent indium tin oxide (ITO) is the industry standard material to provide electrical conductivity to glass and polymeric films. However, the performance of ITO suffers when applied to plastic. These thin film coatings are fragile and are readily damaged during bending or other stress inducing conditions. As an ITO replacement for certain applications, our conductive polymer material Plexcore™ will yield a conductive flexible film with high transparency in the visible region, reducing the manufacturing cost and improving the reliability of resistive touch screens.
- "Mobile Multi-Layered IPSec: Advanced Services"
The Pennsylvania State University and Accipiter Systems
Security in a wireless environment is an important and challenging issue. The limitations of mobile devices must be weighed against the inherent insecure nature of the wireless media when designing a solution. Furthermore, many wireless performance enhancing algorithms require access to packet headers, which is in direct conflict with the basic Internet end-to-end security model using IPSec.
In a recently completed PDG/TTC-funded project called Mobile Multi-Layered IPSec (MML-IPSec) we defined, implemented and evaluated a new protocol that solves many of these problems. MML-IPSec allows a mobile access router, for example a Mobile IP Foreign Agent (FA) to decrypt and operate on packet headers to enable performance enhancing algorithms to execute, while protecting the packet payload end-to-end. In this project we addressed key distribution, mobility management, packet transfer and integration with a TCP-based performance enhancement called SNOOP. For mobility management, we designed and implemented two schemes that result in handoff latencies of between 50 – 100 milliseconds, well within a TCP time-out period. Our throughput measurements showed that when compared to running TCP over IPSec in a wireless environment, running TCP with SNOOP over MML-IPSec yielded throughput gains of over 150%.
This proposal will advance MML-IPSec so that it is on par with existing security methods, maintains high performance and has improved flexibility for supporting applications. We will deliver a specification and implementation of IKEv2 extensions to support key generation, re-keying and mobility in MML-IPSec; design and implementation of a key distribution protocol in a multicast environment based on GDOI; design and implementation of ROHC with MML-IPSec and dynamic MML-IPSec. We will also support PDG companies by writing an IETF Draft in support of submitting MML-IPSec for standardization. We expect this project to benefit companies in the data networking, sensor, and security industries.
- "Design of Efficient RFID Systems"
The Pennsylvania State University
Radio Frequency Identification (RFID) systems are expected to have a large impact with diverse potential applications, from supermarket check-outs, to inventory tracking, to authentication of tagged components and containers that are shipped from all over the world.
This work aims to significantly improve the performance of RFID systems by introducing simple, yet novel solutions that will result in guarantees for successful tag information collection with high reliability. The RFID systems we envision have a large number of tags present in the range of a reader that can possibly move in and out of the reader range. In addition, we envision that in the near future, each tag might be asked to report information related the object it is attached to perhaps in a periodic fashion, that is, tags may report sensory information as well as ID information of the object. The designs we will come up with will render the deployment of dense RFID systems feasible and lead to their realizing the application potential in the near future.
Tag collision is a known problem in RFID systems. Collisions occur when more than one tag sends its information back to a reader at the same time, thereby making it impossible for the reader to interpret data from any tag successfully. Methods proposed up to date for resolving these conflicts at the reader do so either by invoking retransmissions, or by having the tags respond to the reader one at a time, collision-free. These methods are variants of collision resolution schemes proposed several decades ago for packet networks that have relatively infrequent simultaneous transmissions and are expected to perform well only when a relatively small number of tags are present in the range of the reader. When a large number of tags are to be read, they result in a fairly large delay in reading the tags, in view of small amount of exchanged data. This will not be acceptable as the use of RFID systems would grow, with the concurrent and utmost need of reading multiple tags in the least possible time.
The aim of this project is to develop adaptive medium access control mechanisms that will result in significant improvement over the methods currently suggested, in terms of the total read time of tags and throughput. Specifically, we will introduce a class of mechanisms that we term adaptive slotted aloha protocols (ASAP), and a class of protocols that will improve binary tree search, called optimum protocol tree (OPT). These novel methods we propose do not require any modification at the tag level, and can be implemented easily at the reader with virtually no additional complexity.
- "Sensor Stabilization and Geo-Positioning Platform"
RE2, Inc., Valley Technologies, Inc., and Carnegie Mellon University
There is a need in the small Unmanned Air Vehicle (UAV) market for a light-weight, inexpensive sensor mount that is able to remove vibration effects from the mounted sensor while providing geo-location information (knowledge of where the sensor is pointing with respect to the earth). No such system exists today. Our proposed platform addresses the market need. We have identified a customer base for the proposed sensor mount, one of which is matching development funds in cash. This customer, Carnegie Mellon University, will also serve as the beta-tester for the proposed sensor platform product. We also have customers in the defense industry willing to purchase the end product. However, our commercialization vision extends beyond the defense UAV market. There are several other markets that stand to benefit from this technology, including the utilities industry performing power line inspection and remote monitoring. Such commercial industries currently rely on full-scale helicopters and expensive camera platforms. Using a less expensive, small UAV with our sensor platform provides an alternate, more cost-effective solution. Finally, the unmanned ground vehicle (UGV) and unmanned marine vehicle (UMV) industries can also benefit from this technology. Cost reduction and manufacturing for the commercial sector will be one of the primary focuses of our 6 month development effort. To mitigate risk, this program is divided into two phases. The first phase key milestone is the demonstration of a 1 degree-of-freedom (DOF) prototype on the bench.
This bench prototype will demonstrate how the platform will isolate vibration and provide the ability to geo-position the sensor in 1 DOF. At the end of this program, we will deliver a pre-production prototype weighing only 15 lbs that will be demonstrated in flight on a small UAV. A camera will be mounted to the platform, showing how vibration effects are removed from the transmitted image as viewed from a ground control station. The ground control station will also be able to command the sensor to lock onto a particular position on the ground during flight.
- "Passive, Stealthy, Low Cost, Light Weight, High Resolution, Durable, 3D Ranging Sensor"
SEEGRID, Inc., Applied Perception, Inc., re2, Steel River Systems, and Compunetix
Recent, swift advances in CMOS, CCD, DSP and FPGA technologies have created a near term opportunity to commercialize a next-generation stereo ranging camera that provides passive, high resolution, 3D ranging at disruptive price-points (~$200 for a complete sensor) and with break-through capabilities for the military, and industrial markets. SEEGRID is partnering with four other TTC members to leverage our combined expertise into producing this sensor in a timely and cost-effective manner.
While many laser and radar ranging technologies currently exist and more are being invented, only stereo ranging is passive (i.e. no energy is emitted from the sensor). Passive ranging provides certain important capabilities that cannot be achieved through any other type of ranging sensor. These include:
-- no heat signature or EM signature on ranging, providing stealth capability for military use,
-- ranging from very close up (cm’s) to very far away (potentially 100’s of meters from a compact unit) with very, very low power draw, and
-- inherent eye safety and environmental safety. It is important to note that, no other proposed sensing technology provides these important capabilities for the military and industrial markets.
Because of these unique capabilities, work in computer stereo vision has been ongoing for decades. Cost and technology limitations (embedded processor capabilities) have kept stereo ranging, largely, an academic pursuit, or have restricted it to specialized applications requiring low weight, low power and high reliability ranging. (e.g. the ranging sensors on NASA’s Mars Rovers are stereo cameras). SEEGRID currently manufactures a stereo ranger that suffers from the above limitations.
By rethinking the problem in the context of advances in CMOS, CCD, DSP, and FPGA technologies, and implementing a complete ready-to-use stereo ranging system in a compact self-contained unit, SEEGRID can produce a new sensor that is:
1) Price-compatible with both the military and the industrial markets, 2) Performance capable of being used in demanding industrial vision and embedded mission critical applications, 3) Rugged and durable to work reliably in extreme/military and real-world/industrial commercial environments. 4) Simple to implement by keeping product designers shielded from the complexities of calibrating, rectifying, and processing stereo range measurements.
|  |
|
