Path Diversity over Packet Switched Networks: Performance Analysis and Rate Allocation - Computer Science > Networking and Internet ArchitectureReportar como inadecuado




Path Diversity over Packet Switched Networks: Performance Analysis and Rate Allocation - Computer Science > Networking and Internet Architecture - Descarga este documento en PDF. Documentación en PDF para descargar gratis. Disponible también para leer online.

Abstract: Path diversity works by setting up multiple parallel connections between theend points using the topological path redundancy of the network. In this paper,\textit{Forward Error Correction} FEC is applied across multiple independentpaths to enhance the end-to-end reliability. Network paths are modeled aserasure Gilbert-Elliot channels. It is known that over any erasure channel,\textit{Maximum Distance Separable} MDS codes achieve the minimum probabilityof irrecoverable loss among all block codes of the same size. Based on theadopted model for the error behavior, we prove that the probability ofirrecoverable loss for MDS codes decays exponentially for an asymptoticallylarge number of paths. Then, optimal rate allocation problem is solved for theasymptotic case where the number of paths is large. Moreover, it is shown thatin such asymptotically optimal rate allocation, each path is assigned apositive rate \textit{iff} its quality is above a certain threshold. Thequality of a path is defined as the percentage of the time it spends in the badstate. Finally, using dynamic programming, a heuristic suboptimal algorithmwith polynomial runtime is proposed for rate allocation over a finite number ofpaths. This algorithm converges to the asymptotically optimal rate allocationwhen the number of paths is large. The simulation results show that theproposed algorithm approximates the optimal rate allocation found byexhaustive search very closely for practical number of paths, and providessignificant performance improvement compared to the alternative schemes of rateallocation.



Autor: Shervan Fashandi, Shahab Oveis Gharan, Amir K. Khandani

Fuente: https://arxiv.org/







Documentos relacionados