Helping First Responders
Video system facilitates needs of police and transit administrators
- By Nathan Needel
- Mar 01, 2012
Managers of the Shanghai Metro System were very concerned
about what could happen if there was an accident. With so
many people using the underground rail system, responding
to a calamity and getting the injured rescued could be a major
There are many challenges in creating such as system. First of all, because the
system is so vital, it needs to stay in either semi-automatic or full-automatic mode
all the time. Thus, the video system itself must be monitored continually to ensure
that all aspects of it are running in real-time and that recorded video is available to
the surveillance center as well as the police at all times.
Adding to the challenge, a huge number of cameras are deployed. To date,
the system uses more than 1,000 cameras and approximately 60 networked matrix
switchers. Another problem is that as trains go over the rails, they create electromagnetic
interruptions that can impact the video cameras and other surveillance
devices. These factors presented a tough challenge for the transmission design.
Meeting the Challenge in a City of 19 Million
In a metro system, camera positioning is paramount. If the person on the video
can’t be recognized, the system is of no value as a deterrent. Therefore, in any one
visit to a Shanghai metro station, a passenger will be viewed by four different cameras,
all from the front, and will appear in six different surveillance images.
For instance, both the entrances and exits of metro stations have fixed cameras,
each capable of viewing a person from the front as he or she enters a ticket-check
station. In each of the metro halls, an integrated high-speed dome camera surveys the entire facility. Secluded corners have additional fixed cameras so that these
areas are not out of sight to the surveillance system. Cameras also are on the platforms
to give rail system administrators an overview of passenger flow and help
train drivers make sure that all passengers are on or off the train before the train
is put into motion.
To understand how the system works, it helps to visualize the flow of video.
The video from the camera is transmitted to a video equalizer in the train station’s
control room. Once equalized, it is transmitted to a video distributor where
functional icons are overlaid on the video image. From there, the video goes to the
matrix switcher and DVR, respectively. One channel of the DVR is looped back
to the matrix so that the matrix can call up the video being recorded in the DVR.
Three keyboards, installed in the train control room, the security room and the
maintenance room are used to run their own individual applications. For instance,
the train control room is most interested in traffic flow while the security room
concerns itself with safety and criminal concerns. Each is connected to the RS232
port of the matrix, and each has its own priorities.
The control center and each station are connected by an Ethernet network,
used solely by the video surveillance system, to offset the aforementioned problems
of electromagnetism. A matrix in any one of the metro stations uploads eight
channels of images through eight video coders. Eight video decoders in the control
center then take the images collected from the front matrix and transmit them to
the central control matrix. All matrixes also are connected via Ethernet, each provided
with its own IP address.
The central control keyboard also is connected into the Ethernet network instead
of the central control matrix. Its control signal is transmitted from a keyboard
to the video management server. Once the server translates the codes collected
from the keyboard, it identifies the priority of the keyboard and transmits
the corresponding control demand that controls each station, central matrix and
codec to perform image switching and PTZ operation for the domes in each station.
Control and Priority Levels
Although complex, the system is easy for operators to run. For instance, control
center operators can call up and control video in each station via one of the keyboards
connected to the Ethernet network. However, they can’t do anything solely
at will because control procedures and priorities are managed uniformly by a sole
management server within the network.
For example, if the central operator wants to call up video from a specific station,
codes are transmitted to the server, which makes a judgment and sends out
commands to ask the station video matrix to switch the desired video to a specific
video coder. Simultaneously, it asks a specific video decoder to decode the coder
address of that station. The decoded image is then transmitted to the central matrix.
There, the server sends its commands to the central matrix, asking it to output the image uploaded by the decoder
onto a chosen monitor.
Likewise, the control signals for the
domes are identified by the server, comparing
their priority with other keyboards,
including the keyboards within
the network and the keyboards in each
station. At this point, the server transmits
the control command into the matrixes
in each station.
Each keyboard can set up 20 priority
levels to any camera. They do not
need to be the same; each camera can
have different priority levels. Priority
levels are further customized to the
needs of each metro station.
Within a single station, cameras are
divided into two groups. The first group,
situated at the entrances and exits, are
for the police. Obviously, their control
priority is determined by the police.
The second group, comprised of the
metro station domes that survey the operation
in general and passenger flow in
particular, is subject to the needs of the
metro transit authorities. As part of a
total system, there is a crossed priority
among the two groups and, depending
on circumstances, one group can have
priority over another.
To do all of this, the system must
meet three requirements.
- The matrix must be able to change
the original keyboard setting priority
for a camera. As previously noted,
each keyboard can provide up to 20
different priorities for each camera.
- The video surveillance server needs
to identify the priority comparison
of the fixed cameras set up by the
different keyboards in the control
center. As the keyboards in the
control center send out code to the
server through IP, the server identifies
the different keyboards by their
IP addresses, which provides the
server with each camera’s priority
level. At that point, the server
transmits the control commands
collected from the keyboards to the
- Although the server can compare
priorities among the keyboards by
their differing IP addresses, it is
still difficult to compare the priorities
of the IP keyboard and that of
a keyboard used in a metro station.
That’s because the server cannot get
the data from the station keyboard.
Thus, another change needs to be
made to the matrix. Once the station
keyboard controls its front end camera,
the matrix promptly uploads the
current dome’s priority—controlled
by the station keyboard—onto the
central server via IP. The server then
backs-up the collected data in its
database. When and if necessary,
the backup data is read for priority
The Shanghai Metro Transit System leverages
Infinova’s classic use of an Ethernet
matrix for metro surveillance systems,
yielding easy connection, simple
system expansion, strong compatibility
As an example, although of different
protocols and brands, the matrixes
and DVRs are integrated. By using a
server, internal and external ports are
unified, enabling easier system expansion within the transit system—such as the surveillance connection between stations—
and externally—such as from a metro station to the police—simply by using
the server to exchange information.
The matrix system can collect alarm information from every alarm—fire control
or theft alerts—and all video loss alarm information for transmission on to
the central management server. When the central management server receives such
alarms, it can pop up the alarm information window, discharge the alarm bells
and flicker the alarm light. This capability can be provided to other devices in the
network, as well.
The central management server also collects standard clock signals and calibrates
the time for the video matrix switchers, DVR, character overlapping device,
client end and other devices within the LAN to ensure the entire surveillance system
The matrix main control keyboard is provided with a network-control DVR function
by entering the corresponding hotkey that switches the keyboard to function as
the control network DVR. As a result, central users can control any DVR within the
network directly through the matrix keyboard, calling up recorded files from remote
sites to simplify both real-time surveillance and evidence collection. Compared to
systems that can only control the local DVR via RS 485, RS 232 or other ports, controlling
DVRs as the Shanghai Metro System does makes much more sense.
System expansion is simplified. By merely adding one network matrix and
video coder, then adding the corresponding IP address and channel connection
mode into the central server, expansion is complete. This lets the central keyboard
send out code to the server. Once the server identifies its priority and transmits the
control code to the matrix, it is able to control the newly added matrix switch and
control front-end pan and tilt rotation.
Matrix network integration also is simplified. Due to the control server, other
surveillance systems, such as the Shanghai police’s system, can get access to the
system easily via a seamless connection. For instance, the police video system only
has to transmit control commands to the transit system’s management server,
where they are translated, identified and then sent to the appropriate matrix to call
up the desired video. Plus, all is done in Windows, avoiding the need for the police,
or others, to have any other systems or software. They only need to provide the
appropriate protocol to get access.
As a result, Shanghai authorities are able to pinpoint any accidents and determine
their severity to ensure the right numbers of first responders get to the right
place as quickly as possible. The transit authorities can determine if they need to
run more or less train cars in relation to passenger needs, and
the police are provided with evidence that will help them capture
wrongdoers and get them prosecuted and convicted.
This article originally appeared in the March 2012 issue of Security Today.