Project topics and materials
Get Complete Project Material File(s) Now! »
Network Coding for Multimedia Applications
In the previous section, the subject of network coding has been approached from a graph and information theory perspective. From the networking perspective, the OSI or the TCP/IP stack protocols offer several options regarding the level where NC should be implemented. In content distribution networks NC can be implemented at the application layer with little effort. In self-organizing networks it can be used for the wide dissemination of important data, as a substitute for routing at the network layer. In general packet networks there is a high interest in integrating NC with the current TCP/IP protocol, while investigating the coding opportunities at the physical layer in wireless networks can be an interesting topic of research. For multimedia applications though, the recent trend is to build protocols that are implemented as a cross-layer between the application and network layer, so that they can be tailored to the specific media content to be delivered.
The NC approaches can also be divided in two broad categories. If coding is allowed only among packets belonging to the same multicast session, those problems are also known as intra-session network coding and are based on PNC. The other category, which includes the butterfly network, allows for the combination of packets from different multicast or unicast sessions, thus it is referred to as inter-session network coding. In this section we discuss some of the existing NC approaches for multimedia applications.
NC-based Distributed Storage Systems
Distributed storage systems are a solution for the next generation multimedia applications that require increased storage space, ease of access and reliable recovery. As an example, more and more companies offer cloud services for distributed computing and storage. In order to allow information to be spread over multiple, unreliable nodes situated at different locations, such systems have to ensure the redundancy needed for reliable recovery in case of node failures. Hence, the bandwidth a replacing node requires in order to obtain the information needed for the reconstruction of lost data, also known as the repair bandwidth, is an important parameter of the system. Other parameters that influence the system’s capability to recover, such as the delay in access and transmission, can be reduced by taking into account the geographical distribution of resources. In a distributed storage system the storage elements are placed into nodes which function independently of each other and thus exhibit independent failure patterns. These nodes are often connected through a network with arbitrary topology, and the information objects are stored in specific nodes according to a mapping function. The use of NC for distributed storage systems has been studied by Dimakis, Wu et al. in several papers [DGW+10, Wu10] and has been compared to the results obtained by using traditional erasure codes. Since in storage systems that are distributed over networks the nodes may fail or leave the system quite often and the codes have to be maintained over time, the problem rests in finding a trade-off between the storage capability and repair bandwidth.
NC for Multimedia Applications in Wireless Networks
Wireless technologies have been adopted quickly thanks to their advantages in terms of mobility, low cost equipment and ease of deployment compared to wired networks. In addition to maximizing throughput, network coding can offer other advantages in wireless networks, as it can be used to reduce the energy consumption by reducing the number of transmissions. The advantages of NC are more evident in wireless ad-hoc networks, as in the simple scenario of two nodes wanting to exchange messages through one relay node presented in Figure 1.3. Each of the nodes A and B safely transmits its message to the relay R which will afterwards broadcast the XOR of the two messages. This scheme allows each node to decode the desired message after three time slots as compared to the traditional four time slots. This is therefore an important result, as the energy efficiency –the amount of battery energy consumed to transmit bits across a wireless link– is a critical design parameter for wireless ad-hoc networks.
Table of contents :
R´esum´e
Introduction
1 Network Coding
1.1 Network Coding Fundamentals
1.1.1 The Butterfly Network
1.1.2 Max-Flow Min-Cut Theorem for Network Coding
1.1.3 Linear Network Coding
1.1.4 Practical Network Coding
1.2 Network Coding for Multimedia Applications
1.2.1 NC-based Distributed Storage Systems
1.2.2 P2P-based Content Distribution
1.3 NC for Multimedia Applications in Wireless Networks
1.4 Cross-Layer Optimization
1.5 Error Resilient Network Coding
1.5.1 Fundamentals of Rank Codes
1.6 Conclusion
2 Network Coding for Multiple Description Video
2.1 Multiple description video coding
2.2 A framework for joint Multiple Description Coding and Network Coding
2.2.1 Experimental results
2.3 Scheduling for streaming of Multiple Description Video over wireless networks
2.3.1 Experimental results
2.4 Conclusions
3 Network Coding for Multi-view Video Streaming
3.1 Multi-view video
3.1.1 Multi-view video representation
3.1.2 Multi-view video compression
3.2 Proposed contribution
3.3 Experimental results
3.4 Conclusions
4 Distributed Social Caching using Network Coding
4.1 Introduction
4.2 Related work
4.3 Proposed contribution
4.3.1 System model
4.3.2 Proposed method
4.4 Experimental results
4.5 Conclusions
5 Towards Low-Overhead Network Coding with Blind Source Separation
5.1 Blind source separation over finite fields
5.2 Error-detecting code based separation algorithm
5.2.1 Analysis of the discriminating power of odd-parity bit-codes
5.2.2 Experimental results
5.3 Hashing Based Separation Algorithm
5.3.1 Experimental results
5.4 Conclusions
Conclusion and Perspectives
Publications
Bibliography
