| dc.description.abstract | Traditionally in the field of electronics, hardware is designed, developed, and improved on in various methods, whether it be increased storage capabilities or smaller models. Software applications lagged because of the hardware requirements to operate software, but increasingly, software tools are replacing technology that relied heavily on hardware components where applicable because of the abilities to both modify the technology easily and to consolidate tasks in an automated fashion. This research focuses on the networking space and aims to replace hardware architecture with software, as well as write algorithms to intelligently allocate incoming data. To orchestrate this architecture and the algorithms, the modern tools of software-defined networking and software-defined radios were combined. This created a network capable of transmitting packets over-the-air, with the network itself having separated the data plane and control plane in the software-defined networking standard. The control plane is written entirely in software, allowing modifications to be made across the whole system relatively simply. In this research, two software-defined radios were used to represent a base station and a field multi-sensor collector respectively. The field sensor transmits real sensor data from a web database that represents readings of the resistance of a gas over time from fourteen sensors. The base station radio can only receive a single packet at a time from the secondary radio due to bandwidth constraints, and so, using a software-defined controller, the various scheduling policies are compared to develop the most efficient means of processing the individual data packets. The final algorithm started from basic round-robin before evolving into weighted round-robin, with measurable results in terms of root-mean-square error values for each sensor and one for the total transmission period. The weighted round-robin was upgraded a step further to have real-time weight updates at regular intervals based on the accuracy of prediction for the next value in the sequence, per sensor. The contrast between the three stages of development for the round-robin algorithms is plain to see, with steady improvement between basic round-robin and weighted round-robin, and drastic improvement between weighted round-robin and the smart algorithm. The results from the research project yield a final draft of communication between the software-defined radios that produced an effective and efficient manner of software-defined networking. | |
