Exploiting multiple description coding for intermediate recovery in wireless mesh networks
Journal of Network and Computer Applications
| Hi-index | 0.00 |
In this paper, we first propose a multiple descriptionmesh-based motion coding method, which generates twodescriptions for mesh-based motion by first subsamplingthe nodes of a right-angled triangular mesh and thendividing them into two groups. Motion vectors associatedwith the mesh nodes in each group form the mesh-basedmotion field for each description, and are transmittedover distinct network channels. With the nodes in eachgroup, two other regular triangular meshes besides theoriginal one can be constructed, and three differentprediction images can be reconstructed according todescriptions available. The proposed multiple descriptionmesh-based motion coding method is then combined withthe pairwise correlating transform proposed in [Multiple description image coding for noisy channels by pairing transform coefficients], and anew and complete multiple description video codingscheme is proposed. Further measures are taken toreduce the mismatch between the encoder and thedecoder that occurs when the decoder receives only oneof the two descriptions and has to use a differentreference frame for motion compensation other than theencoder does. Simulations were carried out to evaluatethe performance of the proposed scheme, and the resultsshow, compared to the MDTC video coding method in [Multiple Description Coding for Video using Motion Compensated Prediction],the proposed scheme has achieved lower redundancy ratedistortion; in the scenario of packet loss, the proposedscheme outperforms the MDTC video coding method.