Simulation methodology and performance analysis of network coding based transport protocol in wireless big data networks

Abstract

The Multi-Path, Multi-Hop (MPMH) communications have been extensively used in wireless network. It is especially suitable to big data transmissions due to its high throughput. To provide congestion and end-to-end reliability control, two types of transport layer protocols have been proposed in the literature: the TCP-based protocols and the rateless coding based protocols. However, the former is too conservative to explore the capacity of the MPMH networks, and the latter is too aggressive in filling up the communication capacity and performs poorly when dealing with congestions. To overcome their drawbacks, a novel network coding scheme, namely, Adjustable Batching Coding (ABC), was proposed by us, which uses redundancy coding to overcome random loss and uses retransmissions and window size shrink to relieve congestion. The stratified congestion control strategy makes the ABC scheme especially suitable for big data transmissions. However, there is no simulation platform built so far that can accurately test the performance of the network coding based transport protocols. We have built a modular, easy-to-customize simulation system based on an event-based programming method, which can simulate the ABC-based MPMH transport layer behaviors. Using the proposed simulator, the optimal parameters of the protocol can be fine-tuned, and the performance is superior to other transport layer protocols under the same settings. Furthermore, the proposed simulation methodology can be easily extended to other variants of MPMH communication systems by adjusting the ABC parameters.