The storage behind intelligent video surveillance

By Khoo Boo Leong | Feb 22, 2009

The move from analog CCTV with analog VCR to intelligent IP surveillance has opened up opportunities for the video surveillance market to grow its revenues of about US$13.5 billion in 2006 to a projected $46 billion in 2013, according to ABI Research.

While security remains the obvious application, promising markets such as transportation and retail are using video surveillance to prevent legal liability, analyse customer behaviour and design stores. For example, they use facial recognition software to analyse shoppers’ behaviour within stores by tracking eyeball movements as shoppers view product displays.

ABI Research also projected that global spending on video surveillance for transportation markets will jump from about $630 million in 2006 to $2 billion in 2013, while retail will account for a spending rise from about $1 billion in 2006 to almost $4 billion in 2013.

“Video surveillance is suitable for a wide range of industries including government, telecommunication, healthcare, finance, education and manufacturing,” said Vincent Low, the regional director of United Information Technology (UIT), a manufacturer of IP-SAN, FC-SAN and NAS solutions that are critical to video surveillance implementation.

“For instance, using wireless IP connectivity, you can monitor an entire railway line for trespassers. And in China, the government is trying to monitor entire metro areas in an effort to reduce crime. We’re integrating our hardware with our partners’ software for continuous monitoring and video forensics.”

The cost of a networked video surveillance system can be attributed to three main components: Front-end surveillance equipment such as security cameras, monitors, digital video recorders, encoders and decoders; a surveillance configuration and management server; and a video data storage system.

High capacity and high availability are key storage considerations as UIT’s storage systems serve increasingly large-scale surveillance deployments that involve more security cameras, higher frame rates, and megapixel-quality images.

High availability

The company’s video surveillance distributed storage solution, for instance, consists of a high-availability storage system such as the BM3600 Fibre Channel/Serial Attached SCSI (FC/SAS) array.

More importantly, it embeds the surveillance application software in the storage device. After being encoded, data is transmitted directly to the distributed video surveillance storage without having the read-write processing of the surveillance data first passing through a separate video server.

In the BM3600, a Snapshot function creates near-instant point-in-time copies of volumes independent from host and application. Users can replicate data speedily without affecting system performance because Snapshot processes the data at the array and block level to leverage the computing power of the storage system.

The BM3600’s array structure combines physical drives into disk groups which are further subdivided into logical disks to create arrays. These logical disks can be combined or stripped together to form volume structures, enabling storage resources to be utilised efficiently and hard disks to be better managed. One of the benefits is that it automatically recovers lost disk members of a disk group to a new array when multiple drive failures corrupt the RAID array of a logical disk or volume.

To ensure the integrity of video data and high availability, UIT’s video surveillance storage system supports up to 48 SATA/SAS hard disks, RAID 0/1/5, hot swappable hard disks, power supplies, fans and controllers, and an environment monitoring unit.

The BM3600 accommodates up to 16 SAS or SATA-II or a mixture of both hard drives in a single enclosure and it allows JBOD expansion to be attached to increase storage capacity efficiently. Driven by an Intel IOP341 processor with up to 1300MB/s throughput, it accommodates four 4Gb/s FC ports or four SAS 12Gb/s ports for either SAN or DAS deployment respectively.

UIT’s solution receives data streams from the encoder through the network, and then stores the video data in the back-end storage for users to search through and play back the videos. With management, storage, video viewing and recording as well as redistribution, content downloading, playback and networking integrated in a single entity, this solution can be scaled up from building surveillance to city-scale environments.

Further, it supports video recording through a maximum of 100 channels while simultaneously allowing on-demand playback of another 100 channels of recorded video files.

Managed services

Such advances have enabled telecommunication companies to offer managed video surveillance services. “China Telecom has started offering video surveillance installation and management services to businesses, allowing them to specify requirements such as storage capacity and the number of cameras to be installed,” said William Chan, vice-president of UIT. “Businesses do not have to bear equipment expenditure or build up relevant expertise. It is also feasible for companies to monitor their facilities across multiple countries centrally with the advent of broadband IP.”

“Telcos can sell to their mobile phone subscribers 1TB to 2TB of storage. The subscribers then install IP cameras which will feed surveillance content continuously into the storage and [these content can be streamed to] their mobile phones,” said Low. “The recommended frame rate for such surveillance is 16fps although 30fps is ideal but more storage is required the higher the frame rate. [At 16fps,] this will take probably a couple of hundred megabytes of storage a day.”

However, viewing images from multiple observation cameras that are operational 24 hours a day and 7 days a week is an overwhelming challenge in itself. It would take years to view one week’s worth of recorded video material fed from 50 cameras.

Forensics

One of UIT’s partners, Observision, addresses this challenge with its video forensics capability.

“Every frame and every pixel of each frame recorded is assigned a number which represents an attribute such as colour or direction,” Low explained. “So, I could search for a person in a red shirt running down a corridor in a certain direction. The resultant clips could be 5 seconds or 10 minutes as defined by the user.

“You can refine your search, for example, by filtering out people who walk in and out a particular door multiple times within a length of time carrying something of a defined size.”

Observision’s technology analyses image information from security and observation cameras completely and directly and then records them in the database. One of its applications is to detect movements and persons who loiter for lengthy periods in a pre-defined area such as in the vicinity of a cash dispenser or car park. If, for example, someone takes ample time to have a look inside five cars that are parked near one another in a car park, then the system will filter out these images.

UIT’s foray into the world of video surveillance systems began with the massive deployments for last year’s Beijing Olympics. The company and its partners won the bid to equip the Beijing International Airport, the Beijing Olympics athletes’ village and the iconic cube-shaped swimming complex.

UIT provided the backend storage component and its partners provided the software, monitoring and camera. “All in all, there are more than 100 sites using our solution for video surveillance,” said Low.