In this article we will learn how to estimate a bandwidth in a professional and detailed way.
By Germán Alexis Cortés*
The images we achieve with our video systems are traditionally intended for four (4) sites: Monitors, video recorders, servers for video analysis and communications networks to do any of the first three processes but in a remote site.
The best image is almost always achieved in the Monitors. Obtaining the best static resolution (pixels) and dynamic resolution (frames per second) of the entire system; as long as an efficient system is designed and configured. Almost always the streams (data streams) of video, arrive with a quality very similar to the servers that do video analytics (if they exist) and that record the information.
It is within the video recorders, where the configuration parameters are most modified to achieve that all the necessary video fits inside our hard drive ... and of those errors we have already talked on previous occasions ... but I think we will agree, that the most critical video stream, is the one we send to the data networks (LAN / WAN), because we know that the more quality we send, the more bandwidth we require from that network and that can mean greater investment in operating expenses.
And then comes the million-dollar question... how much bandwidth do I need?
Although I know some magicians in our guild, who dare to say figures immediately and others who commit to invented numbers (... perhaps because the question generates a challenge and we are afraid to say "I don't know"...); in this article we will learn how to estimate a bandwidth in a professional and detailed way. My suggestion is that you have a simple calculator handy and keep in mind the process I describe below.
Remember that video streams are made up of data that represents moving images. Therefore it is logical to assume that the more detail there is in each image, the greater the amount of data that our video will have. The starting point for calculating the bandwidth, or Band Width (BW) of a video signal, is the same data that is required to calculate the space of a hard disk (DD).
Remember that to calculate the DD, we calculate the space needed to store one (1) second of video and then multiply it by the number of seconds, hours or days we need to store. From the past articles, we can extract that to know how much is required to store a second of video, we must analyze and investigate some factors: 1. Speed at which we want to record (FPS), 2. Resolution, quality and compression algorithm that we use. This will give us the size of a video frame, on average (Bytes) and 3. The percentage of activity of the scene, which tells us how much one picture changes with respect to another.
Once we have these three data, we apply the formula:
Space for 1 second of video = FPS x Bytes x %Activity
And then if we express this result in bits, we will get the number of bits we need in a second of video... this can be said as "bits per second" (bps), and it is precisely the unit used to express the BW, therefore we can say that:
BW = space for 1 second of video x 8
Yes, it's that simple.
I must clarify that the same must be done for each video stream that we want to send through our communications channel. And then we deduce that the total effective bandwidth we need is the sum of all the bandwidths of each video stream.
Example:
Suppose we have 8 network (IP) cameras, in a commercial establishment and we want to see them simultaneously, from our office. What bandwidth do I require?
Let's assume the following data: A. Because our application is monitoring only, we require video at only 10 FPS. B. Once consulted the factory of the IP camera, we conclude that using an H.264 compression algorithm, in medium quality and using a D1 resolution (720x480 pixels), the average of each image is 9KB. C. Conducting an activity study, we see that on average the scenes of the cameras have an activity of 60%.
Then applying the formulas mentioned above, the EFFECTIVE bandwidth that the data network needs, to see one (1) cameras is:
BW = 10 FPS x 9KB x 0.6 x 8 = 432Kbps
And therefore the bandwidth to simultaneously view the 8 cameras will be:
Total BW = 432Kbps x 8 cameras = 3,456 Mbps
Although it seems to us under this number, I actually suggest emphasizing the word CASH, because the speed of our "Broadband" service, which is contracted with an internet service provider (ISP), is a nominal speed and is always much higher than the effective speed I really need.
Effective Speed vs Rated Speed
When we talk about a telecommunications service like the one that ISP companies (Internet Service Provider) offer us, we must take into account three aspects, to verify the speed we need to hire: OSI Model, Synchronism, Reuse Level.
Let's start by defining what we're talking about:
The nominal speed is the one mentioned in the contract that is made with the ISP. That is, the one that is advertised and comes to light, almost always and at this time (2013), in whole numbers expressed in Mbps (millions of bits per second). However, this speed is not effective, that is, we cannot use the entire nominal speed we hire, because several factors occur (technical and commercial) that affect its performance.
Effective speed is the number of bits per second we can actually use for our video signal.
The effective speed is always less than the rated speed. And it depends on the aspects we just mentioned.
a. OSI Model. Without going into detail, the way video signals are transmitted on a network that uses Internet Protocol (IP), follows several guidelines of the OSI model, for data transmission. Each layer of the model adds some control characters and delivers the new data frame to the lower layer and so on... so that when the video stream travels through the chosen transmission medium, it is actually increased by a percentage that varies between 15% and 40%, depending on the communication protocol used.
b. Synchronism. It refers to whether the download speed (i.e. receiving data from the internet or DownLoad) is equal to the upload speed (sending data to the internet or UpLoad). In the vast majority of cases, the download speed is much higher than the upload speed. The values in our Latin American region can vary at ratios of 1:4 or 1:2, that is, the upload speed is four times lower than the download speed or sometimes half the speed. When the upload speed is equal to the download speed, the channel is said to be synchronous and almost always occurs when the channel is dedicated.
c. Reuse. This term refers to the ability to share the contracted channel with other users. ISP companies must make the most of the investments they make, therefore they assume that not all users require the maximum speed and amount of data simultaneously. Even though everyone can be connected indefinitely and simultaneously.
To better understand the concept think of an avenue that passes in front of your house, state does not make it exclusively for you, many more can use it and at a certain time they could pass from 1 car to 4, simultaneously depending on the number of lanes, but also if the avenue is of good specifications, cars could pass in a more orderly way, safe and fast, therefore in a second many more cars could pass. Obviously at peak times, traffic becomes heavier and overall speed slows down... it's absolutely normal.
Exactly the same thing happens in the communication channels to the Internet, under the service that we call "Broadband", it is understood that it is shared. Reuse is associated with the maximum number of users with whom we must share the same channel, without strongly sacrificing the qualities in communication. For Latin America the factor can vary from 1:12 times to 1:4 times. Although in some countries, there are already services without reuse (1:1). It all depends on the region where you are and the parameters your ISP works with.
d. When the channel is fully dedicated, it is not shared with anyone else, it is almost always synchronous and then the vast majority of nominal bandwidth is guaranteed to be effective.
Then there is a formula for finding the nominal velocity from the effective velocity.
Effective BW x Synchronism Factor x Reuse Factor
Nominal BW = __________________________________________________________________________________
(1 – (%) Protocol Factor)
Notice that communications contracts are made with the speed of Download, because more than 90% of the traditional flow of the internet is to view, investigate and download public or private information; However when we are using a video surveillance system we must remember that we do not download data from the internet but send it, therefore the speed that really matters to us is that of Upload in the place where the cameras are connected.
It is also important to clarify that the reuse factor becomes critical when we are transmitting a video in the hours of greatest traffic on the internet and this depends on where we are located and what neighbors we have. The common slogan is that the maximum peaks of use, are presented a few minutes before or after the hours of departure and entry, of the users of the nearby properties. It is understood then that it is very different to have a school as a neighbor, than a shopping center or a residential building.
However, let's remember that we cannot design an insecure security system, therefore it is always better to assume the most critical case and design with those parameters, so at other times we sub-use the system and the contracted services.
Let's continue with the Example:
We had already found that the effective bandwidth needed by the merchant establishment to simultaneously view 8 cameras is 3.5 Mbps approx.
Suppose that after researching with our ISP (traditional service in Latin America), we obtain the following configurations: Transmission in TCP/IP protocol, with 30% of real use. Download at twice the speed of Upload. Reuse level 1:4; then:
Nominal BW = 3.5Mbps x 2 x 4 / 0.7 = 40Mbps
Yes, you read that right. We need to hire a broadband service of 40 Mbps !!!, to be able to see all 8 cameras, simultaneously, with good quality, even at peak times.
And so, is this possible?
Technically YES. What happens is that it's simply going to cost more money than the service you had in mind originally.
The mistake of almost the entire industry is to ignore these situations and never do the calculations. Therefore it is intended that through a traditional channel of homemade type (traditionally today, about 4 Mbps), go all the information of the 8 cameras.
After configuring the two links (transmission and reception), the engineer responsible to see that some images actually arrive, at a much lower speed than desired, then resorts to lower the resolution of the stream and "plays" with some parameters that ultimately take away quality and safety of the signal, so that the transmitted image looks acceptable. However, the only thing that the end user perceives is poor quality and a feeling of frustration, (even with the entire industry) because it was not what they wanted.
We were not sufficiently aware of our Latin American reality, where the average bandwidths are at 4 Mbps, when in Tokyo or Seoul, for the same money, isps offer them bandwidths of 150 Mbps, at the residential level.
But the bad thing is not our technological backwardness or our state policies. The bad guys are ourselves, for not doing the proper calculations and not explaining to our end user the limitations and strengths of the systems we are delivering. When the end user realizes, that to improve their remote image they must pay more, in the vast majority of cases they do.
Possible solutions
Continuing with the example, let's look at the Possible Solutions:
to. Hire a 40Mbps broadband service. It can cost about 6 times more than average.
b. Hire a communications service with a DEDICATED channel. This implies a guarantee that the bandwidth is not shared, almost always the download speed is the same as the upload speed and finally the ISPs assign in most cases, a fixed public IP address. A channel of BW=3.5Mbps x 1 x 1 / 0.7 = 5Mbps should then be contracted. It can cost about 4 times more than the base value.
c. Decrease the number of video streams I want to watch remotely. That is, why do I have to simultaneously see all 8 cameras? I could watch them all, but one at a time, in that case I only need 1 video stream, therefore the BW I require is 8 times lower, then I would need a channel of only 5Mbps, which can be easily achieved at an acceptable cost. You could even order a dedicated channel of only 600Kbps at a fairly low cost and still ensure good quality in streaming.
d. I could ask PSI to make my channel synchronous. Sometimes they authorize and configure it at no additional cost. In that case the BW of the channel goes down.
and. You could change the transmission protocol (for example go from TCP to UDP), in this case the latency and control characters drop significantly and the refresh rate could go up, but sacrificing security in sending the data. It is not the best alternative for security, however it is acceptable when the use of the system is only to monitor manually from time to time.
f. Finally it is important NOT to modify the resolution of the image, nor the speed of fresh of the video. In this way quality is not sacrificed.
g. A more sophisticated and efficient compression algorithm could be used, however in many cases, it involves changing the video recording hardware or camera, and is not a simple or economical option.
Personally it seems to me that option c) may be the most acceptable, however it all depends on what the designer has really thought and the operational use that is given to the video surveillance system.
To conclude, I suggest that all these calculations, estimates and research, practice them beforehand and perform them together with a suitable and experienced person; before the sale of the equipment. In such a way that when the end user buys the solution, all parties are aware of the responsibility that is acquired, to achieve local and remote images of good quality.
* Germán Alexis Cortés H. is a Colombian Electronic Engineer, with a postgraduate degree in Engineering Management Systems. Consultant of the National Institute of Standards and Technology – NIST, in Electronic Security and Building Automation. CCP of ASIS, and CISSP of ISC2. He has over 20 years of experience in the electronic security industry. Renowned lecturer and university professor on high-tech issues at the Latino level. He has been a director of several companies in the sector and is currently a partner and directs Insetrón Ltda, an engineering and technical consulting company in electronic network projects for security, communications and automation. He has successfully participated in more than 150 projects at the Latino level. Contact him in the email [email protected]
Leave your comment