Intelligent Video Surveillance That Operates on Cellular Networks

[Ana Monsanto]

The IVM™ 3G, a Mobile Intelligent Video Surveillance, which operates in any Environment was introduced for the first time at ASIS. This unique device can be used for deploying intelligent video surveillance in any urban environment, such as city-wide surveillance, covert surveillance, mobile intelligent video surveillance, and fixed installations. It utilizes existing cellular infrastructure for communication. It also accepts any analog video input including color, black and white, low light, day night, and thermal video inputs providing IP video compression and video analytics instantly at the same time providing 12VDC output to power any fixed camera.

The IVM™ 3G is contained in an IP67 enclosure enabling deployment in any outdoor environment. It is no longer restricted by the reach of wired networks, the IVM™ 3G can operate on 800/900 MHz EV-DO or 850/1900/2100 MHz UMTS/HSDPA/HSUPA cellular networks. With intelligent video analytics, the IVM™ 3G can conserve bandwidth transmission of video keeping video traffic minimal for bandwidth conservation. Video can be streamed on a tripwire or an object left behind event. Alerts can be triggered on virtual tripwire, object left behind, camera tampering, and many more events. Events and video are transmitted through HTTP Web Services offering the most scalable deployment through diverse networks and systems. The IVM™ 3G is managed through the MIVS™ Manager specifically designed to manage wireless video and wireless nodes for event monitoring, live video and recordings. Optional accessories include a battery backup and solar panel kits.

[Walt Charleston]

Interesting. We have been doing this for years. From the old ISDN and dialups ---yes dial ups-- through the old CDPD techniques to the EVDO Rev C and wireless mesh networks. We have even used P2P laser links as well as millimeter wave RF. The key is video intelligence at the camera level. You don't want to have to rely on a wide area wireless network for recording data. Thats the biggest mistake of most installs. A smart camera on a smart and robust network is what is required. You really only need bonded Rev A networks + smart cameras for excellent results---for most investigations. We have even enabled Wi-fi hotspot (with admin approvals) servers to act as virtual clients to smart cameras. This allows 30 FPS at 3.1 Megapixels (h.264 or DCT)---global shutter. We use MeOH Fuel cells to recharge our batteries. This is more covert than using solar panels especially wih bad insolation numbers. We are actively working on smart vector MP ip cameras that can produce virtual +/- Z plane images (3D). Our Ai suite includes over 12 functions, selectable by user on the fly.

Good luck! Our enclosures are submersible in icy salt water or boilng water. Great for High Def HAZMAT recovery.

[Walt Charleston]

BTW we don't need a van or other large enclosure for our gear. Each camera, 940nm IR illuminators, and ancillary equipment incl batteries and fuel cell , routers switches and TXRX 2x weighs less than 5 -10 pounds max depending on mission. Each "go Kit' is 99.5 + % autonomous. "Set it and forget it". NO BABY SITTING! All covert and effective.

[Walt Charleston]

Wow, I almost forgot, IR illuminators. We have one that is continuous wave at 940nm with a continuous duty cycle of 15 hours per day. Our elliptical beam pattern can light up an area 30 feet wide by 20 feet at over 600 yards. Its so "bright" that we have to stop down the low QE CMOS MP cameras to 50% at 600 yards from the target, that's how strong reflection we CAN get (usually objects that are naturally green--like green plants can also have pigments that reflect at near IR. We found that if you implement multispectral IR emissions from 880nm-1200nm, you can get better results. Of course we have experience with the lower power 940nm 1 watt laser illuminators (flood) that most are familar with. They do consume lot of energy, but we can project wide area emissions which are effectively intense for our cameras to 10-12 miles for detection of vehicles and humans. We don't need a microbolometer for these surveys. BTW we meet all FDA-BRH, EPA, OSHA standards. These devices are NOT for sale and are for investigational and scientific use only within the CONUS.

[Walt Charleston]

I missed the last sentence---a question.

"The question is- What could the drawbacks be from a device that requires the user to pay a fee for utilizing the network?"

We would NEVER do that! Who is responsible for quality YOU or someone else? If YOU are responsible, never let them know how you set up the network (they can get an idea from your billing). We solved this problem by using VPNS, as well as MULTIPLE DDNS's for each camera. A minimum of three per camera should be used---UNLESS you are using dedicated ip addressess. Most DDNS's usually only ping once per 5 minutes at the fastest. We have solved the "lost ip" problem by hammering the server with at least 3 requests per 5 minutes, sometimes more. But there are new issues. The Terms of Sevice of some wireless carriers (AT&T mobility) restrict the use of automated DDNS. They are looking for the DNS port pinging that occurs during the process. They shut down cameras all the time.

We solved the problem by using encrypted VPN's and dedicated ip's on the networks that don't care. Why VPN? Look at the TOS of most wireless carriers. ALL OF THEM have regulations about sending voice over data cards. They don't want you setting up hotspots for VOIP, which could cut into their revenues. Why voice? Voice is absolutely a requirement. You can learn MORE from "voice"/audio than video--in many cases. Why? Because audio bounces around and you can pick it up 360 degrees, AND some information produced at audio is more important---sounds of vehicles, garage door openers, etc etc. Our RULES -->We NEVER EVER use it for speech quality interception of human voices. That is treading on legal issues---KAtZ 1961 Sup Ct on R.E.P. Once you tunnel and encrypt using decicated ip addreses or multiple DDNS's you start having control. You didnt mention a very important issue. That is commercial wireless networks are not very robust. When they fail, what do you tell the customer? We never ever tell them how we do it---its none of their business. Our fees are high, and they don't ask---when we NEVER give them excuses or reasons for any camera failure. I can't believe that has never happened. The way you solve that problem is through redundancy. Lets say carrier A has taken down an MTSO/Switch/hub/tower for maintenance at 1 am---happens all the time. Most providers conduct their maintenance/ repairs late at night, so they won't interfere with day users. Here is the problem, lets say they switch the site to another network temporarily---if they have the capacity, this will happen. Okay, in the few seconds--minutes---hours of "no network" what do you do? Oh you didnt know this---it happens all the time ---read your TOS---"subject to availability of the network". If you use a "back up system" ON A DIFFERENT TOWER, you can tunnel back through a separate VPN to get your camera. Our new systems do this automatically. Why do we use different towers? Because if power goes out at a TOWER with multiple providers, all the providers will drop the network in that cell/ microcell . In those cases, we use disguised Yagis which are within our local wireless mesh at 5.8 Ghz or so to reach a wireless cell carrier miles away. We can hit to the 19 mile nomograph using elevated yagis at the periphery of the target area. None of these engineering designs are trivial. Even with multiple services and multiple towers, you can lose a camera---sometimes there are problems in whole regions---usually at night and on the weekends. Bottom line the CONUS wireless cell and PCS data networks are very patchy and are not reliable for important missions. We must use them, since they are relatively cheap and easy to use, but they are NOT the final answer. We have been watching EVDO Rev C (Qualcomm) and others which may improve matters.

Bottom line answer--we never bill components of the system. We have ONE NON- itemized bill---for services. If they don't like it they can go somewhere else. "Under promise and over deliver" makes everyone happy. Customers do NOT want to hear excuses. So don't give them excuses. Cover all your bases and charge accordingly. The price for excellence is high. Most ip camera systems on a public shared network will die, its only a matter of WHEN. Our PORTABLE systems have run continuously for 9 months unattended. Why only nine months? a network switch knocked out service to a REGION. Why tell your customer HOW you built the system? Do you want to put yourself out of a job? You will, if they hire engineers to figure you out. It all starts with a bill of materials and services. If you don't have them , it will cost them big bucks to reverse engineer your system. prices? Our 10 watt laser IR illuminator will cost them $40-80k/year. We use OUR OWN design cameras from the ground up--no other camera is like them. They are Never for sale. We sell what the camera can do.---ie services. We use personal services contracts, and NEVER competitively bid on a job ---why tell a competitor how to do it?

Customers?---they come to us.

[Walt Charleston]

What? I read that your IVM 3G is for composite cameras? Help? We have found ONLY 3 different composite type cameras are useful for remote surveillance. Two of them were designed for our specification. One of them is SWIR-VIS at 640x480---CONUS only. The other two are low visible light down top .0002 LUX and high QE to 1200nm. All other cameras are either modified machine vision USB cameras or network cameras we have either modified or had built for us. We are trying to get away from the old LOW resolution cameras. Our favorite for remote surveillance are at QXGA (2048x1536). Megapixel ip cameras with a minimum of 1280x960 ---not true SXGA with GLOBAL electronic shutters AND progressive scan. Most MP ip cameras make "pretty pictures" in bright light. What happens in poor visible light or no light? Low QE and rolling shutters are characteristics of CMOS sensors typically used in security type network cameras. Instead of "pretty colorful movies" you get "pretty colorful frames". Color does provide SOME advantage in bright controlled ambien lightng, but RESOLUTION is KING. Try InGAAS. When coupled with good QE to 1200nm---or higher? and with global shutter and progressive scan. Watch out!

[Walt Charleston]

Look at these links: They have an autonomous micro NVR using WWAN and megapixel IR cameras! It works unattended. No one has to stare at a screen for hours,days,weeks while drinking coffee and munching on sandwichs (or donuts-your choice). No requirements for a van, SUV, or anythng larger than a volkswagon. You can set it---and forget it. You can watch it from your office or lounge while you are having some "cool" ones. You don't even need to do that. It pages you with Twitter and auto records when the sytem is activated. It also verifies with a hot mic and GPS coordinates. The software analytics is by vidonetics. The whole shebang will run on Linux embedded. We are working on an SOC to make it even smaller! There is NOTHING like it anywhere. The system is compatible with the



http://www.megarecorder.com/8.html

http://www.megarecorder.com/index.html