Design and Implementation of Load-balancing Algorithm for Software-Defined Multi-channel Wireless Mesh Network

碩士 === 國立臺北科技大學 === 電子工程系 === 106 === Wireless mesh networks (WMNs) are suitable to serve as a wireless backbone network for the Internet of Things, as a large number of devices require communication service. The major routing protocol is the proactive optimized link state routing (OLSR). In recent...

Full description

Bibliographic Details
Main Authors: You-Ming Yeh, 葉又銘
Other Authors: Po-Hsuan Tseng
Format: Others
Language:zh-TW
Published: 2018
Online Access:http://ndltd.ncl.edu.tw/handle/87bf2r
Description
Summary:碩士 === 國立臺北科技大學 === 電子工程系 === 106 === Wireless mesh networks (WMNs) are suitable to serve as a wireless backbone network for the Internet of Things, as a large number of devices require communication service. The major routing protocol is the proactive optimized link state routing (OLSR). In recent years, access points supporting WMNs have adopted multiple network interface cards to utilize multi-channels at the same time. However, the problems of the OLSR protocol in the multi-channel include: (1) mesh nodes (MNs) only choose the best channel instead of distributing the load to multi-channels; (2) the distributed nature may result in heavy-load paths when multiple MNs are passing packets. This thesis combines WMN and software-defined network (SDN). We implement the MN and Open vSwitch (OVS) using Raspberry Pi 3, and collects user/link information in WMN through open network operating system (ONOS) controller. To support the transmission of multi-users in wireless mesh access point (WMAP), a centralized load balancing algorithm, which considers the priority based on the number of users with WMAPs, is proposed. By monitoring the link state and the number of loads of each node, the non-overlapping paths are discovered globally. The planned routing paths are passed to the OVSs via the OpenFlow protocol. Besides, a distributed load balancing algorithm is proposed, so that when the MNs is not connected to the controller, multiple channels can be utilized. The experimental results show that the centralized load balancing algorithm can utilize the advantage of multiple paths in WMN to have a higher transmission rate while reducing the delay compared to the OLSR.