In recent years, with frequent meteorological disasters, great harm has been caused. Due to the short time resistance of occurrence and unpredictable hazard, it is urgent to release early warning information to the public rapidly and accurately. In spite of the variety of distribution means for early warning information transmission at present, there are still many problems, such as single transmission mode, transmission delay and so on.
Digital Audio Broadcasting (DAB) is based on digital technology and is the
third generation broadcasting (Dong, 2001). It is now
widely used in early warning information transmission field for its powerful
data broadcasting performance. However, DAB can only transmit audio information
and it cant meet the visualization transmission for early warning information.
With the development of DMB (Digital Multimedia Broadcasting), audio, text and
video information can be transmitted simultaneously based on DMB standard. Thus,
people can require warning information by multiple ways (audio, video and message)
in the shortest possible time. In addition, DMB has superior transmission performance,
such as anti-noise, anti-interference and high-speed. Based on above advantage
of digital multimedia broadcasting, a DMB pre-processing system for early warning
information is designed. The system is mainly composed of information compression,
TS (Transport Stream) combination, error correction and intelligent stream rate
control. The whole designed DMB pre-processing system has been validated at
present. Experimental results have shown that pre-processed DMB early warning
information has good error correction capability and stable transmission performance.
DESIGN SCHEME OF THE SYSTEM
The DMB pre-processing system for early warning information is consisted of
information compression, TS combination, error correction and stream rate control.
The whole realization of the system is shown in Fig. 1. Early
warning information is compressed firstly. Then, the compressed audio and video
information are combined into TS packets based on the DMB transmission standard.
In order to improve the reliability of data transmission, error correction is
realized through RS (Reed-Solomon) coding and external convolution interleaving.
After that, by inserting error protection information into the combined TS,
security of information transmission is improved.
|| Preprocessing system of DMB early warning information
|| Implementation process of TS (Transport stream)
Finally, intelligent stream-rate control is realized to make the transmission
stream rate stable and constant.
DESIGN AND IMPLEMENTATION OF THE SYSTEM
Multimedia warning information compression: In order to transmit visualization
information for early warning information, efficient multimedia information
compression is necessary. Early warning information is mainly consisted of audio
compression and video compression. By using advanced technology of band recovery
and parametric stereo, coding performance of AAC (Advanced Audio Coding) core
encoder is greatly improved. For audio compression of early warning information,
the MPEG-4 HE AAC codec algorithm (Xu et al., 2009)
is used. Because H.264 algorithm outperforms MPEG and other classical video
compression algorithms (Chivukula et al., 2012)
in both compressed ratio and visual quality, the H.264/AVC codec algorithm (Wei
et al., 2011) is used for video compression of early warning information.
Implementation of TS (transport stream): According to DMB standard,
multimedia information transmission is realized through TS multiplexing form
(ETSI, 2005). So, compressed multimedia information should
be combined firstly and then it will be packaged to transport stream to output.
The implementation process is shown in Fig. 2.
|| TS (Transport stream) structure of early warning information
During multimedia information transmission, supplementary description information
is essential, such as relevant channels, configuration and synchronization information,
which is mainly used to decode TS file. Description information should be also
multiplexed in TS package in form of data segment. Thus, multimedia information
which is included description information, video information and audio information
is first sent to the SL (Synchronization Layer) package to realize synchronization.
Then, output synchronization information is cut into the appropriate length
with added header to form the PES (Packetized Elementary Stream) package. After
that, PES package is put into the useful loading of TS package. Finally, TS
packages are formed into transport stream to output.
According to the transmission demand of early warning information, a basic TS package structure is as followed in Fig. 3. Firstly, with the header part of SL package, compressed early warning information is added to part of SL payload data to form SL package. Then, with the added PES header, a whole PES package is constructed. After that, PES package is divided into several blocks whose length is fixed to 188 bytes according to the adaptive grouping demand for early warning information transmission. Finally, the header of TS package is added to each block to form TS package.
Among the whole process of TS implementation, how to realize synchronization
between audio and video warning information is important. A synchronization
method based on time stamp is proposed. As is known, time stamp is provided
in SL header of SL package. There are two kinds of time stamps (You
et al., 2010), Decoding Time Stamp (DTS) and Composition Time Stamp
(CTS). Furthermore, DTS indicates the decoding time of current block of elementary
stream in the synchronization package and it is used as bi-directionally predictive
frame (ISO/IEC, 2000, 2004). Similarly,
CTS represents the composition time of current block of elementary stream. Because
bidirectional prediction is not used in the whole system, CTS is adopted to
get time stamp information. According to time stamp information and system clock,
cycle time of audio and video information can be calculated firstly. Then, on
the basis of respective cycle time of audio and video information, synchronization
can be realized by controlling corresponding coding time.
Error correction coding: To improve the reliability of warning information
transmission, error correction is designed. Concatenate RS coding (Yang
and Wang, 2010) and external convolution interleaving method are put forward
to improve error correction performance.
According to RS coding (Claus and Jiang, 2009), 16
checking bytes are added to TS package in order to remove interference (Zhang
and Liang, 2011). So RS (204,188) coding is used in the first step of error
The field generating polynomial of RS coding is as follows:
According to above expression, various elements of the Galois Field GF (28) coding and its corresponding decimal values can be calculated. Due to above values, the generator polynomial of our coding is as follows:
If the information polynomial of TS package is expressed as follows:
Then the polynomial of coded information results as:
|| Flowing chart of convolution interleaving coding
In the above expression 4, x16 μ (x) corresponds to the 188 bytes of data information and x16 μ (x) mod g (x) corresponds to the 16 bytes of checking part.
After RS coding, external convolution interleaving is used to improve error
correction further. By convolution interleaving, the continuous error can be
divided into several TS packets (ETSI TS 102 427) and errors can be corrected
by RS coding (Lee et al., 2005) in the decoder.
The depth of convolution interleaving is 12 bytes. Thus, the maximum length
of continuous error which can be corrected in the decoder is 96 (8*12) bytes
(Ueng et al., 2010). The flowing chart of external
convolution interleaving is shown in Fig. 4 and the whole
process is as follows:
||Initialize static variable num = 0
||Allocate storage space
According to the interleaving depth (I = 12) and the length of the data packet
(L = 204), the whole L*I bytes storage space called int_buf is allocated. Thus,
data package to be interleaved is stored into int_buf and corresponding assigned
address is int_buf+204*num.
||Convolution interleaving (I = 12)
There are two cycles in convolution interleaving. One cycle is to ensure no
delay of the first byte (sync-byte) in each package. The other cycle is to realize
convolution interleaving for other bytes of each package. The Implementing expression
is as follows:
As above expression 5, i represents interleaving depth, j and l denotes circulation
intertwined variables. The output result of convolution interleaved is stored
into array inl_code.
||Let num = (num+1)%12
Return step 3 to continue next convolution interleaving until the whole convolution
interleaving is finished.
Rate control: Because output stream rate of multimedia warning information
is inconstant, in order to meet further requirement of early warning releasing,
a method of inserting null packets is proposed to achieve intelligent stream
rate control. Inserted null packets are undeclared and special transmission
stream (Zhang et al., 2011). According to the
different ID number, null packet and practical information transmission packet
can be easily distinguished. The detailed steps of rate control are as follows:
||Set the stream rate detection cycle T
According to practical multimedia warning information, timing cycle is set
for the current transmission frame. Once the periodic T is set, the output rate
is checked every T ms.
||Calculate the stream rate
Let ts_num is the number of the output packet, the rate of output stream br in the detection cycle T is calculated as follows:
In above expression 6, L is the length of the TS package.
||Calculate the number of inserted null packets
According to the calculated bit rate br, null packets are inserted if the calculated bit rate is less than initial bit rate. On the basis of the following formula 7, the bit rate of one TS packet in periodic T is 1.7 kb sec-1, the number of the inserted null packages n is as follows:
By inserting null packets to TS, stream rate control is realized and output rate of TS is stable at 384 kbps. This constant transmission rate meets the further requirement of early warning releasing. Meanwhile, only simple addition and division are used in the whole implementation of inserting null packets, it is suitable for real-time implementation due to its low computation.
In order to meet the visualization transmission demand, a DMB pre-processing
system for early warning information is designed. According to DMB standard,
multimedia information of early warning is combined to transport stream and
realizes synchronization between audio and video information. Meanwhile, concatenate
Reed Solomon coding and external convolutional interleaving methods are adopted
to achieve error correction. By inserting null packets, output TS rate is stable
at 384 kbps and this constant transmission rate meets the further requirement
of early warning releasing. The whole system has been validated by experiments
and has outperformed former DAB preprocessing system in both reliability and
This study was supported by National Postdoctoral Science Foundation (2012T50510), Jiangsu Provincial Natural Science major projects (12KJA510001) and was funded by the Priority Academic Program Development of Jiangsu Higher Education Institution, China.