Vehicular ad hoc networks, a type of Mobile Ad Hoc Networks (MANET), build wireless networks between vehicles and road side units to potentially provide safer driving experiences and many useful non-safety applications. To deploy this emerging technology in real life, governments, automobile industries, and academic research community have paid considerable attention over recent years. Moreover, the 75 MHz of spectrum in the 5.9 GHz band has been delegated as Dedicated Short Range Communication (DSRC) by the U.S. Federal Communications Commission (FCC) in 1999 [1]. On the other hand, for the process of protocol standardization, the IEEE 802.11p and IEEE 1609 series of standards are also proposed by IEEE to address this requirement [2][3].
Also, many existing VANET routing research results [4][5] have demonstrated that the unique characteristics of road traffic environment make the conventional MANET routing protocols, such as AODV, DSR, and GPSR, inefficient and unproductive.
Several researchers have also presented diverse routing paradigms to improve the performance of information distribution by taking the characteristics of the traffic environment and road structure into account. Among the prior research efforts, VADD [6] looked at both the traffic statistics information and the topology of the local road map to construct a packet delivery path with minimum possible delivery delay.
Unlike normal data transmission, video streaming could let drivers include more suitable applications for human cognitions: (1) providing driver assistance and navigation by collecting and displaying the surrounding view to help drivers make better driving decisions (2) enabling video conferencing/conversation between passengers of different vehicles (3) video surveillance [7] helps a country’s
transportation department to monitor the road traffic regulations (4) video advertising can be integrated with location-based services by local stores to offer advertisements to nearby vehicles (5) entertainment applications such as games [8][9], movies, and shows may also serve as feasible means for relaxing passengers during long distance travel. However, streaming video over VANET presents many challenges about the problems caused due to characteristics of wireless vehicular networks and video streaming. First of all, the rapidly-changing network topology and relatively higher vehicular mobility not only frequently breaks the network connections, but also makes a huge negative impact on network maintenance. Second, the strict video decoding deadline constraint is not easy to satisfy if the quality of the network is unstable. In addition, the bursty traffic, larger packet size, and variable bit rate (VBR) transmission nature of video streaming are also difficulties which make the challenges of video delivery even hard to overcome and solve. In spite of the aforementioned difficulties, [10] and [11] indicate the advancement of wireless networks and video compression technologies, thus making the non-trivial idea of video streaming over VANET a reality. The emerging IEEE 802.11p can support data transfer rates up to 54 Mbps between vehicles and road side wireless infrastructures. Next, the multi-channel communication ability of IEEE 802.11p greatly improves the attainable throughput by adding the frequency diversity. Lastly, the H.264/SVC introduces time, space, and quality scalability, and significantly increases the video coding efficiency [12], and therefore this technology is a good choice to alleviate the effects of the error-prone channels [11].
In MANET routing, research efforts about video streaming have already gathered some momentum. We noted that more and more researchers related to this topic have been changing the considered metrics from network-centric factors to multimedia-centric factors [12]. The former comprises of hop count, end-to-end delay,
jitter and network bandwidth, the metrics concerned with classical MANET routing protocols; and the latter is about the user received video quality, which is more suitable for the streaming problem. According to [12], the multimedia-centric routing protocols typically utilize a cross-layer design approach while the network layer protocols make use of the application layer metrics.
In the works of [13][14][15], the authors proposed routing algorithms for wireless ad hoc networks, by considering the possible video distortion to find the optimal routing path to deliver video streaming. According to their works, we know the distortion may be caused by packet transmission error and video deadline expiration.
The probabilities for these events are calculated by simplified MAC retransmission assumptions and queuing theory. Their works inspired us to apply the concepts into our problem.
Based on the mentioned research, we come up with a solution to deliver video streaming over VANET no matter for urban or highway scenarios. We divide our algorithm into two stages: the first stage is to estimate the video distortion values for vicinity road segments based on some traffic statistics information (in other words, we guess the possible video distortion if the video packets pass through the specific road segment), and we can find the routing path with minimum distortion by Dijkstra’s shortest path algorithm; after getting the path plan, the second stage uses a Markov Decision Process (MDP) based forwarding scheme to find suitable neighbor as the next hop.
The remaining sections are organized as follows. Chapter 2 reviews related works.
Chapter 3 demonstrates the proposed distortion estimation model and the MDP-based packet forwarding scheme. Finally, chapter 4 shows the simulation results and chapter 5 gives the conclusion of our work.