Transcript 下載/瀏覽
Joint Multiple Description Coding and Network Coding for Wireless Image Multicast 出處: Image and Graphics, 2009. ICIG '09. Fifth International Conference on 報告者:黃群凱 2016/7/13 1 Introduction Multiple description coding (MDC) [5] generates different encoded versions for the same source. There are several ways to generate multiple descriptions. In this paper, a three-description lattice vector quantization (3D-LVQ) based image coding scheme is considered. 2016/7/13 2 Introduction Allows mixing of data at the intermediate nodes. Deducing data at the receiver end, whereas the traditional store-and-forward method at intermediate nodes keeps different packets separate. 2016/7/13 3 Introduction Try to avoid collisions of these packets. With network coding, an intermediate node performs network coding on the packets it has received and multicast a mixed one to several neighbors with different needs within a round time instead of sending packet for each neighbors individually, thus saving time and energy 2016/7/13 4 Scenario description The wireless ad hoc network can be modeled as a stochastic directed hypergraphH = (X,E) Where X is the set of nodes in the network and E is the set of hyperedges. Nodes can adjust their transmission power knowing the location of their neighbors, therefore every wireless directed hyperedge e(i, J) A cost function fiJ reflecting energy consumption and a loss rate liJ representing loss probability. 2016/7/13 5 Scenario description In this model, transmission radius of all the nodes equals R. Two nodes in the network are neighbors if the distance between them is no longer than R. We assume that several nodes (denote as destination nodes) in the networks are interested in image information that captured and stored in one single node (known as source node) and each node is aware of its neighbors’ storage. 2016/7/13 6 Joint MDC-NC 3.1 Multiple description coding 2016/7/13 7 Joint MDC-NC The index number N denotes the number of fine lattice points in a Voronoi cell of a sublattice point. 2016/7/13 8 Joint MDC-NC 3.2 Network coding combined with geographic routing • A greedy routing algorithm is proposed in [2] which forwards the packet to the node that is closest to its destination among the neighborhood in every hop. 2016/7/13 9 Joint MDC-NC After receiving a packet from its neighbor in the previous round, an intermediate node stores the packet or recover and then store if receiving a mixed one. It then performs network coding to generate a mixed packet based on the packets in the storage. By associating each packet Pi A coefficient gji 2016/7/13 10 Joint MDC-NC The intermediate node then sends the newly generated mixed packet to a neighbor according to the position-based greedy algorithm. 2016/7/13 11 Joint MDC-NC 3.3 Decoding procedure If 3 descriptions are recovered, the central decoder can produce the best reconstructed image quality. If two descriptions are recovered, twodescription-based side decoder yields reconstructed image based on the vector 2016/7/13 12 Joint MDC-NC • In the case of one description recovered, the one-description based side decoder produces a coarse version of image based on the vector. 2016/7/13 13 Experimental Results 2016/7/13 14 Conclusions The simulation results has clearly validated that the effectiveness and efficiency of the scheme in terms of better decoded image quality, lower failure rate and less energy consumption compared with MDC/SDC with routing schemes. 2016/7/13 15