Traffic Intensity Based Efficient Packet Schedualing

Abstract

Minimum end to end delay, packet delay variations or packet jitter, and packet drop for given throughput in networks plays well known important role not only in achieving better Quality of Service but also have evident effects on network utilization, efficiency and better quality of experience for services being carried over networks. Traffic intensities (TI) are generally used for dimensioning and capacity planning in telecom networks. Usually TI depends on point of presence(POP) and its characteristics like area served, customer density, usage pattern and grade of service that service provider intends to offer for given economic constraints. Most of time, for TI in busiest hour is used to determine required capacities of network resources. In this paper, we have extended original concepts of traffic intensity to period-to-period base rather than busiest hour and utilized period based traffic intensities to make packet scheduling decisions. This provides an opportunity to manage packet scheduling decisions based on dynamics of planned or actual or policy based traffic intensities for given POP and enables us to optimize scheduling decisions and hence end to end delay, jitter, drop behavior of network nodes and increases the interface utilization. This paper provides complete details of methodology and improvements in QoS and Network utilization. To evaluate effects of packet scheduling based on TI, using NS2 we have simulated a model network, and used Traffic Intensity Based Packet Scheduling (TIPS) algorithm to schedule packets and evaluated results of end delay, jitter, and throughput and drop behavior of various network nodes. We have found that TIPS offers optimized results for better QoS and network utilization as compared to existing packet scheduling approaches like Drop Tail, Round Robin, Deficit Round Robin, RED and SFQ. This research paper shall provide details of packet scheduling methodology, TIPS algorithm, simulation results and their comparison with some commonly used packet scheduling algorithms.