Disjoint Multipath QoS Routing

Abstract

Disjoint Multipath QoS (quality of service) Routing algorithm is aimed at selecting multiple paths for a flow in a peer to peer network, which are mutually disjoint w.r.t. bottleneck links. The Widest Disjoint Paths (WDP) algorithm [3][4] assumes a static set of available paths for each source-destination pair a priori, from which disjoint paths w.r.t. bottleneck links are selected. The Shortest Widest Path (SWP) algorithm [9] dynamically determines one path for each source-destination pair, which is shortest among widest ones. The proposed Disjoint Multipath QoS Routing algorithm dynamically finds paths mutually disjoint w.r.t. bottleneck links, which are shortest among widest ones. Thus, the algorithm generalizes both WDP and SWP algorithms.In many important multimedia applications such as video over IP in telehealth application, the application-and network-layer must collaborate in order to provide some necessary QoS guarantee. The network-layer should provide certain level of QoS measures, based on which the application-layer is able to compensate for deficiency of achieved QoS level. Those QoS measures may include end-to-end bandwidth, delay and packet loss rate, etc. The Disjoint Multipath QoS Routing algorithm is designed to achieve certain QoS level the application-layer can easily work with. In this paper, we only deal with the network-layer part and will focus on a scenario of one source and one destination. Simulation results and performance analysis demonstrate that the algorithm converges, offers lower end-to-end packet loss rate and higher throughput in comparison with SWP and Dijkstra Shortest Path algorithm as network traffic grows.

M2FEC: An effective FEC based multi-path transmission scheme for interactive multimedia communication

“…Some approaches suggest increasing path diversity under the heuristic of topology, such as earliest divergence rule (EDR) [16], disjoint multi-path QoS routing (DMQR) [17] and widest disjoint paths (WDP) [18]. Their basic idea is to establish overlay path which takes a different underlying physical routine from default IP-layer path between the sender and receiver.…”

“…(3) By certain multi-path overlay routing algorithm such as [17,28], sender and receiver have already select several overlay paths with good performance between them, and this paper only investigates the coding and transmission strategies between the communication pairs. (4) The loss and delay varying behaviors of the two selected paths are approximate uncorrelated with each other.…”

Maximizing Bandwidth Using Disjoint Paths

“…The problem to generate (s,t) paths in CN with bandwidth QoS has been addressed in [16][17][18][19]. Reference [16] attempts to find one path that satisfies both the bandwidth and delay constraints.…”

“…However, the solutions in [16][17][18] cannot be used to generate a DP with maximum bandwidth. Reference [19] proposed a routing algorithm that select paths to maximize bandwidth, minimize delay and packet loss rate. Since [19] considers all the three QoS, the paths selected may not have the maximum bandwidth.…”

“…Maximizing the bandwidth in a CN is of practical importance [5,19] for bandwidth-intensive applications like video conferencing, video-on-demand, file downloads from the Internet and file transfer using FTP. Since the bandwidth of an (s,t) path is limited by the minimum bandwidth of the edge in the path, the most viable solution to increase the total (s,t) bandwidth in the network is to use multiple paths.…”