Introduction

In this article we are going to look at some of the technological trends in DVR technology, where this is leading us and what the practical implications are in choosing a system. Like any modern technological product, it helps enormously to have a basic understanding of the fundamentals to help you choose a product or solution which best suits your needs and your budget. It will also help you cut through some of the jargon used in the industry and to filter the information some over-eager sales people might throw at you. Luckily there are many expert consultants and sales professionals out there, and hopefully this article will help you tell the difference.

The Fundamentals

The next few sections look at some fundamental technical factors which are essential for specifying and choosing a DVR.

What matters most of all is the quality and rate of images which are recorded, plus the evidential robustness of the total system. Ease of use and convenience of data export are also vital and we will return to these two issues later. Of far less importance are a whole range of add-on features such as behavioural analysis and scene analysis, which add to costs but are of limited practical use for general-purpose applications (see “Active Systems” below).

Note that some systems can have excellent display quality, but poor playback quality. For a DVR, the playback quality is obviously the most crucial issue. A number of factors affect the actual usefulness of the recorded image, with the most obvious being the sharpness and clarity of the picture and the number of pictures captured per second. The sharpness and clarity are affected both by recorded resolution and by image compression methods [See sidebars for detail on these topics]. Other factors which affect picture quality include contrast, colour accuracy, picture noise and picture stability. These are factors which are difficult if not impossible to divine from a spec sheet, so the old saying “seeing is believing” could not be more apt for these products.

Fields, Frames And Capture Rates

As we shall see later on, there is a significant trend towards higher and higher frame rates as technology progresses, up to so-called “real-time” or “full-motion” video. Of course, at what rate a video signal actually becomes real-time is somewhat arbitrary, but convention has it that it means 25 frames per second, as this is the maximum frame rate for normal video (or 30 frames per second for the American NTSC).

Let’s be clear about the difference between fields and frames, as the significance and effect of the interlaced picture system usually not properly understood by DVR buyers.

The video signal standards used in CCTV are based on TV standards. A TV picture is built up from 625 horizontally scanned lines, of which only 576 are actually used for the picture. A video frame is composed of two half-frames or fields, with the first field being the odd-numbered horizontal lines (1, 3, 5 etc.) and the second field being all the even-numbered lines. A field consists of 288 horizontal lines, whilst a frame consists of 576 horizontal lines, giving rise to the resolution figures discussed earlier.

So an odd field and an even field are put together form a frame. This is called interlaced video, and can blur moving objects because there is a 1/50^th sec time interval between the capture of each field. Most DVRs (and indeed time-lapse VCRs) cannot properly handle full-frame interlaced video and so capture fields only.

(For a full and detailed explanation of interlaced video and how this affects the performance of digital CCTV systems, please contact the author by e-mail).

Thus most DVRs quote their capture rates in fields per second (fps), as only single fields (i.e. half vertical resolution) are captured. Usually this figure is quoted as the total capture rate (i.e. across all channels) and not the individual channel capture rate. Other DVRs are capable of capturing full-frame video and therefore quote their rates in frames per second (also, unfortunately, abbreviated to fps). All of this is OK, but the customer must beware of how systems are specified, and what those specifications mean. It is important that you understand what the maximum capture rate per camera channel is, and whether this is a field rate only or full-frame capture.

“Specmanship”

A word of warning : A number of more devious manufacturers quote high resolution and high frame rates in the specification sheets for their DVRs, but neglect to state that you can only have one or the other, but not both, or that this cannot be achieved simultaneously across all camera channels.

Under this heading, it is a good time to discuss CCTV tender specifications, which have become something of a joke in the industry with the advent of DVR technology. We can all recognise when a specific manufacturer has been “favoured” by specifying so tightly, down to irrelevant or unimportant details, that only one machine will fit the job. Is this giving the end-user the best choice or the best deal ? Or too often, someone has done a cut-and-paste job from several different DVR spec sheets, not realising that the resulting specification is actually self-contradictory.

If you are an end-user, beware of letting your adviser specify more detail than is necessary to meet your needs. Also make sure you are comparing apples with apples when comparing quotes, as one manufacturer may have quoted to the letter of the specification, whilst another may take great liberties with it, and not be quoting what you need. When you find out, it may be too late. This is an increasing and unfortunate trend we have observed in the last couple of years.

Storage

We will look at storage trends in a minute, but just a quick look at storage fundamentals is worthwhile. Nearly all DVRs record onto hard disks. This is the perfect medium for video storage – it’s fast, efficient, relatively cheap and disks are getting cheaper and bigger by the day. There are some larger systems that still use digital tape juke boxes for back up and massive storage, but this really a Heath-Robinson way of doing things now.

Having said that, the hard disk is the weakest link in the DVR chain and the most likely component to fail. Careful design of the DVR is necessary to mitigate effects which stress hard disks (power, temperature, vibration, disk filing system type).

Note : this is one reason why so many low-end DVR systems, especially those from the Far East, provide a removable cradle for you to supply your own hard disk. If the disk fails – it’s your problem ! Responsible DVR manufacturers who take a more serious and professional approach to support matters handle things in a different way . . . . .

Disk drive technology is advancing rapidly, not just in capacity, but most importantly in reliability, power consumption and interface type. Until relatively recently, the choice was between IDE drives (also knows as ATA) and SCSI, with SCSI being very much more expensive, but designed for higher duty-cycles. A new type of drive interface, Serial-ATA or SATA is now available. SATA will eventually replace IDE entirely. SCSI disks are now rarely used for DVR applications as they do not have the capacity of IDE and SATA drives. The largest SCSI drive available today is 180GB, whilst IDE and SATA drives are up to 400GB, with 500GB drives just round the corner, so storage just keeps on getting cheaper and cheaper per GB.

There is also a trend at the higher end of the market towards external storage systems linked through SCSI, Firewire or Ethernet links, but using arrays of low-cost IDE or SATA drives. However, very often we see such systems specified with RAID5 being an absolute must.

Another word of caution : too many specifications are written without a proper understanding of storage systems. RAID (Redundant Array of Inexpensive Disks) is a term used for using a group of disks together for either increased speed or redundancy against disk failure and there are several RAID formats. Space does not allow a deep discussion of these, but the commonly-specified RAID5 is worth more detail.

In a RAID5 configuration, the data are spread across the disks in such a way that the loss of any one disk can be tolerated, and the array rebuilt with the original data when a new disk replaces the faulty one. However, rebuilding can take considerable time and the performance (i.e. maximum recording speed) of the array can be vastly reduced during this time. However, if a second disk were to fail before the first one was replaced (and remember that we are talking about the real world where Sod’s Law rules), then EVERYTHING is lost. Were two disks to fail in a normal non-RAID disk array of, say eight disks, then only 2/8^ths or one quarter of the information would be lost.

For applications where the recorded data is potentially sensitive enough to consider using RAID storage at all, then the best option is RAID0 which is effectively mirroring, or recording everything twice on separate disks. However, even with such a system, if the DVR power supply were to fail for example, then the recording stops, RAID or not. The most effective and secure way of recording for high-sensitivity applications is dual-redundant recording systems, where everything is duplicated. Believe it or not, this can often work out cheaper than using expensive external RAID systems.

Trends

With the basics out of the way, and having touched on storage trends, let’s widen this out and see where DVR technology is heading.

The original reason for using time-lapse recording in video surveillance was to squeeze a number of cameras onto one tape and to stretch the tape out to cover 24hrs, so it would only have to be changed once per day. Early DVR systems simply mimicked the VCR/MUX combination. In recent years, however, the image resolution, capture rate and storage capacity of DVRs have improved dramatically.

Today, the only reason for not recording all cameras at the highest resolution in real-time for 31 days are considerations of cost , but there are several systems around now which are capable of this and the trend in moving towards this goal is strong and inevitable. The larger disk drives available now enable multi-disk systems to store very high quality images at high frames for reasonable cost.

There is a definite trend, therefore, towards higher image quality at higher frame rates. This is a very good thing for the CCTV industry, as for too long the results produced by the average CCTV system have been virtually useless for the reliable identification of people and vehicles. Such identifications are, after all, of prime importance in solving crime and should be a fundamental expectation of a high-quality CCTV installation.

At the other end of the market, lower end JPEG time-lapse units have become a commodity item. Prices have fallen to such a level that they are rivalling the cost of a VCR a of a few years ago and are almost becoming disposable. This is good news for small shops and other applications where anything other than a quad-splitter and a VCR has previously been too expensive.

The danger is, of course, that you get what you pay for, and many installers are starting to use these cheap systems in totally inappropriate applications where there are considerations other than price. Data Protection Act compliance, image quality, reference clocks, ease of data transport, remote access, integration with other systems, audit trails for evidential robustness and of course, longer-term technical support are all matters which should be considered carefully.

PC-Based versus Non-PC Based DVRs

DVR systems can be classified into two types : those which are based around standard Windows PC systems (in various guises) and those which are purpose-designed dedicated hardware systems. There are some significant advantages to dedicated hardware systems in terms of cost and reliability, but what is lost is often some ease of use (due to the lack of a graphical interface) and flexibility. There are an increasing number of extremely low-cost dedicated hardware DVR systems now flooding in from the Far East which address the small-system end of the market. This trend will continue.

At the highest-end of the market (high-quality, high resolution at high frame rates), DVRs are all PC-based systems with only one or two very notable exceptions. It is likely that, over time, more of these specialised dedicated hardware systems will be developed which do not make any compromises on performance and which address the high-end of the market.

Ease Of Use and Evidential Issues

You would expect that with DVR’s now being a mature technology that they would have become much easier to use. In general, unfortunately, this is not the case. A feature “arms-race” seems to be happening as some manufacturers try to differentiate their products from the competition. These can make systems more complex than they need to be to operate.

On the evidential side, it is still amazing how many DVRs so not have a simple, straightforward and robust method of taking recordings off the DVR for evidential purposes. This is one of the most important features of a DVR and whatever method is employed should be thoroughly understood and approved by the user before purchase. As a minimum, any DVR should comply to the recently-released Home Office specifications for digital recording systems (http://www.crimereduction.gov.uk/digitalcctvleaflet.pdf).

Integration

DVRs are rapidly becoming the core of the CCTV system. Integrated PTZ control, alarm notification, site-maps, system management and even basic matrix functions are being built in to medium and higher-end systems. Whilst there is the obvious danger of putting too many technological eggs in one basket, and possibly ending up with systems which are too unwieldy for straightforward applications, the trend is there and unstoppable.

Integration doesn’t stop there. Some DVR systems integrate fully with POS (Point Of Sale or electronic till) systems, effectively allowing data searches and filters on till rolls to combat fraud by synchronising the video footage with what is being entered on the till. DVRs and Access Control systems are beginning to work together, providing a very powerful tool for keeping track of not only what access ID cards were used and when, but who used them (was it the authorised person, who was with them, what they were carrying and what did they do when they were in the controlled area ?). Other integration trends include Automatic Number Plate Reading (ANPR) systems and Radio Frequency Identification (RFID) tags. The latter would allow a logistics company, for example, to track their goods through their system and pull up video footage of each time a package passed a scanning point, thus allowing them to follow what happened to it from that point on, perhaps by moving to replay from other cameras linked to the system.

Many different types of biometric systems are being developed due to current global pressures and threats, whether real or perceived. In due course, these will be integrated with DVRs.

Integration is a key trend for DVR technology and for CCTV generally. It vastly improves the overall power of the total security system.

IP Video

IP video systems, or networked video systems, are seen by most people as one of the most obvious and major trends in the CCTV industry. However, there is still more hype than substance, and unfortunately it is right up there with image compression in being generally misunderstood.

The topic of IP Video systems has been the subject of a huge number articles in the Security Trade Press, and a few of them have even been quite good. It’s a vast topic in itself, so only a few salient points will be raised here.

IP Cameras : if you choose to use an IP camera, just remember that you will never have the vast range of choice of camera type that you have with normal CCTV cameras. There are a good reasons why manufacturers offer such a range of camera types for different applications. If you really need to get your video onto your network, it may be better to choose the right conventional camera and use an external codec or network server box.

To get large numbers of cameras streaming over a network requires compromise – are you getting the same resolution, image quality and frame rate that you would from a conventional system ? If so, that requires a very large network bandwidth and in many cases you may have to arrange to do it over your IT manager’s dead body.

IP Video is ideal for some applications – video can be made available anywhere on the network, can be recorded anywhere and played back anywhere. It is the ultimate in flexibility. Some IP video equipment manufacturers have spent a great deal of effort trying to convince people (including themselves, I suspect) that IP video is more cost-effective. For some specialist applications, it can be very cost effective and very good. For general applications this is usually not the case, and although it might be one day, even this is not certain.

If you need an IP video component in your system, ensure that it properly integrates with your conventional video systems. Specifically, your NVRs (Network Video Recorders) and your DVRs should work together and either be the same system or be 100% compatible and seamlessly linked. Your system operators should not have to think about two separate systems.

What will drive IP video forward in the future is the development and exploitation of non-video (i.e. very high resolution) digital sensor arrays, wireless network improvements and a new generation of IT managers who are in charge of all security matters, not just IT. Whether we think this is a good thing or not (because they are not security professionals), it is happening already. How many security managers nowadays report to the IT manager simply because of technology convergence ?

Active Systems

Finally, another, less significant, trend is the advent of ”active” or so-called “intelligent” systems. These are software-based systems which monitor and analyse the video from the cameras for various purposes such as detecting packages left behind, which direction someone is walking down a corridor, whether groups of people are gathering and so on. The trouble with these systems is that they are expensive, tricky to set-up, use and maintain, and their performance and usefulness is questionable. You really don’t want to have to call in an expensive specialist engineer every time a camera is moved, changed or added to the system.

Facial recognition systems could also be classified in this way. Such systems are really at the stage where ANPR was many years ago : a bit flaky. ANPR is now a mature, useful and highly reliable technology and so it will be, in time, with these more advanced systems, but they have a long way to go.

Conclusions

We are seeing a strong trend towards higher frame rates and better image quality as the technology improves, storage capacity increases and costs come down. IP Video systems have a part to play, but must be deployed into the right applications for the right reasons, and not just because someone has a product they want to sell you.

Some understanding of the technology is essential in order to properly specify and select the best DVR for your purposes. Beware of sales people trying to make your needs suit their favoured product, rather than helping you to match a product to your specific needs. Independent advice is always a good idea, but make sure it’s truly independent.

When choosing a system, concern yourself with the fundamentals : the recorded image quality, ease of use, evidential robustness, data transport, reliability, and not least the service and support available to back up the product.

Introduction

A hot topic in the CCTV industry at the moment is analogue vs digital recording. It is generally accepted that future CCTV systems will employ digital recording, but the question is how quickly will the industry in general convert to digital ? A case can be made that analogue VCRs will be around for many years to come, but the advantages of digital recording are now so overwhelming that analogue systems may disappear far more quickly than some think (especially VCR manufacturers !).

This issue is a crucial one for those making investments in CCTV now and the near future, and so we shall take a deeper look at the fundamental differences between analogue and digital systems for CCTV. You might say that as we are company selling digital systems our views are bound to be biased, but the intention is to be as objective and informative as possible and let the facts speak for themselves.

Scope

First of all, in comparing analogue and digital technologies, there is much more to consider than just the recording aspects - this is far too narrow a viewpoint and obscures many of the true advantages and flexibility of digital systems. Further, digital technology is leading us down a path which will result in quite radical changes, not only in system performance and capability, but in the very structure of the industry. Those who ignore digital CCTV systems over the coming years do so at their peril ! The benefits of all-digital systems will be so overwhelming that analogue systems may virtually disappear altogether. Having said that, some of the wider aspects of digital systems and where the technology is leading us must wait for a second article (in preparation). What I want to discuss in this article are the benefits and limitations of systems which are available right now.

Image quality

We would be the first to agree that a many people selling digital recorders talk a lot of nonsense about image quality. Just because images are captured and stored digitally does not make them better quality ! Both digital recorders and well-maintained analogue VCRs with new tapes are capable of recording high-quality images. Any recording system is only as good as the weakest link, and so often poor lighting, inadequate lenses, cheap cameras and shoddy cabling are responsible for the terrible image quality we see on so many systems. We know that many of us in the industry must cringe with horror when we see some of the truly awful CCTV recordings often shown on crime-related TV documentaries.

It has to be said though, that very often it is the VCR which is at fault. Not because VCRs are inherently poor quality, but because they are so poorly maintained. At many sites video tapes are used over and over again, sometimes so much so that you can practically see-through the tape ! To keep costs down (and often due to ignorance), cheap tapes are used, so the recorded quality starts off badly and goes downhill from there. Obviously a well-maintained VCR system will be much better, but maintenance is expensive and requires a rigorous routine - often defeated by the vagaries of human nature. The advantage of an entirely automatic digital system in this respect is that you will always get consistent quality with effectively zero-maintenance, no matter how often the system is used or how often the images are copied.

Tape management, hard disks and & storage capacity

Video tape management problems are not discussed in any detail in this document, but if you think the issues are trivial, then think again ! Just take a look at any CCTV trade magazine and see how many companies advertise products for aiding tape management, or talk to anyone who has been given the task of locating a particular recording !

The first question any prospective user of digital recording systems is "for how long can I record ?"

This is a crucial point, and until quite recently has been a big problem with digital video recorders (DVRs). Now that very large hard disks are available cheaply, digital systems can offer very long recording times.

Ignoring the fact that digital systems are capable of considerably better image resolution, what are the real differences in storage capacities between the two technologies ? In fact, the comparison of analogue video tape versus digital storage is not a straightforward matter. Why ? The reason is that you have to look at the wider operational picture of archiving, image management, and system application to appreciate the flexibility and added benefits of digital storage. No, we’re not trying to weasel out of giving a straight answer about the differences in storage capacity – we’ll give you some concrete figures – but asking how long you can record on a digital or analogue system is a bit like asking "how long is a piece of string ?". It depends on so many factors, such as the number of cameras, the average frame rate, the image quality required, and the number of hours recorded per day.

Let us look at the example given in the aforementioned previous article. A single video tape will store approximately 500,000 fields, and it was calculated that this would be equivalent to about 10 Gigabytes of storage on a digital system. We might argue about the figures a little, but let’s accept them as being roughly correct.

If we take a simplistic approach to storage capacity comparison then, we might say that a video tape costs perhaps £1 in bulk, and a 250Gb hard disk costs about £100. However, when we consider that a digital system might contain 250Gb of on-line hard disk space and store 10 to 15 days worth of recording at a rate equivalent to the 24-hour mode of video tapes, we can see that a digital system could run automatically with no changing of tapes ever. After 10 days, the digital system will simply record over the oldest files. At any particular time, the previous 10 days are always available. (Note : the numbers used here are arbitrary - some digital systems are capable of such storage capacities or even better, whilst others are severely limited).

The 24hrs-a-day, 30-days archive mindset

One of the main advantages of digital recording is that it can free the user from the need for regular archiving, or tape-changing. For this, there must be sufficient disk capacity to give the user adequate recording buffer time. Recording buffer time could be defined as the time required to know when an event has occurred. This event can be copied onto a permanent-archive (e.g. for a prosecution). In reality a few days is enough for most applications, although there is a mind-set in the industry that 7, 14 or even 30 days must be archived at 24hrs per day. Admittedly, in some applications this is a real, even a legal, requirement, but so often this traditional procedure is followed blindly with little thought given as to what the recording system is designed to achieve. (See also points on event-driven recording below).

It may be a controversial viewpoint, but we think that digital systems which record to DAT (a digital tape format), which require daily changing of the tape cartridges, and do not permit instant access to even very recent recordings, are a complete waste of time. Digital systems can provide instant playback, free the customer from the tedious (and error-prone) problems of video tape management, and provide so many other advantages as discussed below, so why not exploit this ?

The wider view

We must accept, however, that 10 or 15 days completely automatic unattended recording may still not be enough for many applications such as those described above. Either some form of digital archiving is required (perhaps weekly) or, it must be said, analogue recorders may still be the most suitable option for some applications, at least for the present.

The discussions above, however, do not take into account the crucial benefits offered by the power and sheer flexibility of a well-designed digital recording system. Features such as event-driven recording, variable frame rates per camera, many and varied recording schedules, instant on-line playback, remote access, and playback-while-record offer a huge advance on traditional tape recording. So let's have a look at some of these points in a little more detail.

Event-driven recording

Far from hiding the limitations of digital video, as was argued in the last article, event-recording actually illustrates the flexibility of a digital system and in contrast, shows up the limitations in analogue tape systems. After all, the poor start/stop characteristics of analogue tape systems are the reason we have mindless 24hr taping in the first place, recording hours and hours of nothing at all. There are many good reasons for event-driven recording, and as many ways to achieve it. Event or alarm-driven systems can provide a higher frame rate of the activity you want, and a more efficient use of storage capacity to avoiding archiving. Most importantly, the filtering of all the hours of non-events means a far more efficient and less tedious task of retrieving a particular incident. This dramatically increases the usability of the security system as a whole. We must all be familiar with the stories of video tapes with important incidents stored on them, but neither the police nor anyone else having the manpower and resources to view them and extract the crucial information.

Note that we are not arguing that event-driven recording be used to the exclusion of time-lapse recording. Far from it - the combination of event-driven and time-lapse recording further illustrates the flexibility of a good digital system and should be exploited. Let's take the example of a CCTV system in a large shopping centre. It makes sense to time-lapse record cameras covering the main car parks and the shopping halls during the day, but what about at night when the shopping centre is closed ? At night these cameras could be recorded only when triggered by simple motion detectors, door sensors or burglar alarm systems. What about cameras in fire-exit stairwells ? Surely these need only be recorded when activity is sensed, even during the day ?

Recording schedule flexibility

A properly-designed digital system will provide a simple and efficient way of setting up many different recording schedules for different times of the day and different days of the week. Each recording schedule can vary in terms of the cameras included or excluded, the frame rates for each, the active alarm channels, and the period for which each schedule is valid. What all this means is that the CCTV recording system can respond to alarms, to varying levels of activity, and record exactly what is required, every hour of the day and every hour of the week, completely automatically. This is vastly more flexible than traditional analogue multiplexer/VCR combinations. This way of doing things is so powerful, and so efficient, that a security manager need never go near the recording system, except when important events occur.

Playback

So what happens when a crime occurs or some other significant event is recorded ? Obviously the first thing which is needed is to playback and review the event. With a tape system, you must first find the right tape, and then either stop recording on a VCR which is in use, or have a separate playback system (including multiplexer). You then wind the tape forwards or backwards to index the camera and time which you want to view. With a digital system, you simply choose your date, time and camera on screen, press a button, and you are playing back within seconds. Furthermore, with a good system, designed for use with standard personal computers, you can playback from any location within networking distance of the recording systems, all without interrupting recording. A digital system will allow all the usual playback controls : Play; Pause; Fast-forward; Play backwards; Frame

forward etc., but also very powerful features to help find particular sequences such as : Skip to next event; Next day; Previous Day; or even to show a list of recordings, dates and times.

Not every digital system will have all of these features, however, but most provide some useful functions over and above analogue systems.

System compatibility

It is well-known that it can be awkward to find a compatible VCR and multiplexer combination to play back a particular tape, and some digital systems also have a corresponding problem. Many digital system manufacturers use their own proprietary hardware and software formats which prevents playback on anything but another identical system. This may suit the interests of these manufacturers, but hardly meets the needs of the users. Any digital system could, and should, produce files which can be played back on common Windows-based desktop computers.

Which digital system ?

Finally, a few remarks about the variation of digital recording systems on offer. Just as with analogue systems, there are good and bad digital systems. As a representative of a digital recording system manufacturer, I obviously cannot make any unbiased recommendations in an article such as this. What I would suggest, however, is that any prospective user of digital recording systems must carefully go through the features and benefits of the systems on offer, and judge whether it meets their own specific requirements, perhaps using the comparative table shown here as a checklist.

Trends

Analogue systems have a huge installed base, and so will be around for some time yet. Digital systems as yet account for less than 5% of the market for CCTV recording systems in the UK, and yet one major CCTV distributor is predicting a rise to 50% by 2001. The figures may be debated, but the trend is obvious and irresistible.

Conclusions

Analogue systems have their place, especially when very long archiving time is required combined with 24hr recording irrespective of activity levels. They are still cheaper than digital systems, but the gap is closing rapidly. The power and flexibility of good digital systems will more than pay for the extra cost, by reducing tape management problems, lowering maintenance costs and improving overall operational efficiency. Not only that, but digital storage capacity is increasing in leaps and the prices are coming right down. So when will the digital age arrive ? You’re looking at it . . .