NEXT GENERATION DIGITAL VIDEO SURVEILLANCE: RESEARCH AND DESIGN ENGINEERING IMAGERY INTELLIGENCE BY WAY OF SECURE WIRELESS TRANSPORT, WITHOUT THE NEED FOR SHORE POWER AND CONVENTIONAL CONNECTIVITY MEDIUMS INHERENT TO CLOSED CIRCUIT TELEVISION Technical POC: Gregory Perry Phone: (813) 505-7713 Email: gperry@adv-wireless.com Fax: (813) 741-3033 5450 Bruce B. Downs Blvd. #313 Wesley Chapel, FL 33543 Administrative POC: Daniel C. Roy Phone: (617) 956-0888 Email: droy@adv-wireless.com Fax: (617) 956-0880 155 Federal Street, 5th Floor Boston, MA 02110 Cost: Total Base (3 years) - $2,971,812 Year 1 (base): $1,635,781 Year 2 (base): $847,416 Year 3 (base): $488,615 Option 1: $639,933 Option 2: $2,443,361 Technical Topic Areas: High Fidelity, Ultra Low Power Video Surveillance Hardware Encrypted Wireless Transport Link for Remote High Security Applications Small Form Factor for Covert Installations Distributed Back End Processing Network for Imagery Data Cost Sharing: Cost sharing on the hardware platform design by AWA is estimated at approximately $370,000 total for the base and options and includes labor, waiver of fee, and the use of the AWA data collection network and design lab facilities. Includes use of hardware and software engineering resources (as appropriate). 1 Submitted by: ADVANCED WIRELESS AUTOMATION, INC. (AWA) (Type of business - "Video Surveillance Hardware and Software Provider") Table of Contents SECTION 1: ADMINISTRATIVE INFORMATION 1 SECTION 2: TECHNICAL PROPOSAL A. INNOVATIVENESS OF THE PROPOSED RESEARCH B. TECHNICAL RATIONALE AND APPROACH C. DELIVERABLES D. STATEMENT OF WORK E. MILESTONES F. TECHNOLOGY TRANSFER G. COMPARISON WITH OTHER ONGOING RESEARCH H. KEY PERSONNEL I. FACILITIES/RESOURCES J. COST 3 3 4 10 X X X X X X X SECTION 3: ADDITIONAL INFORMATION (X Related Papers) Attached - "Sample Paper 1” 2 A. INNOVATIVENESS OF THE PROPOSED RESEARCH This proposal from Advanced Wireless Automation, Inc. focuses on advanced video surveillance technology intended for use within imagery intelligence (IMINT) gathering applications, with certain unique features that allow users to perform visual and other forms of monitoring in remote and rugged locations previously unsuitable for deployment. By integrating existing COTS wireless transmission products, high performance CMOS and other image sensors, recently commercialized digital signal processing components, aggressive power saving and data compression methods, military grade public and conventional keyed cryptography, and custom software running on a robust and efficient embedded Linux operating system platform, AWA's IMINT platform will operate off of the power grid and without the need for hardwired network connectivity. The proposal concentrates on three related thrust areas: (1) high-resolution video imaging hardware capable of instantly locking and acquiring imagery data; (2) secure wireless mesh networking technology to relay said imagery, telemetry, and sensor data from endpoint devices to a centralized observation network for subsequent viewing and storage; and (3) robust compression and power management disciplines to provide a compact form factor hardware solution that can be deployed in rugged and remote locations without the need for shore power supply and conventional network connectivity mediums. Our proposed effort involves a variety of COTS product offerings used in conjunction with AWA’s proprietary video imaging technology, with the intent of creating a unified IMINT acquisition platform capable of being deployed in a wide variety of configurations, both overt and covert in nature. AWA personnel will build upon their core areas of expertise in the fields of wavelet video compression, encryption and network security, embedded hardware research and development, and network design engineering, to create an open architecture solution to be used for near real time acquisition of remote imagery and sensor data in harsh environmental conditions. B. TECHNICAL RATIONALE AND APPROACH AWA's approach to imagery intelligence realizes that in many situations streaming real-time video surveillance such as that delivered by legacy CCTV systems are expensive to deploy, and overvalued with regard to intrusion detection capability, post mortem forensic recovery, confidentiality of imagery data, and power consumption. Traditional video surveillance products rely on a cumbersome physical infrastructure of cabling and heavy current electrical power supply, and the data product of CCTV and other conventional video surveillance systems are copious in volume, low resolution analog in format, and require expensive human attention to be of value to the end consumer. Addressing these problems in tandem requires a systemic approach to design and architecture, and yields a product that bears little resemblance to what we are accustomed to in traditional video surveillance systems. The first step to solving this problem set is to efficiently remove surveillance data from the analog domain at the first opportunity, i.e. as close to the image sensor as possible. This is 3 accomplished in the human visual spectrum by use of highly integrated CMOS image sensor arrays, whose performance are now approaching that of legacy CCD technology while offering large price and implementation advantages. The second barrier to entry is that of producing an imager that is ready to operate with very small time-to-acquire latency, while providing a reasonably swift frame rate; in addition, the imager array must be compatible with extended battery and/or fuel cell operation, and capable of operating asynchronously to the main platform host processor. This is accomplished by the use of a lightweight supervisory processor, interrupted awake via various sensors, external stimulus, or internal schedules, and in control of a programmable logic device. An FPGA in turn clocks the image sensor and provides address generation and image data transfer to a large pool of static memory, which can later be interrogated by a host controller or DSP to further process and transport resulting frame data. A further advantage of employing an FPGA for image sensor drive is that multiple sensors can be driven by one FPGA, with sensor selection easily accomplished by supervisory processor configuration of the FPGA. A second processor hosting the platform operating system and other peripherals is booted on demand by the supervisory processor. Once an image sensor frame has been captured, frame buffer memory is made available to this host for transfer to main memory. Other scheduled platform tasks to be accomplished by the main processor can also be initiated by the supervisory processor, improving the operational characteristics of the image capture system and reducing overall power consumption. This system is designed for deployment regardless of local infrastructure, while at the same time exploiting locally available low cost communications networks. Wireless transmission methods vary and will continue to do so depending on deployment contexts. Possible topologies include point to multipoint wireless networks; self healing wireless peer to peer mesh networks; stand alone single platform installations using LEO satellite links and VSAT terminals on a one-to-one basis, etc. Most wireless communications methods to be employed by the AWA system are of relatively low bandwidth. High bandwidth communications require advanced modulation techniques that consume large amounts of power relative to typically available current. Further, when highpower transmission methods are unavoidable, it then becomes imperative that data payload is limited, especially for high-resolution imagery data. AWA uses a very aggressive waveletencoding scheme to compress image data; although this is an extremely computationally intensive compression technique, recent plunges in DSP prices coupled with sharp reductions in power consumption and equally dramatic increases in performance afford sufficient horsepower to accomplish wavelet transformation of image data. The advent of dual core chipsets incorporating general purpose microcontroller and signal-processing functionality mean that this class of algorithm can now be deployed in previously unavailable market segments, with compression ratios exceeding 120:1 at 24-bits per pixel image data. Once image data has been acquired and compressed, data security is provided by first negotiating 4 a DSA certificate exchange, followed by a Diffie-Hellman public key exchange, after which an encrypted communications tunnel is established by way of 3DES or AES block cipher. The use of certificate-based authentication provides AWA with an absolute mechanism for granting and revoking access to wireless network resources on a per-endpoint basis, in the event that a device becomes inaccessible due to theft or compromise. In addition, the use of robust encryption technology provides for (3) layers of protection for acquired data: Confidentiality – the use of AES/3DES block ciphers provide industrial strength privacy for communications traversing the wireless network link, preventing eavesdropping and recovery of network traffic via signals intelligence and/or TEMPEST-related attacks Authenticity – communications are guaranteed to have originated from the appropriate corresponding endpoint device, an important factor to consider when preserving forensic quality video stills Integrity – imagery data is guaranteed integrity from endpoint to backend collection network by way of SHA-1/MD5 one way hashing functions, which prevents attempts to manipulate or alter image data after the acquisition process AWA's platform consumes miniscule quantities of electricity in idle mode, and the system can be powered by primary or secondary batteries for extended periods of time. A caveat here is that if the platform is deployed in a high traffic area, thus excessively increasing the duty cycle of the electronics, some means of supplying a modest amount of power must be present. This may take the form of a small PV panel. In operation the products are under the immediate control of a supervisory processor, which monitors time and sensor inputs to awaken the remaining electronics to perform various sampling tasks, which include acquiring images, measuring temperature, or running other sensors as per customer demands. The supervisory processor also periodically opens a receive window via whatever communications method is present in a given deployment context, allowing remote configuration and control of the device, as well as network synchronization. Potential applications for AWA's platform include gate monitoring, access control, area and perimeter surveillance, environmental sampling, and numerous other uses where access to power and network connectivity is difficult or impossible due to geographic dispersion. Further, the underlying architecture can be mutated and rearranged to accomplish a variety of other goals. Some features and applications of the technology employed by AWA are described below. 5 Low Latency and Low Power Image Capture System Design objectives: Ability of imager to lurk with low power consumption and low latency to acquire first image Ability of imager to capture multiple frames, buffer until such time as a transmission channel is open Ability of imager to reduce image data to minimum size commensurate with forensic quality restoration Implementation: A low power MCU typically in sleep mode monitors via hardware interrupts intrusion sensor inputs including loop contacts and PIR motion detectors. An interrupt from a sensor input activates the MCU, causing it to generate a signal to an FPGA (field programmable gate array). The FPGA then generates signals to operate a solid state CMOS image sensor, and provides a path to transfer pixel data from the image sensor to SRAM (static random access memory), generating address signals to control pixel storage location in SRAM. One or several frames may be captured over a time period determined by configuration of the MCU program. As and when complete image frames are available, the FPGA switches control of the SRAM address and data bus to a host controller, which retrieves frame data for further processing leading to compression of image data. Compressed image data is then returned to the host controller board for intermediate storage prior to transmission. Another embodiment has the FPGA switch the SRAM address and data bus directly to the bus of a DSP controller board, allowing direct access of image data by a DSP controller. Another embodiment allows connection of SRAM to a DSP controller board via a DMA controller on a host board supporting the DSP controller and imager support card. Description of Fixed/Remote Host Synchronization, Minimization of Power Consumption: Design objectives: Maintain a transmission availability window with minimum power consumption. Discussion and implementation: In order to allow a back channel for control of a battery powered remote wireless device, it is necessary to minimize receiver power consumption at the remote device while maintaining a predictable receive window for a command to arrive at said device. The controller hardware and protocol necessary to do this with minimum power consumption are incompatible with the hardware and protocol needed for actual bulk data transmission between devices. Thus it is necessary to provide a means of switching hardware and protocol controlling a radio data modem as necessary for the immediate needs of the device using the radio data modem. A low power controller on the remote device is equipped with a solid-state switch capable of asserting control of electrical signals to a radio data modem. This controller periodically turns on a radio receiver, opening a receive window for reception of data. A data concentrator and network controller periodically transmits a beacon signal, and this signal is used to adjust a timer 6 on the remote controller so as to synchronize the data modem activation window with the beacon signal. At times, a command embedded in the beacon signal or local activity at the remote device may necessitate a hardware and protocol change of control of the radio data modem. When this occurs, the controller deasserts control of electrical connections to the radio data modem, allowing another controller to assume control of the radio data modem and simultaneously changing protocol imposed on the radio channel. Secure Deployment of Unattended Devices Design Objectives Provide a means of detection of device theft or unauthorized removal leading to inactivation of said device Discussion & Implementation AWA's equipment is intended for deployment in various locations, mostly unattended. For various reasons, it is desirable that if devices in the network are tampered with or removed that they should inert themselves and wipe out any surveillance or telemetry data they may contain. Various means are provided for this. In the context of a micro cellular deployment involving a master host, a device that is removed from the service area of the master controller will inert itself after detecting that the master is unavailable for a period of time and other conditions. The remote device looks for a characteristic signature generated by the master and embedded in the beacon signal broadcast by the master. The signature will be rendered immune to replay attacks by being composed of a SHA-1 message digest of the content of the beacon signal extraneous to the signature itself. The signature is created using a private key, while the signature analysis is performed using a public key. Telemetry/Surveillance Network Design Objectives Provide a means of gathering image and other sensor data in locations lacking power and data Provide a network that is largely self-configuring and is capable of self-healing Provide a network that is agnostic with regard to uphaul to a central data collection point Network may consist of 1.n devices Discussion and implementation The field side of the system consists of a common suite of core hardware and software supporting low level network functions, augmented by specialized peripheral and memory configurations designed to support functions specific to each type of equipment participating in the network. Hardware characteristics that are likely generic and are important to the application include: the aggressive use of micropower operation methods; combinations of high endurance/renewable power sources; reduction of component counts to a low number consistent with demanding physical endurance goals; highly robust and inconspicuous packaging. 7 Proprietary hardware methods include: the use of multiple image sensors operating semiautonomously in conjunction with a single host processor so as to provide high performance wide angle scene coverage with minimum power consumption and expense; initiation of image sensor operation and completion of scene acquisition in advance of host processor availability; combinations of image sensors, multiport storage devices, digital signal processors and host processors so as to afford pipelining of image acquisition, compression, analysis, and transmission; simple arithmetic logic control of memory devices to afford high performance; low power hardware based data reduction; combinations of the aforementioned methods. Important software characteristics include: the use of wavelet data compression techniques to simultaneously compress and secure image data; wavelet scene feature recognition. Important network characteristics include: autonomous device configuration with regard to peer mesh and uphaul route discovery; autonomous failure management; autonomous identification and lockout of compromised or unauthorized devices. 8 Definition of Terms 3DES – Triple DES AES – Advanced Encryption Standard CCD – Charged Coupled Device CCTV – Closed Circuit Television CMOS - Complementary Metal Oxide Semiconductor COTS - Commercial Off-the-shelf DES – Data Encryption Standard Diffie-Hellman - Public key cryptography based on calculating logs in modular arithmetic DMA – Direct Memory Access DSA – Digital Signature Algorithm DSP – Digital Signal Processor FPA – Focal Plane Array FPGA – Field Programmable Gate Array GPS – Global Positioning Satellite IMINT – Imagery Intelligence LEO – Low Earth Orbit Satellite MCU – Microcontroller MD5 – Message Digest Algorithm RFC1321 PIR – Pyroelectric Passive Infrared PV – Photovoltaic solar panel SHA-1 – Secure Hash Standard SRAM – Static Random Access Memory VSAT – Very Small Aperture Terminal satellite communications system Wavelet - Mathematical function useful in digital signal processing and image compression 9 C. DELIVERABLES Key deliverables for this project are described in the sections below. 1. Endpoint Device A prototype of the endpoint device will be delivered during year 2 of the project. This will include a single board version of AWA’s wireless surveillance platform, and will include: (1) Spread spectrum, fast frequency hopping 900 MHz ISM radio section with antenna array (1) 1280x1024 CMOS imager (1) 320x200 uncooled thermal imaging FPA (1) Light sensor (2) PIR motion sensors (1) Integrated GPS core (1) Temperature and humidity sensor (1) 3-axis magnetometer (1) Covert pinhole lens array 10