Virtual network traffic shaping (VNTS) technique to ensure fairness in virtual WiMax networks

Ronak Daya
2009
This thesis presents the results of an experimental study on wireless network virtualization. The design and evaluation of virtualization methods for a WiMAX base station, a wireless technology of growing importance for emerging "4G" wide-area data services is performed using an experimental WiMAX base station which has recently been deployed at WINLAB, Rutgers University as part of the national GENI experimental network for future Internet research. The goal of this project was to benchmark
more » ... was to benchmark the performance of this WiMAX base station with and without virtualization over a range of realistic scenarios, and to evaluate methods for improving fairness and isolation between virtual networks in the system under study.A set of mobile client experiments (without virtualization) show that the base station is capable of operating at peak rates of around 16 Mbps indoors, and that the achievable bit-rate at a mobile client varies considerably as a function of received signal strength (RSSI) at the mobile client. The next step was to perform virtualization of this device, using Kernel Virtual Machine (KVM); with multiple slices having the ability to share the same base station. Experimental results are performed for multiple virtual networks operating in realistic scenarios with varying received signal strength at mobile devices. The results show that virtual networks can have significant coupling between their throughput performance, and that unfairness is further amplified by autorate and scheduling algorithms in the WiMAX base station, which have the effect of moving resources to mobile clients with low signal-to-noise ratio.As a solution to the above mentioned problem, a Virtual Network Traffic Shaping (VNTS) technique is proposed and evaluated. An algorithm is developed using feedback from the base station's control interface to assess the current channel utilization for each slice. A control parameter in the form of downlink data rate is defined and this value is regulated dynamically for each slice, using an imple [...]
doi:10.7282/t3j67h3b fatcat:savb3ofajne4no52dz37bsyhpm