26 April 2010
Security goes digital
System on chip integration is driving video over IP into the security market.
Historically, cctv security has been a small and relatively specialised sector. Analogue technology has provided few opportunities to increase functionality beyond basic capture of images and sound, to be studied in real-time and stored on video tape or DVD.
This is changing quickly as digital imaging and internet based technologies, such as Video over IP (VoIP), promise dramatic cost reductions while enabling network storage, remote access and value added features such as image analysis and event detection.
Equipment for network based digital security benefits from significant economies of scale by taking advantage of existing consumer multimedia and pc technology. Among these, new generations of networkable security cameras and devices such as digital video recorders (DVRs) are emerging, using consumer oriented SoCs and software to deliver advanced functionality at highly competitive prices.
Since images can be distributed over lans, wans or the internet, infrastructure costs are reduced compared to conventional analogue cctv systems that require high grade coaxial cable. Where dedicated wiring is necessary, this can be implemented using Ethernet cable. Moreover, Power over Ethernet (PoE) technology can save the need for a separate power source for the camera, not only simplifying design, but also speeding installation and allowing cameras to be positioned wherever they are required.
At the same time, IP transport increases the potential reach of a video surveillance system: security images can be viewed at the desktop, at a remote site or anywhere internet access is available.
Like any network based application, VoIP can suffer from latency and packet loss if the network is heavily loaded; although inconvenient for web applications, these effects can seriously challenge a security system. Network based security attacks, which may seek to intercept or corrupt data or to disable the system, are also a threat. Measures to combat delays and packet loss include optimising Quality of Service (QoS) settings at the network management level and applying techniques such as traffic shaping.
Bandwidth hungry traffic, however, will suffer losses and delays, particularly on overloaded networks. Considering that uncompressed full motion video can have a raw data rate of more than 150Mbit/s, depending on pixel resolution, the use of high compression signal coding standards is essential to reduce demand for network bandwidth, leading to better real time performance. Compressed video also reduces the demand for bandwidth and the amount of space needed to store digital security video.
Mpeg4 and H.264, the two leading coding/compression standards, are closely related and H.264 – which enables the highest signal compression – has recently been incorporated within the latest release of Mpeg4 (section 10). This compression must be applied to image data before transmission, which has implications for the design of IP video cameras.
An IP security camera has two main functional blocks: capture and manipulation of image data; and managing the interface with the network. Manufacturers such as Texas Instruments, Analog Devices, Freescale and NXP offer suitable processors for these applications and support them with reference designs optimised for security and surveillance applications.
An Analog Devices reference design for a camera capable of H.264 compression and suitable for security surveillance or machine vision applications contains two Blackfin processors: a BF526C host processor and a BF561 coprocessor. The BF562 runs an embedded Linux OS and handles networking and control functions, while the BF561 performs video processing tasks, including H.264 encoding.
Captured video is input to the BF561, which processes and compresses the bitstream before sending it to the BF526C, which coordinates sending the video through the network to the receiving device for decoding, displaying or storage. Communication between the two processors is implemented through a high speed synchronous serial port.
Supporting software implementing common IP camera features is also provided as part of the reference design, and enhanced product level software suitable for direct use in the final product is also available.
NXP can draw on a large body of IP for consumer multimedia applications to position solutions for professional high end video including surveillance.
The TriMedia cpu core, for instance, forms the basic building block for many video processing solutions, including the PNX1005, a software programmable core for high definition video processing. This benefits from the 400MHz TriMedia TM3282 cpu core, whose instruction set is optimised for media processing tasks such as picture optimisation, video compression and motion estimation.
NXP also has the Nexperia media processor, such as the PNX1500 and PNX1700, which are ideal for security DVR applications. Capable of multiresolution, multichannel encoding and decoding of digital video, these devices feature an efficient PCI/XIO architecture that provides enough bandwidth for applications such as analytics, motion detection, gait analysis and facial recognition, even while streaming multiple video channels. Hard disk and flash memory storage is supported.
Developing solutions based on these devices also brings the benefit of using the Nexperia Media Processor Software Development Kit (SDK), which includes tools for compiling and debugging code, analysing and optimising performance, and simulating execution of applications entirely in C/C++. The SDK can also be used to develop customised security applications for IP camera configurations.
Texas Instruments' Digital Video Software Development Kit (DVSDK) supports designers using devices such as TI's DaVinci digital media processors. Included are the TMS320DM365, which has strong features for use in IP security cameras, and the TMS320DM64x, targeted at DVR opportunities.
The DVSDK accelerates design completion by providing elements such as board support packages, video, audio, speech and image codecs and multimedia processing components. Tools such as DSPLink for interprocessor communication and the Digital Media Application Interface helps designers take advantage of software multimedia capabilities.
The DM365 is optimised for applications such as IP networked cameras requiring low power consumption. The device combines an ARM9 core with a video processing subsystem and two video image coprocessors supporting a variety of encoding and decoding capabilities. The video processing subsystem provides numerous features including a hardware face detect engine, real time image processing, image sensor interfaces including support for cmos image sensors and a glueless interface to common video decoders.
Freescale's i.MX27 multimedia applications processor provides similar capabilities for use in IP security cameras. It has an ARM9 core, hardware for Mpeg4/H.264 encoding/decoding and image sensor and Ethernet interfaces. The i.MX27 also incorporates Freescale's Platform Independent Security Architecture, which provides a blended hardware/software solution with features such as tamper detection, ic identification, secure ram and security monitoring and a runtime integrity checker to help guard against exploits such as malicious service attacks, unauthorised data access, and cloning.
While the adoption of VoIP for security applications brings new opportunities for system integrators and end users, it also brings internet related challenges, such as bandwidth management and vulnerability to attack.
Developers of products such as IP surveillance cameras and DVRs can take advantage of proven, high performance silicon with built in security and content protection features – and supported by software, online knowledge, development hardware and reference designs – to deliver flexible, powerful solutions quickly, and at competitive prices.
Alistair Winning is technical editor for Farnell UK.