Design and implementation of medium-range outdoor wireless mesh network with openflow in raspberry Pie

Abstract

This thesis has designed and implemented the prototype of software-defined wireless mesh network (SDWMN) testbed with in-band control approach for road traffic monitoring system on Phaya Thai road between Rama 1 and Rama 4 roads. Wireless mesh nodes for this outdoor SDWMN testbed are composed of 6 waterproof boxes, 6 Raspberry Pi’s, 6 cameras, and 6 power banks and 2 Intel NUC computers. Ad-hoc based IEEE 802.11 WiFi standard is used to send the captured image from Raspberry. Two gateways are installed at the traffic police boxes and two wireless mesh nodes are installed at each crossover bridge on Phaya Thai road. The total distance between two gateways is 1100 meters. On Phaya Thai road, the average distance between adjacent crossover bridges is 250-350 meters. In summary, the main contributions of this thesis are as follows. Firstly, we have designed and developed all components in preparation for the actual installation SDWMN testbed. The software parts include the installation of OpenVswitch, RYU, driver for external WiFi adapter in all wireless nodes and routing for outdoor SDWMN. Linux kernel version 4.4 has been used with the driver for applied antenna in this thesis. A waterproof box is designed for installation on the crossover bridges on Phaya Thai road. The primary route and the alternative route are built by predefined forwarding rules based on minimum hop path. The primary route is installed by predefined forwarding rules at bootstrapping stage in all wireless nodes and the alternative route is established by predefined backup forwarding rules from RYU controller when one of the wireless mesh nodes is failed with the usage of standard OpenFlow configuration request message. Based on our measurement of network performance, OpenFlow control traffic requires around 12 kbit/sec when all wireless mesh node are connected to RYU controller and requires at least 20 kbit/sec when one of the wireless mesh nodes is disconnected from RYU controller. The failure of wireless mesh node is investigated by manually rebooting the wireless mesh node. From our results, the required largest time to reroute for is 46 seconds and for data plane is 30 seconds. The actual restoration time for individual failure cases depends on the actual physical location where outdoor SDWMN is installed and the nodes that fail Finally, we have integrated the intended traffic monitoring application and SDWMN network. We have provided to traffic police for usages of traffic monitoring system for 16 hours on 26th November 2018 and the status of control plane during this practical operation of a traffic monitoring system is investigated. During network operation with traffic monitoring application in winter season of Thailand, the temperature of wireless mesh node is lower than the maximum operable temperature which is 85-degree Celsius. Testing in the warmer seasons is left as a future work together with the testing of resultant temperaturedependent node inoperability and SDWMN reliability. The current network testbed will be a baseline for those future implementation verification of large-scale SDWMN of road traffic monitoring network.