The Digital CCTV Enigma
by Phil Mailes
Not a month passes these days where no articles appear in trade magazines detailing digital CCTV. Very few of these actually discuss the merits of Digital CCTV technology over Analogue technologies in layman's terms. Most articles seem to focus on the technology and rely on jargon, misinformation and hype, which is weighted in the favour of a manufacturers product, rather than focusing on the more important aspects, such as ease of use, cost of ownership, operational considerations etc.
Similarly manufacturers representatives are relying on user ignorance to hoodwink customers, rather than rising to the challenge and educating end-users and thus allowing the customer to make an informed decision.
As an example consider this scenario, a potential user of Digital Video Recorders (DVR's) had requested two quotations from two manufacturers, the requirements specification called for 32 cameras recording simultaneously for 30 days. The resultant quotations varied significantly, one was 30% more expensive than the other.
The end-user was about to place an order with the first manufacturer, hereafter called (ABC Ltd), as they felt they were getting a bargain, fortunately however the second manufacturer, hereafter called (XYZ plc), requested an opportunity to explain their offering as they felt sure their solution could not be that much more expensive.
Evaluation showed that XYZ plc had quoted 3 x 120Gb hard disk drives to achieve this requirement; they had also quoted a refresh rate of 3.125 frames per second. ABC Ltd had stated they could store 32 cameras for 30 days on a 20Gb hard disk drive. This was impossible, as current technology would not allow this at the same refresh rates as quoted by the first company.
Careful scrutinisation of the quotation revealed small print at the bottom of their specification stating that their storage estimate was based upon a recording at a rate of 1 frame every 2 minutes. In a security environment, a lot can happen in a 2-minute interval and so effective historical analysis would not have been possible utilising this solution.
The ABC Ltd was then asked to revise their quotation based upon a refresh rate of 3.125 frames per second, they did this and it was found that their digital video recorder was not capable of supporting hard disks with sufficient capacity to meet the specification. Consequently ABC Ltd was forced to utilise an external archive device, this drove up the price of their offer to twice that of XYZ plc's original offer.
This is typical of what we call "spec sheet selling" which seems to be prevalent in the industry at the moment, and seems to follow the introduction of new technology in any market where a buyer can be hoodwinked by misinformation. This format dictates that in order for an industry to mature, customers have to learn from their mistakes, which is an expensive option.
This article is intended to dispel some of the myths and cut through the hype surrounding digital video recording, and provide guidance to those considering adopting this exciting technological advance to enhance their operations.
Advantages of Digital Video
The advantages of digital video over analogue are becoming established and now known. To reiterate video images can be saved on a hard disk by using Digital Video Recorders (DVRs), instead of utilising videotapes that degrade with use and have many moving parts. Digital video techniques enable enhanced and more efficient search facilities. It is easy to search the video imaged stored on the hard disk drives via time and date criteria. Additionally multiple copies of the images can be produced without degradation. Transmission of images over LAN/WAN technology enhances the ease of distribution, and the adoption of video player software into operating systems allows the viewing of video clips without proprietary viewing software.
Sending Video Images Over the Network
There are several factors to consider when planning the deployment of digital video over a network. The available bandwidth of the computer network is a key factor in all network video transmission. If bandwidth is limited, the amount of information being sent over the network must be reduced, either by longer update (refresh) intervals or lower image quality. The factors that affect the file size are:
· Compression Algorithm
· Refresh Rate
· Resolution
Compression Algorithm
Compression is the amount of redundant video information that can be stripped out of a video image before storage and transmission. The higher the compression, the more information is lost. There are many compression algorithms available such as H263, Wavelet, MJPEG, MPEG etc. For this article it is not intended to justify the merits of each different compression technology as there are many manufacturers shooting it out in the trade press as it is. However, it is worth explaining that there are basically two methods of compression to enable the transmission of high-quality video over a network. Either a complete image is transmitted every time (JPEG, Wavelet) or a reference image is transmitted and updated little by little (MPEG, H263) until a major change requires that a new reference image should be transmitted. Depending on the application, either method is suitable.
The real issue is what does this mean to the user of a system? Different compression techniques and rates will affect the storage requirements and the bandwidth utilisation. This, in turn, will affect the capital cost and cost of ownership of the system. As in all walks of life, there is always a trade off to be made as we cut the cloth to suit our budget.
Refresh Rates
How often the image is updated is known as the refresh rate and is generally quoted in Frames Per Second (fps). It would appear that when selling Digital CCTV systems, salespeople are quoting Frames Per Second as being the benchmark of a superior system. To confuse matters even further, manufacturers are quoting in various formats. Some quote Frames per second (25 Frames per second provides real time live video), others will quote Fields per second. There are 2 Fields per frame and so many will quote 50 Fps leaving the unsuspecting buyer with the impression that this unit has twice the refresh rate of a 25 frames per second unit. Other manufacturers will quote pictures or images per second. The refresh rate has nothing to do with the quality of the image; it only affects the smooth movement of objects within the field of view. All of this obviously disguises the true picture, it is time that the industry got together and utilised a common format to assist buyers.
Although 25 Frames per second will provide real time live video there is a high cost to pay in utilising such refresh rates. The more frames that are recorded the larger the file size. Large file sizes will require more storage devices and utilise increased bandwidth on a network. Studies have shown that even though many units available on the market are capable of recording at 25 frames per second almost 75% of the installed base of such units are throttled back to a level that provide a good compromise between usability and cost overhead. 3.125 fps seems to be the standard that most companies are utilising, this provides adequate surveillance whilst reducing storage requirements, and minimising the impact on a network. However, most potential users are still requesting 25 fps within specifications, without realising the impact, as this is the information they have been force fed by unscrupulous suppliers.
Just in these paragraphs alone we have been banding about figures that to the uninitiated, are likely to confuse and so lets put some of this into context. For a start, most users of analogue CCTV systems tend to use a 16 way multiplexer and a time lapse VCR set at 24 hours, this combination will provide a maximum of 0.54 frames per second. Therefore, by increasing the refresh rate to 3.125 frames per second, customers are getting a 500% increase in the refresh rates they are used to at present.
Additionally, we tend to think linearly rather than logarithmically, thus we assume that a refresh rate of a system operating at 25 frames per second is twice as good as a system operating at 12.5 frames per second. Although in theory it is, our eyes and brains cannot actually see twice the difference. Try counting to 25 or even 12 in a single second to understand how difficult it is for us to process this amount of information. Indeed, many of the early cartoon animations were manufactured at 15 frames per second. This allowed the animators to draw 15 individual cells rather than 25, thus reducing the time required to generate, and the cost of producing an animated film. When viewed we saw these as real time moving pictures as the receptors in our eyes could not process the information quick enough for us to see the difference.
When evaluating a system, it is important to analyse the differences between the refresh rates. This should be undertaken by visiting the manufacturer and comparing several cameras all set up at different refresh rates and all looking at the same source. When doing so ensure that there is movement in the field of view, which is likely to represent your actual conditions. Having done this simple test, you will be able to settle on a refresh rate that meets your operational requirements. You should also request a quotation based on two scenarios, one at the optimum refresh rate and another at a rate which you would be prepared to accept. Set some parameters for the quote i.e. 32 cameras, all recording at x fps with storage for 30 days.
Resolution
Resolution is the measure of image quality and is measured in pixels. Remember that the higher the image quality the more data. Images with a lot of data will take longer to transmit, will utilise more network bandwidth and will require more space on the hard disk to store. For example a 320 x 240 pixel image (approx VHS quality) is ¼ the size of a 640 x 480 pixel image (approx S-VHS).
To give you some idea consider the two pictures below, both were taken using a Minolta Dimage F100 digital camera. One was captured using the following settings: 2272 x 1704 pixels (High Resolution) with no compression, whilst the other was captured using 640x480 (Medium Resolution) with high compression. The first file was 11.352 Mb whilst the later was 90Kb, can you spot the difference?
Fig. 1 Fig. 2
As can be seen there is very little difference between the two images however if you were to blow up the two images the following would be the result:
Fig. 3 Zoom In using Fig1 Fig. 4 Zoom In using Fig2
Here the results of using the lower resolution can be clearly seen however when using CCTV cameras if they are positioned correctly the need to zoom in is negated and the overall cost of ownership of the system can be drastically reduced.
Other Considerations
Hard Disks
Digital CCTV recorders are basically a PC with 1 or more hard disks included for the storage of the video images. As detailed above, dependant upon the compression algorithm used, the amount of compression selected, the refresh rate utilised and the resolution adopted the storage requirements will differ widely. This will mean that more hard disks will be required as the specification is increased.
Hard Disks are now quoted in terms of the number of transactions written to them before failure, rather than the length of time before failure. Recording video is a very transaction intensive process, as information is continuously being written to the disk/s. The law of averages dictates that the more disks utilised for storage the more likely it is that one will fail. As Digital CCTV systems mature in the market place we are starting to see that the hard disk is the most common point of failure. Given this scenario it is important to select a DVR with the largest capacity disk drives available as it is better to have fewer large disks than more small capacity disks. Additionally DVR's with easily removable disks that can be removed and replaced from the unit without having to strip down the unit should be considered.
Some manufacturers will treat 3 hard disks in a digital recorder as a single volume, thus if one disk should fail all of the stored information is lost on all 3 drives. Other manufacture's treat each disk as a separate volume, therefore if one drive fails only the information on that drive is lost, whilst the information on the others remains intact.
To overcome the disk failure issues many manufacturers utilise RAID storage devices, these will allow for disk failure without loss of video information.
Archiving
From the information detailed above it can be deduced that DVR's can only store a limited amount of information. The number of hard disks that can be mounted within the units limits this. Even though some DVR's can accommodate up to 1 Terabyte of hard disk space this will soon fill up. Most DVR's are set up to have two areas of storage, the first being for event driven "locked" video and the other being for continuously recorded video. The system is continuously recording, or caching, the video and when an event occurs it will "grab" the video associated with that event from the cache, according to pre and post alarm video configuration, and place it in a locked file. The locked files cannot be overwritten by the system unless the operator manually purges them. The cache is overwritten on a first in first out basis. The ratio of locked video space to free space on a disk is dynamically allocated and so after a period of time the disk will fill up with locked video.
When this situation occurs it is necessary to purge the files, in which case all of the information is lost or clips can be archived. Should the user wish to access this information at a later time it will be necessary to archive the video. Similarly, where there is a requirement to store large amounts of continuous video, it too can be archived, either automatically or manually.
Information can be archived to two types of media, either tapes (video juke boxes) or arrays of disks. Tapes tends to be slow in terms of information throughput and selection and loading of the tape containing the relevant clips and contain many moving parts and so require regular maintenance, but they are a cheaper solution than hard disks. Disks on the other hand offer virtually instant access to video clips but at greater cost. As Hard Disk Drive costs are continually coming down and sizes are growing this difference in cost is becoming smaller and smaller.
Most DVR's rely on these being located next to and attached to the field DVR's. If for example the customer has 3 digital video recorders they would require 3 archive storage devices.
Smart manufacturers are starting to deploy central archiving capabilities meaning that video images from remote DVR's are transmitted across the network and stored centrally on either tape or disk. The likes of IBM have been doing this for pure data for many years, not only for simplified storage and retrieval, but also for disaster recovery. They are now offering their considerable experience and technology solutions to the industry to the point that they can store petabytes of video information and allow access to many simultaneous video streams thus enhancing retrieval and analysis of video clips by multiple people.
Bandwidth Utilisation
Bandwidth utilisation is the amount of the "pipe" taken up by the video information being transmitted over it. The frames per second, the algorithm and the compression ration will all effect the utilisation. When planning a system, the buyer should get some estimates on the utilisation, based upon the scenarios they have selected. Some systems are designed to be good network citizens and take bandwidth utilization into consideration from the outset, whereas some systems are designed to be analogue VCR replacements. There is room in the market for both products, but they are two separate solutions for two separate needs. A networked Digital recorder can be used as a VCR replacement but would tend to be expensive. A digital VCR replacement, would struggle to effectively sit on a Wide Area Network.
DataBases
Another issue that is rarely discussed is the utilisation of a central database. Many manufacturers produce units with a database included, this is fine for a site with a single unit, however if a site has two or more units then it will need two or more databases i.e. 1 per unit. The utilisation of a central database means that all recorders can be configured from a single computer, it is possible to run a report or get status, event or configuration information from every unit in the system from a single computer. It means that the management of events can be accomplished from a single location including the archiving of multiple recorders back to a single archive server. Thus a single database reduces management complexity, increases usability and therefore reduces the overall cost of ownership.
Integration
Within the industry, there still seems to be two distinct camps: that of CCTV and that of Access Control, neither of these two camps really gets it. The fact is that it is really just data acquisition.
CCTV manufacturers are merely moving along producing a plethora of digital VCR replacements. Access Control manufacturers are still producing their products with an RS232 port to send ASCII commands to the CCTV system, to drive the system upon an event.
Very few have really embraced the two technologies and produced a single holistic solution. Even as a result of the recent spending spree on company purchases within the industry, we have seen holding companies purchasing both Access Control and CCTV manufacturers, but we are yet to see any true integration of the two technologies.
For example, the benefits to the user of a single platform that would allow programming, control and monitoring of both the Access Control and CCTV system from a single screen would be considerable. Consider the cost savings of such a system utilising a single user interface i.e. one lot of training, a single database to administer, a single application to upgrade, duplication of servers and workstations eliminated, etc etc.
Historical analysis utilising two discreet systems, albeit interfaced, can be cumbersome. To track an event, information has to be reviewed on an Access Control system and a report generated, giving the date and time of the event in question. An operator will then have to review that date range on the Digital CCTV system to access video clips. As the two separate systems operate on two separate PC's, the clocks on both are likely to drift out of synchronisation, for example 13:30 on one system, may be 13:50 on the other, further hampering investigation and adding to user frustration.
True integration will provide the ability to attach video clips to alarms and events on a single workstation, thus an operator should be able to click on an alarm and launch the video clip of the event that caused that alarm. Using a single database on a single server means that the times of the alarm occurrence and video clip will always be in sync.
Several manufacture's have launched Digital Video Servers that are capable of running other applications, for example, it is now possible to have one PC (Video Server) running Digital Video, Access Control, ID Management, Guard Tour, etc. therefore again lowering the cost of ownership.
Conclusion
It can be seen from the above that the decision to embrace digital CCTV is a wise choice, however, the selection of the technology is about what works for you in your environment and within your budget. After all, video is a visual media and it is a simple task to view the resultant results of the various factors that affect digital CCTV and determine for yourself the best solution rather than have inappropriate and thus potentially expensive technology forced upon you.