The multi-DSP circuitry behind the optics of this company’s camera, which is used for perimeter intrusion detection, provides the computational equivalent of 10 high-end computer servers.
Claims and counterclaims are made about whether one company’s video analytics are all on one chip, or how much competitors have compromised their algorithms by placing them on one chip.
“Our video analytics are all on one chip,” asserts Dvir Doron, vice president of marketing for ioimage, Herzliya, Israel. “The heat consumption is low. Texas Instruments was amazed with what we can do.”
Object Video also maintains its analytics are on one chip. “We’ve been successful in putting a full suite of analytics capabilities on a DSP and utilizing only 20 percent of that chip,” Troha maintains.
He notes that enough room must be left on a chip for all the other camera functions, such as compression. The tie-breaker that is hard for security dealers and systems integrators to evaluate without additional information is the efficiency of the algorithms placed on the chip.
“If the algorithms are very efficiently designed for the mass market, they will certainly run multiple channels on a single DSP chip,” maintains Carolyn Ramsey, director of program management for Honeywell Systems Group, Louisville, Ky. “When they put them on a chip, they are going to learn the unique capabilities of that chip, and optimize to that chip. So the hard work on the hardware side is trying to pick the chip that has the best power.”
But iOmniscient Corp., New York, plays by different rules, maintains Rustom Kanga, Ph.D., CEO. “We don’t use the traditional video motion detection algorithms that everyone else uses — we have a unique type of artificial intelligence algorithms that are very ‘computing-like’ so we can put full algorithms on the edge device,” Kanga maintains. “We don’t have to downgrade them at all.”
SightLogix, Princeton, N.J., manufactures edge-based intelligent video analytics sensors that they say require several DSPs to be truly effective.
DSPs are programmable chips that function like minicomputers, Ramsey explains. They are not preprogrammed like application-specific integrated circuits (ASICs). She compares ASICs to CDs from record companies that can only be listened to and DSPs to rewritable CD-ROMs.
An advantage Ramsey cites of DSPs is that they are small enough to be placed in cameras or on a PCI card that will fit in some recently installed DVRs to upgrade their analytic capabilities.
Products that employ a frequency hopping technique don’t operate well with other manufacturer’s products in the same band space and yield a much lower aggregate data capacity.
Tips to Avoid Radio Frequency Interference
Radio frequency (RF) communication technologies provide a cost-effective means to deploy wireless data transmission across a variety of geographic and weather conditions.In the United States, the Federal Communication Commission (FCC) defines the rules and regulations for the use of domestic data telecommunications using the RF spectrum. The agency has devised a simple, two-pronged approach to assign and regulate RF spectrum: licensed versus unlicensed.
Telecommunication giants such as AT&T, Verizon and Sprint pay billions to license their own exclusive slice of RF spectrum at the appropriate frequency for the devices being supported.
For example, by paying for the 1850-1910 MHz spectrum in Dallas, Verizon can advertise that their GSM cell phone has a high probability of success working in that location when a call is made.
For this reason, the licensed spectrum is a hot commodity. Many companies bid on the spectrum at auction whenever available. Once secured, the telecom firms vigorously defend their RF “real estate” to protect the quality of service provided to their customers.
Furthermore, in the United States with millions in annual leasing fees at stake, the FCC is strongly motivated by these telecom giants to ensure that others not paying the license fees do not encroach upon the spectrum. Violation of licensed spectrum can exact heavy fines.
By contrast, the unlicensed portion of the RF spectrum requires that manufacturers submit their radio products to rigorous testing to receive an approval to sell their products operating in a specific unlicensed band.
Moreover, any modifications made to radio products by an end-user typically void the manufacturer’s FCC approval and could subject the user to legal action. Such modifications often include installing a wireless system with uncertified antennas or use of external power amplifiers to increase the radio transmitter’s output.
As a result of the dramatic increase in the use of wireless technologies over the last five years, the unlicensed radio spectrum is becoming overcrowded in urban areas.
Three factors should be considered when determining the appropriate spectrum utilization for maximum RF interference avoidance in this crowded wireless world.
The first is to make sure the preferred spectral architecture is a fixed-frequency, non-overlapping channel model, whereby the RF band is divided into smaller subsections or “channels.” Under this model, each non-overlapping channel supports its own discrete communication link. This channelized approach yields maximum data capacity of the band.
By contrast, products that employ a frequency hopping technique, whereby the entire band is occupied by a single transmission device, yield a much lower aggregate data capacity for the band, because each hopping radio causes interference to all new and existing band users.
A second consideration is using a system with a channelized radio transceiver with adaptive frequency agility (AFA). AFA is a useful feature in crowded RF environments, because it provides the wireless system with the ability to sense an increase in the noise level and relocate to a quieter channel automatically as a result of interference in a specific channel within the band.
AFA allows the radio to be perpetually self-healing as it adapts to other spectral users that may enter the band in the future.
Finally, when possible, use directional or sector antennae to focus the RF energy at the target to avoid transmitting or receiving undesirable RF pollution.
By only sending and receiving the signal in the preferred direction, you are able to improve your own transmission success rate while simultaneously helping your neighbors, because you are less likely to pollute their RF air space. —Contributed by Ray Shilling, vice president of sales and marketing, AvaLAN Wireless, Mountain View, Calif.
Sidebar: Receiver Gains UL 864 9th Edition Listing
Bosch Security Systems Inc., Fairport, N.Y., received listing for UL 864 9th edition for the company’s Conettix D6600 communications receiver/gateway and D6680 Ethernet network adapter. The listing validates that the products remain up-to-date with current revisions of the National Fire Alarm Code, NFPA 72.The listing provides dealers with expanded sales opportunities for the D6600, an integrated receiver for telephone, network and wireless alarm communication technologies, and the D6680, a two-channel network adapter that connects the D6600 to an Ethernet LAN/WAN.
The D6600 is designed for large monitoring centers. It features processing power for alarm and event data transmission over IP-based networks or through traditional PSTN networks. The D6600 can receive up to 32 phone line signals simultaneously and support up to 3,200 network accounts with the D6680 network adapter.
