Stepping Back From the Edge
Editor / Provider: a&s International | Updated: 12/5/2011 | Article type: Tech Corner
The advantages of HD over SD are evident, but even after two years of “evangelization,” issues that hinder the overall performance of HD surveillance systems still linger. Transmission and storage are complex but easily resolved when enough money is thrown at the problem. Things are a bit different on the management and display side, where manufacturers still have a bit of work to do to optimize the software and hardware for better visibility and increased interoperability.
Newer IP-based video systems, regardless of image resolution, have many useful functionalities, but also require significant technical and business operation knowledge to ensure they can achieve the user 's goals, warned Mrinalini Lakshminarayan, Video Solutions Strategist for Motorola Solutions. “Moreover, there are many different components in these systems (different cameras, VMS, diverse networks), which could lead to more complex compatibility issues. Even the claims of being standardscompliant may not be totally correct. Reading the fine print and looking at what the specs are not saying are critical in ensuring proper operation of the system.”
Users and integrators should be careful when designing systems with these products, Lakshminarayan continued. “Open and flexible designs are critical to futureproof systems. Vendor specification and marketing messages could be misleading. Design and implementations based primarily on vendor specification could pose challenges unless the whole system is architected to fit the environment and needs of the user.”
Furthermore,the growing coexistence of analog and digital technologies presents consequences involving bandwidth, storage, retrieval and image management , said Mark Wilson, VP of Marketing for Infinova. “There are different uses and purposes of stored video, various locations where images can be stored, a myriad of storage types, and business and operations issues to consider.”
Today's VMS and cameras can use H.264, handle multiple megapixel streams and run onboard analytics on video, and it is very common for vendors to market all of these features.
However, network cameras are, in essence, small devices that could be likened to computers. When you have too many applications consuming the available resources, the system becomes sluggish and unstable, eventually leading to system failure. Network cameras use computer-like chips or accelerators for video, and they tend to perform similarly, Lakshminarayan explained. “Specifically, you should be careful about megapixel streams and video analytics; both can take huge amounts of resources and can place significant constraints relative to the marketing claims presented.”
Superior image quality is, in part, fulfilled through more advanced image sensors, higher frame rates and better low light performance; more processing power is needed to implement and utilize these improvements, as they add load to hardware components such as the image pipe, scaler, encoding and network communication, said Johan Paulsson, CTO of Axis Communications.
The Weight of Analytics
Running video analytics inside of the DVR offers a lot of potential, albeit at a cost. Analytics add cost whether onboard the camera or independently on the system, Lakshminarayan cautioned. In addition to considering channel counts and the types of video analytics, what is more important is to consider the user's goals to avoid wasting resources, both monetary and system-wise.
Video analytics require processing power and possibly dual-stream video, and these limit the number of network cameras you may be able to support, Lakshminarayan said. “The more activity there is on your cameras, the less you may be able to manage. Network cameras can send multiple streams, use multiple codecs and run a variety of video analytics. Analytics still require additional server-side processing. The design and implementation have to take these into considerations.”
This can be quite difficult because it depends on the tools provided in the products used, Lakshminarayan continued. “In cameras and hybrid DVRs, you are dependent on what the manufacturer gives you. In most of these products, there are few to no tools. With IP video surveillance software, even if the software does not provide you with good tools, you can still use standard monitoring applications provided with the OS.” [NextPage]
One of the most attractive claims of using software is that you can greatly expand the number of cameras you can record on a given server. However, flexibility also introduces complexity, Lakshminarayan warned.
“The configuration, calibration and regular maintenance of the system are required to ensure it performs as specified. Vendors often talk about recording hundreds of cameras per server; they usually mean SD cameras using MPEG- 4 and only motion detection. Once you start using multimegapixel cameras, H.264 codecs and video analytics, the number of cameras per server drop dramatically.”
According to Lakshminarayan, key performance issues with video surveillance systems include:
? Dropped frames: Even if you set the frame rate to 15 fps, the camera may only be able to send 8 fps or the recorder may only be able to record at 7 fps. The network may not be designed to handle the data jitter and delay characteristics. The PC and video processing may not be able to deliver the horsepower required.
This may also lead to false alarms from video analytics, which could be very painful for users and cause them to ignore or turn off the analytics.
? Unresponsive system: The storage unit, recorder or camera may become very slow or unresponsive due to network or server-processing issues. The system may crash or need to be manually rebooted. During this time, if reliability, redundancy and resilience are not built into the design, there could be a lot of loss of data.
The IT world has had years of experience dealing with failing systems, and has long since realized that all systems will indeed fail. “The best way to solve these problems is by a careful, up-front design and building TCO into the system architecture. It is important to work with a design architect who understands the pitfalls and compatibility issues of the systems,” Lakshminarayan said.
The glaring bottlenecks of HD systems are network bandwidth and the required storage capacity, said Craig Howie, Commercial Director for Visimetrics. “Larger image resolutions coupled with high frame/bit rates, potentially low compression and probably multiple streams from each camera can leave a significant impact on network bandwidth, storage capacity and display performance.” With higher bandwidths, the network configuration or architecture becomes an essential element in the project design.
Bandwidth consumption from HD video is often several times that of its SD counterpart, said Xiang-Qun Ying, IP Camera Director for Hikvision Digital Technology. “For example, a 4-CIF image flows smoothly at just 1.5 Mbps, whereas a 2-megapixel HD camera needs more than 4 Mbps to make the clearer images useable. This places stress on both bandwidth and storage.” While this may not be a problem for smaller setups,networking becomes a headache when a large number of channels is required, and will need professional technicians to design and maintain the network.
HD network cameras often use 720p or 1,080p. In practice, these cameras require 2 to 8 Mbps of bandwidth for an acceptable and practical stream. When a system comprises several hundreds of these, the tremendous amount of data will need a well-designed network to avoid performance issues.
Regardless of equipment manufacturers, the most congested point in the network in terms of bandwidth usage within an HD system is between the cameras and NVR, Howie said. “For this very reason, it is always best to design the network architecture using a distributed storage methodology where the storage systems are located in relatively close proximity to the cameras, keeping the network congestion to as few hops as possible and definitely away from the core of the network.”
In addition, the cost of fiber optics is continuously dropping. A gigabit local network completely comprising fiber is much more achievable than a few years ago. The significance of bandwidth issues will decrease, as transmission technology is constantly and rapidly improving.
The increasing adoption of IP-based video surveillance and the move to high-resolution cameras translate to more demanding storage needs for security and business intelligence applications. Frost & Sullivan estimates that this market will expand from US$123.1 million in 2010 to $181.1 million by 2016, citing demands for DAS, NAS and SAN solutions.
According to Frost & Sullivan, “IT storage for the physical security industry is still emerging, and dominant IT-based suppliers are modifying their existing enterprise IT storage offerings to suit physical security needs. As end users become more aware of these IT storage solutions and vendors gain greater expertise in serving the specific needs of the security industry, these solutions are poised to gain rapid momentum toward the end of the forecast period.”
However, the recent flooding in Thailand has hindered worldwide production, as the country churns out 40 percent of the world's hard drives. Average drive prices have already jumped about 20 percent because of the flooding, according to Bloomberg News . In a prepared statement, Stephen Luczo, CEO of Seagate Technology, said it will take a year, until the end of 2012, for production to recover to preflood levels. [NextPage]
Interoperability and Implementation
Video technology is moving from proprietary to standardsbased solutions, Lakshminarayan said. “There are many advances that have driven manufacturers to either comply with these standards or publish APIs to ensure that their products integrate with the existing multipurpose systems. Mix-andmatch has become very common as these video systems grow to satisfy a multitude of business goals and requirements.”
However, with all the promises of interoperability out there — ONVIF, PSIA and HD-SDI — “truly open” systems are still simply not there yet. Even HD-SDI, which is supposed to provide true plug-and-play functionality and analog simplicity, is plagued with incompatibility issues among brands, which were highlighted in the September 2011 issue of a&s International.
That said, current standards do indeed guarantee at least some level of interoperability and peace of mind, and they are constantly being improved. If customers follow standards, they reduce the possibility of facing future compromises, Wilson said. Vendors whose products comply with these standards can be found on the respective websites of the various organizations.
HD in security video surveillance has quite a few obstacles to overcome before it matures. While some are psychological factors, there are also some very practical issues to be resolved, Ying said. ”Installing plug-and play analog surveillance systems requires little more than hard labor. However, simple as it may seem at first glance, configuring IP addresses and selecting software GUIs require more technical skills that are currently lacking in the average security installer.”