Publication

Abstract : In this paper, a new prediction based dynamic scheduling mechanism is proposed to ensure quality of service especially in internet of things (IoT) by dynamically allocating bandwidth. The scheduling scheme is analyzed for two different applications: (1) to accommodate various streaming data of IoT applications or for medical applications. Traffic is segregated based on different traffic characteristics such as bit rate, packet loss rate and tolerable delay into different classes and they are prioritized based on these characteristics; and (2) to consider continuous long time data from all lightweight devices like sensors. The algorithm allocates bandwidth such that its utilization can be optimized for different applications. The scheduler decision is divided into two steps: (1) calculating average increase in queue length of high priority and medium priority services at current time slot from the previous time slot and, (2) adding these values to previous weight values for corresponding services and multiplying with a coefficient. The performance is carried out under different system loading and scenarios. The results illustrate the improvement of the advised scheduler with respect to end to end delay distributions, packet drop and end to end jitter values. © 2018 Springer Science+Business Media, LLC, part of Springer Nature