Video systems for mobile use: the specific challenges
Certain applications simply need to store the data generated (video, audio and metadata such as GPS coordinates) in the end devices used - the data can then be copied and analysed by the cameras or recorders following an accident or other incident. If data protection regulations require, the data is encrypted before storage so that it can only be read and processed by authorised persons. In many cases, however, direct access to and the live display of data are also needed. As the installed devices are mobile and constantly changing position - i.e. not part of an installed network - wireless data transmission is required.
It seems logical to use the infrastructure of existing mobile networks for this transmission. These networks have excellent regional coverage, so there are no restrictions on the distance between or relative position of the various devices used. A Web-based server collects and displays the data from the individual end devices. The data can therefore be accessed over the Internet anywhere and from a wide range of devices (smartphones, touchpads and laptops).
Yet this approach also has a major drawback. Existing mobile networks which use 3G/UMTS technology provide only a limited, and not always constant, bandwidth. This limits how much data can be sent live to the central server. A live overview may be output with one to two images a second at a resolution of up to 720p, but can be subject to latency of 20 to 30 seconds. Whilst this may be good enough to get a general picture, it is nowhere near the real-time capability which is essential in many safety-critical applications.
NGMN (Next Generation Mobile Networks), which are based on 4G/LTE (Long Term Evolution) technology, are set to improve the situation by introducing higher bandwidths. Their launch is already under way. However, it will be some time before sufficient network coverage is guaranteed. Another way to avoid a data bottleneck is to divide up the data for transmission over multiple channels. The technology for this technique is already available; it is used for live television broadcasts. It is, however, expensive to buy and adds to running costs.
How, therefore, can mobile surveillance solutions deal with these restrictions?
In the solution presented here, the end devices generate images, audio data and metadata (GPS coordinates and speed), and send this information continuously to a central server. The server has a live display with one to two images per second (depending on the bandwidth), and the data is simultaneously saved in real time - at up to 25 images per second - to integrated storage media in the end devices. MicroSD or CF cards with a data volume of up to 64GB are used. In line with the selected solution and the image speed, this offers a recording period of anything between one and several days.
The live view of the NVR (Network Video Recorder) gives the user an overview of what is currently happening at all connected devices. If GPS data is available, there is also a map view showing the current position. Selecting the symbol for a device in the map view displays the corresponding live image.
If the user requires further details on a device, he can check the history view. The history view saves all images sent and compiles them to produce a video stream. The user can now browse the history using a timeline to search for relevant events (the corresponding GPS coordinates are also displayed). If an event of interest is selected, the user selects the relevant time period by highlighting the required start and end point. The real-time video stream for this period is then downloaded from the end device for further analysis.
The data transmission and data management requirements outlined above are not the only specific demands on mobile video systems and the devices used.
The systems must be as compact as possible with maximum integration, and extremely robust. Resistance to mechanical shock, vibrations and extreme g-forces is a prerequisite for reliable operation.