Improvement in Quality of Service in Iot Enabled Applications

Abstract

The scope of internet is expanding beyond computing and computer devices being connected. It is envisaged to provide advanced level of services to society, businesses, etc. Internet of Things connects everyday ‘things’ embedded with sensors and electronic software to the internet enabling them to collect and exchange data. IOT will lead to unification of technologies viz. cloud computing, low power embedded systems, machine learning, big data and networking; and an era of ubiquity which means connectivity at anyplace and anytime. An enormous amount of data would be generated by billions of users which has to be handled and processed which requires a much wider and more complex network than the present day internet. The heterogeneous networks, gigantic number of links and information exchange within edge nodes in IOT makes the system very complex and hence creates hurdles for satisfaction of the dynamic QoS requirements. In such enormous smart devices connectivity it is of utter important to maintain better quality of service (QoS). This thesis proposes weighted fair queueing based packet scheduling along with dynamic bandwidth allocation based on average queue length to improve the quality of service in IoT enabled applications. This predictive model employs AEWMA (Adaptive Exponentially Moving Average) method for the calculation of average queue length which follows a nonlinear method to find the average queue length so that accidental network bursts can also be considered. The scheduler composes of two different steps: firstly, the weighted fair queueing policy will take the previous slot average queue length into consideration for the calculation of current slot average queue length. Then, the bandwidth will be allocated by multiplying the average queue length (which is calculated based on the current queue size and the previous average queue size) with a coefficient. The proposed approach is tested on network model where we have data ranging from few kilobits to megabits based on which the packet loss, delay and jitter performance of the proposed scheme is measured.