The Technology Collaborative
Round 14 Awards
1. "Battery Operated ZigBee Router Development"
Wellspring Acquisitions
Wellspring will develop a low-power-consuming router application for ZigBee, the new open standard for machine-to-machine data networks. Like Wi-Fi and Bluetooth before it, ZigBee is an IEEE affiliated open protocol that improves performance and lowers cost of wireless communications in building controls, security monitoring, and meter reading. It enables low cost patient tracking, condition sensing, maintenance monitoring, capacity control as well as other two-way wireless robotic-enabling applications. ZigBee is especially exciting because nodes self configure in a mesh network that self-heals in response to environmental dynamics that defeat traditional one-way point-to-point radio systems. Wellspring is an early adopter of ZigBee in its submetering product line, and is now recognized as a world leader in ZigBee applications. This project will assist Wellspring in leveraging its ZigBee lead by developing lower power routers that may be battery operated - making wireless network nodes wireless in a second dimension. This new approach to ZigBee will be used by Wellspring, its design-service clients and other members of the Technology Collaborative to lower the installed cost of ZigBee networks.
2. "Surveillance Goes Cellular"
Intrigue Technologies
Constant concern for increased security, emergence of new technologies, and the rising acceptance of cameras in everyday life are driving the video surveillance market to a nearly limitless future of growth. This project proposes to develop an IP (Internet Protocol) camera for surveillance and monitoring applications. The camera is based on the RoboRetina® camera core, currently being developed under TTC sponsorship. While the Video-over-IP is not a new idea, the surveillance market is just now beginning to displaying mainstream need for it. There are many applications for the IP camera including monitoring of homes, businesses, and schools. In addition, the cameras can aid in gathering customer statistics, inventory control, and traffic analysis.
The project will develop and add networking capabilities for universally deployable video, including video over cellular infrastructure. Compared to “hard” IP infrastructure, the cellular network has lower bandwidth, but offers a low cost and easily deployable alternative that covers nearly 99% of the world’s population. To enable real-time network transmission, the project will prototype hardware and firmware for three video compression approaches supported with commercially available processing chips. The main product differentiation for the firm, still lies with the RoboRetina® core that, unlike conventional camera, can deliver superior images under vastly changing illumination conditions.
3. "Toolkits for Blocking Remote Code Execution Attacks"
DayZero Systems Incorporated
DayZero is a Pennsylvania company providing internet security systems and software built with the purpose of identifying and neutralizing previously unknown, or zero-day, malicious attacks and content. The present project will create tool kits vital to the deployment of SigFree: a scalable, real-time, out-of-the-box, and signature-free buffer overflow attack blocker to protect enterprise servers and clients.
Today, both enterprises and individuals face increasing Cyber attacks that are becoming more creative and destructive. Buffer overflow is a root cause for most of these Cyber attacks. SigFree includes novel techniques to distinguish malicious code from valid data. SigFree’s signature free feature allows it to block new and unknown buffer overflow attacks while providing transparency to the systems protected by SigFree. DayZero’s SigFree has a low deployment cost, with no need to upgrade or replace the legacy components of an enterprise system, and operates with negligible throughput degradation. Experimental studies have demonstrated SigFree’s ability to block all types of code injection attack packets without impairing normal client requests. The end result of this project will pave the way to enterprise deployment of this new tool to thwart Cyber attacks.
4. "Electronic Design Automation and the Hardware Engineer"
Design Advance
The implementation of Field-programmable gate arrays (FPGAs) at the PCB is a time-consuming and largely manual step in the design of electronic products. Nearly every hardware engineer in tandem with a PCB designer incorporates FPGA’s into their designs. These complex logic devices can range from 144 to 2,000 pins per device. The hardware engineer responsible for FPGA design needs to evaluate the placement of his component (aka floor planning) and analyze pin assignments (aka pin swapping) in several stages of the PCB design process. The goal of the FPGA design is to support the design of a routable (and cost-effective) circuit-board layout. The pin-swapping analysis allows the designer to produce pin assignments that shorten and uncross signal paths. The floor-planning part of the analysis allows the designer to identify opportunities to adjust the board layout so as to improve the results obtained by pin swapping.
5. "Face-Centric Video Compression"
Pittsburgh Pattern Recognition
The distribution and playback of video content has become commonplace in a number of applications, with growing interest in better quality with limited bandwidth. Despite technological advances, widely distributing real time video remains challenging. At broadcast television quality, real time video can push the limits of both consumer-grade Internet connections and third generation wireless services. To compensate, applications must often compromise on video quality. Widespread adoption of video- based services will require further advancements in video compression technology: advances that enable lower bit-rate video streams without compromising perceived quality. We believe that high accuracy face detection can dramatically improve video compression capabilities for a specific class of video: sequences in which human faces are of prime interest. Our state-of-the-art face detection software efficiently locates and tracks the size and position of faces in video sequences and feeds this information to an encoding engine. Our technique can be an effective means of reducing bandwidth requirements while allocating more bits to face regions. Preliminary studies with still images suggest that this face-centric compression strategy can dramatically reduce bandwidth requirements by a factor of two. We plan to implement a software library for face-centric video compression and market this library as a tool for halving the bandwidth requirements of real-time video for a given resolution.
6. "Low Cost Obstacle Detection for Outdoor Vehicles"
Sensible Machines
This proposal seeks to develop low-cost obstacle detection for autonomous machines performing slow to moderate-speed outdoor operations. This capability is fundamental in achieving economically viable autonomous operation for many tasks; however, deployment of automated machines has been greatly constrained by the lack of a sufficiently inexpensive and reliable means of detecting near obstacles. An ideal sensor would be solid state so it could be low-cost, highly reliable and provide high data rates. We will develop such a sensor using structured light to detect objects in the environment. Structured light involves projection of a coherent pattern of light (typically infrared) into the scene. The distortions of the pattern are used to determine the shape of the environment and deduce the presence of obstacles. Until recently, it was not possible to use structured light in environments with natural lighting because sunlight easily drowns out the projected pattern when energy is kept within eye safety limits. This new sensor employs technology closely related to Sensible Machines’ low cost localization system technology.
7. "Hardware Fabrics: Embedded Solutions for High Performance and Micro Power"
University of Pittsburgh
Field Programmable Gate Arrays (FPGAs) are high-volume, low-cost products targeting various markets, including consumer wireless products for which power consumption may be a primary concern. The market for FPGAs has recently increased due to its use as cheaper Application Specific Integrated Circuit (ASIC) alternatives. University of Pittsburgh proposes to create a new reconfigurable SuperCISC Fabric and C-based design flow that provides the performance of custom hardware acceleration with ultra low power consumption that can be programmed from C code. The SuperCISC Fabric is a simpler, FPGA replacement for System-on-Chips (SoC) that provides similar performance to existing FPGA solutions with a much lower energy/power requirement. The SuperCISC Fabric may be used as an eFPGA IP block as a solution to SoC customers with hardware design components that are not well defined at the design start, but may require hardware acceleration and/or low power. It also can be applied to systems requiring upgradability to new algorithms with hardware performance and low energy requirements. The SuperCISC Compiler and SuperCISC Architecture implemented within an FPGA is able to achieve significant performance improvements. With addition of low power embedded processing with low-energy SuperCISC Fabric, similarly significant energy improvements are expected.
8. "Botnet Detection and Mitigation"
University of Pittsburgh
Bots are Trojan horses that propagate like worms or viruses and have a channel through which they can be remotely controlled by a botmaster. Botmasters can use bots in a variety of attacks, including identity or intellectual property theft, sending denial-of-service (DoS) attacks or spam, or committing click fraud. It is estimated that 7% of Internet hosts are infected by at least one bot, and 250,000 new bots are detected each day. Our research program will prototype and evaluate two appliances, respectively called Local and DoS BotBuster, to detect and mitigate botnets. The Local BotBuster can detect bots, disable their communication with bot controllers, and notify the owners of the infected hosts. The DoS BotBuster can detect and block DoS attacks that mimic flash crowds. It can also be configured as a front-end for enterprise server farms or Web hosting. Current DoS mitigation equipment is unable to detect such attacks because they look like naturally occurring peaks in legitimate client traffic.
9. "Portable Sensor-based Coverage for Sustained
Geophysical and Radiological Survey"
Red Team Inc.
Red Team, Inc. recently completed its first commercial project by using an autonomous unmanned vehicle to perform a successful UXO survey of nearly 2000 acres in the southwestern United States within a sixty-five day operational window. The project realized a significant improvement in productivity levels over the traditional manual means of performing such surveys with much-improved accuracy and repeatability. During the initial project, the system operated with full autonomy mode over 90% of the time. From time to time, however, in order to avoid obstacles or traverse unexpectedly complex terrain, the autonomous navigation system was over-ridden with remote control. Resuming autonomous operations then caused areas of coverage to be missed and modification were a time-consuming and relatively complex task. Another concern is that no single survey platform will support geophysical and radiological surveys in all terrain. Also, the current software requires a programmer familiar with the internals of the technology to define issues such as the area of coverage or the path to be taken by entering commands. This is a relatively sophisticated and tedious operation, especially for irregularly shaped operating areas. Red Team, Inc. plans to develop a more user-operable, dynamic coverage planning capability that can be readily adapted onto different sized robotic platforms for semiautonomous survey, patrol, and other coverage operations.
