PTP: Path-specified Transport Protocol for Concurrent Multipath Transmission in Named Data Networks

Autor: Ye, Yuhang, Lee, Brian, Flynn, Ronan, Murray, Niall, Fang, Guiming, Cao, Jianwen, Qiao, Yuansong
Rok vydání: 2018
Předmět:
Druh dokumentu: Working Paper
Popis: Named Data Networking (NDN) is a promising Future Internet architecture to support content distribution. Its inherent addressless routing paradigm brings valuable characteristics to improve the transmission robustness and efficiency, e.g. users are enabled to download content from multiple providers concurrently. However, multipath transmission NDN is different from that in Multipath TCP, i.e. the "paths" in NDN are transparent to and uncontrollable by users. To this end, the user controls the traffic on all transmission paths as an entirety, which leads to a noticeable problem of low bandwidth utilization. In particular, the congestion of a certain path will trigger the traffic reduction on the other transmission paths that are underutilized. Some solutions have been proposed by letting routers balance the loads of different paths to avoid congesting a certain path prematurely. However, the complexity of obtaining an optimal load balancing solution (of solving a Multi-Commodity Flow problem) becomes higher with the increasing network size, which limits the universal NDN deployments. This paper introduces a compromising solution - Path-specified Transport Protocol (PTP). PTP supports both the label switching and the addressless routing schemes. Specifically, the label switching scheme facilitates users to precisely control the traffic on each transmission path, and the addressless routing scheme maintains the valuable feature of retrieving content from any provider to guarantee robustness. As the traffic on a transmission path can be explicitly controlled by consumers, load balancing is no longer needed in routers, which reduce the computational burden of routers and consequently increase the system scalability. The experimental results show that PTP significantly increases the users' downloading rates and improved the network throughput.
Comment: journal, 21 pages
Databáze: arXiv