COQRC: A Rateless Video Transmission Solution
Lingyu Liu , Jun Wu
∗
, and Jian Wu
College of Electronics and Information Engineering, Tongji University, 201804, Shanghai, China
email:liulingyu 009@126.com, wujun@tongji.edu.cn, wujiantongji@126.com
Abstract—The explosively growing wireless video traffic de-
mands high video quality and transmission efficiency. A new
scheme for video transmission over channel with noise is pro-
posed in this paper. Like standard video transmission, the
scheme includes entropy coding followed by powerful channel
coding. Unlike conventional schemes, we use a rateless coded
modulation scheme instead of adaptive modulation coding (AMC)
in traditional digital video transmitting to solve the dilemma
of inaccurate channel estimation. The proposed rateless scheme
does not need the explicit channel condition, it matches channel
condition implicitly. We optimize the proposed system over GF(2).
Performance comparisons are made with respect to the state-of-
the-art video coding and channel coding scheme used in Long
Term Evolution (LTE) mobile network. Experiments show that
the proposed system outperforms LTE by 2-3dB.
Index Terms—Wireless Video Communication, Rateless Cod-
ing.
I. INTRODUCTION
According to Cisco VNI Forecast[1], mobile video traffic
accounted for 60 percent in 2016. Mobile data traffic caused
by video will account for over 3/4(78 percent) of the worlds
mobile data traffic by 2021, increasing 9-fold than 2016.
Conventional digital video transmission over noisy channels is
designed based on Shannon’s source-channel separation theo-
rem. Video traffic is implemented by a linear transformation
like Discrete Cosine Transform, followed by quantization. The
quantized bits are entropy coded and transmitted after channel
coding. Due to the waterfall behavior of the channel coding,
fluctuation in the transmission channel can cause dramatic
degradation of the decoding Bit Error Rate (BER), causing a
large number of packets retransmission and bandwidth-wasting
[2].
In this paper, we consider the utilization of the Rateless
Coded Modulation scheme put forward in [3] to deal with
the problem of wireless video transmission. The proposed
technique, referred to as Channel Optimized Quantization
with Rateless Coding(COQRC), maps the entropy coded bit
sequence directly into modulation symbols instead of conven-
tional channel coding and modulation. Unlike conventional
scheme based on concatenating sources coding and chan-
nel coding which are optimized independently, the proposed
scheme optimizes these two components jointly. Research has
shown that the quantization parameters as well as channel cod-
ing schemes are essential for maintaining a good transmitting
performance. In general, there are a variety of rateless coding
schemes aiming at adapting noisy channel conditions without
∗
Jun Wu is the corresponding author(email:wujun@tongji.edu.cn).
feedback [3][4][5]. The transmitter progressively produces
symbols with fine-grained bit energy allocation. A survey of
the rateless coding schemes is provided in section II.
If the transmission channel is Additive White Gaussian
Noise(AWGN), COQRC achieves near optimal performance
given by Shannon’s theorem with the restriction of finite block
length and affordable decoding complexity. COQRC has a
better PSNR performance than the conventional source and
channel coding.
The rest of the paper is organized as follows. Section II
introduces backgrounds and related work. Section III includes
the framework of the proposed system. We choose Rateless
Coded Modulation (RCM) codes as the channel code of the
COQRC scheme in view of their rate flexibility. Section IV fo-
cuses on the optimization for the specific COQRC application.
Section V introduces the de-convolution belief propagation
algorithm used by the receiver. In section VI, we evaluate the
performance of COQRC and compare with the state-of-the-art
video transmission scheme presently used by LTE. Finally, we
conclude the paper in section VII.
II. BACKGROUND AND RELATED WORK
Our work is related to the work on the rate-adatption video
coding and rateless coding.
Rate Adaption: Wireless channels are time-varying because
of the Multipath Effect and Doppler Effect of the movement.
Choosing the lowest rate of channel coding and modulation
guarantees the transmission stability but reduces efficiency.
With Adaptive Modulation Coding (AMC), transmitter ad-
justs coding rate and modulation scheme according to channel
conditions. However, in reality, this technology is not very
efficient because of two reasons: first, it is difficult to ensure
the accuracy of channel estimation since estimations base on
a finite number of reference signals. Second, there is staircase
effect in AMC. The number of encoding and modulation rate
is limited either it is a local area network (LAN) 802.11
or mobile communication network LTE, thus the throughput
is not smooth. Even if with accurate channel estimation,
AMC can only achieve discrete staircase-like rate adjustment
(throughput curve like a staircase). The discontinuity between
staircases lowers the efficiency of communication.
Hybrid Automatic Repeat Request (HARQ), first introduced
in[6], encodes the bitstream using a low-rate code and trans-
mits subpackets in turn using an automatic repeat request
(ARQ) algorithm, until the receiver decodes the bitstream
correctly.