Improving Traffic Surveillance with Intelligence, Integration and Efficiency

The traffic surveillance market already hungers for IP cameras and superior video quality, but with these systems now incorporating intelligent video analytics and recognition, server platforms face fresh challenges.

Worldwide, users are replacing or upgrading their H.264 cameras and NVR/DVR equipment, accelerating the trend that began in the last few years, when adoption of video surveillance equipment based on the H.265 video codec first took off.

At the Security China 2016 exhibition held in Beijing, the top three Chinese surveillance vendors, Hikvision, Dahua, and Uniview, unveiled a series of H.265 based IP cameras and Network Video Recorders (NVR) products. VIVOTEK, Taiwan’s top vendor of surveillance products, had already launched its H.265 solution, including IP cameras, digital video storage, and surveillance system, as well as video management software at the end of 2015.

Traffic surveillance systems in developed countries must meet basic requirements such as high- resolution video (minimum 1440p, ideally 4K), high image quality (starlight level low-light, super wide dynamic range, dynamic video optimization), and high efficiency compression (H.265). Other must-haves include video analytics and smart image recognition functions.

New Intelligent Video Analytics (IVA) technology performs basic real-time video analysis functionality for traffic conditions, traffic volume, and pedestrian flow, such as detecting traffic accidents and abnormal congestion. IVA also enables intelligent searches after a crime or the ability to detect and prevent crimes before they happen. Basic traffic surveillance video analytics functions include logging pedestrian and traffic flow, face detection, and vehicle color and model detection from each camera. Advanced functions include preliminary analysis to obtain critical frames for image recognition. The frames aid the search for stolen vehicles, wanted criminals, and the presence of suspicious persons or vehicles near a crime scene.

Use Case: Taking an Intelligent Traffic Control Center into the Future


The customer issued the following requirements for the construction, upgrade, and future planning of an intelligent traffic control center:

  • Old and failed products are to be replaced with new H.265 IP cameras with a resolution 1440p or greater.
  • Some H.264 IP cameras and video management systems (VMS) within their warranty period will be retained.
  • New 4K high-definition IP cameras are to be added at important public venues, road junctions, road sections, and urban viaducts.
  • Legacy VMS software:
    • Can remain in use on the new server.
    • Must be compatible with the new server platform and support future system upgrades over the next three years.
  • The new VMS and central management system (CMS) software must:
    • Incorporate intelligent video analytics and recognition functions.
    • Have a high availability (HA) architecture and redundancy.
  • The new server platform must:
    • Comply with the system planning requirements.
    • Be able to satisfy the performance requirements of related application software.
    • Be energy efficient to ensure that the upgraded equipment will reduce overall system power costs.
  • The new intelligent traffic surveillance system must be compatible with 4K HDMI monitors so that it can comply with the requirements of a high-definition video wall.
  • The new VMS/CMS must be able to simultaneously process at least 64 video streams in real-time (H.265, 720p, 15 fps) for display on a video wall.

Figure 1: Intelligent video surveillance solution

Based on this intelligent traffic surveillance system’s application and cost-performance requirements, ADLINK proposed its MCS-2080 Media Cloud Server Platform, a high density integrated video platform with 16 independent systems in eight MCN-1500 compute nodes.

Incorporating Intel® Iris Pro Graphics P580 and the Intel® C236 Chipset, the MCS-2080 platform has two Intel® Xeon® Processor E3-1585 v5 per node. An integrated GPU assures high computing performance and video processing power.

Each of the platform’s two 10GbE switch modules is equipped with four fibre ports that can transmit data at up to 40Gbps. 10GbE fibre ports ensure the stability of video streams from IP cameras to prevent system anomalies due to network latency or cabling quality. An additional two 1GbE RJ-45 ports are available for the integration of a DIDO controller, IoT network sensor, and gateway equipment. Short bursts of data and video streams that must not be interrupted therefore pass through separate ports—an approach that prevents mutual interference during transmissions and improves overall transmission quality.

Two-way audio-video conferencing for crime prevention or the reporting of traffic accidents and violence can be supported on one of the systems without adversely affecting other security processing being performed on the platform.

To enhance the usability and flexibility of surveillance displays in the control center, the platform can be integrated with a video wall consisting of multiple computers and monitors. In addition, information from IoT sensor modules can be integrated with the surveillance management system logs to provide integrated alarms. By making it possible to tag alarm video files, the MCS-2080 aids quick searches through videos of related incidents.

Figure 2: Built-in network switch application architecture

Balancing performance and energy efficiency, the MCS-2080 includes two redundant power supplies, each with a maximum output of 1600 watts. The entire system equipped with 16 processors has a maximum power consumption of 1380W and occupies significantly less rack space than 16 stacked 1U server systems.

Its eight PCI Express x8 expansion slots can be used for Fibre Channel (FC) adapter cards to connect to SAN storage. High-speed network cards with speeds of over 10 GbE are also supported for connecting to NAS storage. If large amounts of video recognition data need to be buffered, then the expansion slots can be populated with SSD cards to provide additional onboard storage.

Figure 3: High density design to save the server room space

The MCS-2080 system architecture is designed for flexibility and can be treated as a server platform with 16 Intel® Xeon® Processor E3 servers with dual redundant power supplies, functioning as 16 independent systems. The clustering functions of Windows Server 2012 or Linux operating systems can also be used to combine the 16 servers into one super high performance operating platform.

Some of the systems can be for installing an existing VMS, with the new server platform boosting the performance of the legacy video management software. CMS software or VMS software can also be installed on two systems simultaneously for an HA implementation. Using an HA architecture with VMS or intelligent video surveillance software for important locations will improve the reliability of the video analytics system and avoid the risk of shutdowns that can occur with a single system.

The MCS-2080 can support LCD video walls up to 4×4 in size with its 16 4K HDMI video outputs. ADLINK has tested software from its partner VIVOTEK, a leading provider of surveillance solutions, that allows a single system to process 64 real-time H.265 IP camera channels at once (720p, 15 fps). This means the MCS-2080 with 16 systems can process in real time up to 1024 channels of H.265, HD 720p, 15 fps video and output it to a video wall. This level of performance is superior to server systems with standalone graphics cards. The risk of a single machine failure causing the entire video wall to become inoperable is also reduced.

Figure 4: Integrated storage device application via PCIe expansion slots to save the recording files


Figure 5: Video wall application

Integrated verification testing conducted by ADLINK demonstrates a high level of compatibility between the MCS-2080 Media Cloud Server platform and major types of video surveillance system software. In addition to more efficient use of server room space and better energy-efficiency management, it also allows the flexible integration of new and legacy surveillance systems. The ability to hot-swap individual systems reduces the risk of lost surveillance coverage during shutdowns. Server expansion to keep pace with the number of cameras managed by the system is possible. System migration in the event of failure is also supported for an HA architecture.

The MCS-2080 Media Cloud Server is an integrated platform that implements the hardware architecture of distributed servers while offering the advantages of centralized high physical density management. If the surveillance system or software is upgraded in the future, there is no need to replace the entire unit because components can simply be re-purposed. For video recognition software that requires high processing power, support for clustering at the operating-system level harnesses the maximum performance of the platform’s 16 processors, using software programs that support parallel processing.

Andy Wu is currently the President and CEO of A-MAXIM Intelligent Communications Tech (A-MIOCT), a new startup company focused on integrating new IoT, machine learning, deep learning, and AI technologies. Mr. Wu has more than fifteen years’ experience in the surveillance industry. Before establishing A-MIOCT, Mr. Wu was Product Manager, VMS/CMS software, and Sales and Brand Manager of networking cameras and NVR. He is also the consultant for ADLINK Technology surveillance marketing and government surveillance project deployment. You can reach him at

Share and Enjoy:
  • Digg
  • Sphinn
  • Facebook
  • Mixx
  • Google
  • TwitThis